Réponses des poissons et des macroinvertébrés aux

variations rapides des conditions hydrauliques à l’aval

des centrales hydroélectriques gérées par éclusées
Clarisse Judes

To cite this version:

Clarisse Judes. Réponses des poissons et des macroinvertébrés aux variations rapides des conditions
hydrauliques à l’aval des centrales hydroélectriques gérées par éclusées. Ecologie, Environnement.
Université de Lyon, 2021. Français. �NNT : 2021LYSE1201�. �tel-03638828�

HAL Id: tel-03638828

https://theses.hal.science/tel-03638828v1
Submitted on 12 Apr 2022

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
N° d’ordre NNT : 2021LYSE1201

THESE de DOCTORAT DE L’UNIVERSITE DE LYON

opérée au sein de
l’Université Claude Bernard Lyon 1

Ecole Doctorale N° 341

Evolution, Ecosystème, Microbiologie, Modélisation (E2M2)

Spécialité de doctorat : Écologie

Discipline : Hydroécologie

Soutenue publiquement le 08/10/2021, par :

Clarisse Judes

Réponses des poissons et des

macroinvertébrés aux variations
rapides des conditions hydrauliques à
l’aval des centrales hydroélectriques
gérées par éclusées

Sous la direction de :
Gouraud Véronique, Ingénieur de Recherche, EDF R&D Encadrante
Capra Hervé, Directeur de Recherche, INRAE Co-directeur de thèse
Lamouroux Nicolas, Directeur de Recherche, INRAE Directeur de thèse

Devant le jury composé de :

Sylvain Doledec (Professeur des universités), CNRS, Lyon Président
Eva Enders (Chercheure), Fisheries and Oceans Canada, Winnipeg Rapporteure
Hélène Roux (Maître de conférences), IMFT, Toulouse Rapporteure
Franck Cattanéo (Professeur), HEPIA, Jussy (Suisse) Examinateur
Gouraud Véronique (Ingénieur de Recherche), EDF R&D Encadrante
Lamouroux Nicolas (Directeur de Recherche), INRAE Directeur de thèse
2
 Université Claude Bernard – LYON 1

Président de l’Université M. Frédéric FLEURY

Président du Conseil Académique M. Hamda BEN HADID
Vice-Président du Conseil d’Administration M. Didier REVEL
Vice-Président du Conseil des Etudes et de la Vie Mme Céline BROCHIER
Universitaire
Vice-Président de la Commission de Recherche M. Petru MIRONESCU
Directeur Général des Services M. Pierre ROLLAND

COMPOSANTES SANTE

Département de Formation et Centre de Recherche Directrice : Mme Anne-Marie

en Biologie Humaine SCHOTT

Faculté d’Odontologie Doyenne : Mme Dominique SEUX

Faculté de Médecine et Maïeutique Lyon Sud - Charles Doyenne : Mme Carole BURILLON
Mérieux
Faculté de Médecine Lyon-Est Doyen : M. Gilles RODE
Institut des Sciences et Techniques de la Réadaptation Directeur : M. Xavier PERROT
(ISTR)
Institut des Sciences Pharmaceutiques et Biologiques Directeur : M. Claude DUSSART
(ISBP)

COMPOSANTES & DEPARTEMENTS DES SCIENCES & TECHNOLOGIE

Département Génie Electrique et des Procédés (GEP) Directrice : Mme Rosaria FERRIGNO
Département Informatique Directeur : M. Behzad SHARIAT
Département Mécanique Directeur M. Marc BUFFAT
Ecole Supérieure de Chimie, Physique, Electronique (CPE Directeur : Gérard PIGNAULT
Lyon)
Institut de Science Financière et d’Assurances (ISFA) Directeur : M. Nicolas LEBOISNE
Institut National du Professorat et de l’Education Directeur : M. Pierre CHAREYRON
Institut Universitaire de Technologie de Lyon 1 Directeur : M. Christophe VITON
Observatoire de Lyon Directrice : Mme Isabelle DANIEL
Polytechnique Lyon Directeur : Emmanuel PERRIN
UFR Biosciences Administratrice provisoire : Mme Kathrin
GIESELER
UFR des Sciences et Techniques des Activités Directeur : M. Yannick VANPOULLE
Physiques et Sportives (STAPS)

UFR Faculté des Sciences Directeur : M. Bruno ANDRIOLETTI

3
4
 REMERCIEMENTS
Tout d’abord, je remercie mes trois directeur·ices de thèse Véronique, Nicolas et Hervé.
Merci pour vos conseils avisés et le temps que vous m’avez consacré. Vous n’avez pas toujours
eu les mêmes avis mais c’est ce qui a rendu ma thèse très riche. Nos discussions m’ont
énormément enrichi, et grâce à vous j’ai amélioré mes connaissances sur le fonctionnement des
écosystèmes de rivières. Merci aussi à tous les trois, de m’avoir motivée, de m’avoir convaincue
que je n’avais pas un mauvais jeu de données et que je pouvais en faire quelque chose de bien.
Véronique, je tiens à te remercier particulièrement pour tes encouragements et ton regard
bienveillant sur mon travail. Nico et Hervé, je tiens à vous remercier de m’avoir aidé à prendre
certaines décisions importantes pour le cours de ma thèse (et aussi pour ma vie personnelle),
j’ai beaucoup râlé mais aujourd’hui je vous en remercie grandement !

Je remercie également Franck Cattanéo et Sylvain Doledec d’avoir accepté d’être

membre de mon jury et en particulier Eva Enders et Hélène Roux d’avoir accepté d’être
rapporteures.

Je remercie tous les membres de mon comité de thèse : Christoph Hauer, Dominique
Courret, Loïc Teulier, et Sylvain Doledec. Vos conseils ont été très riches et pertinents, ils
m’ont permis d’avancer sur les bons rails.

Je remercie Nico Batz et Christine Weber d’avoir accepté de travailler avec moi. Merci
pour les discussions très riches que nous avons pu avoir, cela a été un réel plaisir de travailler
avec vous.

Un grand merci à Martin Shletterer et Christoph Hauer d’avoir accepté de collaborer

avec moi pour le volet Autriche de ma thèse. Ce projet n’a pas encore pu être valorisé pour le
moment mais j’espère qu’une publication verra bientôt le jour. Martin Shletterer un merci
spécial pour l’aide précieuse que tu m’as apportée sur le terrain sur la rivière d’Inn et cela
malgré les conditions climatiques difficiles.

Je remercie Agnès Barillier pour m’avoir donné des conseils et fourni des données. Je
remercie également toutes les personnes ayant contribué de près ou de loin à l’élaboration du

5
jeu de données de mon premier article ainsi toutes les personnes présentent sur le terrain pour
l’expérimentation in situ sur la rivière d’Ain.

Je remercie très chaleureusement mes deux équipes : EcoFlowS à INRAE et P76 à EDF.
J’ai eu énormément de chance d’être aussi bien entourée. J’ai été très bien accueillie que ce soit
à Lyon ou à Chatou. Ces deux équipes débordent de merveilleuses personnes qui m’ont apporté
un immense soutien à chaque étape de ma thèse. Coté EDF, je remercie en particulier Anthony
pour sa présence, son soutien et ses conseils lors de mes débuts en thèse, mon démarrage en
thèse n’aurait pas été le même sans toi. C’est toi qui m’a motivée à faire une thèse, je te dois
beaucoup alors merci infiniment pour tout. Merci aussi à Léo qui pareillement a été un immense
soutien lors de ma première année de thèse, de collègues nous sommes devenus amis et j’espère
pour longtemps. Un grand merci également à Maria pour son soutien tout au long de ma thèse,
Ecoflows a beaucoup de chance de t’avoir dans l’équipe, tu apportes énormément de
bienveillance et de gentillesse. Merci également de m’avoir donné l’opportunité de faire la
collaboration ave Nico et Christine. Merci à l’équipe technique de choc : Maxence, Bertrand et
Guillaume, vous êtes géniaux ! Vous m’avez apporté énormément de conseils et d’aide durant
la thèse. Cela a commencé par le terrain et ses débuts catastrophiques (c’est en partie grâce à
vous si tout s’est déroulé sans encombres) puis après ça a été toutes les petites questions et
dérangements à votre bureau. Merci aussi à vous trois pour votre bonne humeur, vos
encouragements et le soutien moral que vous m’avez apporté. Je remercie également Nuria,
Sophie, et les ex-stagiaires en particulier Alexia et Colin (merci aussi pour ton aide sur le terrain
Colin !) pour leurs soutiens, les sorties et les pauses cafés.

Je remercie l’ensemble des doctorants et post doctorants de INRAE. Je remercie en

particulier Noëlle pour tout le bonheur qu’elle m’a apporté, son soutien sans faille et sa
bienveillance. Tu as été mon rayon de soleil, ma confidente et ma conseillère scientifique. Merci
infiniment pour tout Noëlle ! Je remercie aussi Jake, ça a été un réel plaisir de te côtoyer au
quotidien. Merci aussi pour tes encouragements et d’avoir cru en moi, ton soutien a été précieux.
Merci aussi à Juliette, ça a été un réel plaisir de partagé ce bureau avec toi. J’en profite pour
m’excuser pour le désordre que j’ai mis parfois au bureau. Je remercie également Maxime, Julie
et Laura, les ancien·nes, vous avez été de vrais modèles et sources d’inspiration pour moi, j’ai
eu énormément de chance de vous avoir auprès de moi lorsque je suis arrivée à INRAE. Merci

6
pour toutes les discussions, les pauses cafés, votre soutien et vos conseils. Je remercie
également Maxime Gauthier pour toute la bonne humeur et les joies qu’il m’a apporté, ça a été
vraiment génial de partager un bout de chemin avec toi, tu as laissé un gros vide à INRAE. Je
remercie aussi Manu, la relève, un immense merci à toi pour ton soutien lors de l’écriture de
ma thèse tu as été d’une aide précieuse au moment où je n’avais vraiment pas la forme. Merci
d’avoir rayonné autant dans ma vie à ce moment-là et encore pour un moment je l’espère. Merci
également à Térésa, Romain, Hanieh et Mathis, votre bonne humeur et vos personnalités ont
enchanté mon quotidien au labo.

Un immense merci à Héloïse, Juliette, Maïlys et Charlotte. Je n’ai pas assez de mots
pour vous dire à quel point je suis reconnaissante pour tout ce que vous m’apporté. Vous avez
été un soutien énorme à chaque étape de ma thèse mais aussi dans ma vie personnelle. Vous
êtes des femmes extraordinaires et une réelle source d’inspiration pour moi, je vous aime très
fort. Maïlys, je te remercie en particulier d’avoir partagé ma vie pendant un peu plus de 2 ans.
Je n’aurais pas pu avoir meilleure colocataire que toi. Je te remercie de m’avoir pris sous ton
aile dès mon arrivée à Lyon, ma vie ici aurais été totalement différente (et certainement
beaucoup moins belle) sans toi. Un grand merci aussi à Fériel, Ade, Corentin, Thomas, Alice,
Anllaoui, Michelle et Lise pour toute la joie que vous m’avez apportée. Merci pour toutes les
soirées, les discussions téléphoniques, les sorties et les vacances. Vous m’avez permis de penser
à autre chose que ma thèse, vous m’avez encouragé et soutenu, j’ai de la chance d’avoir des
ami·es comme vous. Merci aussi à Ibrahim, tu as été un grand soutien pour moi au début de ma
thèse, c’est toi qui m’a poussé à poursuivre mon rêve de faire de la recherche quand je pensais
que ce n’était plus possible.

Je remercie très chaleureusement Alex d’avoir été là durant la dernière année de ma

thèse. Malgré mes sautes d’humeur durant la rédaction du manuscrit, tu as su m’apporter tout
le soutien et la motivation dont j’avais besoin. Merci pour tout.

Je remercie mon père, mon frère et ma grand-mère. Merci de m’avoir permis de suivre
mes rêves et de m’avoir encouragé dans tout ce que j’ai entrepris.

7
 RÉSUMÉ
Les éclusées sont des variations artificielles, fréquentes et rapides du débit permettant de
répondre à des pics infra-journaliers de la demande d'électricité. Les variations hydrauliques
qu'elles génèrent peuvent provoquer l'échouage-piégeage ou la dérive forcée des organismes ce
qui peut entrainer des changements dans la structure des communautés aquatiques. Jusqu’à
maintenant, que ce soit à l’échelle de tronçon (~ 100 m) ou du microhabitat (~ m2), peu d’études
ont mis en évidence des liens transférables entre les variables liées aux éclusées (p.ex. gradient
de hausse de vitesse) et les réponses biologiques (p.ex. densité des espèces, sélection de
l’habitat). Il est souvent difficile (1) de traduire les débits en conditions hydrauliques proximales
et (2) de tester l’influence relative des éclusées par rapport aux autres déterminants
environnementaux. Afin de combler ces lacunes de la littérature, dans notre premier article,
nous avons évalué les effets des variations hydrauliques provoquées par les éclusées sur les
populations de poissons en réalisant une combinaison originale d'analyses spatiales (sur 45
tronçons de rivières dont six groupes de tronçons géographiquement proches) et temporelles
(sur 3 à 17 ans). Nous avons montré que l'influence des éclusées était secondaire par rapport à
la structuration le long des gradients longitudinaux et aux influences négatives des crues sur les
densités annuelles. Cependant, les analyses spatiales et temporelles ont suggéré que les éclusées
défavorisent les espèces de poissons typiques des rivières de taille moyenne au profit des
espèces typiques des petites rivières (truite, vairon, chabot). Pour mieux comprendre les effets
des éclusées, dans notre second article, nous avons étudié la sélection de l’habitat des poissons
et des macroinvertébrés dans une rivière de taille moyenne (la rivière d’Ain) à l’échelle du
microhabitat. Pour ce faire, nous avons utilisé des observations des abondances de poissons
dans 1180 microhabitats (507 échantillonnés par pêche électrique, 673 par observations
subaquatiques) et des abondances de macroinvertébrés dans 36 microhabitats (hyporhéiques et
benthiques) ainsi qu’un modèle hydraulique 2D pour estimer l’hydraulique passée des
microhabitats (pendant les 15 jours avant l'échantillonnage). Globalement, l'hydraulique passée
a influencé la sélection des microhabitats, avec un effet plus fort sur les macroinvertébrés que
sur les poissons et une influence directe plus forte de l'assèchement (3-15 fois moins de
macroinvertébrés dans les microhabitats qui s’assèchent) que des fortes vitesses passées. Les
poissons suivent leurs conditions d’habitat favorables et évitent donc les fortes vitesses tandis
que les macroinvertébrés restent et se cachent. A partir des résultats de ces deux analyses, des
perspectives opérationnelles sont proposées, notamment pour intégrer les résultats dans les
modes de gestion.

8
 ABSTRACT
Hydropeaking is the frequent and rapid artificial variations in flow to meet sub-daily peaks in
electricity demand. The hydraulic variations it generates can cause the stranding or forced drift
of organisms, which can lead to changes in aquatic community structure. To date, at either the
reach (100 m) or microhabitat (~ m2) scale, few studies have demonstrated transferable links
between hydropeaking variables (e.g. upramping rate of current velocity) and biological
responses (e.g. species density, habitat selection). It is often difficult (1) to translate flow into
proximal hydraulic conditions and (2) to test the relative influence of hydropeaking versus other
environmental drivers. To address this knowledge gap, in our first paper, we evaluated the
effects of hydraulic variations from hydropeaking on fish communities. In particular, we
conducted an original combination of spatial (over 45 river reaches including six groups of
nearby reaches) and temporal (over 3 to 17 years) analyses. We found that the influence of
hydropeaking was secondary compared to well-known spatial variations in fish assemblage
structure along longitudinal gradients, and negative influences of floods on annual densities.
However, spatial and temporal analyses suggested that hydropeaking may disfavour fish species
from medium-sized rivers compared to species from small rivers (trout, minnow and sculpin).
To better understand the effects of hydropeaking, in our second paper, we investigated habitat
selection by fish and macroinvertebrates in a medium-sized river (Ain River) at the microhabitat
scale (~ m2). To do so, we used observations of fish abundances in 1180 microhabitats (507
sampled by electrofishing, 673 by snorkeling) and macroinvertebrates abundances in 36
microhabitats (hyporheic and benthic) as well as a 2D hydraulic model to estimate the past
hydraulics of microhabitats (for the 15 days before sampling). Overall, past hydraulics
influenced microhabitat selection, with a stronger effect on macroinvertebrates than fish and a
stronger direct influence of dewatering (3-15 times fewer macroinvertebrates in microhabitats
that were dewatered) than high velocities. Fish follow their suitable habitat conditions and thus
avoid high velocities while macroinvertebrates stay and hide. Based on these two analyses,
operational perspectives are proposed, in particular to integrate the results into management
models.

9
 ARTICLES

Judes et al. (A) :

Consistent but secondary influence of hydropeaking on stream fish assemblages in space and
time.
Judes, C, Gouraud, V, Capra, H, Maire, A, Barillier, A, Lamouroux, N.
Article accepté par Journal of Ecohydraulics

Judes et al. (B) :

Past hydraulics influence microhabitat selection by macroinvertebrates and fish in
hydropeaking rivers.
Judes, C, Capra, H, Gouraud, V, Pella, H, Lamouroux, N.
Article soumis à River Research and Applications

Batz et al. (Soumis) :

Habitat dynamics in hydropeaking rivers: the impact of flow frequency and duration on patch-
scale habitat dynamics.
Batz, N.*, Judes, C.*, Weber, C.
Article soumis à River Research and applications
*Les auteurs ont contribué à parts égales

10
 TABLE DES MATIÈRES
I. Synthèse
Introduction : les éclusées, une perturbation à plusieurs échelles spatiales, temporelles, et à
plusieurs niveaux biologiques 17

1. Influences des éclusées sur les densités de poissons (échelle des populations) 24
1.1 Etat de l’art : un manque d’analyse multisites décrivant des processus communs 24
1.2 Influence cohérente mais secondaire des éclusées sur les communautés de poissons des rivières dans
l'espace et le temps (Judes et al. A) 26
1.3 Perspectives scientifiques 29

2. Influences des éclusées sur la sélection du microhabitat (échelle de l’individu) 31

2.1 Etat de l’art : des connaissances parcellaires et contradictoires 31
2.2 Les variations hydrauliques influencent la sélection du microhabitat des macroinvertébrés et des
poissons dans les rivières à éclusées (Judes et al. B) 33
2.3 Perspectives scientifiques 40

3. Perspectives opérationnelles 47
3.1 Considérer les connaissances sur la sélection du microhabitat pour mieux quantifier l’influence des
éclusées à l’échelle du tronçon 47
3.2 Cohérence des résultats de Judes et al. (B) avec les mesures de gestion des éclusées prises sur la basse
rivière d’Ain 53
3.3. Résumé des perspectives opérationnelles de la thèse 54

Bibliographie 58

1. Judes et al. (A) : 71

1.1 Abstract 72
1.2. Introduction 73
1.3. Materials and methods 76
1.4. Results 83
1.5. Discussion 90
1.6. Conclusion 94
1.7. Acknowledgements 95
1.8. Bibliography 95
1.9. Appendix 99

2. Judes et al. (B) : 101

2.1. Abstract 102
2.2. Introduction 103
2.3. Materials and methods 106
2.4. Results 115
2.5. Discussion 124

11
 2.6. Acknowledgments 126
2.7. Bibliography 126
2.8. Appendix 130

3. Batz et al. (Soumis) : 140

3.1. Abstract 141
3.2. Introduction 142
3.3. Habitat shifts at the reach scale 146
3.4. Habitat shifts at the patch scale 149
3.5. Challenges for patch-scale metrics 160
3.6. Conclusions 161
3.7. Perspectives 163
3.8. Acknowledgments 166
3.9. Bibliography 167

12
 LISTE DES FIGURES
Figure I.1. Débits horaires enregistrés dans différents tronçons de rivière soumis à éclusées.
(A, B) la rivière Ume au niveau de la centrale hydroélectrique de Harrsele et la rivière
Vindel (à écoulement libre) au niveau de Granåker près de Vindeln dans le nord de la
Suède. ............................................................................................................................... 18
Figure I.2. Exemple de perturbations générées par les éclusées : échouage de poissons sur la
rivière d’Ain (Photo N. Lamouroux, 2018)...................................................................... 20
Figure I.3. Cascade d’influences des effets des éclusées de l’individu à la population. .......... 21
Figure I.4. Résumé et liens entre les articles de la thèse. ......................................................... 23
Figure I.5. Résumé de l’article de Judes et al. (A). .................................................................. 28
Figure I.6. Échantillonnage des poissons et des macroinvertébrés à plusieurs débits sur l’Ain
et carte des vitesses de courant obtenue à partir du modèle hydraulique 2D (Judes et al.
B).. .................................................................................................................................... 35
Figure I.7. Résumé des résultats de Judes et al. (B).. ............................................................... 39
Figure I.8. Résumé des principaux résultats de la thèse. .......................................................... 40
Figure I.9. Résultats d'un modèle 2D et classification en cinq types d'habitats simples (classes
de vitesse du courant) pour un tronçon de la rivière Aare près d'Innertkirchen en Suisse.
.......................................................................................................................................... 50
Figure I.10. Evolution des pertes relatives de surface mouillée par m3 s-1 sur l’ensemble du
linéaire étudié à Pont d’Ain sur l’Ain. Tiré de Malavoi (2019). ...................................... 54
Figure II.1.1. Schematics of the three approaches used to study the relative influence of
natural environmental drivers and hydropeaking events on fish assemblage structure.. ......... 75
Figure II.1.2. Reach locations in continental France and schematics of nearby reaches (from
six rivers) with different hydropeaking pressure. ............................................................. 77
Figure II.1.3. Results of the CoA analyses on the fish and environmental datasets. (A):
Between-reach spatial analysis (B): Within-reach temporal analysis on “trout reaches”.
(C) Within-reach temporal analysis on “cyprinid reaches”.............................................. 84
Figure II.1.4. Comparison of fish assemblages between nearby reaches.. ............................... 86
Figure II.1.A1. Example of a discharge time-series from the reach “Verdon_D1” and of
identified increasing (in red) and decreasing events (in blue). ................................................ 99
Figure II.1.A2. Interannual average of fish species density (number of individuals per 100 m2
sampled) in each reach.. ................................................................................................. 100
Figure II.2.1. Location of the Ain River study reach and maps of its flow velocity for the
maximum and the minimum flow discharge encountered over the study period. ......... 107
Figure II.2.2. Hourly discharge at Pont d’Ain (7.5 km upstream the study reach) during the 15
days preceding sampling and during sampling (2018). .................................................. 109
Figure II.2.3. Relationship (A) among present hydraulic variables, (B) among past hydraulic
variables and (C,D) between present and past hydraulic variables for benthic
macroinvertebrates ( ), fish sampled by electrofishing ( ), and fish sampled by
snorkeling ( ). .............................................................................................................. 119
Figure II.2.4. Fits of M2 models relating guild abundance to present hydraulics and including
an additive effect of dewatering (A) or past velocity (B)............................................... 122
Figure II.2.5. Comparison of the average preferred velocity (AVG_V, corresponding to model
M1) obtained in this study with data from Plichard et al. (2020) for fish and from
Forcellini et al. (in press) for macroinvertebrates. ......................................................... 123
Figure II.2.A1. Fits of M2 models relating guild abundance for hyporheic macroinvertebrates
to present hydraulics and including an additive effect of dewatering (A) or past velocity
(B) for the hyporheic macroinvertebrates data. ............................................................. 138

13
Figure II.3.1. Comparison of two neighboring mountain rivers in Switzerland, with the Aare
being affected by hydropeaking (station 2019; FOEN 2020), whereas hydrological
alterations for the Lütschine can be neglected (station 2109; FOEN 2020). ................. 145
Figure II.3.2. (A) shows the habitat dynamics at the patch scale considering a typical day or
week for a river section with a natural versus hydropeaking flow regime and (B) the
related frequency and relative duration of dewatering. .................................................. 153
Figure II.3.3. Results from a 2D model and classification in five simple habitat types (current
velocity classes) for a channelized, hydropeaking-affected reach with artificial groynes in
the Aare River close to Innertkirchen in Switzerland. ................................................... 156
Figure II.3.4. Comparisons of habitat shifts within patches between hydropeaking and natural
regime during a few exemplary days in autumn 2012 for patch 10 (A-B) and patch 20
(C-D).. ............................................................................................................................ 157
Figure II.3.5. Comparisons of spatial shifts of habitats between hydropeaking and natural
regime during a few exemplary days in autumn 2012. .................................................. 158
Figure II.3.6. Conceptual model summarizing the effect of altered flow regime components on
the habitats at reach and patch scale. .............................................................................. 163

14
 LISTE DES TABLEAUX

Tableau I.1. Résumé des perspectives opérationnelles qui découlent des résultats de la thèse.
.......................................................................................................................................... 56
Table II.1.1. Description of the environmental variables considered. ..................................... 81
Table II.1.2. List of the thirteen species studied. ..................................................................... 85
Table II.1.3. Reach characteristics (see Table II.1.1 for variable codes). Reaches considered
for the nearby reaches analysis are linked using brackets. Note that the two reaches
indicated by a * correspond to a unique reach but with different dates. .......................... 87
Table II.2.1. Fish guilds, species, taxa codes, with total abundance and occurrence in
microhabitats. ......................................................................................................................... 116
Table II.2.2. Benthic macroinvertebrate guilds, family, genus, taxa codes, with total
abundance and occurrence in microhabitats. .................................................................. 117
Several significant additive effects of past hydraulics were observed (indicated by asterisks in
Figure II.2.4.). ................................................................................................................ 120
Table II.2.3. Fitting statistics for models M0 (no microhabitat selection), M1 (effects of
present hydraulics) and M2 (additional effect of past hydraulics), including AIC and
Spearman Rho values.. ................................................................................................... 121
Table II.2.A1. Abundance per taxa in the benthic and hyporheic area. ................................. 134
Table II.2.A2. Benthic macroinvertebrate guilds, family, genus, taxa codes, with total
abundance and occurrence in microhabitats. .................................................................. 137
Table II.2.A3. Fitting statistics for models M0 (no microhabitat selection), M1 (effects of
present hydraulics) and M2 (additional effect of past hydraulics), including AIC and
Spearman Rho values.. ................................................................................................... 139
Table II.3.1. Summary statistics for the habitat shifts within patches for hydropeaking and
natural regime over the entire winter season (Dec.-Feb.) for six consecutive years (2012-
2017)....................................................................................................................................... 159
Table II.3.2. Summary statistics for the comparisons of spatial shifts of habitats between
hydropeaking and natural regime for the entire winter season (Dec.-Feb.) for six
consecutive years (2012-2017; see also Figure II.3.1 and Figure II.3.3). ...................... 159
Table II.3.3. Open questions for future research and practice to better account for the effects
of temporal flow regime components (i.e. frequency and relative duration) on ecological
processes on different spatial scales (i.e. reach to patch scale) in hydropeaking rivers. 165

15
Première partie
Synthèse

16
INTRODUCTION : LES ECLUSEES, UNE PERTURBATION A
PLUSIEURS ECHELLES SPATIALES, TEMPORELLES, ET A

PLUSIEURS NIVEAUX BIOLOGIQUES

Le mode de gestion des ouvrages hydroélectriques par éclusées consiste à lâcher les
eaux retenues dans un réservoir afin d’adapter la production d’énergie aux fluctuations de la
consommation d’électricité ou de compenser les variations de production des énergies
intermittentes. La production d'hydroélectricité est, ainsi, la seule énergie décarbonée
permettant de suivre la demande en électricité et de soutenir le réseau. En France, la production
par éclusées vient compléter l'électricité produite par les centrales thermiques et nucléaires qui
fournissent 80% de la production (RTE 2019), mais qui ne peuvent pas répondre rapidement
aux fluctuations de la demande. Les centrales d’éclusées sont au nombre de 140 pour 4 000
MW de puissance et 14 TWh de production annuelle moyenne (UFE 2017).

Le mode de gestion par éclusées génère des fluctuations de débit infra journalières
rapides et fréquentes (Bruder et al. 2016). Leur fréquence quotidienne est de deux ordres de
grandeur plus élevés que dans les systèmes naturels (Archer et al. 2002, Greimel et al. 2016) et
le débit de pointe peut-être jusqu’à dix fois plus élevé que le débit de base quotidien (Figure
I.1) et ainsi provoquer des variations de hauteur d’eau de plus de 3m (Moog 1993, McManamay
et al. 2016).

Chaque espèce est adaptée pour survivre et exploiter le régime hydrologique naturel (c.
-à-d. durée, fréquence, timing, intensité des variations de débit) d'où elle a évolué. Leurs cycles
de vie, leurs morphologies ou leurs comportements leur permettent de résister aux crues, aux
étiages ou aux variations saisonnières du débit lorsqu’elles sont dans les gammes de débit
naturel et qu’elles arrivent de manière prévisible à une période précise de l’année (Encadré 1).
Les éclusées étant plus rapides, plus fréquentes et plus intenses que les variations journalières
naturelles, elles peuvent agir comme une “perturbation” sur les écosystèmes de rivière, c’est-à-
dire qu’elles agissent comme "une force, un agent ou un processus physique, abiotique ou
biotique, provoquant une perturbation dans une composante ou un système écologique” (Rykiel

17
1985). Cette perturbation sur le système écologique peut être visible à différents niveaux
biologiques (ou résolutions) (Figure I.2). L’impact écologique de ce mode de gestion doit donc
être quantifié pour pouvoir mettre en place des mesures d’atténuation adaptées à son intensité.
Cela est d’autant plus urgent que la proportion de linéaires de rivières subissant des éclusées
pourrait augmenter à l’échelle mondiale pour pallier l'intermittence de production des énergies
vertes (Kougias et al. 2019).

Figure I.1. Débits horaires enregistrés dans différents tronçons de rivière soumis à
éclusées. (A, B) la rivière Ume au niveau de la centrale hydroélectrique de Harrsele et la
rivière Vindel (à écoulement libre) au niveau de Granåker près de Vindeln dans le nord de
la Suède. (C, D) la rivière Colorado à Lees Ferry avant et après la fermeture de la centrale
hydroélectrique gérée par éclusées de Glen Canyon dans l'Utah, USA. Les graphiques (A)
(B) sont sur une année tandis que (B) et (D) sont sur une semaine. Tiré de Bejarano et al.
(2018).

18
 Encadré 1. Sélection naturelle et régime hydrologique

Le régime hydrologique agit comme un filtrage environnemental, il sélectionne au fil des

générations des comportements et/ou une morphologie permettant aux organismes de
survivre et exploiter les variations de débit. Généralement, cela implique une
synchronisation des événements du cycle de vie, tels que la reproduction et la croissance
avec les variations temporelles du débit naturel. Par exemple, la perche dorée
(Plectroplites ambiguus) originaire des rivières arides d'Australie a besoin des crues pour
frayer, sinon, les œufs en développement se résorbent dans le tissu somatique (Mackay
1973).
Tous les régimes hydrologiques ne favorisent pas la sélection des traits permettant de
survivre aux variations de débit. La théorie des traits d'histoire de vie prédit que
l'amplitude, la fréquence et la prévisibilité des événements extrêmes, peuvent affecter la
façon dont les organismes évoluent ou non (Lytle & Poff, 2004). Lorsque les débits
extrêmes sont fréquents, importants et prévisibles, la sélection naturelle favorise les cycles
de vie qui sont synchronisés pour éviter ou exploiter les événements de débit extrême. En
revanche, si les débits extrêmes sont fréquents, de grande intensité mais imprévisibles, la
force de sélection est faible et cela même s’ils peuvent engendrer des fortes mortalités.
Dans ce cas, c'est plutôt une stratégie de minimisation des risques (ou “bet hedging” en
anglais) qui est adoptée, les parents produisent différents types de descendance qui
correspondent à différents futurs possibles (Cohen 1966, Lytle et al. 2004).
Les effets des éclusées sur une espèce donnée dépendent donc du régime hydrologique
historique auquel l’espèce a dû faire face et pour laquelle elle a été sélectionnée au cours
des générations (Biggs et al. 2005).

A une échelle plus locale, les variations de débit induisent des variations des conditions
hydrauliques locales sur quelques mètres carrés (ci-après, échelle du “microhabitat”). L’habitat
hydraulique des organismes aquatiques varie dans l’espace à chaque variation de débit. Batz et
al. (Soumis) ont montré sur la rivière Aare en Suisse que dans une parcelle de quelques mètres
carrés de rivière la vitesse peut changer de classe huit fois par jour (une classe tous les 0.25 m
s-1) dans le cas d’un régime soumis à des éclusées contre seulement 0.2 fois par jour si
l’hydrologie avait été naturelle. Dans cette même rivière, ils ont aussi montré qu’un habitat
hydraulique donné peut se déplacer de 15 mètres en moyenne par jour dans une rivière à
éclusées contre seulement 0.4 mètre par jour en moyenne en hydrologie naturelle (Batz et al.
soumis). Ainsi, il peut y avoir une sélection des espèces et des stades de vie qui sont capables
soit de bouger ou de se cacher rapidement lors des variations de débit soit de résister à des
conditions hydrauliques locales extrêmes (assèchement ou forte vitesse). Les organismes qui
n’en sont pas capables peuvent mourir par échouage-piégeage lorsque la lame d’eau se retire

19
trop rapidement (Figure I.2), ou être entraînés de force par dérive lors des augmentations de
vitesse. Dans la rivière de Alta (Ukraine), Jensen et al. (1992) ont trouvé environ 1,5 poissons
(toutes espèces confondues) échoués par m2 après une seule baisse des débits. Dans la rivière
Kootenai (USA), Perry & Perry (1986) ont montré que le nombre de macroinvertébrés en dérive
était huit fois plus élevé lors de l’augmentation des débits par éclusées (de 4,1 à 283 m3 s-1) que
lors du débit de base. La modification des conditions d’habitat local agit aussi sur le
comportement de frai et peut compromettre l’éclosion des œufs par décapage ou générer
l'assèchement des nids. Par ailleurs, les variations hydrauliques peuvent induire des variations
de température modifiant les cycles de vie des organismes, la date d’émergence et la dérive des
macroinvertébrés (Hogg et al. 1996, Céréghino et al. 1997).

Figure I.2. Exemple de perturbations générées par les éclusées : échouage de

poissons sur la rivière d’Ain (Photo N. Lamouroux, 2018).

Puisque ces processus peuvent entraîner la mort (dérive ou échouage-piégeage) ou

perturber la reproduction, les éclusées peuvent affecter la fitness (survie x reproduction) des
organismes ce qui peut induire des changements à l’échelle des populations (Figure I.3). Ainsi,
souvent dans les tronçons à éclusées, il y a une perte de diversité, d’abondance et de biomasse
(Bain et al. 1988, Bain 2007). Par exemple, Hayes et al. (2021) ont montré que la biomasse en
ombres (Thymallus thymallus) était environ huit fois plus élevée dans les tronçons sans éclusées
que dans les tronçons soumis aux éclusées. Cependant, les processus agissant à l’échelle de
l’individu ne se traduisent pas toujours par des effets sur la population. Par exemple, Hedger et
al. (2018) et Sauterleute et al. (2016) ont montré par modélisation que l’influence de

20
l’échouage-piégeage des juvéniles de saumons atlantiques pouvait être compensée
partiellement par l’effet de la densité dépendance (agissant comme un mécanisme de rétroaction
négative qui atténue les changements d'abondance à l’échelle de la population).

Figure I.3. Cascade d’influences des effets des éclusées de l’individu à la population.
La thèse traite uniquement ce qui est entouré en rouge. L'influence des températures
ou les interactions avec d’autres éléments du paysage tels que la présence d’abris, ou de
confluences n’ont pas pu être traitées. Repris de Barillier et al. (in press).

Des lacunes de la littérature empêchent de développer des mesures d’atténuation des

éclusées transférables d’une rivière à l’autre et adaptées aux communautés en place. D’un point
de vue méthodologique jusqu’à présent la plupart des études décrive les éclusées par des
descripteurs hydrologiques. Les descripteurs hydrologiques ne permettent pas de tenir compte
de la morphologie et empêche la transférabilité des liens entre éclusées et réponses biologiques.
D’un point de vue écologique, il manque des connaissances (1) sur la sensibilité des espèces ou
groupe d’espèce en fonction de leurs préférence d’habitat ou de leurs trait d’histoire de vie, (2)
sur l’influence des autres déterminants environnementaux sur le lien entre éclusées et les
réponses biologiques et (3) sur les mécanismes, c’est-à-dire sur la compréhension des
phénomènes à l’échelle du microhabitat pour expliquer et comprendre les résultats observés à
plus large échelle spatiale.

21
 L’objectif de ma thèse est de répondre à ces lacunes de la littérature pour mieux
comprendre l'influence des variations hydrauliques induites par les éclusées sur les poissons et
les macroinvertébrés. Pour cela, à l’échelle de la population (échelle du tronçon), les liens entre
les éclusées et les densités de poissons ont été étudiés par une analyse multisites (45 tronçons)
à l’échelle de la France sur plusieurs années (Judes et al. A). Ensuite, pour s’affranchir des
effets confondants multiples des analyses multi-sites et obtenir une compréhension plus
mécanique des effets des éclusées sur les populations, la sélection du microhabitat par les
organismes a été étudiée. Pour cela, 1180 microhabitats de poissons et 72 pour les
macroinvertébrés ont été échantillonnés dans un secteur de 6 km de la basse rivière d'Ain
(rivière de taille moyenne, affluent du Rhône) (Judes et al. B). Concrètement à ces deux échelles
spatiales, la variabilité des conditions hydrauliques vécue par les populations où les individus
avant l’échantillonnage (Judes et al. (A): 1 an avant, Judes et al. (B): 15 jours avant) a été liée
à l’abondance des taxa du tronçon ou du microhabitat (Figure I.4). Pour obtenir des relations
les plus transférables possibles, l’influence relative des variations hydrauliques induites par les
éclusées a été étudiée par rapport aux autres déterminants environnementaux principaux. Ces
autres déterminants environnementaux sont les conditions hydrauliques extrêmes et la
structuration longitudinale (de la source à l’embuche) des peuplements de poissons dans Judes
et al. (A) et les conditions hydrauliques présentes (ou actuelles) dans Judes et al. (B). Les
interactions avec des éléments du paysage comme la présence d’abris ou de confluence n’ont
pas pu être considérées.

Les deux premières parties de cette synthèse de thèse résument les apports des études
de Judes et al. (A) et Judes et al. (B) pour la littérature et abordent les perspectives scientifiques.
La troisième partie s’appuie en partie sur un troisième article dont je suis coauteur, Batz et al.
(Soumis), et aborde les perspectives opérationnelles à tirer des résultats majeurs de la thèse.
Cette partie aborde en particulier comment à partir des connaissances acquises à l’échelle du
microhabitat on peut estimer l’impact des éclusées sur les populations de poissons.

22
Figure I.4. Résumé et liens entre les articles de la thèse. La synthèse de thèse est organisée
de la manière suivante : Partie 1 : recherche l’influence des éclusées sur les communautés de
poissons (densité des espèces), Partie 2 : recherche des mécanismes par lesquels les éclusées
peuvent affecter les communautés de poissons en s’intéressant à l’influence des éclusées sur la
sélection du microhabitat, Partie 3 : résumé des perspectives opérationnelles et zoom sur les
façons dont on peut extrapoler les connaissances acquises à l’échelle du microhabitat pour
estimer les perturbations à l’échelle du tronçon (échelle utilisée pour la gestion).

23
1. INFLUENCES DES ECLUSEES SUR LES DENSITES DE

POISSONS (ECHELLE DES POPULATIONS)

1.1 Etat de l’art : un manque d’analyse multisites décrivant des

processus communs

Les effets mesurés à l'échelle de la population sont souvent jugés préoccupants pour la
gestion car ils se traduisent par une hausse de mortalité pour un ou plusieurs stades de
développement (recrutement, juvénile ou adulte) qui sont les paramètres démographiques
déterminant de la viabilité des populations. Ainsi, déterminer les causes de variation des
densités de poissons est un objectif essentiel pour le développement des connaissances aussi
bien que pour la gestion des ressources (p. ex. Lobón-Cerviá, 2009).

La grande majorité des études sur l’impact des éclusées sur les populations de poissons
sont faites autour d’un seul aménagement par une comparaison : amont/aval,
référence/influencé (Crips et al. 1983, Travnichek et al. 1994, Connor et al. 2004, Enders et al.
2017), ou avant/après changement du mode de gestion (protocole BACI Before-After-Control-
Impact, p. ex. Conquest, 2000). Le problème de ce genre d’études comparatives sur peu de
tronçons est que les résultats sont peu transférables et les conclusions sont parfois différentes.
La majorité des études montrent une diminution de l’abondance (Parasiewicz et al. 1998,
Macnaughton et al. 2015, Mihalicz et al. 2019) ou de la diversité (Schmutz et al. 2015) en
poissons et en macroinvertébrés dans les tronçons soumis aux éclusées. Cependant,
Smokorowski et al. (2011) ont trouvé que la composition, l'abondance des macroinvertébrés et
la biomasse des poissons étaient égales ou supérieures dans un tronçon de rivière soumis à
éclusées par rapport à un tronçon de rivière sans éclusées. De façon similaire, il y a aussi des
incertitudes concernant les espèces les plus sensibles. Par exemple, Garcia De Jalon et al. (1994)
ont constaté qu'après la construction d'une centrale hydroélectrique à éclusées sur la rivière Tera
(Espagne), les populations de presque toutes les espèces de cyprinidés ont diminué alors que la
population de truite a persisté. Cependant, sur le Rio Duraton (Espagne), Camargo & Garcia de
Jalon (1990) ont constaté l'effet inverse : les populations de cyprinidés ont persisté ou même
augmenté alors que la truite a disparu. Ces différences sont difficiles à expliquer et dépendent
probablement de la gestion des éclusées, de la période d’échantillonnage (p. ex. la saison, après

24
une crue), de leur emplacement le long du gradient longitudinal et/ou des caractéristiques
morphologiques des tronçons. Enfin, l’analyse d’une seule rivière ne permet pas d’isoler les
influences des différents paramètres de gestion des éclusées, tels que les gradients de
hausse/baisse de débit ou la valeur du débit de base.

Il est donc nécessaire de faire des études à large échelle, sur plusieurs rivières et suivies
pendant plusieurs années pour avoir des relations générales et transférables d’une rivière à une
autre. Cependant ces études sont rares puisque qu’il est difficile (1) de réunir des données
hydrologiques, biologiques et des traductions hydrauliques sur de nombreux tronçons suivis
pendant plusieurs années, et (2) de démêler l’influence des éclusées par rapport aux autres
déterminants environnementaux structurant les communautés de poissons. En effet, la
structuration des communautés de poissons longitudinale de la source à l’embouchure (Encadré
2) et les événements extrêmes tels que les crues et les étiages (Huet 1949, Verneaux 1973,
Nicola et al. 2009, Alonso et al. 2011, Bret et al. 2015) sont parfois difficiles à considérer
puisque cela nécessite de réaliser à la fois une étude spatiale et temporelle.

Il existe quand même quelques rares études multi-tronçons (Hayes et al. 2021 (Ntronçons
=74), Schmutz et al. 2015 (Ntronçons =69), en Autriche). Ils ont permis d’identifier des gradients
de baisse de hauteur d’eau à partir desquels les communautés piscicoles étaient fortement
dégradées (12-24 cm h-1 pour les ombres (Hayes et al. 2021) et 15 cm h-1 (Schmutz et al. 2015)).
Cependant, ces études ont leurs limites : (1) elles n'impliquent pas de variables hydrauliques
proximales, comme le gradient de diminution de la largeur mouillée (lié directement au risque
d’échouage) ou le gradient de hausse de la vitesse de courant (lié directement au risque de
dérive) (Moreira et al. 2019), (2) en utilisant une variable représentant la communauté
biologique dans son ensemble ou uniquement sur les densités d’ombre, ces études ne permettent
pas de savoir quelle espèce est la plus impactée, (3) elles n’ont pas étudié l’importance relative
des éclusées par rapport aux autres déterminants environnementaux.

Encadré 2. La structuration des communautés le long du gradient longitudinal

De nombreux concepts expliquent la distribution des espèces le long du réseau

hydrographique. Ces concepts se basent sur les changements physiques et énergétiques
(Huet 1949, Illies et al. 1963, Verneaux 1973, Vannote et al. 1980, Statzner et al. 1986).
De l’amont vers l’aval la pente diminue, la largeur de la rivière et la hauteur d’eau
augmentent.
Huet (1949) postule que les deux principaux déterminants de la distribution des poissons
dans les eaux vives sont la vitesse du courant et la température de l’eau (effet majoritaire),
tous deux dépendant de la pente. Ainsi Huet a fait correspondre quatre zones piscicoles à

25
 des gammes de pente. Les limites de pente d'une zone piscicole dépendent de la largeur
des rivières : plus la rivière est petite, plus la pente tolérée par une espèce de poisson
donnée est grande. La typologie de Verneaux (1973) augmente le nombre de zones afin
d’adapter la zonation aux rivières françaises.
Vannote (1980) introduit l’influence de l'énergie disponible pour expliquer les successions
biologiques le long du gradient longitudinal des milieux tempérés (le concept de la
continuité fluviale). La taille des particules organiques diminue vers l’aval, du fait que
chaque espèce est adaptée pour exploiter un certain type de matière organique, cela
conditionne leur répartition longitudinale.
D'autres concepts sont venus après compléter celui de la continuité fluviale. Parmi eux, on
peut citer le concept du régime de crue ou le concept de la discontinuité. Ce dernier intègre
l’effet des grands barrages et des réservoirs (Ward et al. 1983). En effet, de par les
modifications anthropiques les rivières ne sont pas des entités continues. Les
discontinuités modifient les composantes physico-chimiques et changent la structure des
communautés biologiques. Le concept du régime de crue (“flood pulse”) ajoute au concept
de continuité fluviale l’importance de considérer les crues. Ces dernières permettent un
apport en particules organiques et jouent donc un rôle important dans la structuration des
communautés biologiques (Junk et al. 1989).

1.2 Influence cohérente mais secondaire des éclusées sur les

communautés de poissons des rivières dans l'espace et le temps (Judes
et al. A)

Grâce à un large jeu de données piscicoles (45 tronçons suivis de 3-17 ans), Judes et al.
(A) ont testé l’influence relative des variations hydrauliques induites par les éclusées par rapport
à la structuration le long du gradient longitudinal des communautés et aux événements extrêmes
de forts et faibles débits (Figure I.5) sur les densités de plusieurs espèces de poissons communes
en France. Pour cela, à l’aide d’analyses multivariées (de type Co-inertie (Doledec et al. 1994,
Mérigoux et al. 1999)) ils ont combiné trois approches : (1) une analyse spatiale des effets des
éclusées sur les communautés de poissons entre tronçons (pour peser l’influence relative des
éclusées par rapport à la structuration amont-aval des peuplements), (2) une analyse temporelle
de ces effets au sein des tronçons (permet de s’affranchir de l’influence spatiale pour peser
l’influence relative des éclusées par rapport aux événements extrêmes), et (3) une comparaison
impliquant six groupes de tronçons voisins soumis à différentes intensités d’éclusées (permet
de s’affranchir à la fois des effets spatiaux et temporels). Chaque variation de débit a été traduite
en variation des conditions hydrauliques grâce à des modèles de géométrie hydraulique (Morel
et al. 2020). Ces modèles permettent de décrire les variations des conditions hydrauliques

26
(hauteur d’eau, largeur mouillée, vitesse du courant) en fonction du débit dans l’espace et le
temps. Ils fournissent les meilleures estimations de l'hydraulique des tronçons pour un débit
donné en l'absence de mesures hydrauliques détaillées sur le terrain (Morel et al. 2020).

L’analyse spatiale a montré que la structuration longitudinale des communautés de

poissons masque l’influence des éclusées. Sur un site donné, le suivi de la composition du
peuplement sur plusieurs années a montré que les événements extrêmes de fort débit sont plus
structurants que les variations hydrauliques liées aux éclusées. Ainsi, la prise en compte des
événements de fort débit est importante à intégrer dans les études d'impacts des éclusées afin
d’aboutir à des diagnostics fiables de l’état des peuplements. De façon secondaire à ces
événements extrêmes de fort débit, l’analyse temporelle de Judes et al. (A) a montré que les
éclusées pourraient favoriser les espèces typiques des petites rivières (truite fario, vairon,
chabot). Le caractère secondaire des effets des éclusées peut être lié aux gradients hydrauliques
subjournaliers modérés au sein des tronçons du jeu de données. Enfin, la comparaison des
tronçons voisins a montré que, pour les tronçons situés dans des rivières de taille moyenne, une
plus forte intensité d'éclusées était associée à des densités plus élevées d'espèces de poissons
typiques des petites rivières et à des densités plus faibles d'espèces typiques des rivières de taille
moyenne (gardon, barbeau, chevesne, hotu) (Figure I.5). Ainsi ces résultats sont cohérents avec
ceux de l’analyse temporelle et Judes et al. (A) qui concluent que de façon secondaire par
rapport à la structuration longitudinale des peuplements et aux événements extrêmes de fort
débit, les éclusées pourraient défavoriser les espèces de poissons typiques des rivières de taille
moyenne par rapport aux espèces typiques des petites rivières.

27
Figure I.5. Résumé de l’article de Judes et al. (A). La figure en bas à droite représente les
scores des tronçons sur le premier et le deuxième axe de l'analyse spatiale de coinertie entre
tronçons. Les flèches indiquent la position des paires de tronçons proches, les flèches vont du
tronçon le moins influencé par les éclusées au tronçon le plus influencé. Les flèches bleues
montrent les paires pour lesquels le tronçon le plus influencé présente des densités plus élevées
d’espèces typiques des petites rivières et des densités moins élevées des espèces typiques des
rivières de taille moyenne que le tronçon le moins influencé.

28
1.3 Perspectives scientifiques
A l’avenir, les liens entre la gestion des éclusées et la structure des peuplements piscicoles
pourraient être approfondis par les actions suivantes :
- Couvrir un gradient d’intensité de gestion des éclusées plus important (en termes
de fréquence d’éclusées et de gradient de hausse et de baisse de débit). Judes et
al. (A) ont considéré principalement des tronçons avec une intensité de gestion
des éclusées qui était relativement faible par rapport à ce que l’on peut trouver
ailleurs en Europe (Schmutz et al. 2015). Ainsi, par la suite il faudrait compléter
leurs analyses avec des tronçons issus des autres pays européens subissant des
intensités de gestion des éclusées plus importantes.
- Augmenter le nombre de tronçons proches mais subissant différentes intensités
de gestion des éclusées. Cela permettra de réaliser une étude quantitative mettant
en lien les différences d’intensité des gestions des éclusées avec les différences
de peuplement piscicole dans des contextes d’événements hydrologiques
extrêmes similaires.
- Développer des modèles d'atténuation des gradients de vitesse et de largeur
mouillée avec la distance d’évolution. Ces modèles devront considérer la
géomorphologie moyenne du linéaire comme la largeur mouillée, la hauteur
d’eau au module et la pente. Cela permettrait de considérer des tronçons de
pêche plus éloignés des stations de mesure de débit et donc de pouvoir
augmenter drastiquement la taille du jeu de données. Judes et al. (B) avait
imposé une distance <25 km et l'absence d'obstacles majeurs ou d'affluents entre
le jaugeage et la station de pêche poissons.
- Considérer les variations de température ou des éléments du paysage comme la
présence d’abris ou de confluence. Il y a généralement un compromis à faire
entre précision des données et quantité pour réaliser une analyse multisites.
Judes et al. (B) par exemple ont décidé de ne pas considérer les variations de
températures car elles étaient disponibles sur une proportion trop faible de
tronçon. A l’avenir, l’idéal serait d’avoir un large jeu de données avec des

29
descriptions précises de l'environnement, du paysage et des poissons (taille,
poids) sur le long terme.

30
2. INFLUENCES DES ECLUSEES SUR LA SELECTION DU

MICROHABITAT (ECHELLE DE L’INDIVIDU)

2.1 Etat de l’art : des connaissances parcellaires et contradictoires

Pour être en mesure de comprendre les mécanismes qui expliquent la structure des
communautés aquatiques (poissons, macroinvertébrés), il est nécessaire de descendre au niveau
des réponses individuelles. La sélection du microhabitat fait partie des réponses individuelles
qu’il est intéressant d’étudier pour comprendre comment les éclusées affectent le quotidien des
organismes et les risques associés. Ce genre d’étude peut permettre de répondre à des questions
telles que : Est-ce que les organismes se déplacent pour suivre leurs conditions d’habitat
favorables (induisant des pertes énergétiques) ? Est-ce qu’ils utilisent ou évitent des habitats à
risques (dérive et échouage) ? Est-ce qu’ils restent et subissent des conditions hydrauliques
défavorables (ex. diminution de l’efficacité de nourrissage) ?

A l’échelle du microhabitat, il a déjà été montré que les conditions hydrauliques (hauteur
d'eau, vitesse du courant) peuvent expliquer les effets des éclusées. Par exemple, Halleraker et
al. (2003) ont montré que l'échouage de truites brunes (Salmo trutta) juvéniles se produit
lorsque les gradients verticaux de hauteur d’eau sont supérieurs à 0,10 m h-1. En revanche, on
sait peu de choses sur la façon dont le processus de sélection du microhabitat est influencé par
les variations des conditions hydrauliques à cette échelle.

Concernant les poissons, les conclusions des précédentes études divergent et on ne sait
pas bien si lorsque le débit varie, les poissons (1) suivent leurs conditions d’habitat favorables
dans l’espace ou (2) restent sur place et subissent des conditions d’habitat pouvant être
défavorables. Kemp et al. (2003), en milieu expérimental, et Korman et Campana (2009), par
pêche électrique in situ, ont constaté que peu d'individus (Salmon parr et Oncorhynchus mykiss
respectivement) cherchent à se maintenir dans un habitat favorable lorsque le débit varie. Deux
à trois fois plus de poissons étaient pêchés proche des berges à bas débit par rapport à haut
débit. A l’inverse, Robertson et al. (2004) et Harby et al. (2001) ont montré que les juvéniles
de saumons atlantiques se déplaçaient à chaque variation de débit pour suivre leurs conditions
d'habitat favorables. Par ailleurs, certains organismes seraient capables de percevoir leur
environnement et de sélectionner les habitats les moins variables. C’est ce qu’ont montré Capra

31
et al (2017) par télémétrie sur le Rhône en France sur trois espèces de poissons (barbeau,
chevesne et silure) : les organismes évitaient les zones qui s’assèchent lors des baisses de débit.
De même, Geist (2008) a montré en laboratoire que les esturgeons sélectionnent les habitats les
moins variables en termes de vitesse.

Concernant les macroinvertébrés, des études ont montré (1) qu’ils étaient moins
abondant le long des berges à cause des assèchements fréquents (Fisher et al. 1972, Troelstrup
et al. 1990, Blinn et al. 1995) et (2) que la zone hyporhéique (dans le substrat) pourrait jouer un
rôle de refuge lors des fortes vitesses ou des assèchements (Dole-Olivier et al. 1997,
Stubbington et al. 2019). Cependant, peu d’études ont comparé la réponse des macroinvertébrés
benthiques à celle des macroinvertébrés hyporhéiques dans les rivières à éclusées pour
confirmer cette dernière hypothèse.

Malgré les études montrant que la dynamique spatio-temporelle des habitats pouvait
influencer le comportement de sélection du microhabitat des poissons et des macroinvertébrés,
peu d’articles ont étudié de manière quantitative l’influence des assèchements et encore moins
celle des fortes vitesses de courants passés sur la sélection du microhabitat des poissons et des
macroinvertébrés (Capra et al 2017). Cela est certainement dû à la difficulté de décrire les
conditions hydrauliques passées des microhabitats car cela nécessite des modèles
hydrodynamiques précis et spatialement explicites, tels que des modèles bidimensionnels (2D)
qui permettent de cartographier les hauteurs d’eau et les vitesses du courant (moyennées sur la
verticale) à différents débits. L'étalonnage (hauteur d’eau pour un ou des débits donnés) et la
validation (hauteur d’eau et en vitesse du courant) des modèles 2D prend beaucoup de temps et
sont donc très coûteux. De plus, chercher les liens entre la sélection du microhabitat et les
conditions hydrauliques passées nécessite un travail de terrain important, afin d'échantillonner
les organismes à plusieurs débits dans un nombre suffisant de microhabitats et avec diverses
combinaisons de conditions hydrauliques passées et présentes (p. ex. pour limiter la corrélation
négative attendue entre la fréquence d'assèchement et la hauteur d’eau).

32
2.2 Les variations hydrauliques influencent la sélection du microhabitat
des macroinvertébrés et des poissons dans les rivières à éclusées (Judes
et al. B)

Judes et al. (B) ont étudié l'effet des éclusées de la basse rivière d'Ain (Encadré 3) sur
la sélection du microhabitat des poissons et des macroinvertébrés principalement dans le but de
(1) quantifier l’influence de l’hydraulique passée (assèchement et forte vitesse) sur la sélection
de l'habitat et (2) de montrer si la sélection du microhabitat dans une rivière à éclusées était
différente de la sélection du microhabitat dans des tronçons de rivière non influencés par des
éclusées. Pour cela, Judes et al. (B) ont observé l'abondance des taxa de poissons dans 507
microhabitats par pêche électrique et 673 microhabitats par observations subaquatiques et
l'abondance des taxa de macroinvertébrés benthiques et hyporhéiques dans 36 microhabitats le
long d'un tronçon de 6 km en conditions d’éclusées. L'échantillonnage a été réalisé à différents
débits pour réduire la corrélation entre les conditions hydrauliques présentes et passées (p. ex.
la corrélation négative attendue entre la fréquence d'assèchement et la hauteur d’eau). Un
modèle hydraulique 2D calibré sur le tronçon a permis de simuler l’hydraulique passée des
microhabitats. Pour chaque microhabitat le passé hydraulique a été représenté par deux
variables : l’assèchement et les vitesses de courant élevées (>1,3 m s-1) sur les 15 jours
précédant l’échantillonnage. Ces deux variables étaient qualitatives : condition rencontrée
pendant plus de 10 heures ou moins de 10 heures. L’influence des conditions hydrauliques, a
été testée sur des guildes de taxa représentant les préférences d’habitat. Pour les poissons, la
guilde “chenal” regroupe les taxa utilisant des habitats à fortes vitesses de courant et profond,
et la guilde “berge”, les taxa utilisant les habitats peu profonds et des faibles vitesses de courant.
Judes et al. (B) ont regroupé les macroinvertébrés préférant les faibles vitesses de courant dans
la guilde “limnophile” et ceux ayant une affinité pour les fortes vitesses de courant dans la
guilde "rhéophile". Seule la comparaison de la sélection du microhabitat dans l’Ain avec les
rivières sans éclusées a été faite sur les taxa individuellement quand des données étaient
disponibles dans la littérature. Chaque abondance de guilde a été liée à l’hydraulique passée et
présente des microhabitats (mesurée lorsqu'elle est disponible, modélisée sinon) par des GLMs.
Pour éviter une surparamétrisation et en raison d'une puissance statistique limitée, Judes et al.
(B) ont dû considérer des modèles impliquant des combinaisons d'une seule variable
hydraulique présente (hauteur d’eau/vitesse) et d'une seule variable hydraulique passée
(assèchement/forte vitesse).

33
Encadré 3. Zoom sur la rivière d’Ain.

La rivière d’Ain est le plus important affluent du haut Rhône, elle

draine un bassin versant de 3630 km2 et parcourt 200 kilomètres de
sa source sur le plateau de Nozeroy (Jura) à sa confluence avec le
Rhône (St Maurice de Gourdans). La rivière d’Ain a un régime
pluvial (ou pluvio nival), elle est soumise à des crues rapides et
fortes (les crues cinquantennales sont de 2500 m3 s-1). De par son
lit constitué de galets et de graviers et ses berges très érosives
(galets, sables, limons), elle se transforme à chaque crue.
Il existe six centrales hydroélectriques sur la rivière d’Ain, de
l’amont vers l’aval : Vouglans, Saut Mortier, Coiselet, Moux
Charmines, Cize-Bolozon et le barrage d’Allement. Le tronçon d’étude de Judes et al. (B.)
est situé à 20 km à l’aval du barrage de l’Allement dans ce qu’on appelle la Basse Vallée de
l’Ain.
Le débit moyen annuel sur ce site d’étude est de 103 m3 s− 1 avec un débit moyen journalier
compris entre 14 m3 s-1 (dépassé 95% du temps) et 278 m3 s-1 (dépassé 5% du temps). Le
tronçon d'étude est soumis à de fréquentes variations de débit infra-journalières, avec un débit
de base généralement compris entre 14 m3 s-1 et (occasionnellement) 150 m3 s-1 et un débit
de pointe généralement compris entre 40 m3 s-1 et 200 m3 s-1. Le débit est plus stable pendant
les week-ends et les périodes
de faible débit (juin-
septembre).
La Basse Vallée de l’Ain est
connue pour sa faune
piscicole variée, on y compte
de nombreuses espèces de
cyprinidés (barbeaux,
chevesnes, vairons, hotus…)
mais aussi des salmonidés,
l’ombre commun et la truite.
Elle était jusqu’en 1994 la seule rivière en France à être classée « principalement peuplée
d’ombres communs ». Elle possédait de ce fait une grande renommée au niveau européen
pour la pêche. Dès 1939, Vibert évoquait l’impact des éclusées sur les peuplements de
poissons et la biodiversité sur l'Ain et exposait la perte économique encourue si l’image
halieutique de la rivière d’Ain se dégrade. De 1994 à 2011, la convention « Frayères » passée
entre EDF et la fédération de pêche de l’Ain garantissait un débit minimum de 28 m 3 s-1
pendant la période de reproduction des salmonidés (01/12 - 01/05), alors qu’il est
réglementairement de 12,3 m3 s-1 le reste de l’année, en sortie d’Allement. Le but de cet
accord était de limiter l’exondation des zones de fraie après la ponte. Une nouvelle
convention (la convention « Salmonidés ») a été signée en janvier 2013. Encore maintenant,
les associations de pêche de la basse vallée de l’Ain négocient avec EDF pour limiter au
maximum l’impact des éclusées sur les poissons.

34
Figure I.6. Échantillonnage des poissons et des macroinvertébrés à plusieurs débits sur
l’Ain et carte des vitesses de courant obtenue à partir du modèle hydraulique 2D (Judes
et al. B). (A) Echantillonnage de macroinvertébrés hyporhéiques (avec un tube bou-rouch).
(B) Echantillonnage de macroinvertébrés benthiques (avec un bethométre). (C)
Echantillonnage de poissons par pêche électrique. (D) Echantillonnage de poissons par
observations subaquatiques. (E) Débit horaire à Pont d'Ain (7,5 km en amont du tronçon
d'étude) pendant les 15 jours précédant l'échantillonnage et pendant l'échantillonnage (2018).
Les flèches indiquent les jours d'échantillonnage à midi. A noter que les forts débits (~100
m3 s-1) du 17 septembre au 28 septembre sont des lâchers d’eau pour le soutien des centrales
nucléaires du Rhône. (F) Cartes de la vitesse de courant pour le débit maximal et le débit
minimal rencontrés sur la période d'étude. La flèche indique le sens du courant.

35
 Judes et al. (B) ont montré que l’hydraulique passée, en particulier les assèchements
passés, influence grandement la sélection des microhabitats des macroinvertébrés. Les
macroinvertébrés étaient 3-15 fois moins abondants dans les microhabitats qui se sont asséchés.
Les macroinvertébrés ne semblent pas capables de suivre leurs conditions d’habitat favorables
puisque leurs préférences pour l'hydraulique présente étaient différentes de celles observées
dans les rivières sans éclusées (Figure I.7). Ces observations amènent à se questionner sur le
terme de « sélection du microhabitat » lorsque l’on parle d’organismes à faible mobilité dans
des rivières très dynamiques telles que les rivières à éclusées (Encadré 4). Par ailleurs, Judes et
al. (B) ont montré que les macroinvertébrés benthiques et hyporhéiques ont répondu de la même
manière aux assèchements. Au vu de ce résultat les macroinvertébrés dits « hyporhéiques » ont
probablement été échantillonnés encore dans la zone benthique même si les prélèvements se
sont faits à 30 cm de profondeur dans le substrat. Il semblerait, en tout cas, qu’à cette profondeur
le substrat ne soit pas saturé en eau de façon pérenne. Les futures recherches devront tester s’il
existe une zone hyporhéique pérenne sur les bords de la rivière d’Ain, si oui à quelle profondeur
elle se trouve et si elle joue un rôle refuge à l’asséchement pour certain taxa de macroinvertébrés
(certainement les plus petits).

Encadré 4. La sélection du microhabitat dans les milieux fortement dynamiques.

La sélection de l’habitat est définie comme un choix actif (inné ou appris) d’utiliser un
habitat plutôt qu’un autre disponible autour de lui (Krausman 1999). L'abondance est utilisée
comme un proxy de la sélection de l’habitat puisque la préférence d'habitat se traduit par
l'utilisation disproportionnée de certains habitats par rapport à d'autres (Morrison et al. 1992).
Lorsque l’hydrologie est naturelle (moins variable que dans les rivières à éclusées),
l'hypothèse d’un choix actif est certainement vraie mais cela peut poser question dans les
rivières à éclusées, en particulier pour les taxa ayant de faibles capacités de déplacement (ex.
les macroinvertébrés). Dans un milieu fortement variable, la part de processus actif et passif
expliquant l'abondance d’un taxon dans un microhabitat dépend de la durée entre le
changement de débit et l'échantillonnage par rapport aux capacités de déplacement et de
résistance du taxon (à l'assèchement ou aux fortes vitesses de courant). Par exemple, si un
organisme a de faibles capacités de déplacement, les abondances observées à l’instant t
peuvent refléter un choix actif ayant eu lieu à l’instant t-1 puis une résistance aux conditions
de l’instant t (choix passif). Plus les conditions de ce que l’on observe sont là depuis
longtemps (durée dépendante du taxon considérée), plus il est probable que les abondances
observées soient issues d’un processus actif de sélection de l'habitat.

Judes et al. (B) ont montré que les poissons suivent leurs conditions d’habitat favorables
(Figure I.7). Ces résultats s'opposent à certaines études ayant montré que les poissons étaient
réticents à se déplacer pour suivre leurs conditions d’habitat favorables avec le débit (Kemp et
al. 2003, Korman & Campana, 2009). Les différences de réponses dépendent certainement de

36
l'équilibre entre le gain/perte associé au fait (1) d’être toujours dans les conditions hydrauliques
favorables (gain: optimisation de l'alimentation et du repos ; perte: coût énergétique et risque
de prédation associé au déplacement) et (2) de rester au même endroit (gain: garder le territoire
et éviter la prédation pendant le déplacement ; perte: difficulté à se nourrir ou à se reposer et
coût énergétique lié au fait de subir des conditions hydrauliques défavorables) (Korman et al.
2009). Ces différences sont aussi largement attribuées à des facteurs (1) physiques, par exemple
la présence de refuges de vitesse et la dynamique spatio-temporelle des habitats et (2)
biologiques, par exemple la variabilité inter et intra-individuelle ou la présence de prédateurs
(Krimmer et al. 2011, Taylor et al. 2014).

Judes et al (B) ont montré que l'hydraulique passée avait une influence moins forte sur
la sélection de l’habitat des poissons par rapport aux macroinvertébrés. Néanmoins, les poissons
de la guilde « berge » sélectionnent positivement les microhabitats qui se sont asséchés et à
l’opposé les poissons de la guilde “chenal” les évitent. Ces différences s'expliquent par les
préférences d’habitat plus strictes des poissons de la guilde “berge”, ils sont obligés d’utiliser
les microhabitats qui s'assèchent pour rester dans leurs conditions d’habitat favorables. La
relation positive s’explique par le fait que les poissons se réfugient sur les bords lors des
augmentations de débits. Les résultats de Judes et al. (B) fournissent donc un aperçu
mécanistique permettant en partie d’expliquer comment les espèces de poissons des rivières de
taille moyenne sont affectées par les éclusées : les poissons suivent leurs conditions d’habitat
favorables et ces déplacements provoquent certainement des pertes énergétiques, de la fatigue
et augmentent le risque de prédation. Certaines études ont montré qu’effectivement les poissons
se déplacent plus dans les rivières à éclusées que dans les rivières sans éclusées (Harby et al.
2001, Harvey-Lavoie et al. 2016) mais d’autres n’ont pas trouvé de différence (Bunt et al. 1999,
Berland et al. 2004). Néanmoins, ces études impliquent des suivis individuels par télémétrie et
ignorent généralement les déplacements et mouvements des petits organismes (plutôt guilde
« berge ») qui semblent les plus concernés par ces déplacements fréquents.

Judes et al (B) n’ont pas montré d’influence significative des fortes vitesses passées
chez les poissons et chez les macroinvertébrés limnophile. Les poissons de la guilde “berge”
restent proches des berges et ne subissent donc pas les fortes vitesses alors que les poissons de
la guilde “chenal” arrivent certainement à les supporter ou se cachent dans les refuges
hydrauliques. Les macroinvertébrés rhéophiles auraient tendance à sélectionner positivement
les microhabitats soumis à des fortes vitesses passées. Ils ont certainement la capacité de résister

37
aux fortes vitesses en raison de leurs adaptations morphologiques ou comportementales aux
fortes vitesses de courant. Holomuzki & Biggs (2000) ont étudié la réponse du comportement
de quatre espèces de macroinvertébrés de Nouvelle-Zélande (l'escargot Potamopyrgus
antipodarum, l'éphémère Deleatidium spp, et deux espèces de Tricoptère : Pycnocentrodes
aeris et Hudsonema amabilis) aux fortes vitesses de courant (1,4 m s-1). Ils ont montré que ces
espèces pouvaient “sentir” l’augmentation des contraintes de cisaillement et y répondre
rapidement. L’ensemble des espèces ont réalisé de petits mouvements entre les couches de
substrat pour chercher des refuges hydrauliques, en s'enfonçant dans le substrat ou en se
repositionnant à la surface du substrat (Lancaster et al. 1997). Par ailleurs, certains taxa ont des
adaptations physiologiques pour s'accrocher aux substrats et tenir lors des fortes vitesses de
courant. Par exemple, les Trichoptères se fixent aux substrats à l’aide de leurs soies, l'éphémère
grâce à ces griffes tarsiennes et l'escargot par son pied.

Les principaux résultats de la thèse sont résumés dans la Figure I.8.

38
Figure I.7. Résumé des résultats de Judes et al. (B). (A) Ajustements des modèles (ou
courbe de préférence d’habitat) reliant l'abondance des guildes à l'hydraulique présente et
incluant un effet additif de l'assèchement (courbe rouge). Les étoiles (**) indiquent que
l'effet de l'hydraulique passée était significatif P<0,01. (B) Adéquation de la préférence
moyenne pour la vitesse obtenue dans Judes et al. (B) avec les données de la littérature sur
des rivières majoritairement non-soumises aux éclusées. Les régressions des axes (lignes
pleines) sont représentées ainsi que les lignes y = x (lignes pointillées). (C) Tableau des
effets des variables de l’hydraulique passée sur l’abondance des guildes de taxa. Les flèches
rouges représentent un effet négatif, les flèches bleues un effet positif. Le nombre de flèches
représente la significativité des résultats : une flèche pour P<0,05 ; double flèche pour
P<0,01. Les résultats sur les poissons concernent uniquement les données de pêche
électrique (les observations subaquatiques n’ont pas révélé d’effets significatifs de
l’hydraulique passée).

39
 Figure I. 8. Résumé des principaux résultats de la thèse.

2.3 Perspectives scientifiques

Quelles sont les conséquences des mouvements fréquents des poissons sur leurs
bilans bioénergétiques ?

Judes et al (B) ont montré que les poissons de la guilde « berge » (globalement des
juvéniles) se repositionnent à chaque changement de débit pour suivre leurs conditions d’habitat
favorables. Les déplacements fréquents de ces poissons provoquent certainement des pertes
énergétiques, de la fatigue et augmentent le risque de prédation.

Plusieurs études ont comparé de manière qualitative le poids par rapport à la taille
(condition physique), la masse corporelle, ou le taux de croissance des organismes entre des
conditions avec ou sans variations artificielles de débit (en laboratoire ou in situ). Malgré
quelques exceptions (Puffer et al. 2015), la majorité des études ne semblent pas avoir montré
d’effets des éclusées sur ces paramètres (Enders et al. 2017, Finch et al. 2015, Flodmark et al.
2006, Puffer et al. 2017). Par exemple, Puffer et al. (2017) et Foldmark et al. (2006), dans des

40
chenaux artificiels, ont montré uniquement des effets mineurs et non significatifs des éclusées
sur les performances de croissance des salmonidés (saumons atlantiques et juvéniles de truites
brunes). Ainsi cela suggère que le bilan bioénergétique est équilibré (c.-à-d. que l'énergie
dépensée est égale à l'énergie gagnée) : il ne semble ne pas y avoir d'augmentation ou de
diminution des réserves internes d'énergie en réponse aux éclusées. Cependant, puisque tout
comportement utilise du temps et de l'énergie (Cuthill et al. 1997), soit (1) la dérive des
macroinvertébrés, plus forte durant les éclusées (Schulting et al. 2016), permet de compenser
les pertes énergétiques pour certains taxa, soit (2) pour compenser les poissons sont obligés
d'augmenter leur temps à rechercher de la nourriture. Puisque les organismes sont sujets à des
plus forts risques de prédation lorsqu’ils sont en recherche de nourriture que lorsqu’ils se
reposent (McNamara et al. 1992) cela peut affecter leurs survies. Il est donc important d’étudier
si le temps de recherche de nourriture est plus important dans les rivières à éclusées. Pour cela,
je suggère deux approches.

Premièrement, une approche empirique visant à comparer le temps à rechercher de la

nourriture par les poissons avec ou sans éclusées. Bien qu’aucune ne soit idéale pour le moment,
plusieurs méthodes sont possibles. Parmi elles, l’expérimentation ex situ ou in situ via des
vidéos (Boisclair, 1992, Enders et al. 2005, Sabo et al. 1996, Shamur et al. 2016). Les approches
basées sur la vision fonctionnent bien lorsque l'eau est claire et que les poissons sont dans le
champ de vision de la caméra. Lorsque ce n’est pas le cas, il sera peut-être bientôt possible
d’utiliser des électrocardiogrammes puisque Shen et al. (2021) ont développé un algorithme
permettant de détecter les phases de nourrissage par les changements du rythme cardiaque.
Cependant, utiliser le rythme cardiaque en tant qu'indicateur d'activité peut être parfois difficile
puisque la fréquence cardiaque est influencée par d'autres variables (température de l’eau, stress
etc.). Pour l’instant, il n’est pas possible d’étudier les comportements de recherche de nourriture
par des suivis individuels de poissons par télémétrie en rivière car la précision des localisations
ne permet pas de les différencier du repos.

Deuxièmement, une approche théorique qui consisterait à convertir les distances entre
les deux habitats favorables successifs (Batz et al. soumis) en coût énergétique associé à la nage.
La conversion des mouvements des poissons en coûts nécessite des relations entre les
caractéristiques des mouvements et les dépenses énergétiques. Les relations sont généralement
développées à partir d'expériences de respiromètrie réalisées en laboratoire (Enders et al. 2005,
Enders et al. 2003, Guensch et al. 2001, Sabo et al. 1996). Une fois les pertes énergétiques

41
connues, elles pourraient être traduites en temps supplémentaire à rechercher de la nourriture
grâce à des modèles de recherche de nourriture. Les modèles de recherche de nourriture sont
des simplifications du mode de vie dans lequel les poissons trouvent et consomment leurs
nourritures. Ces modèles prennent en considération des facteurs tels que le comportement de
recherche de nourriture, la vitesse de nage, la densité et la taille des proies, la distance et
l'efficacité de capture des proies (et leur dépendance par rapport à la clarté de l'eau) (Guensch
et al. 2001, Hayes et al. 2007).

Comment compléter l’étude de Judes et al. (B) ?

Pour confirmer les hypothèses formulées par Judes et al. (B) il faudrait tout d’abord
tester si l’effet de l’hydraulique passée sur la sélection du microhabitat des poissons et des
macroinvertébrés est transférable et généralisable à d’autres rivières. Pour cela, il faudrait
réaliser la même expérimentation que Judes et al. (B) dans une rivière à l’hydrologie et à
l’intensité de gestion des éclusées contrastées par rapport à la rivière d’Ain.

Concernant les macroinvertébrés, plusieurs approches sont possibles pour compléter

l’étude de Judes et al. (B). Premièrement, Judes et al. (B) ayant échantillonné uniquement à fort
débit (~ 100 m3 s-1) et au débit de base (14 m3 s-1), il faudrait échantillonner à des débits
intermédiaires (p. ex. 28/50/70 m3 s-1) afin de tester la réponse des macroinvertébrés à d’autres
patterns d’assèchement (p.ex. différentes durées entre deux remises en eau et différentes
fréquences d’assèchement). Deuxièmement, il faudrait tester l’influence de la durée depuis la
remise en eau puisqu’il s’agit un facteur important pour la recolonisation des microhabitats
asséchés (Kjaerstad et al. 2018). Pour cela, il faudrait échantillonner des microhabitats remis en
eau depuis différentes durées (p. ex. 15 jours, 1 mois, 2 mois). Sur l’Ain, pour obtenir une
longue période de fort débit il faudra échantillonner à la fin du printemps. A priori, Judes et al.
(B) ont montré qu’une période de huit à neuf jours de fort débit (lâché prolongé de 100 m3 s-1
pour le fonctionnement des centrales nucléaires du Rhône, Figure I.6) ne suffit pas pour
permettre la recolonisation complète des microhabitats souvent asséchés. Troisièmement, il
faudra tester l’influence de l’hydraulique passée vécue sur une période plus longue avant
l’échantillonnage que les 15 jours testés par Judes et al. (B) (c.-à-d. 1 mois ou 2 mois avant
l’échantillonnage). L’été 2018 (juste avant l’échantillonnage de Judes et al. (B)) a été très sec,
les débits n’ont pas dépassé 100 m3 s-1 de la mi-juin à la fin août. Cette longue période de faible

42
débit a pu expliquer en partie la réponse des macroinvertébrés à l’asséchement de Judes et al.
(B). Enfin, il faudrait répliquer l’expérimentation de Judes et al. (B) au printemps pour avoir le
maximum de taxa échantillonnés et identifiés jusqu’à l’espèce. En effet, Judes et al. (B) ayant
échantillonné en automne ils n’ont pas pu étudier l’ensemble des taxa de l’Ain car la plupart
émergent au printemps ou été (Brabec et al. 2021, Buffagni et al. 2021, Graf et al. 2021 a, b).
Judes et al. (B) avaient dû faire un compromis entre l’hydrologie (avoir des débits contrastés)
et la période idéale des échantillonnages biologiques.

Concernant les poissons, pour confirmer que les poissons de la guilde “chenal” évitent
les microhabitats qui s’assèchent, il faudra d'abord tester l’influence de la disponibilité des
habitats sur la sélection de l’habitat (Encadré 5) (Pert et al. 1994). Cela n’a pas pu être fait par
Judes et al. (B) à cause d’un manque de pouvoir statistique lié au fait de devoir diviser le jeu de
données en deux (faible et fort débit) pour le tester. Par ailleurs, l’échantillonnage sur des
poissons de la guilde “chenal” devra être complété par des observations subaquatiques à fort
débit ou des suivis par télémétrie. L'avantage de l’observation subaquatique est qu’elle permet
d’apporter de grandes quantités de données rapidement et à faible coût, mais pourrait être
limitée par des vitesses de courant trop fortes. La télémétrie est sans doute la seule méthode
permettant d’observer les organismes dans les habitats à fortes vitesses de courant, mais elle
peut être sensible à la profondeur (radio) ou à la turbulence (acoustique). Cette méthode
permettrait cependant de tester si les poissons de la guilde « chenal » utilisent des habitats à
plus fortes vitesses de courant lors des forts débits.

Par ailleurs, il serait intéressant de tester l’influence de l’hydraulique passée sur la

sélection du microhabitat par les poissons par d’autres méthodes que les modèles corrélatifs
développés par Judes et al. (B). Parmi les autres méthodes existantes, il y a l'écologie du
mouvement (Nathan et al. 2008) qui vise à étudier les interactions entre les individus et
l’environnement dans leurs déplacements par l'analyse des trajectoires observées des
organismes (p. ex. vitesse, orientation, changements de direction). Cette approche pourrait être
mise en œuvre par télémétrie acoustique (précision dans la localisation des poissons < 5 m)
dans un tronçon pour lequel un modèle hydraulique 2D est disponible. La télémétrie acoustique
permet de calculer la vitesse et les directions des déplacements des poissons, de détecter les
phases d'accélération et les changements de direction à échelle fine quasiment en temps réel.
La première étape consistera à proposer des indicateurs de trajectoire (vitesse absolue, vitesse

43
angulaire, accélération, etc.) pour discriminer les différents types de comportements
(statique/mouvement). Ensuite, grâce au modèle hydraulique 2D, il s’agira de décrire comment
l’hydraulique présente et passée (même variables testées que Judes et al. (B)) influence les
changements de comportement. L’hypothèse serait que pour les poissons de la guilde
« chenal », la probabilité de rester dans un microhabitat qui s’assèche est plus faible que de
rester dans un microhabitat qui ne s’assèche pas alors même que les conditions hydrauliques
présentes sont semblables. Concernant les poissons de la guilde « berge », il serait intéressant
d’étudier le temps de réaction pour fuir les microhabitats qui vont s’assécher lors des baisses
de débit.

Encadré 5. Influence de la disponibilité de l’habitat sur l’interprétation des résultats

des études sur la sélection de l’habitat

Lorsque des différences de comportement de sélection de l’habitat sont observées à différents

débits, elles peuvent être liées à une différence de disponibilité de l’habitat. En effet, certains
organismes peuvent choisir différentes conditions d’habitat en fonction de la gamme de
valeurs disponibles autour d’eux
(Belaud et al. 1989, Arthur et al.
1996). Si en effet, les organismes
suivent la disponibilité des habitats
alors il peut y avoir des erreurs
dans l'interprétation des résultats,
c’est-à-dire de conclure à un
évitement des microhabitats qui
s’assèchent à tort.

La figure ci-contre illustre les

courbes de sélection de la
profondeur d’eau à bas débit (14
m3 s-1) et à haut débit (214 m3 s-1)
pour la guilde “chenal” issues des
données de Judes et al. (B). Les
valeurs de disponibilité de l’habitat de moins d’un mètre dans le tronçon d’étude de Judes et
al. (B) à bas et à haut débit sont aussi représentées (obtenues à partir du modèle
hydrodynamique 2D). Dans cet exemple, à bas débit, 91% des poissons se trouvent dans des
habitats de moins d’un mètre alors que ces derniers représentent 77% de l’habitat disponible.
Dans ces habitats de moins d’un mètre on ne retrouve plus que 57% des poissons
échantillonnés à haut débit mais ces habitats ne représentent plus que 34% de l’habitat
disponible. Les habitats de moins d’un mètre à haut débit sont en grande majorité soumis à
des assèchements fréquents (81 %). Ainsi si l’on ne considère pas la disponibilité et que l’on
met dans une même analyse les données issues des échantillonnages à haut et à bas débit, il
est fort probable que la conclusion soit que les poissons évitent les zones d’assec alors qu’en
réalité ils les sélectionnent positivement (plus nombreux dans les zones d’assec par rapport
à leur disponibilité).

44
 Judes et al. (B) ont étudié l’hydraulique présent par deux variables corrélées, la hauteur
d’eau et la vitesse moyenne du courant. A l’avenir, il faudra considérer la hauteur d’eau et la
vitesse simultanément dans les modèles statistiques. Mieux encore, pour ne pas surparamétrer
les modèles et parce que les organismes perçoivent la hauteur d’eau et la vitesse de courant
simultanément (Coarer 2007, Muñoz-Mas et al. 2012, Plichard 2018), il faudrait utiliser une
variable qui combine la hauteur d’eau et la vitesse comme le nombre de Froude (FR=V/[g x
H]0.5 ou g est l’accélération due à la gravité, sans diminution) (Lamouroux et al. 2002). Par
ailleurs, les conditions hydrauliques ne sont pas les seuls paramètres qui décrivent l’habitat des
poissons et des macroinvertébrés. Les futures études devront tenir compte de la température et
du substrat. D’une part, ces paramètres influencent la sélection du microhabitat par les poissons
et les macroinvertébrés (Bradford et al. 1997, Choi et al. 2017) et cela peut possiblement
masquer les effets de l’hydraulique passée. D’autre part, ils influencent la susceptibilité à
l'échouage-piégeage et à la dérive et ainsi peuvent influencer le choix d’occuper ou non un
microhabitat qui s’assèche ou qui est soumis à des fortes vitesses de courant. Par exemple,
l’échouage des poissons semble être plus élevé dans les substrats grossiers (>7,6 cm) par rapport
à des substrats plus fins (Beck et al. 1989), et durant les basses températures (Bradford et al.
1997).

Y a-t-il une adaptation comportementale aux variations des conditions

hydrauliques ?

En général, le processus de sélection du microhabitat est adaptatif car il repose sur des
signaux (p. ex. conditions hydrauliques) qui, au cours de l'évolution, ont été corrélés de manière
fiable avec la survie et le succès reproducteur. Cependant, si l’environnement change
soudainement (p.ex. apparition des éclusées), ces signaux jusqu’alors considérés peuvent ne
plus être en corrélation avec le résultat attendu et, par conséquent, la sélection du microhabitat
peut ne plus être adaptative (Williams et al. 1984). Dans le cas des rivières à éclusées, par
exemple, suivre ses préférences d’habitat n’est plus adaptatif si l’organisme n’est pas capable
de fuir à temps les conditions hydrauliques extrêmes pour échapper l’échouage-piégeage ou la
dérive forcée. Dans ce cas, le comportement le plus adaptatif serait de sélectionner des habitats
certainement moins favorables à l’instant t mais qui ne subissent pas de changements
hydrauliques extrêmes (microhabitats qui ne s’assèchent pas ou qui ne sont pas soumis à des
fortes vitesses de courant). Un tel changement de comportement pourrait apparaitre par

45
plasticité phénotypique (basé sur l’apprentissage) ou par sélection naturelle (sélection sur la
variation génétique existante) (Kokko et al. 2001, Schlaepfer et al. 2002).

Judes et al (B) ont étudié la réponse d'individus de l’Ain subissant des éclusées depuis
1930, il est donc possible que ces individus soient adaptés (sélection naturelle) ou acclimatés
(plasticité phénotypique) aux variations rapides des conditions hydrauliques. En particulier, les
poissons de la guilde « chenal » sont peut-être adaptés ou acclimatés pour éviter les
microhabitats qui s’assèchent et les poissons de la guilde « berge » sont peut-être adaptés ou
acclimatés pour changer rapidement de position lors des baisses ou des hausses de débits. Pour
tester ces hypothèses, je comparerais en condition expérimentale les réponses
comportementales aux variations hydrauliques artificielles d'organismes issues de rivière à
éclusées avec ceux issus de rivière non influencée par les éclusées. Concrètement, il s’agirait
de regarder le temps de réponse pour fuir les microhabitats qui vont s’assécher en fonction de
l’origine des poissons, ou les préférences moyennes pour la vitesse et la hauteur d’eau. A priori,
Judes et al. (B) suggèrent que les préférences pour les vitesses moyennes pourraient être plus
faibles dans les rivières à éclusées (Figure I.7), il s’agit peut-être d’une adaptation (ou
acclimatation) pour éviter de se faire emporter par les fortes vitesses de courant lors des hausses
de débit. Une autre manière de faire serait d’étudier génération après génération comment la
sélection du microhabitat des taxa évolue dans une rivière récemment soumise aux éclusées.

46
3. PERSPECTIVES OPERATIONNELLES

3.1 Considérer les connaissances sur la sélection du microhabitat pour

mieux quantifier l’influence des éclusées à l’échelle du tronçon

L'échelle du microhabitat est celle qui représente le mieux les conditions d'habitat qui
sont perçues et exploitées par les organismes. Il s’agit donc de l’échelle la plus pertinente pour
essayer d’identifier les mécanismes expliquant les observations faites à des échelles plus larges.
Cependant, les mesures opérationnelles ignorent généralement l’influence des perturbations à
l’échelle du microhabitat (Pringle et al. 1988) et concernent principalement l'échelle du tronçon.
Il est donc nécessaire de disposer de méthodes qui intègrent les résultats obtenus à l’échelle du
microhabitat dans des approches qui permettent d'estimer l’influence des perturbations sur les
populations à l’échelle du tronçon (Winemiller et al. 2010).

Quantifier la dynamique spatio-temporelle des habitats

En partant du constat que les poissons et les macroinvertébrés ne répondent pas de la

même manière aux éclusées (mouvement ou statique), Batz et al. (Soumis) suggèrent une
approche pour quantifier l’influence de dynamique spatio-temporelle des habitats sur les
organismes selon un trait d’histoire de vie : la mobilité.

On a vu précédemment que les poissons suivent leurs conditions d’habitat favorables et

que ces déplacements peuvent entrainer des pertes énergétiques (dues à la nage) et/ou
augmentent le risque de prédation. Ainsi, Batz et al (Soumis) ont proposé des métriques
quantifiant les distances de déplacement nécessaires pour se maintenir dans une condition
d’habitat particulière à chaque changement de débit (Figure I.9). L’hypothèse étant que les
poissons se déplacent dans le nouveau microhabitat favorable le plus proche de lui à chaque
changement de débit. Cependant, Shirvell (1994) a montré par télémétrie qu’un poisson se
déplaçait en moyenne de 6,8 m à chaque changement de débit alors que les conditions
hydrauliques se déplaçaient latéralement en moyenne de 2,0 à 2,9 m. Ce résultat contraste avec
les précédentes études montrant que les poissons sont capables de percevoir leur environnement
et d'adapter le comportement de sélection de l’habitat le moins contraignant possible (Geist et
al. 2005, Capra et al. 2017). De plus, le suivi strict des conditions d’habitat favorables concerne

47
certainement plus les poissons de petite taille (guilde « berge ») qui ne peuvent pas être suivis
par télémétrie. Les futures études devront creuser les liens entre déplacements des organismes
et déplacements des conditions d’habitat (Conallin et al. 2014) mais cela risque d’être
compliqué à réaliser puisque les poissons se déplacent souvent dans différents habitats selon
leurs activités (p. ex. repos, recherche de nourriture).

A l’inverse, les macroinvertébrés, du fait de leurs faibles capacités de déplacement, sont

particulièrement affectés par les variations des conditions d’habitat au sein des microhabitats.
Pour rendre compte de la perturbation vécue par les macroinvertébrés, Batz et al (Soumis)
proposent de délimiter des patchs (zones géographiquement fixes (Pickett et al. 1997) entre 0
et 100m2 (Pringle et al. 1988, Winemiller et al. 2010) et d’examiner les variations temporelles
des conditions d’habitat au sein de ces patchs. Concrètement, ils ont délimité cinq classes
(toutes les 0.25 m s-1) de conditions de vitesse (ou de hauteur d’eau) et ont quantifié le nombre
de changements de classe sur une période donnée. Ils proposent aussi de quantifier la
persistance temporelle d’une condition d’habitat dans les patchs alors que la persistance était
jusqu’alors utilisée pour décrire les tronçons (Freeman et al. 2001). A l’échelle du microhabitat,
Judes et al. (B) suggèrent que la persistance de mise en eaux semble importante pour les
macroinvertébrés en général et la persistance des faibles vitesses de courant importante pour les
macroinvertébrés limnophiles.

La persistance des faibles vitesses pourrait aussi être favorable pour les poissons pour
des stades de vie précoces comme les œufs puisque la persistance des conditions de vitesses
faibles semble essentielle pour limiter le décapage. Par exemple, Bartoň et al. (2021) ont montré
que le site de frai des poissons ne devrait pas dépasser 0,7 m s-1 (sur les œufs de Leuciscus
aspius sur la rivière de Želivka en Réplique tchèque).

Avant que ces métriques soient utilisées pour la gestion il sera nécessaire de (1) tester
leurs pertinences biologiques et (2) de trouver des méthodes pour intégrer les résultats dans des
mesures écologiques significatives à l'échelle du tronçon. Cependant, les gestionnaires qui
adopteront la perspective des organismes (mobilité) telle que décrite ci-dessus auront l’avantage
de permettre une approche spatio-temporelle et de proposer des schémas de gestion affectant la
dynamique et la structure de la mosaïque d'habitats (Pringle et al. 1988, Pickett et al. 1997,
Hitchman 2017). De plus, une telle approche à l'échelle locale facilitera une comparaison

48
quantitative et transférable au sein et entre rivière (Pringle, 1988). Bien que les écologistes du
paysage aient, une longue expérience de la quantification des formes, des tailles, des
configurations, de la diversité, de la densité et de la connectivité des habitats, il manque un
ensemble de mesures d'évaluation pertinentes écologiquement pouvant être utilisées par les
gestionnaires. L'apprentissage collaboratif (Weber et al. 2017) entre la recherche et la pratique
sont nécessaires pour développer des approches de gestion qui tiendront compte de la
dynamique spatio-temporelle des habitats.

49
Figure I. 9. Résultats d'un modèle 2D et classification en cinq types d'habitats simples
(classes de vitesse du courant) pour un tronçon de la rivière Aare près d'Innertkirchen
en Suisse. (A) Distribution des conditions de vitesse au débit de base (5 m3 s-1), avec deux
surfaces d’habitat homogène de référence marquées par des contours blancs (« patch »). (B)
Distribution des habitats au débit de pointe (45 m3 s-1), le déplacement des habitats de
références précédent est illustré par les contours en pointillés blanc, et leurs nouvelles
positions par des contours pointillés blanc et noir. (C) Etapes intermédiaires entre (A) et (B).
Notez qu'avec l'augmentation du débit, le type d'habitat 1 se contracte ; tandis que le type
d'habitat 2 s’est déplacé vers les berges. Tiré de Batz et al. (Soumis).

50
Utiliser les modèles pour lier les connaissances sur la sélection du microhabitat
et les effets sur la population

Les modèles d’habitat

Les modèles d’habitat hydrauliques (Lamouroux et al. 2017) sont certainement les plus
utilisés pour lier les connaissances sur la sélection du microhabitat des organismes et la
perturbation attendue à l’échelle du tronçon. Les modèles d'habitats couplent un modèle
hydraulique du tronçon avec les préférences d’habitat connue d’un groupe taxonomique pour
fournir une valeur d'habitat potentielle (VH) à chaque microhabitat (compris entre 0 et 1 (bonne
qualité)). En sommant le produit des VH par la surface des microhabitats on obtient la « surface
pondérée utile » (SPU) du tronçon, interprétée comme une surface d’habitat favorable. Le
calcul de la SPU pour plusieurs conditions de débit permet d’obtenir une courbe de SPU. Cette
courbe permet d'estimer la réponse de la population selon le débit. Il s’agit d’une méthode très
utilisée pour aider à la gestion des débits réservés (débit minimal obligatoire) (Lamouroux et
al. 2015), mais jusque-là peu pertinent pour aider à la gestion des débits éclusées. En effet, les
modèles de sélection du microhabitat ne considéraient pas jusqu’alors la dynamique temporelle
des habitats. Les travaux de Capra et al (2017) et Judes et al (B) sont un premier pas vers le
développement de modèles de sélection du microhabitat dynamiques. De tels modèles
permettront de mieux estimer l’habitat disponible dans les rivières à éclusées et ainsi d’établir
des préconisations de gestion plus pertinentes.

Les modèles bioénergétiques

Les modèles bioénergétiques déterminent les préférences d’habitat en reliant les

caractéristiques de l'habitat à des corrélats de la fitness directe, tels que l’apport net énergétique
sur la base de connaissances fondamentales de la physiologie ou du comportement du taxon
cible (Hayes et al. 2007, Kearney & Porter, 2009). Cette approche utilise comme paramètres
d'entrée, les sorties d'un modèle hydraulique (et de température si disponible), les observations
sur le terrain des caractéristiques de la populations (p.ex. distribution des tailles et poids), la
densité de macroinvertébrés (généralement en dérive) et une connaissance fondamentale de la
physiologie ou du comportement du taxon cible (Kearney et al. 2009). De la même façon, que
les modèles d’habitat, les modèles bioénergétiques peuvent fournir une valeur d’habitat à

51
chaque type d’habitat sur une échelle de 0 à 1 reflétant l’apport énergétique net et ainsi obtenir
une « surface pondérée utile ».

Ce type de modèle pourrait être adapté pour être applicable aux rivières à éclusées. Pour
cela, il faudra lier l’apport énergétique net à l’hydraulique passée. Par exemple, l'assèchement
passé réduit l’apport énergétique des organismes se nourrissant de macroinvertébrés
benthiques. De plus, il faudrait aussi inclure l’influence des éclusées sur la dérive des
macroinvertébrés puisque la dérive augmente avec les hausses de débit éclusées (Schülting et
al. 2019).

Il s’agit d’une approche à développer car les modèles biogénétiques établissent un lien
plus direct entre la condition physique des organismes et l'habitat biotique (compétition inter-
et intra-spécifique et le risque de prédation) et abiotique. Ils sont réputés pour être une approche
plus robuste et transférable pour prévoir les réponses biologiques aux changements d'habitat
que les modèles d’habitat (Guensch et al. 2001, Cuddington et al. 2013, Naman et al. 2019).

Les modèles individus centrés

La modélisation individu centrée simule l’ensemble des comportements individuels et

des relations entre individus pour prédire les réponses à des niveaux biologiques plus grands
(Grimm et al. 2016). Elle a donc l’avantage de fournir en sortie l’abondance par classe d’âge ce
qui est plus concret pour les gestionnaires que le concept de « surface pondérée utile »
(Railsback 2016). La modélisation individu centrée permet aussi d’intégrer une partie de la
complexité des systèmes écologiques réels et de comparer les résultats avec des modèles
empiriques afin de tester des théories sur l’origine et les motivations des comportements
observés (« Pattern-oriented modeling» voir Railsback & Harvey, 2002 et Goto et al. 2015).

Les modèles individus centrés peuvent être utilisés pour modéliser les processus de
sélection de l'habitat des organismes en se basant en partie sur la façon dont la probabilité de
survie (prédation, échouage piégeage, dérive forcé, famine) et de croissance (disponibilité en
nourriture, capacité à attraper la nourriture, compétition, coût énergétique) varie en fonction des
habitats disponibles. L’évolution de la population est prédite en intégrant la réponse de
l’ensemble des individus de la population (en termes de croissance, de reproduction et de
mortalité).

52
 Ce type d’approche pourrait être utile pour la recherche scientifique pour aider à
comprendre les origines et motivations des comportements de sélection du microhabitat
observés. Ils permettront, entre autres, de tester si les organismes sont adaptés ou acclimatés
aux variations spatio-temporelles des habitats dans les rivières à éclusées. Pour cela il faudra
réaliser deux modèles, un modèle qui considère que les organismes se basent en partie sur
l’hydraulique passée pour choisir le microhabitat qu’ils occupent et à l’inverse un modèle qui
considère qu’ils se basent uniquement sur l’hydraulique présente. Enfin, il s’agira de tester
lequel des deux modèles reproduit au mieux ce qui est observé sur le terrain (p.ex. les modèles
empiriques Judes et al. B). Concernant la gestion, ce type d’approche pourrait être utilisé pour
estimer l’abondance des taxa dans les rivières à éclusées en fonction de différents scénarios de
dynamique spatio-temporelle d’habitat (p. ex. fréquence et temps d'assèchement ou de forte
vitesse de courant, déplacement des conditions hydrauliques). Cela pourrait permettre
d'identifier des seuils à partir desquels il y a une diminution nette de la taille des populations et
ainsi être un outil précieux pour la gestion.

Cependant, malgré ces nombreux avantages puisqu'il y a énormément de paramètres

d'entrée, ces modèles sont souvent extrêmement difficiles à paramétrer et sont compliqués à
mettre en œuvre pour la gestion. Souvent les gestionnaires disposent de peu de temps et de
ressources et la complexité des calculs, l'étendue et le détail des informations requises pour faire
de tels modèles peuvent être un frein à leurs développements.

3.2 Cohérence des résultats de Judes et al. (B) avec les mesures de
gestion des éclusées prises sur la basse rivière d’Ain

Les résultats de Judes et al. (B) peuvent être discutés au regard des mesures mises en
œuvre par EDF pour réduire l'effet des éclusées sur la basse rivière d’Ain ces dernières années.
Ces mesures visent en particulier l’atténuation du risque d'exondation des frayères et du risque
d'échouage-piégeage des juvéniles de truites et d’ombre. Concernant l’exondation des frayères,
le débit réservé a été élevé à 28 m3 s-1 une partie de l’année (durant la période de reproduction
et de croissance larvaire : 01/12 au 01/05) au lieu de 12,3 m3 s-1 le reste de l'année à l’aval
immédiat du barrage de l’Allement. La figure I.10 montre la relation entre le débit et la surface
immergée : la surface immergée augmente nettement avec le débit de base jusqu’à environ 50
m3 s-1. Puisque Judes et al. (B) ont montré que les macroinvertébrés se trouvent majoritairement

53
dans les habitats qui restent toujours en eau cette mesure semble pertinente pour augmenter
l’abondance des macroinvertébrés dans la basse rivière d’Ain. Concernant les gradients de
baisse de débit, pour une baisse de débit compris entre 12 <Q< 80 m3 s-1 le gradient de baisse
est limité à 5,8 – 15 m3 s-1 h-1 en 2016 et pour une baisse de débit comprise entre 12 <Q< 120
m3 s-1 le gradient doit être entre 4 – 15 m3 s-1 h-1 en 2018-2019. Ces limitations de gradient sont
appliquées de mars à fin mai (dates variables). Une proposition a été faite pour 2020 d'appliquer
un gradient très atténué (1 m3 s-1 h-1) lors des baisses les plus à risque (après 01/05). Judes et al.
(B) ont montré que les poissons de la guilde « berge » (majoritairement les juvéniles) utilisent
de manière préférentielle les microhabitats qui s’assèchent. Cette proposition semble donc
intéressante et mérite d'être suivie si l’on souhaite minimiser l'échouage-piégeage des juvéniles
de poissons.

Figure I. 10. Evolution des pertes relatives de surface mouillée par m3 s-1 sur l’ensemble
du linéaire étudié à Pont d’Ain sur l’Ain. Tiré de Malavoi (2019).

3.3. Résumé des perspectives opérationnelles de la thèse

Le Tableau I.1 résume les perspectives opérationnelles que l’on peut tirer des résultats
de la thèse. Cependant, il faut prendre des précautions quant à leurs mises en œuvre, des
analyses supplémentaires devront être réalisées pour confirmer leurs pertinences. Les résultats
de Judes et al. (A, B) proviennent de données récoltées in situ, les conditions d’échantillonnage
ne sont pas nécessairement représentatives des modes de gestion des éclusées rencontrées sur
les tronçons d’études ou de façon plus large à ce qui est rencontré en France ou en Europe. Par
exemple, comme on a pu le voir précédemment les échantillonnages de Judes et al (B) sur la

54
rivière d’Ain se sont fait au cours ou juste après des lâchers d’eau pour le soutien d’étiage des
centrales nucléaires du Rhône. Ainsi il y a eu des débits élevés (~ 100 m3 s-1) pendant une
période inhabituelle, ce qui a pu favoriser l’utilisation des microhabitats souvent asséchés par
les organismes (poissons de la guilde « berge »). De même, les modes de gestion des éclusées
sur les tronçons étudiés par Judes et al. (A, B) ne sont pas forcément représentatifs des
conditions rencontrées de manière générale en France et encore moins en Europe ou dans le
monde. Par exemple, dans l’étude de Judes et al. (A) le gradient de baisse de hauteur d’eau
moyen était de 3,2 cm h-1 (max = 10,6 cm h-1) alors que Schmutz et al. (2015) en Autriche avait
montré un effet négatif des éclusées sur les communautés de poissons à partir d’un gradient de
baisse de hauteur d’eau de 15 cm h-1 et Hayes et al. (2021) à partir de 12-24 cm h-1 sur les
populations d’ombre (Autriche). Il est donc important de considérer ces mesures au regard du
fonctionnement des écosystèmes et de la gestion des éclusées en place.

55
Tableau I.1. Résumé des perspectives opérationnelles qui découlent des résultats de la
thèse.

Articles Résultats Perspectives Études

opérationnelles supplémentaires
nécessaires

Judes et al. L’influence des éclusées Être particulièrement Réaliser plus d'études
(A) semble moins importante vigilant sur l’impact sur les cyprinidés et
pour les espèces de poissons des éclusées dans les sur les rivières de
typiques des petites rivières rivières de taille taille moyenne
par rapport aux espèces de moyenne.
poissons typiques des
rivières de taille moyenne.

Judes et al L'abondance des Augmenter le débit Echantillonnage à des

(B) macroinvertébrés et des de base pour limiter débits intermédiaires
poissons de la guilde les surfaces d’habitat (entre 14 et 100 m3 s-
1
"chenal" est plus faible dans soumis à des ) pour les
les microhabitats qui assèchements macroinvertébrés
s'assèchent. pour tester d’autres
L’assèchement a des effets patterns
plus importants sur la d’assèchement.
sélection de l’habitat des Résultat qui demande
poissons et des à être confirmé pour
macroinvertébrés que les les poissons de la
fortes vitesses de courant. guilde "chenal" avec
plus de données et en
considérant la
disponibilité en
habitat.

Judes et al Inclure les conditions Amélioration de Validation biologique

(B) hydrauliques extrêmes l'estimation de par la réplication de
passées améliore les l'habitat disponible si l’expérimentation
modèles de sélection du l’on considère Judes et al. (B).
microhabitat des l’hydraulique passée
macroinvertébrés et des dans les futures
poissons. modèles de sélection
du microhabitat.

56
Batz et al. Il est nécessaire d'utiliser Mieux évaluer la La pertinence des
(Soumis) des métriques qui rendent perturbation vécue métriques doit être
compte de la perturbation par les organismes vérifiée par des
vécue par les organismes aquatiques. observations in situ
selon leurs mobilités. Par Métriques (télémétrie) ou ex situ
exemple, il faut considérer quantitatives et (caméra).
la dynamique spatiale des transférables au sein
habitats pour rendre compte et entre rivières.
de la perturbation vécu par
les poissons.

57
BIBLIOGRAPHIE

Alonso C, García de Jalón D, Álvarez J, Gortázar J, 2011. A large-scale approach can help
detect general processes driving the dynamics of brown trout populations in extensive areas.
Ecology of Freshwater Fish 20:449–460.

Archer D, Newson M, 2002. The use of indices of flow variability in assessing the hydrological
and instream habitat impacts of upland afforestation and drainage. Journal of Hydrology
268:244–258.

Arthur SM, Manly BF, McDonald LL, Garner GW, 1996. Assessing habitat selection when
availability changes. Ecology 77:215–227.

Bain MB, Finn JT, Booke HE, 1988. Streamflow regulation and fish community structure.
Ecology 69:382–392.

Bain MB, 2007. Hydropower Operations and Environmental Conservation: St. Mary’s River,
Ontario and Michigan, Canada and USA.

Barillier A, Beche L, Malavoi J, Gouraud V, (in press) Identification of effective hydropeaking

mitigation measures: Are hydraulic habitat models sufficient in a global approach? Journal of
Ecohydraulics.

Bartoň D, Bretón F, Blabolil P, Souza AT, Vejřík L, Sajdlová Z, Kolařík T, Kubečka J, Šmejkal
M, 2021. Effects of hydropeaking on the attached eggs of a rheophilic cyprinid species.
Ecohydrology.

Batz N, Judes C, Weber C, (Soumis) Habitat dynamics in hydropeaking rivers: the impact of
flow frequency and duration on patch-scale habitat dynamics.

Beck RW, Associates, 1989. Skagit River salmon and steelhead fry stranding studies. Report
for Seattle City Light Environmental Affairs Division. Seattle.

Bejarano MD, Jansson R, Nilsson C, 2018. The effects of hydropeaking on riverine plants: a
review. Biological Reviews 93:658–673.

Belaud A, Chaveroche P, Lim P, Sabaton C, 1989. Probability‐of‐use curves applied to brown

58
trout (Salmo trutta fario L.) in rivers of southern France. Regulated Rivers: Research &
Management 3:321–336.

Berland G, Nickelsen T, Heggenes J, Økland F, Thorstad EB, Halleraker J, 2004. Movements

of wild Atlantic salmon parr in relation to peaking flows below a hydropower station. River
Research and Applications 20:957–966.

Biggs BJF, Nikora VI, Snelder TH, 2005. Linking scales of flow variability to lotic ecosystem
structure and function. River Research and Applications 21:283–298.

Blinn DW, Shannon JP, Stevens LE, Carder JP, 1995. Consequences of fluctuating discharge
for lotic communities. Journal of the North American Benthological Society 14:233–248.

Boisclair D, 1992. An evaluation of the stereocinematographic method to estimate fish

swimming speed. Canadian Journal of Fisheries and Aquatic Science 49:523–531.

Brabec K, Janecek BFU, Rossaro B, Spies M, Bitusik P, Syrovatka V, Schmidt-Kloiber A,

2021. Dataset ‘Chironomidae’. www.freshwaterecology.info - the taxa and autecology database
for freshwater organisms, version 7.0.

Bradford MJ, Taylor GC, Allan JA, 1997. Empirical Review of Coho Salmon Smolt Abundance
and the Prediction of Smolt Production at the Regional Level. Transactions of the American
Fisheries Society 126:49–64.

Bret V, Bergerot B, Capra H, Gouraud V, Lamouroux N, 2015. Influence of discharge,

hydraulics, water temperature, and dispersal on density synchrony in brown trout populations
(Salmo trutta). Canadian Journal of Fisheries and Aquatic Sciences 73:319–329.

Bruder A, Tonolla D, Schweizer SP, Vollenweider S, Langhans SD, 2016. A conceptual

framework for hydropeaking mitigation. Science of the Total Environment 568:1204–1212.

Buffagni A, Armanini DG, Cazzola M, Alba-Tercedor J, López-Rodríguez MJ, Murphy J,

Sandin L, Schmidt-Kloiber A, 2021. Dataset ‘Ephemeroptera’. www.freshwaterecology.info -
the taxa and autecology database for freshwater organisms, version 7.0.

Bunt C, Cooke SJ, Katopodis C, Mackinley RS, 1999. Movement And Summer Habitat Of
Brown Trout (Salmo Trutta) Below A Pulsed Discharge Hydroelectric Generating Station.
Regulated Rivers: Research & Management 403:395–403.

59
Camargo JA, Garcia de Jalon D, 1990. The downstream impacts of the burgomillodo reservoir,
Spain. Regulated Rivers: Research & Management 5:305–317.

Capra H, Plichard L, Bergé J, Pella H, Ovidio M, Mcneil E, Lamouroux N, 2017. Fish habitat
selection in a large hydropeaking river: strong individual and temporal variations revealed by
telemetry. Science of the Total Environment 578:109–120.

Céréghino R, Boutet T, Lavandier P, 1997. Abundance, biomass, life history and growth of six
Trichoptera species under natural and hydro-peaking conditions with hypolimnetic releases in
a Pyrenean stream. Archiv fur Hydrobiologie 138:307–328.

Choi SU, Kim SK, Choi B, Kim Y, 2017. Impact of hydropeaking on downstream fish habitat
at the Goesan Dam in Korea. Ecohydrology 10:1–12.

Coarer Y Le, 2007. Hydraulic signatures for ecological modelling at different scales. Aquatic
Ecology 41:451–459.

Cohen D, 1966. Optimizing reproduction in a randomly varying environment. Journal of

Theoretical Biology 12:119–129.

Conallin J, Boegh E, Olsen M, Pedersen S, Dunbar MJ, Jensen JK, 2014. Daytime habitat
selection for juvenile parr brown trout (Salmo trutta) in small lowland streams. Knowledge and
Management of Aquatic Ecosystems 413.

Connor EJ, Pflug DE, 2004. Changes in the distribution and density of pink, chum, and Chinook
salmon spawning in the Upper Skagit River in response to flow management measures. North
American Journal of Fisheries Management 24:835–852.

Conquest LL, 2000. Analysis and interpretation of ecological field data using BACI designs:
Discussion. Journal of Agricultural, Biological, and Environmental Statistics 5:293–296.

Crips D, Mann HK, Cubby P, 1983. Effects of Regulation of the River Tees Upon Fish
Populations Below Cow Green Reservoir. Journal of Applied Ecology 20:371–386.

Cuddington K, Fortin MJ, Gerber LR, Hastings A, Liebhold A, O’connor M, Ray C, 2013.
Process-based models are required to manage ecological systems in a changing world.
Ecosphere 4:1–12.

60
Cuthill IC, Houston AI, 1997. Managing time and energy. Behavioural ecology 4th ed Oxford:
Bl., pp. 97–120.

Dole-Olivier MJ, Marmonier P, Beffy JL, 1997. Response of invertebrates to lotic disturbance:
Is the hyporheic zone a patchy refugium? Freshwater Biology 37:257–276.

Doledec S, Chessel D, 1994. Co-inertia analysis: an alternative method for studying species-
environment relationships. Freshwater Biology 31:277–294.

Enders EC, Boisclair D, Roy AG, 2003. The effect of turbulence on the cost of swimming for
juvenile Atlantic salmon (Salmo salar). Canadian Journal of Fisheries and Aquatic Sciences
60:1149–1160.

Enders EC, Buffin-Bélanger T, Boisclair D, Roy AG, 2005. The feeding behaviour of juvenile
Atlantic salmon in relation to turbulent flow. Journal of Fish Biology 66:242–253.

Enders EC, Watkinson DA, Ghamry H, Mills KH, Franzin WG, 2017. Fish age and size
distributions and species composition in a large hydropeaking Prairie River. River Research
and Applications 33:1–11.

Finch C, Pine WE, Limburg KE, 2015. Do hydropeaking flows alter juvenile fish growth rates?
A test with juvenile humpback chub in the Colorado River. River Research and Applications
31:156–164.

Fisher SG, Lavoy A, 1972. Differences in Littoral Fauna Due to Fluctuating Water Levels
Below a Hydroelectric Dam. Journal of the Fisheries Research Board of Canada 29:1472–1476.

Flodmark LEW, Forseth T, L’Abée-Lund JH, Vøllestad LA, 2006. Behaviour and growth of
juvenile brown trout exposed to fluctuating flow. Ecology of Freshwater Fish 15:57–65.

Freeman MC, Bowen ZH, Bovee KD, Irwin ER, 2001. Flow and habitat effects on juvenile fish
abundance in natural and altered flow regimes. Ecological society of America 11:179–190.

Garcia De Jalon D, Sanchez P, Camargo JA, 1994. Downstream effects of a new hydropower
impoundment on macrophyte, macroinvertebrate and fish communities. Regulated Rivers:
Research & Management 9:253–261.

Geist DR, Brown RS, Cullinan V, Brink SR, Lepla K, Bates P, Chandler JA, 2005. Movement,

61
Swimming Speed, and Oxygen Consumption of Juvenile White Sturgeon in Response to
Changing Flow, Water Temperature, and Light Level in the Snake River, Idaho. Transactions
of the American Fisheries Society 134:803–816.

Goto D, Hamel MJ, Hammen JJ, Rugg ML, Pegg MA, Forbes VE, 2015. Spatiotemporal
variation in flow-dependent recruitment of long-lived riverine fish: Model development and
evaluation. Ecological Modelling 296:79–92.

Graf W, Lorenz AW, Tierno de Figueroa JM, Lücke S, López-Rodríguez MJ, Murphy J,
Schmidt-Kloiber A, 2021.a Dataset ‘Plecoptera’. www.freshwaterecology.info - the taxa and
autecology database for freshwater organisms, version 7.0.

Graf W, Dahl J, Zamora-Muñoz C, López-Rodríguez MJ, Schmidt-Kloiber A, 2021.b Dataset

‘Trichoptera’. www.freshwaterecology.info - the taxa and autecology database for freshwater
organisms, version 7.0.

Greimel F, Zeiringer B, Hauer C, Holzapfel P, Fuhrmann M, Sc, 2016. Longitudinal

Assessment of Hydropeaking Impacts and Evaluation of Mitigation Measures. 11th ISE 2016,
Melbourne, Australia- 4.

Grimm V, Berger U, 2016. Structural realism, emergence, and predictions in next-generation

ecological modelling: Synthesis from a special issue. Ecological Modelling 326:177–187.

Guensch GR, Hardy TB, Addley RC, 2001. Examining feeding strategies and position choice
of drift-feeding salmonids using an individual-based, mechanistic foraging model. Canadian
Journal of Fisheries and Aquatic Sciences 58:446–457.

Harby A, Alfredsen KT, Fjeldstad HP, Halleraker JH, 2001. Ecological impacts of hydro
peaking in rivers. Hydropower in the New Millennium: Proceedings of the 4th International
Conference on Hydropower Development.

Harvey-Lavoie S, Cooke SJ, Guénard G, Boisclair D, 2016. Differences in movements of

northern pike inhabiting rivers with contrasting flow regimes. Ecohydrology 9:1687–1699.

Hayes DS, Lautsch E, Unfer G, Greimel F, Zeiringer B, Höller N, Schmutz S, 2021. Response
of European grayling, Thymallus thymallus, to multiple stressors in hydropeaking rivers.
Journal of Environmental Management 292.

62
Hayes J, Hughes N, Kelly L, 2007. Process-based modelling of invertebrate drift transport, net
energy intake and reach carrying capacity for drift-feeding salmonids. Ecological Modelling
207:171–188.

Hedger RD, Sauterleute J, Sundt-Hansen LE, Forseth T, Ugedal O, Diserud OH, Bakken TH,
2018. Modelling the effect of hydropeaking-induced stranding mortality on Atlantic salmon
population abundance. Ecohydrology 11:1–14.

Hitchman SM, 2017. Identifying keystone habitats with a mosaic approach can improve
biodiversity conservation in disturbed ecosystems.

Hogg ID, Dudley Williams D, 1996. Response of stream invertebrates to a global-warming

thermal regime: An ecosystem-level manipulation. Ecology 77:395–407.

Holomuzki J, Biggs BJF, 2000. Taxon-specific responses to high-flow disturbance in streams.

Journal of the North American Benthological Society 19:670–679.

Huet M, 1949. Aperçu des relations entre la pente et les populations piscicoles des eaux
courantes [Overview of the relationship between the slope and fish populations in flowing
waters]. Schweizerische Zeitschrift für Hydrologie 11:332–351.

Illies J, Botosaneanu L, 1963. Problèmes et méthodes de la classification et de la zonation

écologique des eaux courantes, considerées surtout du point de vue faunistique. SIL
Communications, 1953-1996 12:1–57.

Jensen J, Koksvik J, Karlsen L, 1992. Rapport fra forsøk med korttidsregulering av Altaelva.
Stensil.

Judes C, Gouraud V, Capra H, Maire A, Barillier A, Lamouroux N, (in press) Consistent but
secondary influence of hydropeaking on stream fish assemblages in space and time. Journal of
Ecohydraulics.

Judes C, Capra H, Gouraud V, Pella H, Lamouroux N, (Soumis) Past hydraulics influence

microhabitat selection by macroinvertebrates and fish in hydropeaking rivers.

Junk WJ, Bayley PB, Sparks RE, 1989. The flood pulse concept. International Large River
Symposium 110–127.

63
Kearney M, Porter W, 2009. Mechanistic niche modelling: Combining physiological and spatial
data to predict species’ ranges. Ecology Letters 12:334–350.

Kemp PS, Gilvear DJ, Armstrong JD, 2003. Do juvenile Atlantic salmon parr track local
changes in water velocity? River Research and Applications 19:569–575.

Kjaerstad G, Arnekleiv J V., Speed JDM, Herland AK, 2018. Effects of hydropeaking on
benthic invertebrate community composition in two central Norwegian rivers. River Research
and Applications 34:218–231.

Kokko H, Sutherland WJ, 2001. Ecological traps in changing environments: Ecological and
evolutionary consequences of a behaviourally mediated Allee effect. Evolutionary Ecology
Research 3:537–551.

Korman J, Campana SE, 2009. Effects of Hydropeaking on Nearshore Habitat Use and Growth
of Age-0 Rainbow Trout in a Large Regulated River. Transactions of the American Fisheries
Society 138:78–87.

Kougias I, Aggidis G, Avellan F, Deniz S, Lundin U, Moro A, Muntean S, Novara D, Pérez-

Díaz JI, Quaranta E, Schild P, Theodossiou N, 2019. Analysis of emerging technologies in the
hydropower sector. Renewable and Sustainable Energy Reviews 113:109257.

Krausman PR, 1999. Some Basic Principles of Habitat Use. Grazing Behavior of Livestock and
Wildlife 85–90.

Krimmer AN, Paul AJ, Hontela A, Rasmussen JB, 2011. Behavioural and physiological
responses of brook trout Salvelinus fontinalis to midwinter flow reduction in a small ice-free
mountain stream. Journal of Fish Biology 79:707–725.

Lamouroux N, Souchon Y, 2002. Simple predictions of instream habitat model outputs for fish
habitat guilds in large streams. Freshwater Biology 47:1531–1542.

Lamouroux N, Augeard B, Baran P, Capra H, Coarer Y Le, Girard V, Gouraud V, Navarro L,

Prost O, Sagnes P, Tissot L, 2015. Débits écologiques : la place des modèles d’habitat dans une
démarche intégrée. Hydroécologie appliquée 20:1–27.

Lamouroux N, Hauer C, Stewardson MJ, LeRoy Poff N, 2017. Physical Habitat Modeling and
Ecohydrological Tools. Water for the Environment: From Policy and Science to

64
Implementation and Management. Elsevier Inc.

Lancaster J, Belyea LR, 1997. Nested Hierarchies and Scale-Dependence of Mechanisms of

Flow Refugium Use. Journal of the North American Benthological Society 16:221–238.

Lobón-Cerviá J, 2009. Why, when and how do fish populations decline, collapse and recover?
the example of brown trout (Salmo trutta) in Rio Chaballos (northwestern Spain). Freshwater
Biology 54:1149–1162.

Lytle DA, Poff NLR, 2004. Adaptation to natural flow regimes. Trends in Ecology and
Evolution 19:94–100.

Mackay NJ, 1973. Histological changes in the ovaries of the golden perch, Plectroplites
ambiguus, associated with the reproductive cycle. Marine and Freshwater Research 24:95–102.

Macnaughton CJ, McLaughlin F, Bourque G, Senay C, Lanthier G, Harvey-Lavoie S, Legendre

P, Lapointe M, Boisclair D, 2015. The Effects of Regional Hydrologic Alteration on Fish
Community Structure in Regulated Rivers. River Research and Applications 33:249–257.

Malavoi JR, 2019. Les éclusées de l’Ain. Volet hydromorphologie.

McManamay RA, Oigbokie CO, Kao SC, Bevelhimer MS, 2016. Classification of US
Hydropower Dams by their Modes of Operation. River Research and Applications 32:1450–
1468.

McNamara JM, Houston AI, 1992. Risk-sensitive foraging: A review of the theory. Bulletin of
Mathematical Biology 54:355–378.

Mérigoux S, Ponton D, 1999. Spatio-temporal distribution of young fish in tributaries of natural

and flow-regulated sections of a neotropical river in French Guiana. Freshwater Biology
42:177–198.

Mihalicz JE, Jardine TD, Baulch HM, Phillips ID, 2019. Seasonal effects of a hydropeaking
dam on a downstream benthic macroinvertebrate community. River Research and Applications
35:714–724.

Moog O, 1993. Quantification of daily peak hydropower effects on aquatic fauna and
management to minimize environmental impacts. Regulated Rivers: Research & Management

65
8:5–14.

Moreira M, Hayes DS, Boavida I, Schletterer M, Schmutz S, Pinheiro A, 2019. Ecologically-

based criteria for hydropeaking mitigation: A review. Science of the Total Environment
657:1508–1522.

Morel M, Booker DJ, Gob F, Lamouroux N, 2020. Intercontinental predictions of river

hydraulic geometry from catchment physical characteristics. Journal of Hydrology 582:124292.

Muñoz-Mas R, Martínez-Capel F, Schneider M, Mouton AM, 2012. Assessment of brown trout

habitat suitability in the Jucar River Basin (SPAIN): Comparison of data-driven approaches
with fuzzy-logic models and univariate suitability curves. Science of the Total Environment
440:123–131.

Naman SM, Rosenfeld JS, Neuswanger JR, Enders EC, Eaton BC, 2019. Comparing correlative
and bioenergetics-based habitat suitability models for drift-feeding fishes. Freshwater Biology
64:1613–1626.

Nathan R, Getz WM, Revilla E, Holyoak M, Kadmon R, Saltz D, Smouse PE, 2008. A
movement ecology paradigm for unifying organismal movement research. Proceedings of the
National Academy of Sciences of the United States of America 105:19052–19059.

Nicola GG, Almodóvar A, Elvira B, 2009. Influence of hydrologic attributes on brown trout
recruitment in low-latitude range margins. Oecologia 160:515–524.

Parasiewicz P, Schmutz S, Moog O, 1998. The effect of managed hydropower peaking on the
physical habitat, benthos and fish fauna in the River Bregenzerach in Austria. Fisheries
Management and Ecology 5:403–417.

Perry SA, Perry WB, 1986. Effects of experimental flow regulation on invertebrate drift and
stranding in the. Hydrobiologia 134:171–182.

Pert EJ, Erman D, 1994. Habitat use by adult rainbow trout under moderate artificial
fluctuations in flow. Transactions of the American Fisheries Society 123:913–923.

Pickett STA, Rogers KH, 1997. Patch Dynamics: The Transformation of Landscape Structure
and Function. In: Bissonette, J. A. (ed.), Wildlife and Landscape Ecology: Effects of Pattern
and Scale. Springer New York, New York, NY, pp. 101–127.

66
Plichard L, 2018. Modélisation multi-échelle de la sélection de l’habitat hydraulique des
poissons de rivière.

Pringle CM, Naiman RJ, Bretschko G, Karr JR, Oswood J, Webster J, Welcomme RL,
Winterbourn M, 1988. Patch dynamics in Lotic systems: the stream as Mosaic. Journal of the
North American Benthological Society 7: 503–524

Puffer M, Berg OK, Huusko A, Vehanen T, Forseth T, Einum S, 2015. Seasonal effects of
hydropeaking on growth, energetics and movement of juvenile atlantic salmon (Salmo salar).
River Research and Applications 31:1101–1108.

Puffer M, Berg OK, Einum S, Saltveit SJ, Forseth T, 2017. Energetic Consequences of
Stranding of Juvenile Atlantic Salmon (Salmo salar L.). Journal of Water Resource and
Protection 9:163–182.

Railsback SF, 2016. Why it is time to put PHABSIM out to Pasture. Fisheries 41:720–725.

Robertson MJ, Pennell CJ, Scruton DA, Robertson GJ, Brown JA, 2004. Effect of increased
flow on the behaviour of Atlantic salmon parr in winter. Journal of Fish Biology 65:1070–1079.

RTE, 2019. Bilan énergetique.

Rykiel EJ, 1985. Towards a definition of ecological disturbance. Australian Journal of Ecology
10:361–365.

Sabo MJ, Orth DJ, Pert EJ, 1996. Effect of stream microhabitat characteristics on rate of net
energy gain by juvenile smallmouth bass, Micropterus dolomieu. Environmental Biology of
Fishes 46:393–403.

Sauterleute JF, Hedger RD, Hauer C, Pulg U, Skoglund H, Sundt-Hansen LE, Bakken TH,
Ugedal O, 2016. Modelling the effects of stranding on the Atlantic salmon population in the
Dale River, Norway. Science of the Total Environment 573:574–584.

Schlaepfer MA, Runge MC, Sherman PW, 2002. Ecological and evolutionary traps. Trends in
Ecology and Evolution 17:474–480.

Schmutz S, Bakken TH, Friedrich T, Greimel F, Harby A, Jungwirth M, Melcher A, Unfer G,

Zeiringer B, 2015. Response of fish communities to hydrological and morphological alterations

67
in hydropeaking rivers of Austria. River research and applications 31:919–930.

Schulting L, Feld CK, Graf W, 2016. Effects of hydro- and thermopeaking on benthic
macroinvertebrate drift. Science of the Total Environment 573:1472–1480.

Schülting L, Feld CK, Zeiringer B, Huđek H, Graf W, 2019. Macroinvertebrate drift response
to hydropeaking: An experimental approach to assess the effect of varying ramping velocities.
Ecohydrology 12.

Shamur E, Zilka M, Hassner T, China V, Liberzon A, Holzman R, 2016. Automated detection

of feeding strikes by larval fish using continuous high-speed digital video: A novel method to
extract quantitative data from fast, sparse kinematic events. Journal of Experimental Biology
219:1608–1617.

Shen Y, Arablouei R, De Hoog F, Hao X, Malan J, Sharp J, Shouri S, D. Clark T, Lefevre C,

Kroon F, Severati A, Kusy B, 2021. In-situ Fish Heart Rate Estimation and Feeding Event
Detection Using an Implantable Biologger. IEEE Transactions on Mobile Computing.

Shirvell CS, 1994. Effect of changes in streamflow on the microhabitat use and movements of
sympatric juvenile coho salmon (Oncorhynchus kisutch) and chinook salmon (O.tshawytscha)
in a natural stream. Canadian Journal of Fisheries and Aquatic Sciences 51:1644–1652.

Smokorowski KE, Metcalfe RA, Finucan SD, Jones N, Marty J, Power M, Pyrce RS, Steele R,
2011. Ecosystem level assessment of environmentally based flow restrictions for maintaining
ecosystem integrity: a comparison of a modified peaking versus unaltered river. Ecohydrology
4:791–806.

Statzner B, Higler B, 1986. Stream hydraulics as a major determinant of benthic invertebrate

zonation patterns. Freshwater Biology 16:127–139.

Stubbington R, Sarremejane R, Datry T, 2019. Alpha and beta diversity of connected benthic–
subsurface invertebrate communities respond to drying in dynamic river ecosystems.
Ecography 42:2060–2073.

Taylor MK, Hasler CT, Hinch SG, Lewis B, Schmidt DC, Cooke SJ, 2014. Reach-scale
movements of bull trout (Salvelinus confluentus) relative to hydropeaking operations in the
Columbia River, Canada. Ecohydrology.

68
Travnichek VH, Maceina MJ, 1994. Comparison of flow regulation effects on fish assemblages
in shallow and deep water habitats in the Tallapoosa River, Alabama. Journal of Freshwater
Ecology 9:207–216.

Troelstrup NH, Hergenrader GL, 1990. Effect of hydropower peaking flow fluctuations on
community structure and feeding guilds of invertebrates colonizing artificial substrates in a
large impounded river. Hydrobiologia 199:217–228.

UFE, 2017. Hydroelectricity at the crossroad: let us give a new momentum to the first
renewable energy White paper on hydroelectricity (INIS-FR--17-0679). France. Paris (France).

Vannote R, Minshall W, Cummins K, Sedell J, Cushing C, 1980. The River Continuum

Concept. Canadian Journal of Fisheries and Aquatic Science 37:130–137.

Verneaux, 1973. Recherches écologiques sur le réseau hydrographique du Doubs. Franche

Comté.

Ward J V, Stanford J a, 1983. Serial Discontinuity Concept of Lotic Ecosystems. Dynamics of

Lotic Systems, Ann Arbor Science, Ann Arbor 29–42.

Weber C, Åberg U, Buijse AD, Hughes FMR, 2017. Goals and principles for programmatic
river restoration monitoring and evaluation : collaborative learning across multiple projects.
Wiley Interdisciplinary Reviews: Water 5.

Williams BK, Nichols JD, 1984. Optimal timing in biological processes. American Naturalist
123:1–19.

Winemiller KO, Flecker AS, Hoeinghaus DJ, 2010. Patch dynamics and environmental
heterogeneity in lotic ecosystems. Journal of the North American Benthological Society 29:84–
99.

69
Deuxième partie
Articles

70
1. JUDES ET AL. (A) :
Consistent but secondary influence of hydropeaking on stream
fish assemblages in space and time.
Judes C, Gouraud V, Capra H, Maire A, Barillier A, Lamouroux N.

71
1.1 Abstract

Hydropeaking corresponds to rapid artificial discharge variations, designed to address sub-daily peaks
in electricity demand. It generates rapid changes in physical habitat (e.g. flow velocity and water depth)
with potential impacts on stream assemblages. For assessing the generality of hydropeaking effects on
fish assemblages, we present an original combination of spatial (among 45 reaches, including six groups
of nearby reaches) and temporal (over 3-17 years) analyses of these effects. Our analyses involved
descriptions of natural and artificial hydraulic variations in reaches, obtained after translating hourly
discharge data into hydraulics. We found that the influence of hydropeaking was secondary compared
to well-known spatial variations in fish assemblage structure along longitudinal gradients, and negative
influences of floods on annual densities. However, the spatial and temporal analyses consistently
suggested that hydropeaking may disfavour fish species typical of medium-sized streams relative to
species of headwater streams (Salmo trutta, Phoxinus phoxinus, Cottus gobio). The magnitude of
hydropeaking effects observed here, as well as an apparent weaker effect of ramping rates than the
frequency of hydropeaks, may be due to lower ramping rates in our data set than in other studies.

Keywords: hydropower, flow management, hydraulic habitat, fish community structure, high flow
variations

72
1.2. Introduction

Hydropeaking corresponds to rapid artificial flow fluctuations downstream of

hydropower plants, designed to address sub-daily peaks in electricity demand. Hydropeaking
management concerns a large number of dams and river reaches (e.g, 144 of about 600 large
dams in France and around 800 km of rivers in Austria; Lauters 1995; Metcher et al. 2017), a
number expected to increase with the development of renewable energies (IPCC 2011). The
flow regime (e.g, flow magnitude, timing and rate of change, frequency and duration of extreme
flow events) is strongly modified downstream of hydropeaking power plants, with more rapid
and frequent flow variations than occurring naturally. Because many characteristics of the flow
regime influence the structure and functioning of river ecosystems (Schlosser 1991; Poff et al.
1997; Humphries et al. 1999; Magoulick and Kobza 2003), an improved understanding of the
ecological effects of hydropeaking is needed. This is particularly true for fish assemblages,
which can be affected by hydropeaking at several levels: increased drift of individuals during
rapid flow velocity increases (Lechner et al. 2016), stranding due to rapid shoreline dewatering
(Leclere et al. 2012; Hauer et al. 2014 Sauterleure et al. 2016), and repeated drying and scouring
of spawning grounds (McMichael et al. 2005; Malcolm et al. 2012, Casas-Mulet et al. 2014,
2016).

An attractive method for estimating the ecological impacts of hydropeaking on aquatic

communities is to perform “pressure-impact” spatial analyses among numerous river reaches
(Gehrke and Harris 2001; Schmutz et al. 2015). Their principle is to build correlative models
involving descriptors of hydropeaking intensity (e.g. frequency of events, ramping rates: rate
of change in water level) and descriptors of fish responses (e.g. species richness or indices of
biotic integrity). The work of Schmutz et al. (2015) is to our knowledge the most comprehensive
study of this type, analysing the impacts of hydropeaking descriptors in 74 reaches of 16 rivers
in Austria, with estimated down-ramping rates typically ranging between 5 and 40 cm h-1. Using
the national Austrian fish index, they found that the ecological status of natural-like streams
was significantly degraded (the index lost two points on a scale of five) for the largest ramping
rates observed. A difficulty with these “pressure-impact” approaches is to appreciate if the
variables finally retained in the models, often after removal of several intercorrelated variables,
are the actual drivers of observed responses. In other words, intercorrelation between
environmental variables along natural and anthropogenic gradients may complicate the
interpretation of such spatial analyses. Another difficulty with these large-scale approaches is

73
that they often lack proximate habitat variables, such as wetted-width ramping rates or rate of
changes in flow velocity (Moreira et al. 2019), because these are rarely available across many
reaches. Nevertheless, these approaches are extremely useful for identifying thresholds beyond
which environmental pressures can become problematic (e.g. hydropeaking ramping rate above
15 cm h-1; Schmutz et al. 2015).

Interpreting spatial analyses of the ecological effects of hydropeaking among reaches

can be facilitated when nearby reaches with contrasting hydropeaking pressures are included in
the analyses. Such groups of reaches ideally have similar general characteristics (e.g,
morphology, temperature, water quality) and differ only by their hydropeaking regime. These
approaches have revealed that hydropeaking can reduce fish biomass and diversity (Liebig et
al. 1999; Freeman et al. 2001; Smokorowski et al. 2010; Enders et al. 2017), with different
responses among species. For example, Bain et al. (1988) compared two nearby rivers and
showed that hydropeaking reduced the abundance of small fish species (and small size classes)
living in shallow and slow-flowing habitats (see also Travnichek and Maceina 1994).
Nevertheless, these comparisons generally involved one or a few reaches only, limiting the
transferability of their results.

Temporal studies (either in situ or experimental) of the ecological effects of

hydropeaking regime may also contribute to better identify thresholds in hydropeaking
characteristics triggering a response of aquatic organisms or assemblages, while limiting
problems associated with confounding environmental factors. For example, Saltveit et al.
(2001) monitored juvenile fish in a river section during rapid down-ramping events (90 cm h-1)
and found that 60% became stranded. Consistently, in artificial channels, Bradford et al. (1995)
reported that the proportions of stranded juveniles of rainbow-trout increased from 6% to 30%
with down-ramping rates increasing from 5 cm h-1 to 60 cm h-1. At the population level,
reducing the frequency of hydropeaks may favour the reproduction of several salmonid species
(Connor and Pflug 2004). Such experimental studies improve our mechanistic understanding
of the ecological effects of the different hydropeak characteristics. However, as for comparisons
of nearby reaches, temporal analyses deserve repeated field validations in multiple reaches.
Difficulties for gathering hydraulic and fish data in many hydropeaking reaches over several
years explain why few or no studies have carried out combined spatial and temporal analyses.

The originality of our study is to combine three approaches: (1) a spatial analysis of the
effects of hydropeaking on fish assemblages between stream reaches, (2) a temporal analysis

74
of these effects within reaches, and (3) a comparison involving groups of nearby reaches
subjected to different intensities of hydropeaking (Figure II.1.1.). Our objective was to test
whether these different approaches reveal consistent effects of hydropeaking on fish
assemblages. For this purpose, we used a unique dataset from 45 reaches of 26 rivers covering
3-17 years. The dataset included six groups of nearby reaches close to hydropower plants,
hourly discharge data, hydraulic descriptions of reaches, and electrofishing surveys. We used
fish species abundances as response variables to account for differential responses among
species (Bain et al. 1988). We expected the spatial analysis to indicate how hydropeaking
influences the classical longitudinal organisation of fish assemblages (Vannote et al. 1980). We
expected the temporal analysis to better indicate the relative influence of hydropeaking on fish
compared to the well-documented influence of high and low flows on fish recruitment (Bischoff
and Wolter 2001; Thieme et al. 2001). We expected the comparison between nearby reaches to
help disentangle the effect of hydropeaking on fish assemblages from those of other
environmental drivers.

Figure II.1.1 Schematics of the three approaches used to study the relative influence
of natural environmental drivers and hydropeaking events on fish assemblage
structure. For each approach, factors assumed to have a major influence on the structure
of fish assemblages are shown in black whereas others are in grey.

75
1.3. Materials and methods

Reach selection and characteristics

The dataset consisted of fish assemblage surveys (n = 318 reach x year combinations)
conducted in 45 French stream reaches between 1990 and 2017, and corresponding hourly
discharge time-series. Fish reaches were selected from two available databases (n = 33 from the
national fish survey database www.naiades.eaufrance.fr; n = 12 surveyed by the national
hydropower company “Electricité de France”, EDF). These reaches had inter-annual median
daily discharges between 2.2 and 128 m3 s-1, slopes between 0.1 and 73.8 ‰ and widths at mean
discharge between 7 and 89 m (Table I.2.3), according to the national extrapolation of
environmental data available in Pella et al. (2012). Most selected reaches were situated
downstream of hydropeaking power plants, except four reaches that were not influenced by
hydropeaking. Two of them were situated upstream from hydropower plants, two others in
bypassed sections not influenced by hydropeaking. These four reaches belonged to a subset of
15 reaches involved in the analysis of nearby reaches. These 15 reaches were themselves
grouped to six groups of two or three nearby reaches, which were used for paired comparisons
(Figure II.1.2).

Reach selection was strongly constrained by the availability of hourly discharge data
measured close to the fish reaches (we imposed a distance <25 km and the absence of major
obstacle or tributary between the gauging and fish reach). We used the database of Pella et al.
(2012), which includes estimates of reach mean annual flow over the French river network, to
check the mean annual discharge of tributaries relative to the fish reach discharge; this
relative discharge was on average 2%, with only 3/45 cases above 10% (kept in the analyses
to avoid removing long fish time-series).

76
 Figure II.1.2. Reach locations in continental France and schematics of nearby
reaches (from six rivers) with different hydropeaking pressure. Reach codes are from
Table II.1.3.

Fish data

Fish were sampled by electrofishing between 1990 and 2017 between July 1 and
October 30 (average sampling date September 5; SD = 25 days), when young-of-the-year
individuals are catchable and easily identifiable. Three sampling methods were used among
surveys. For 10 reaches, generally small ones, the whole reach was most frequently prospected
by wading (De Lury 2014). This method could involve several passages, but we retained only
the first one for consistency among reaches. In 30 larger reaches, fish were mostly frequently
sampled by wading or by boat using point abundance samples (Nelva et al. 1979), i.e. more
than 75 electrofishing points (estimated sampling area of each point: 7 m2) distributed
throughout the reach. For five reaches, fish were mostly frequently sampled in more than 30
larger habitat units of varying areas, distributed between geomorphic units (e.g, a riffle or a
pool) in proportion of their availability within the reach (Vadas and Orth 1993; Lamouroux et
al. 1999). The area of these habitat units depended on the size of distinct habitat elements (e.g,
a group of boulders in the centre of the channel) and the inherent variability of areas sampled
without enclosures (e.g, in the centre channel, the area sampled depended on current velocity).

When the sampling method changed at a reach over years, which occurred in seven
reaches, we split the data as sampled in two different reaches to avoid bias in the temporal

77
analysis. Finally, fish individual length was reported for only 40% of the data, strongly reducing
the statistical power of analyses of size class density variations. Therefore, analyses of size class
variations are not reported here.

The sampled area was reported in all cases, allowing to estimate fish species densities
per survey (number of individuals per 100 m²). These values were log(1+x)-transformed to
approach normality in our analyses.

Environmental data

Defining seasons

We used two seasons for describing annual reach environmental conditions at the most
important periods for fish life cycle. The “Spring” season lasted from 1th March to 31th May,
encompassing most of the spawning period of the studied fish species (except brown trout).
“Summer” was defined as the period from 1th June to 30th September (or to the sampling date
if earlier), during which young-of-the-year fish are actively feeding.

Hourly discharge and temperature data

Hourly discharge data came from the national gauging network (n = 38 reaches,
www.hydro.eaufrance.fr) or from EDF (n = 7). For the latter, discharge was generally estimated
by adding a theoretical bypassed base flow, the turbined discharge and eventually an overflow
discharge over the dam. These reconstitutions could create additional uncertainty on hourly
discharge estimates. Air temperature data came from daily mean estimations based on the
closest air temperature station (SAFRAN, Durand et al. 1993).

Translations of discharge data into hydraulic characteristics and habitat suitability

We translated hourly discharge data into more proximate habitat variables describing
reach hydraulics (wetted width, water depth, flow velocity) and hydraulic habitat suitability
(weighted usable areas) for fish guilds (groups of species with comparable habitat use).

Hydraulic translations were made using the hydraulic geometry of Morel et al. (2020).
In brief, these models improve the classical hydraulic geometry relationships of Leopold and
Maddock (1953). They are based on the analysis of hydraulic data collected at the scale of
stream reaches in 1327 stream reaches of France and New Zealand. Their input variables

78
include climatic, hydrologic, topographic and land use descriptors, all available over the French
hydrographic network (Pella et al. 2012). These models provide the best estimates of reach
hydraulics at a given discharge rate in the absence of detailed field hydraulic measurements.
Morel et al. (2020) used cross-validations to quantify how their models predicted hydraulic
variations between reaches. For example, they showed that the models predicted 86% and 65%
of the variance in observed width and depth among French reaches. However, these models
were less accurate when predicting how the rate of change in width and depth with discharge
varied among reaches. In detail, the models explained only 35% (width) and 13% (depth) of the
exponent of hydraulic geometry for width and depth (Figure II.2.3. in Morel et al. 2020).

Hydraulic habitat translations were made for two fish guilds using the statistical
hydraulic habitat models of Lamouroux and Souchon (2002). These statistical habitat models
predict a “weighted usable area” in the reach, at a given discharge rate, as a function of the
hydraulic geometry of the reach (described above) and an estimate of average particle size (from
Snelder et al. 2011). The weighted usable area is the product of the reach wetted area and an
habitat value for the guild that varies between 0 and 1, depending on the suitability of velocities
and water depth in the reach for the fish guild. In other words, the weighted usable area increases
with both habitat quantity (wetted area) and quality (habitat value). Here, we used weighted
usable area for two fish guilds (Lamouroux and Souchon 2002), the “midstream” fish guild
grouping species selecting deep and fast-flowing microhabitats, and the “bank” guild grouping
species selecting shallow and slow-flowing microhabitats. The “bank” guild is expected to be
more affected by hydropeaking than the “midstream” guild, because hydropeaking generates
frequent dewatering and rapid flow velocity variations in shallow habitats.

Identifying significant flow events (increases and decreases)

To describe hourly changes in discharge, including both natural and artificial events, we
classified each hourly flow variation as “increasing”, “decreasing”, or “stable”. Hours with
variations below 0.1 m3 s-1 h-1 were classified as “stable”. We defined an increasing event
(respectively decreasing event) as a concatenation of successive increases (decreases),
potentially including periods of “stable” hours if shorter than two hours in total. We defined the
peak flow as the highest discharge of the event, at the beginning of the event for decreases and
at the end for increases. In the same way, base flow is the lowest discharge of the event. In all
subsequent analyses, we considered only significant events having a discharge ratio (i.e. peak
to base flow ratio) higher than 30%. We found this empirical threshold appropriate (visually)

79
for selecting major natural changes in flow (typically floods) as well as most hydropeaking
events in all study reaches (Figure II.1.A1). With this threshold, the number of events also
strongly differed between reaches subject or not to hydropeaking.

Seasonal environmental descriptions

For each sampling reach × year combination, we considered a set of environmental

variables (Table II.1.1) potentially affecting annual fish densities and calculated for the spring
and summer seasons preceding sampling. These variables describe air temperature (T), median
hydraulic conditions (variable abbreviations starting by “Med” in Table II.1.1), high flow
hydraulics (starting by “Hi”), low flow hydraulics and habitats (“Lo”), frequencies of flow
events (“Fr”) and ramping rates (“Rr”). We also described temperature and discharge conditions
during fish sampling (“S”). All environmental variables were normalized prior to analyses to
give them equal weights.

Correcting ramping rates (attenuation between the discharge station and the fish reach)

The hydraulic ramping-rates attenuate with the distance from the water release (Nestler
et al. 1989; Hauer et al. 2013; Sauterleute and Charmasson 2014). The attenuation is site-
specific and complicated to characterize, depending for example on flow velocity, bed
morphology and roughness. Here, we decided to correct ramping rates empirically (for 80% of
the studied reaches), as a function of distance between the gauging station and the fish sampling
reach (mean = 3.5 km, SD = 6.0 km, max = 24.8 km). For this purpose, we used seven pairs of
gauging stations located in the same hydropeaking rivers, with no flow obstacles or tributaries
between them (distance between 1 and 30 km, 1/7 pair being located in a river independent
from our dataset). For each pair of stations, we identified the time lag (hours) on the discharge
time-series between both stations by numerical optimization, and identified paired significant
events. We calculated an average attenuation coefficient for each pair of stations, then regressed
attenuation coefficients against distance among pairs (forcing the attenuation to be null at
distance zero). An attenuation of 0.4% per km was found for up-ramping rates of flow velocity
(R2=0.63) and 0.5% per km for down-ramping rates of wetted width (R2=0.45).

80
Table II.1.1. Description of the environmental variables considered.

Variable group Abbreviation Description Unit

Topography Sl Mean reach slope ‰
Temperature T Air temperature °C
Median hydraulic Med_D Median water depth m
conditions
Med_V Median hourly flow velocity m s-1
Med_W Median hourly wetted surface m
High flow velocity Hi_V Maximum hourly flow velocity m s-1
Hi_V_d Average daily duration of hourly hours day-1
flow velocity >0.7 m s-1
Low flow velocity Lo_V Minimum flow velocity m s-1
Lo_V_d Average daily duration of hourly hours day-1
flow-velocity <0.2 m s-1
Low flow weighted usable Lo_HabM Weighted usable area for 100 m m2
area of river, for the ‘midstream’
guild, at low flow 5% quantile
(discharge below which hourly
discharge is for 5% of the time)
Lo_HabB Weighted usable area for 100 m m2
of river, for the ‘bank’ guild, at
low flow 5% quantile (discharge
below which hourly discharge is
for 5% of the time)
Occurrence of flow events Fr_Dec Average daily occurrence of nb day-1
decrease events
Fr_FDec Average daily occurrence of nb day-1
rapid decrease events with
wetted-width down-ramping-
-1
rate >0.6 m h
Fr_Inc Average daily occurrence of nb day-1
increase events
Fr_FInc Average daily occurrence of nb day-1
rapid increase events with flow
velocity up-ramping-rate >0.07
m s-1 h-1
Ramping rate Rr_V Quantile 90% of flow velocity m s-1 h-1
up-ramping-rate
Rr_W Quantile 90% of wetted width m h-1
down-ramping-rate
Fish sampling conditions S_Q Flow during fish sampling m3 s-1
S_T Air temperature during fish °C
sampling

81
Data analyses

Spatial and temporal analyses

To study the link between annual fish densities and seasonal environmental variables,
we used Coinertia Analyses (CoA; Dolédec and Chessel 1994; Mérigoux and Ponton 1999),
which compute successive pairs of environmental and fish multivariate axes being as covariant
as possible. A CoA is a simultaneous analysis of the fish and environmental datasets that is
appropriate when the number of variables (biological and environmental) is relatively high
compared to the number of surveys. Optimizing the covariance implies that the fish and
environmental axes are correlated and simultaneously explain a high variance (i.e. they
summarize variations in fish assemblages and environmental variables among reaches). Here,
the environmental and fish datasets were analysed following a “principal component analysis”
logic. The environmental variables were standardized, but not the biological variables that had
similar units and were only log-transformed.

We performed two different CoAs of the environmental and fish datasets, a “between-
reach” CoA and a “within-reach” CoA, to analyse separately spatial effects (between reaches)
and temporal effects (within reaches). The between-reach CoA is our spatial analysis among
reaches. It is computed on interannual averages of fish and environmental variables. The within-
reach CoA is our temporal analysis of the annual variations within reaches. It is computed on
differences between annual variables and their interannual average (by reach). In other words,
it removes differences between reaches to analyse relative annual changes.

We paid a particular attention to the position of nearby reaches on the between-reach

fish factorial map (spatial analysis) to appreciate how changes in hydropeaking pressure have
influenced the spatial ordination of fish assemblages.

Relations between environmental variables and fish densities

We reported the correlation between reach coordinates on the two CoA axes
(environmental and fish axes) to appreciate how environmental variables and fish densities were
related. We also used another measure of the overall similarity using a multivariate extension
of the Pearson correlation coefficient called the RV-coefficient (Robert and Escoufier 1976).
COVV(𝑋,𝑌)
The RV-coefficient (RV(X, Y) = ) is calculated as the total co-inertia (i.e. sum of
√𝑉𝐴𝑉(𝑋)𝑉𝐴𝑉(𝑌)

82
eigenvalues of a CoA) divided by the square root of the product of the squared total inertias
(sum of the eigenvalues) from the separate analyses of each dataset. RV-coefficient ranges from
0 to 1, with a high RV-coefficient indicating a high degree of co-structure. Finally, a
permutation test was conducted on the datasets to check the significance of the co-structure (RV
test). All analyses were performed using the R software (R Development Core Team 2020)
“ade4” package (Dray et al. 2007).

1.4. Results

The dataset concerned 207,386 fish individuals, and we considered the densities of 13
species (Table II.1.2, Figure II.1.A2) with relative survey-averaged density above 1% of the
total survey-averaged density. Fish assemblages were typical of those found along longitudinal
gradients in Europe (Figure II.1.3A), with 17/45 reaches with relatively high densities of Salmo
trutta (>10%; hereafter, “trout” reaches) and the other, larger streams dominated by cyprinids
(hereafter, “cyprinid” reaches).

Significant flow events were mostly hydropeaking events. For example, flow decreases
per day in spring averaged 0.4 (min=0.01, max=1.95) in hydropeaking reaches vs. 0.02
(min=0.01, max=0.03) in others. The average wetted-width down-ramping rate across all
reaches was 0.68 ± 0.42 m h-1 (mean ± sd), the average water depth down-ramping-rate was
0.03 ± 0.01 m h-1 and the flow velocity up-ramping-rate was 0.09 ± 0.04 m s-1 h-1 (see details
in Table II.1.3). Therefore, environmental variables describing the occurrence of flow events
and ramping rates (abbreviations starting by “Fr” and “Rr” in Table II.1.1) mostly described

83
hydropeaking intensity. By contrast, variables describing high flows, low flows and median
conditions (“Hi”, “Lo” and “Med”) were strongly influenced by floods and droughts.

Figure II.1.3. Results of the CoA analyses on the fish and environmental datasets. (A):
Between-reach spatial analysis (B): Within-reach temporal analysis on “trout reaches”.
(C) Within-reach temporal analysis on “cyprinid reaches”. Left panels show the fish
species scores on the first and second fish axes; right panels show the environmental variable
scores on the first and second environmental axes. Environmental variables describing
spring conditions are in grey; summer in black; sampling conditions in blue. Slope (in
orange) is projected as supplementary variable. The dotted boxes on the right illustrate the
interpretation of environmental axes. See Tables II.1.1 and II.1.2 for variable and species
codes

84
 Table II.1.2. List of the thirteen species studied.
Abbreviation Scientific name Common name
SaT Salmo trutta Brown trout
PhP Phoxinus phoxinus Minnow
CoG Cottus gobio Sculpin
LeL Leuciscus leuciscus Dace
BaBu Barbus barbus Barbel
ChN Chondrostoma nasus Nase
LeG Lepomis gibbosus Pumpkinseed
RuR Rutilus rutilus Roach
SqC Squalius cephalus Chub
GoG Gobio gobio Gudgeon
AlB Alburnoides bipunctatus Stream bleak
BaBa Barbatula barbatula Stone loach

Spatial analysis and positions of nearby reaches

The first and second axes of the between-reach CoA (Figure II.1.3A) explained 81%
and 11% of the total inertia, respectively (see Table II.1.4 for the proportion of variance of the
initial datasets taken into account by each CoA axes). Pearson correlation between the two
datasets was 0.64 for the first axis and 0.49 for the second axis, and the two datasets were
significantly related (RV=0.26, P = 0.005). The first axes of the fish and environmental factorial
maps suggested that larger streams (i.e. high median depth and width, high weighted usable
area, low slope) had lower densities of Salmo trutta and Cottus gobio and higher densities of
other species such as Squalius cephalus, Rutilus rutilus, Gobio gobio and Barbus barbus. The
second axis indicated that faster-flowing reaches had higher densities of Phoxinus phoxinus and
Barbatula barbatula and lower densities of Alburnoides bipunctatus.

Accordingly, positions of reaches on the fish factorial map (Figure II.1.4A) indicated a
traditional longitudinal gradient, with smaller “trout” reaches on the right and larger “cyprinids”
reaches on the left. Positions of pairs of nearby reaches on the map (black arrows on Figure
II.1.4A) suggested effects of hydropeaking intensity for cyprinid reaches more than for trout
reaches. In particular, for four pairs of reaches on the left of the map, the reach with higher
hydropeaking pressure was consistently situated towards the top or top-right of the map relative
to its less impacted corresponding reach. This globally indicated, for paired reaches, higher
relative abundance of Phoxinus phoxinus and/or Salmon trutta in the reaches with highest
hydropeaking intensity. Figure II.1.4B highlights these differences for two species: higher

85
densities of Phoxinus phoxinus were observed in reaches with higher hydropeaking intensity
whereas higher densities of Squalius cephalus were found in reaches with lower hydropeaking
intensity.

Figure II.1.4. Comparison of fish assemblages between nearby reaches. (A): Reach scores
on the first and second axes of the fish dataset for the between-reach spatial analysis. Triangles
correspond to “trout reaches” and circles to “cyprinid reaches”. Black arrows indicate the
position of pairs of nearby reaches (same colour), with arrows going from the reach least
influenced by hydropeaking to the most influenced. (B): Density of minnow Php and chub Lec
in the four pairs of nearby reaches. Reach colour codes are from Figure II.1.2. Reach codes are
from Table II.1.3.

86
Table II.1.3. Reach characteristics (see Table II.1.1 for variable codes). Reaches considered for the
nearby reaches analysis are linked using brackets. Note that the two reaches indicated by a * correspond to
a unique reach but with different dates.

Median Mean air

Geographical Rr_Vgrad
Reach code coordinates (X; Y)
discharge Slope temperature Fr_Dec Fr_Inc
m s-1 h-1
Q50 ‰ Spring; Mean±sd Mean±sd
in Lambert 93 Mean±sd
m3 s-1 Summer °C
871774.242;
Ain 6526504.737 70.5 2.0 13.2; 21.0 0.05 ± 0.01 0.43 ± 0.12 0.37 ± 0.12
738589.163;
Allier 6446380.559 15.3 2.1 10.8; 17.9 0.06 ± 0.02 0.75 ± 0.32 0.70 ± 0.35
811952.726;
Ardeche 6384583.096 9.8 0.1 14.1; 22.2 0.09 ± 0.01 0.70 ± 0.25 0.77 ± 0.34
752029.166;
Besbre 6600455.770 4.1 1.1 12.6; 17.1 0.03 ± 0.01 0.20 ± 0.08 0.11 ± 0.05
608335.903;
Cepre 6424801.115 17.7 1.2 14.1; 18.6 0.08 ± 0.04 0.55 ± 0.40 0.55 ± 0.41
669551.775;
Cher1 6588271.080 7.0 0.4 13.1; 19.1 0.08 ± 0.03 0.69 ± 0.26 0.45 ± 0.16
587688.181;
Correze2 6453638.630 13.8 2.3 14.1; 19.2 0.06 ± 0.02 0.26 ± 0.11 0.15 ± 0.05
Creuse_D3 566106.558;
6615406.208 22.0 1.8 13.3; 19.6 0.04 ± 0.01 0.37 ± 0.08 0.32 ± 0.10
545127.844;
Creuse_D4 6622154.201 22.5 0.4 13.0; 19.5 0.03 ± 0.01 0.32 ± 0.09 0.25 ± 0.11
DordA_D1 600865.545;
6424853.320 96.1 0.8 13.8; 19.8 0.04 ± 0.01 0.08 ± 0.10 0.10 ± 0.08
570227.164;
DordA_D2* 6419870.682 115.6 4.1 13.9; 14.7 0.03 ± 0.01 0.08 ± 0.11 0.11 ± 0.11
570227.164;
DordB_D1* 6419870.682 96.1 0.8 13.9; 20.5 0.05 ± 0.02 0.38 ± 0.27 0.37 ± 0.25
558010.860;
DordB_D3 6413463.240 128.0 2.1 14.2; 20.6 0.04 ± 0.01 0.26 ± 0.18 0.23 ± 0.15
608936.299;
Dordogne 6436765.767 82.7 0.9 12.7; 18.3 0.05 ± 0.02 0.24 ± 0.25 0.25 ± 0.24
664807.474;
Dordogne1 6498563.755 3.2 8.2 10.8; 16.1 0.17 ± 0.1 0.62 ± 0.89 0.59 ± 0.87
675890.967;
Dordogne3 6500545.354 2.8 12.3 8.4; 15.8 0.26 ± 0.1 0.21 ± 0.12 0.24 ± 0.17
737411.593;
Dore 6531059.903 13.4 0.9 12.2; 18.5 0.03 ± 0.01 0.33 ± 0.23 0.29 ± 0.21
1005479.587;
Doubs1 6702034.457 32.5 1.3 9.9; 15.5 0.04 ± 0.01 0.44 ± 0.08 0.41 ± 0.08
998685.499;
Doubs2 6692083.909 25.9 2.8 8.7; 14.5 0.07 ± 0.01 0.63 ± 0.08 0.59 ± 0.10
Dur_D1 867476.940;
6299688.683 122.6 2.9 15.5; 24.5 0.12 ± 0.02 0.37 ± 0.12 0.34 ± 0.09
889882.098;
Dur_UP 6293508.414 122.6 2.3 14.7; 24.0 0.12 ± 0.13 0.03 ± 0.03 0.03 ± 0.02
557491.117;
Garonne1 6248020.216 91.8 3.0 13.6; 21.3 0.11 ± 0.03 0.57 ± 0.22 0.54 ± 0.16
513155.364;
Garonne2 6224770.194 47.6 3.5 12.3; 19.6 0.08 ± 0.02 0.51 ± 0.23 0.51 ± 0.22
1223972.662;
Golo 6177578.191 11.2 57.3 13.7; 21.4 0.09 ± 0.01 0.75 ± 0.39 0.71 ± 0.40
790882.892;
Loire1 6471148.745 24.8 3.2 11.0; 17.8 0.05 ± 0.01 0.57 ± 0.27 0.55 ± 0.29
790895.984;
Loire3 6524644.455 33.5 0.5 12.5; 18.3 0.11 ± 0.05 0.74 ± 0.12 0.79 ± 0.14
775925.352;
Loire4 6584819.553 48.1 2.1 12.1; 17.7 0.03 ± 0.02 0.38 ± 0.15 0.31 ± 0.11
638852.847;
Lot 6387661.801 68.0 4.9 12.8; 19.4 0.13 ± 0.03 0.99 ± 0.22 0.95 ± 0.21

87
 616560.143;
Maronne 6441947.071 9.3 2.9 12.9; 18.7 0.09 ± 0.05 0.33 ± 0.37 0.33 ± 0.37
1066121.216;
Roya_BP1 6335287.829 6.4 36.2 11.1; 19.1 0.24 ± 0.13 0.28 ± 0.29 0.28 ± 0.28
1064803.026;
Roya_D 6332576.200 6.6 27.7 11.5; 20.2 0.14 ± 0.04 1.05 ± 0.32 0.94 ± 0.35
1067468.239;
Roya_UP 6340374.076 2.2 73.8 10.4; 19.0 0.08 ± 0.03 0.02 ± 0.02 0.03 ± 0.02
534701.500;
Salat 6230865.410 30.3 2.4 12.2; 18.2 0.06 ± 0.02 0.30 ± 0.12 0.32 ± 0.13
383599.079;
Selune 6841493.378 8.2 3.0 11.6; 17.1 0.06 ± 0.02 0.18 ± 0.08 0.16 ± 0.08
1018695.535;
Siagne 6280195.484 2.8 0.4 16.8; 23.0 0.39 ± 0.07 1.03 ± 0.38 1.14 ± 0.48
579398.131;
Tarn 6308816.030 83.6 1.3 14.9; 20.6 0.04 ± 0.01 0.51 ± 0.15 0.44 ± 0.13
657601.117;
Tet 6171763.877 7.1 10.3 14.0; 21.9 0.08 ± 0.02 0.22 ± 0.17 0.18 ± 0.09
934680.457;
VerdonD_BP1 6299317.267 16.9 1.0 14.0; 23.4 0.1 ± 0.17 0.01 ± 0.01 0.01 ± 0.01
929722.209;
VerdonD_BP2 6297495.920 17.3 5.7 14.5; 23.9 0.11 ± 0.19 0.01 ± 0.01 0.01 ± 0.01
983527.417;
VerdonM_BP1 6311283.309 9.5 8.1 10.2; 19.8 0.11 ± 0.25 0.01 ± 0.03 0.01 ± 0.03
981513.856;
VerdonM_D1 6310766.070 9.6 4.7 10.7; 19.4 0.14 ± 0.02 0.22 ± 0.06 0.24 ± 0.05
976499.250;
VerdonM_D2 6307373.744 9.9 6.2 11.2; 19.8 0.15 ± 0.02 0.22 ± 0.06 0.24 ± 0.05
578479.213;
Vezere 6458829.865 17.1 0.3 14.2; 18.9 0.07 ± 0.02 0.46 ± 0.16 0.42 ± 0.15
570177.247;
Vien_D1 6530340.952 35.2 1.9 12.3; 18.1 0.09 ± 0.03 0.48 ± 0.14 0.45 ± 0.10
552401.5587;
Vien_D2 6528649.95 46.5 1.6 12.3; 18.0 0.05 ± 0.01 0.29 ± 0.07 0.26 ± 0.07
518554.183;
Vienne2 6603719.964 63.3 2.3 12.8; 19.0 0.04 ± 0.01 0.26 ± 0.09 0.25 ± 0.08

88
Temporal analysis

We decided to apply the within-reach CoA separately on “trout” reaches (Figure

II.1.3B) and on “cyprinid” reaches (Figure II.1.3C), because pooling these reaches made the
axes hardly interpretable due to the different species involved.

The first and second CoA axes for “trout” reaches explained 44% and 27% of the total
inertia, respectively (Table II.1.4). Pearson correlation between the two datasets was 0.54 for
the first axis and 0.33 for the second axis and the two datasets were significantly related (RV=0.
05, P = 0.04). The first axis of the fish and environmental factorial maps suggested that the
major temporal effect was a decrease in density of Salmo trutta with high-flow events during
spring and/or summer. The second axis, for which the correlation was low, suggested that the
densities of Phoxinus phoxinus and secondarily Cottus Gobio were positively related to
hydropeaking occurrence.

The first and second CoA axes for “cyprinid” reaches (Figure II.1.3C) explained 71%
and 11% of the total inertia, respectively (see Table II.1.4 for details). Pearson correlation
between the two datasets was 0.48 for the first axis and 0.44 for the second axis. The
relationship between fish densities and the environment was significant (RV=0.14, P = 0.002).
The first axis of the fish and environmental factorial maps suggested that the major temporal
effect was a general decrease in the densities of most fish species with high-flow events during
summer. The second axis suggested that spring high flows could increase the densities of
Leuciscus leuciscus and Chondrostoma nasus relatively to the density of Squalius cephalus.
Summer hydropeaks tended to have the opposite effect.

On both within-reach CoA analyses, high discharge and very cold temperature on
sampling date tended to reduce observed densities of most species.

89
 Table II.1.4. Summary statistics of the CoAs: RV coefficients; P-values of permutation tests;
proportion of variance of each initial datasets taken into account by the two first axes; and Pearson
correlation coefficients between the fish and environmental axes.

Pearson
Environment: Fish density:
Reaches RV correlation
Analysis P-value variance variance
involved coeff. F1envxF1fish;
explained explained
F2env xF2fish
Between-
F1: 30% F1: 47% F1: 0.64
reach CoA All 0.26 0.005
F2: 15% F2: 16% F2: 0.49
(Spatial)
“Trout” F1: 27% F1: 42% F1: 0.54
Within- 0.05 0.01
reaches F2: 22% F2: 20% F2: 0.33
reach CoA
“Cyprinid” F1: 29% F1: 34% F1: 0.48
(Temporal) 0.14 0.002
reaches F2: 18% F2: 22% F2: 0.44

1.5. Discussion

Secondary influence of hydropeaking compared to stream size and high-flow

events

The spatial analysis (between-reach CoA) showed that the influence of hydropeaking
on fish assemblage was secondary relative to the influence of stream size. In this analysis, the
fish factorial map separated Salmo trutta and Cottus gobio preferably found in steep
headwaters, Phoxinus phoxinus and Barbatula barbatula in intermediate-sized streams, and all
other species in larger streams. Such an organization of fish assemblages along the upstream-
downstream gradient has been largely described (e.g. Huet 1949, Vannote et al. 1980).
However, we observed two exceptions to this upstream-downstream gradient on the second
axis: the position of Phoxinus phoxinus and Barbatula barbatula (with high densities in faster-
flowing reaches) opposed to Alburnoides bipunctatus. The computation of the second axis was
certainly influenced by the spatial distribution of the studied reaches, some being within and
others outside the heterogeneous geographic distribution of Alburnoides bipunctatus.

The temporal analysis (within-reach CoA) showed that the influence of hydropeaking
on fish densities was secondary relative to the effects of high flows. In line with previous
studies, we found that high-flow events in spring or summer reduced the densities of most
species (Bischoff and Wolter 2001, Thieme et al. 2001, Cattaneo 2005, Bret et al. 2015). For

90
“cyprinid” reaches, the second axis of the environmental factorial map discriminated the effects
of spring and summer high-flow events. Spring high-flows logically affected earlier spawners
(i.e. March/April), such as Leuciscus leuciscus and Chondrostoma nasus, while summer high-
flows affected species that spawn later (i.e. May/June), such as Gobio gobio and Phoxinus
phoxinus (Lelek and Penaz 1963, Mills 1981, Lascaux et al. 2013).

Since most studies have shown that hydropeaking was linked to adverse consequences
for fish (e.g. stranding, drifting and spawning ground scouring, Halleraker et al. 2003, Connor
and Pflug 2004, Auer et al. 2017), frequent hydropeaks and high ramping rates were expected
to be negatively related to the density of most species. However, our temporal and spatial
analysis brought different findings. We propose three potential explanations for the secondary
influence of hydropeaking observed here. First, within-reach, the variability of hydropeaking
intensity between years was low, limiting our ability to show relationships between fish
densities and hydropeaking descriptors. For example, the annual within-reach standard
deviation for up-ramping rates of flow velocity was 44% of the interannual average. Second,
the hydropeaking intensity in the reaches studied here is relatively low. For example, we found
an average water-depth down-ramping-rate of 3.2 cm h-1 for hydropeaking reaches (max = 10.6
cm h-1) whereas Schmutz et al. (2015) showed an adverse effect of hydropeaking on fish
assemblages for water level down-ramping rates above 15 cm h-1. Halleraker et al. (2003)
observed almost no stranding of juvenile salmons for water-depth down-ramping-rates below
18.6 cm h-1. Although we cannot directly compare these values (our ramping rates are based on
average depths, others on water levels), the down-ramping rates in our reaches were clearly
lower. Third, since most French hydropower plants were built more than fifty years ago, the
fish species that are currently present could have adapted to these highly variable environments.
For example, studies have shown that many fish individuals move in response to temporal
variations in habitat suitability (Pert and Erman 1994, Shirvell 1994, Bond and Jones 2015) and
can adapt their behaviour against hydropeaking risks by using the less constraining habitats
(Capra et al. 2017).

Focusing on young-of-the-year individuals could have revealed a stronger influence of

hydropeaking compared to our analyses made without differentiating life stages. Indeed, young-
of-the-year individuals have lower swimming capacity than adults and mostly use near-bank
habitats (Moore and Stanley 1988), making them highly susceptible to drifting and stranding
(Saltveit et al. 2001, Halleraker et al. 2003, Lechner et al. 2016). However, density-dependence

91
may offset the effects of hydropeaking in some populations (Ratikainen et al. 2008, Puffer et
al. 2019). For example, using a population dynamic model for Atlantic salmon (Salmo salar)
in a hydropeaking river in Norway, Sauterleute et al. (2016) showed that the stranding mortality
of young-of-the-year individuals had weak effects on population dynamics.

Hydropeaking favours headwater species at the expense of medium-sized stream

species, consistently in space and time

The comparison of nearby reaches showed that, for cyprinid reaches, stronger
hydropeaking intensity was associated to higher densities of fish species typical of headwater
streams (Phoxinus phoxinus, Salmo trutta and Cottus gobio) and lower densities of species
typical of medium-sized streams (Rutilus rutilus, Barbus barbus, Squalius cephalus,
Chondrostoma nasus). In contrast, for trout reaches, no clear trend was observed, indicating
that headwater fish species were less influenced by hydropeaking than others. This result is
consistent with the temporal analysis on trout reaches, where years with higher hydropeaking
intensity had higher densities of fish species typical of headwater streams. Several previous
studies have also shown that fish species typical of headwater streams may better withstand
hydropeaking (Garcia De Jalon et al. 1988, Casado et al. 1989, Garcia De Jalon et al. 1994).
Garcia De Jalon et al. (1994) found that, after the construction of a hydropeaking power plant
on the Tera River (Spain), populations of almost all cyprinid species declined while the trout
population persisted. However, on the Rio Duraton (Spain), Camargo and Garcia de Jalon
(1990) found the opposite effect: cyprinid populations persisted or even increased whereas trout
disappeared. These differences are difficult to interpret and likely depend on dam management,
dam location along the river and/or the particular morphological characteristics of each
hydropeaking reach.

Another way to study the influence of hydropeaking is to focus on the individual scale,
for example by comparing the growth of individuals under different hydropeaking intensity.
Such approaches suggest that Atlantic salmon (Puffer et al. 2017) may thrive in hydropeaking
reaches, in contrast to the cyprinid humpback chub (Gila cypha) in medium-sized streams
(Finch et al. 2015). A possible reason for the greater hydropeaking tolerance of headwater
stream species may be their behavioural adaptation to naturally highly variable environments
(Horwitz 1978). They may have developed ecological strategies to cope with frequent and rapid
flow variations (Schlosser 1982, Oberdorff et al. 2001, Lytle and Poff 2004). For example,

92
rainbow trout (Oncorhynchus mykiss) can detect the rising limb of a flood, allowing a rapid
search of hydraulic shelter (Gore et al. 1994). The freshwater sculpin (Cottus gobio) lives in
coarse substrate interstices, limiting the risk of drifting during high flows (Keith et al. 2011).
Minnow (Phoxinus phoxinus) spawns several times a year (Wootton and Mills 1979), reducing
the mortality risk of a whole young-of-the-year cohort after a single intense hydropeaking event.
Furthermore, due to the greater availability of rocks, tree roots and woody debris, headwater
streams have a greater diversity of habitats and shelter than larger streams (Jackson and Sturm
2002, Benda et al. 2005, Gooderham et al. 2007).

Perspectives

We identified several biases that could have blurred the relationships between
hydropeaking descriptors and fish densities. First, variations in electrofishing efficiency
between surveys, related to environmental conditions during sampling, probably introduced
noise in observed fish densities (Cauvy-Fraunié et al. 2020). Consistently, we found that higher
air temperature or lower discharge rates during fish sampling resulted in higher densities.
Second, variations in fish sampling methods between reaches could have influenced our spatial
analysis. We checked that this influence was limited (results not shown) by repeating our
analysis after introducing method-specific correction coefficients (up to a factor 5) on observed
densities. This indicated that our spatial analysis was essentially driven by differences in species
relative densities and little sensitive to the fish sampling method. Third, environmental
descriptions could be improved. In particular, air temperature alone did not allow to evaluate
the potential influence of water temperature variations induced by hydropeaking (known as
thermopeaking, Zolezzi et al. 2011). In addition, although they are relatively accurate for
describing hydraulic differences between reaches, the general hydraulic translations of Morel
et al. (2020) cannot reflect the presence of shelter or particular substrate configurations. More
importantly, hydraulic geometry models are uncertain for describing the rate of change in width
and depth with discharge. Field measurements would have been desirable for improving our
hydraulic descriptions. However, using more accurate descriptions would unlikely affect our
conclusions, because we described many aspects of the frequency and ramping rates of
hydropeaks, all of which had a secondary influence on density annual variations. Finally, all
environmental characteristics of reaches (hydraulic geometry, substrate composition, shelter
availability) could have changed over the study period.

93
 In spite of the large dataset used here, increasing the range of hydropeaking intensity
studied would help to identify thresholds over which significant adverse impacts are observed
on fish assemblages. In particular, it is possible that higher hydropeaking intensity negatively
affects headwater species (Phoxinus phoxinus and Cottus Gobio). Since each country has its
own rules for managing hydropeaking, international comparisons will help to extend the range
of hydropeaking intensity considered. The number of reaches considered could also be
increased by using environmental DNA techniques that allow rapid, semi-quantitative
assessments of fish community structure in a non-invasive way (Lodge et al. 2012; Taberlet et
al. 2012; Pont et al. 2018).

In the future, there is a need to improve our understanding of the interactions between
hydropeaking and other stresses for aquatic biota (e.g. changes in water temperature or quality;
Zolezzi et al. 2011; Bruno et al. 2013; Vanzo et al. 2016) to determine whether their effects are
additive, synergistic or antagonistic. This could be achieved with analyses similar to ours, but
including complementary environmental descriptors (e.g. turbidity, water temperature) and
more reaches with unique or multiple pressures (e.g. Acreman et al. 2014; Bondar-Kunze et al.
2016). Finally, understanding taxa behavioural response to hydropeaking could help identifying
improved management measures (Metcher et al. 2017). For example, the development of
models of habitat selection adapted to highly variable environments, taking into account the
history of habitat conditions, could help to predict the amount of suitable habitat for different
taxa under different management scenarios.

1.6. Conclusion

We found that the influence of hydropeaking was secondary compared to the well-
known organization of fish assemblages along the longitudinal gradient of rivers, or the
frequently observed negative influence of floods on recruitment. Our temporal analyses suggest
that, within the range of hydropeaking intensity considered, small changes in hydropeaking
management should have limited effects on fish assemblage structure. However, further
analyses covering a wider range of hydropeaking intensity are needed to assess possible
threshold effects. Consistently in space and time, we found that hydropeaking affected fish
species typical of medium-sized streams more intensively than those of headwater streams.
Although this result remains to be confirmed with a larger dataset, it suggests that flow should
be managed cautiously in medium-sized streams. Because past studies focused more on

94
salmonid species (Moreira et al. 2019), more attention is deserved on assemblages of larger
streams.

1.7. Acknowledgements
This research is part of the team HYNES built within the EDF-INRAE collaboration agreement
on water engineering and aquatic systems. We thank all the people who have been involved in the
monitoring. Fieldwork was conducted with adequate administrative permits for electrofishing and in
accordance with French laws and ethical rules.

1.8. Bibliography

Arthur SM, Manly BF, McDonald LL, Garner GW. 1996. Assessing habitat selection when
availability changes. Ecology 77:215–227.

Auer S, Zeiringer B, Führer S, Tonolla D, Schmutz S. 2017. Effects of river bank heterogeneity
and time of day on drift and stranding of juvenile European grayling (Thymallus thymallus L.)
caused by hydropeaking. Science of the Total Environment 575:1515–1521.

Blinn DW, Shannon JP, Stevens LE, Carder JP. 1995. Consequences of fluctuating discharge
for lotic communities. Journal of the North American Benthological Society 14:233–248.

Bou C, Rouch R. 1967. Un nouveau champ de recherches sur la faune aquatique souterraine.
C. R. Academie hebdomadaires des sciences 265:369–370.

Bruno MC, Maiolini B, Carolli M, Silveri L. 2009. Impact of hydropeaking on hyporheic

invertebrates in an Alpine stream (Trentino, Italy). Annales de Limnologie - International
Journal of Limnology 45:157–170.

Bruno MC, Siviglia A, Carolli M, Maiolini B. 2013. Multiple drift responses of benthic
invertebrates to interacting hydropeaking and thermopeaking waves. Ecohydrology 6:511–522.

Capra H, Plichard L, Bergé J, Pella H, Ovidio M, Mcneil E, Lamouroux N. 2017. Fish habitat
selection in a large hydropeaking river: strong individual and temporal variations revealed by
telemetry. Science of the Total Environment 578:109–120.

Chamberland JM, Lanthier G, Boisclair D. 2014. Comparison between electrofishing and

95
snorkeling surveys to describe fish assemblages in Laurentian streams. Environmental
Monitoring and Assessment 186:1837–1846.

Dole-Olivier MJ, Marmonier P, Beffy JL. 1997. Response of invertebrates to lotic disturbance:
Is the hyporheic zone a patchy refugium? Freshwater Biology 37:257–276.

Dole-Olivier MJ, Maazouzi C, Cellot B, Fiers F, Galassi DMP, Claret C, Martin D, Mérigoux
S, Marmonier P. 2014. Assessing invertebrate assemblages in the subsurface zone of stream
sediments (0-15 cm deep) using a hyporheic sampler. Water Resources Research 50:453–465.

Forcellini M, Plichard L, Dolédec S, Mérigoux S, Olivier JM, Cauvy-Fraunié S, Lamouroux N.

(n.d.) Microhabitat selection by macroinvertebrates: generality among rivers and functional
interpretation. Journal of Ecohydraulics.

Halleraker JH, Saltveit SJ, Harby A, Arnekleiv J V, Fjeldstad HP, Kohler B. 2003. Factors
influencing stranding of wild juvenile brown trout (Salmo trutta) during rapid and frequent flow
decreases in an artificial stream. River Research and Applications 19:589–603.

Holomuzki J, Biggs BJF. 2000. Taxon-specific responses to high-flow disturbance in streams.

Journal of the North American Benthological Society 19:670–679.

Judes C, Gouraud V, Capra H, Maire A, Barillier A, Lamouroux N. (2021) Consistent but

secondary influence of hydropeaking on stream fish assemblages in space and time. Journal of
Ecohydraulics.

Kemp PS, Gilvear DJ, Armstrong JD. 2003. Do juvenile Atlantic salmon parr track local
changes in water velocity? River Research and Applications 19:569–575.

Kjaerstad G, Arnekleiv J V., Speed JDM, Herland AK. 2018. Effects of hydropeaking on
benthic invertebrate community composition in two central Norwegian rivers. River Research
and Applications 34:218–231.

Korman J, Campana SE. 2009. Effects of Hydropeaking on Nearshore Habitat Use and Growth
of Age-0 Rainbow Trout in a Large Regulated River. Transactions of the American Fisheries
Society 138:78–87.

Lamouroux N, Capra H, Pouilly M, Souchon Y. 1999. Fish habitat preferences in large streams
of southern France. Freshwater Biology 42:673–687.

96
Manly BF, McDonald L, Thomas D, McDonald TL, Erickson WP. 2002. Resource Selection
by Animals: Statistical Design and Analysis for Field Studies. Springer Netherlands.

Patton BW, Braithwaite VA. 2015. Changing tides: Ecological and historical perspectives on
fish cognition. Wiley Interdisciplinary Reviews: Cognitive Science 6:159–176.

Pert EJ, Erman D. 1994. Habitat use by adult rainbow trout under moderate artificial
fluctuations in flow. Transactions of the American Fisheries Society 123:913–923.

Pella H, Lejot J, Lamouroux N, Snelder T. 2012. Le réseau hydrographique théorique (RHT)

français et ses attributs environnementaux [The French theoretical hydrographic network
(RHT) and its environmental attributes]. Géomorphologie: Relief, Processus, Environnement
18:317–336.

Plichard L, Capra H, Mons R, Pella H, Lamouroux N. 2017. Comparing electrofishing and

snorkelling for characterizing fish assemblages over time and space. Canadian Journal of
Fisheries and Aquatic Sciences 74:75–86.

Plichard L, Forcellini M, Le Coarer Y, Capra H, Carrel G, Ecochard R, Lamouroux N. 2020.

Predictive models of fish microhabitat selection in multiple sites accounting for abundance
overdispersion. River Research and Applications 36:1056–1075.

Reebs SG. 1996. Time-place learning in golden shiners (Pisces: Cyprinidae). Behavioural
Processes 36:253–262.

Regis J, Pattee E, Lebreton J. 1981. A new method for evaluating the efficiency of electric
fishing. Archiv für Hydrobiologie 93:68–82.

Saltveit SJ, Halleraker JH, Arnekleiv J V., Harby A. 2001. Field experiments on stranding in
juvenile Atlantic Salmon (Salmo salar) and Brown Trout (Salmo trutta) during rapid flow
decreases caused by hydropeaking. Regulated Rivers: Research & Management 17:609–622.

Schmutz S, Bakken TH, Friedrich T, Greimel F, Harby A, Jungwirth M, Melcher A, Unfer G,

Zeiringer B. 2015. Response of fish communities to hydrological and morphological alterations
in hydropeaking rivers of Austria. River research and applications 31:919–930.

Schooley. 1994. Annual Variation in Habitat Selection : Patterns Concealed by Pooled Data.
The Journal of Wildlife Management 58:367–374.

97
Shirvell CS. 1994. Effect of changes in streamflow on the microhabitat use and movements of
sympatric juvenile coho salmon (Oncorhynchus kisutch) and chinook salmon (O.tshawytscha)
in a natural stream. Canadian Journal of Fisheries and Aquatic Sciences 51:1644–1652.

Spearman C. 1904. The proof and measurement of association between two things.
International Journal of Epidemiology 15:72–101.

Stubbington R, Sarremejane R, Datry T. 2019. Alpha and beta diversity of connected benthic–
subsurface invertebrate communities respond to drying in dynamic river ecosystems.
Ecography 42:2060–2073.

Tachet P, Richoux P, Bournaud M, Usseglio-Polatera P. 2002. Invertébrés d’eau douce:

systématique, biologique et écologie. CNRS Editions.

Tuhtan JA, Noack M, Wieprecht S. 2012. Estimating stranding risk due to hydropeaking for
juvenile European grayling considering river morphology. KSCE Journal of Civil Engineering
16:197–206.

Vila-Martínez N, Caiola N, Ibáñez C, Benejam L, Brucet S. 2019. Normalized abundance

spectra of fish community reflect hydro-peaking on a Mediterranean large river. Ecological
Indicators 97:280–289.

98
1.9. Appendix

Figure II.1.A1. Example of a discharge time-series from the reach “Verdon_D1” and of
identified increasing (in red) and decreasing events (in blue).

99
Figure II.1.A2. Interannual average of fish species density (number of individuals per 100
m2 sampled) in each reach. Fish species and reaches are ordered according to their
position on the first axis (F1) of the between-reach CoA performed on the fish dataset. See
Tables II.1.1 and II.1.3 for species and reach codes.

100
2. JUDES ET AL. (B) :
Past hydraulics influence microhabitat selection by
macroinvertebrates and fish in hydropeaking rivers.
Judes C, Capra H, Gouraud V, Pella H, Lamouroux, N.

101
2.1. Abstract

Hydropeaking hydropower plants are the major source of renewable energy meeting
sub-daily peaks in electricity demand. They induce rapid artificial flow variations, highly
variable velocities, drift and stranding risks for aquatic organisms. An improved mechanistic
understanding of microhabitat selection by organisms is needed to better understand the effects
of hydropeaking on communities. In hydropeaking reaches, microhabitat selection likely
depends on both present and past hydraulics (flow velocity and water depth); this study aims to
assess their relative impact. For this purpose, we used observations of fish abundance in 1,180
microhabitats (507 sampled by electrofishing, 673 by snorkeling) and of macroinvertebrate
abundance in 36 microhabitats (both hyporheic and benthic) in a medium-sized river. We
described past hydraulics of microhabitats over the 15 days preceding sampling, using a 2D
hydrodynamic model, by identifying microhabitats dewatering (drying during > 10h) or with
high-velocity conditions (>1.3 m s-1 during > 10h). Macroinvertebrates habitat guilds
responded significantly to past hydraulics, with abundances 3.5-15.3 times lower in dewatering
habitats. Consequently, their preference for present hydraulics was different from that observed
in rivers without hydropeaking. For more mobile fish, responses were weaker and different,
with a “bank” guild selecting dewatering microhabitats and, secondarily, a “midstream” guild
avoiding them. Their selection of present hydraulics was similar to that observed in rivers
without hydropeaking. Overall, past hydraulics influenced microhabitat selection, with stronger
effects on macroinvertebrates and stronger effects of dewatering than of high past velocities.
However, high past velocities force fish to move and macroinvertebrates to hide.

Keywords: Habitat preferences, Flow variations, 2D hydrodynamic model, Behavior,

Dewatering responses, Hydropower

102
2.2. Introduction

Hydropeaking hydropower plants are the major source of renewable energy that meets
sub-daily peaks in electricity demand. By producing electricity on demand, they create frequent
rapid flow variations known as hydropeaking. Hydropeaking influences the habitats of aquatic
organisms, with strong spatial and temporal variations in point flow velocity, water depth and
shear stress (hydraulic conditions). In particular, near-shore areas may be subject to dewatering
during base flow (i.e., become temporarily dry when turbines are shut down), which may result
in stranding of individuals using shallow and slow-flowing habitats (Saltveit et al. 2001;
Halleraker et al. 2003). Midstream habitats may be subject to high flow velocity during peak
flow (when turbines are on), which may cause forced drift of individuals with low swimming
ability (Bruno et al. 2013).

Hydropeaking often results in a decrease in abundance, biomass and species diversity

in both fish and macroinvertebrate communities (Schmutz et al. 2015; Kjaerstad et al. 2018;
Vila-Martínez et al. 2019). However, these negative effects are not systematic and may depend
on hydropeaking hydraulic characteristics. For example, Judes et al. (2021) showed, in rivers
with moderate hydraulic variations, that hydropeaking could have weaker effects than floods
on fish community dynamics and did not greatly modify the organization of fish communities
along longitudinal gradients. To improve our mechanistic understanding of hydropeaking
influences on biota, we need to better understand individual responses to changes in the spatial
and temporal variations of point hydraulic conditions (i.e. microhabitat scale). The microhabitat
scale (here corresponding to fixed patches of ~7m2 for fish, 0.05 m2 for benthic
macroinvertebrates and 6 L of hyporheic water for hyporheic macroinvertebrates (Stubbington
et al. 2016)) is particularly suited for identifying the key hydraulic drivers of biological response
to hydropeaking. Several studies quantified key hydraulic conditions in microhabitats for fish
and macroinvertebrates drifting and stranding processes (Saltveit et al. 2001; Halleraker et al.
2003; Auer et al. 2017). For example, Halleraker et al. (2003) showed that stranding of juvenile
brown trout (Salmo trutta) occurs with dewatering vertical water level gradients higher than
0.10 cm h-1. By contrast, little is known concerning how microhabitat selection of fish and
macroinvertebrates is influenced by variations in point hydraulic conditions.

103
 In rivers without hydropeaking and associated sub-daily variations, microhabitat
selection by fish and macroinvertebrates is usually predicted from present hydraulic conditions
and substrate size (Forcellini et al., in press.; Lamouroux et al., 1999; Plichard et al., 2020). In
hydropeaking rivers, past hydraulic variability during the weeks preceding sampling is expected
to influence microhabitat selection by aquatic organisms. Indeed, experiments on fish
individual behavior (Green 1971; Roy et al. 2018) indicated that fish can memorize
environmental conditions over durations up to two weeks. Individuals perceive temporal
variations in hydraulic conditions (Patton et al. 2015) and may avoid habitats that are frequently
dewatered or have major variations in flow velocity. For example, using telemetry in the Rhône
River (France) to monitor 18 individual fish, Capra et al. (2017) suggested that fish memorize
spatial and temporal environmental variations during the two weeks preceding sampling so as
to use the “least constraining” microhabitats. Organisms may also have strong habitat fidelity
even after their habitats become unsuitable (Kemp et al. 2003). Several studies showed that
some fish species are reluctant to move across the river during sudden flow increases and
consequently use habitats with higher flow velocity and/or deeper habitats (Pert et al. 1994;
Shirvell 1994; Kemp et al. 2003). This behavior may be explained by the energy cost associated
with displacement and/or a higher risk of predation. Less mobile taxa, which include many
macroinvertebrates, may be unable to respond instantaneously to a rapid shift in habitat
conditions (Blinn et al. 1995). Some macroinvertebrates taxa come be remove from
hydropeaking reaches by passive or active drift during low or high flow (Bruno et al. 2013).
However, in adapting to naturally variable microhabitat conditions, some taxa have developed
morphological and behavioral adaptations to maximize colonization of available habitat and
can withstand extreme habitat conditions (e.g., resistance to desiccation). For instance, Asellus
aquaticus (crustaceans) may pass through a desiccation-resistant life stage, allowing them to
occupy intermittent stream habitats (Tachet et al. 2002). Thus, in hydropeaking rivers, such taxa
could actively select dewatering habitats. Other macroinvertebrate taxa may also colonize
subsurface hyporheic habitats to escape from variable benthic hydraulic conditions (Dole-
Olivier et al. 1997).

Studying the effects of past hydraulics on microhabitat selection by aquatic organisms

from field data is challenging. A first challenge is to describe past hydraulic conditions in

104
microhabitats. Spatially explicit hydrodynamic models, such as two-dimensional (2D) models
that map vertically averaged flow velocities at various discharge rates, are useful for this
purpose. However, calibrating and testing 2D models is time-consuming, expensive, and
requires an appropriate amount of quality field measurements. A second challenge is to sample
organisms at several discharge rates in order to have a sufficient number of microhabitat
samples with various combinations of past and present hydraulic conditions. Addressing this
challenge requires extensive fieldwork.

In this study, we combined extensive biological sampling in microhabitats at various

discharge rate with results of a calibrated 2D hydrodynamic model in a hydropeaking reach to
weight the relative effects of past and present hydraulic conditions on microhabitat selection.
Specifically, we recorded fish taxa abundance in 1,187 microhabitats and the abundance of
benthic and hyporheic macroinvertebrate taxa in 36 microhabitats along a 6 km reach of a
hydropeaking reach of a medium-sized French river (the Ain River). We studied the responses
of taxa that had significant microhabitat selection documented in literature. We grouped them
into a few habitat guilds with comparable habitat selection. The 2D hydrodynamic model was
used to estimate past hydraulics at each microhabitat during the 15 days preceding sampling
(Capra et al., 2017).

For fish, we expected “midstream” species, using fast-flowing and deep habitats, would
avoid dewatering areas and sustain high flow velocity variations to reduce their displacement.
By contrast, fish using shallow and slow-flowing habitats along the banks could be expected to
be forced to use dewatering habitats.

For macroinvertebrates, given their reduced mobility, we expected that past hydraulic
conditions would have a stronger influence on their microhabitat selection than for fish. They
should be disfavored by past dewatering and high flow velocity periods. Consequently, we also
expected macroinvertebrate selection for present hydraulics to be different in hydropeaking
rivers than those without hydropeaking. Finally, we expected different responses in benthic and
in hyporheic macroinvertebrate assemblages, with a weaker effect of dewatering and high flow
velocities on hyporheic macroinvertebrates, the hyporheic zone acting as shelter.

105
2.3. Materials and methods

Study reach

The Ain River in eastern France flows from the Jura mountains to the Rhône River, with
a catchment area of 3,630 km2. The study reach is 6 km long and 110 m wide, with a mean
discharge of 103 m3 s−1, an altitude of 230 m and a Strahler order of 5. It includes a diversity of
morphological units (runs, riffles and pools), secondary channels and tributaries. Substrate is
dominated by cobbles and the reach contains a few woody debris but little macrophyte cover.
The reach is located 20 km downstream from the Allement hydropower plant (46° 06′ 44″ N,
5° 25′ 20″ E), the last of a series of five hydropeaking hydropower plants along the river
between Vouglans and the Rhône confluence. According to data measured by DREAL Rhône-
Alpes (between 1960 and 2019) at Pont-d’Ain (7.5 km upstream of the study reach), mean daily
discharge is generally between 13.5 m3 s−1 (exceeded 95% of the time) and 278 m3 s−1 (exceeded
5% of the time). The study reach is subject to frequent sub-daily discharge variations, with base
flow generally between 14 m3 s−1 and (occasionally) 150 m3 s−1 and peak flow generally
between 40 m3 s−1 and 200 m3 s−1 (Figure II.2.1. ). Discharge is more stable during weekends
and low-flow periods (June-September).

Reach hydraulics: the 2D model

A 2D unsteady hydrodynamic model (mapping flow velocities averaged over the

vertical; Rubar 2D; Bazin et al. 2017) was built, based on a digital elevation model obtained
from topographic and bathymetric LiDAR surveys carried out in 2015 and 2016. Model
calibrations used water surface levels derived from the LiDAR surveys at 16 m3 s−1 and from
field surveys at 90 m3 s−1. Calibrations were made so that the absolute difference between the
water level simulated by the model and that measured in the field was less than 10 cm at any
measuring point. We used the model to translate the hourly flow time-series given by the
gauging station of Pont d’Ain, seven kilometers upstream of the study site (from
http://hydro.eaufrance.fr), into maps of hourly hydraulic time-series (flow velocity, water level
and water depth) over the 403,591 nodes of the 2 m-edge rectangular mesh of the model.

106
Figure II.2.1. Location of the Ain River study reach and maps of its flow velocity and
water depth for the maximum and the minimum flow discharge encountered over the
study period.

Sampling

Fish sampling

Fish species abundance was estimated by electrofishing in 507 microhabitats and by

snorkeling in 673 microhabitats. With both protocols, the sampled surface area of microhabitats
was approximately 7 m², fish being attracted (electrofishing) or observable (snorkeling) within
a radius of about 1.5 m (although this distance can depend on fish size; Regis et al., 1981)

Electrofishing was conducted by a team of three or four operators from a motorboat in

deep areas and by wading in shallow areas, approaching the microhabitat as discreetly as
possible to minimize fish escape. An anode was immersed and held steady, and all fish around
the anode were captured with a landing net, identified, measured and released. Microhabitat
positions were recorded by GPS with a precision of 5 m. Electrofishing surveys covered
different flow conditions (Figure II.2.2.) and microhabitat locations were chosen to cover
diverse combinations of past and present hydraulic conditions. In practice, the 2D model was
used to draw maps of fictive past hydraulic conditions corresponding to a typical hydropeaking
week, to help finding these different combinations of past and present hydraulics in the field.
Sampling at different flows also reduced the correlation between present and past hydraulic

107
conditions (e.g., the expected negative correlation between dewatering frequency and water
depth).

Snorkeling observations targeted only large individuals of species that are difficult to
sample by electrofishing (Plichard et al. 2017). Observations were made along six longitudinal
transects, regularly spaced laterally across the reach, in the direction of flow over the entire
study reach, during base flow only to ensure fish identification (~16 m3 s-1). Two persons drifted
along the transect in a downstream direction: (1) an experienced snorkeler detecting and
identifying fish, followed by (2) an hydrospeeder recording the observations and GPS positions
with a waterproof recorder. As reported by others (Chamberland et al. 2014), we did not
frequently observed strong escape behavior while snorkeling and judged that most fish were
observed by snorkelers. Underwater visibility, estimated as the distance at which snorkelers
could see their bright swimming fins, was 5 m. Fish were observed and identified when in a 1.5
m radius around the snorkeler, and several fish individuals were associated to the same GPS
position when observed simultaneously. Therefore, each GPS position in the dataset was
considered as a microhabitat of ~7m2, containing one or several fish. When drifting, snorkelers
observed large fish (estimated length > 25 cm) of eight species (Barbus barbus, Squalius
cephalus, Chondrostoma nasus, Thymallus thymallus, Perca fluviatilis, Esox lucius, Cyprinus
carpio, Salmo trutta fario) and smaller individuals of Thymallus thymallus only (estimated
length < 25 cm).

All snorkeling microhabitats were presence-only (microhabitats with fish presence),

unlike the electrofishing and macroinvertebrate microhabitats. To enable common statistical
processing of all data, we created fictive snorkeling microhabitats with fish absence (“pseudo-
absence microhabitats”; Manly et al. 2002). In practice, for each taxon × microhabitat
combination, we randomly picked n =10 pseudo-absence microhabitats among the nodes of the
2D model network. At this step, we considered only nodes with water depth >30 cm, as
snorkeling was not conducted in shallower habitats. The choice of n = 10 pseudo-absence
microhabitats was intended to provide a global prevalence comparable with that of the
electrofishing data. We tested the sensitivity of results to this choice by repeating the analysis
for n=20.

108
Figure II.2.2. Hourly discharge at Pont d’Ain (7.5 km upstream the study reach) during
the 15 days preceding sampling and during sampling (2018). The vertical lines represent
indicate sampling days at noon.

Macroinvertebrate sampling

We sampled benthic and hyporheic macroinvertebrates in 36 microhabitats, under

different flow conditions (Figure II.2.2.). Similarly to the procedure used for fish, we used the
2D model to draw maps of fictive past hydraulic conditions corresponding to a typical
hydropeaking week, to help finding different combinations of past and present hydraulics in the
field. We chose microhabitats in six cross-sections with low and six with high velocity,
distributed along the reach to have different present and past flow velocity. Within each cross-
section, we selected three points with low (0≤ depth<0.2), medium (0.2≤ depth<0.5) and high
water depth (0.5≤ depth<0.8) (and thus different dewatering frequencies).

We sampled benthic macroinvertebrates with a Hess sampler (surface area 0.05 m2, 250
µm mesh size) and recorded their positions by GPS. Then, within a radius of one meter around
the Hess sampler, hyporheic macroinvertebrates were sampled with a Bou-Rouch pump (Bou
et al. 1967; Dole-Olivier et al. 2014), which extracted 6L of water from 30 cm below the stream
bed (using a planted steel pipe) to catch organisms living in the interstices of substrate particles.
The water was filtered through a sieve with 63 µm mesh size. Macroinvertebrates were
preserved in 96% ethanol and were sorted, counted and identified to the lowest practical
taxonomic level using a microscope in the laboratory (list of references used for
macroinvertebrate identification in Appendix, list II.2.A1).

109
Classification in species guilds

We grouped fish and macroinvertebrate individuals into a few habitat guilds with
comparable habitat selection according to Plichard et al. (2020) and Forcellini et al. (in press),
who synthesized a large part of the available information on fish and macroinvertebrate
microhabitat selection in a wide range of mostly non-hydropeaking European rivers. This
grouping by guilds was used to increase the statistical power of our analyses and the potential
transferability of our results in rivers with other species composition.

Fish at electrofishing points were first grouped by size class, using the taxa definitions
of Plichard et al. (2020). These taxa combined a specific code and one or several size classes
(1: <8 cm; 2: 8–18 cm, 3:18–30 cm; 4: >30 cm; Table II.2.1. ). For example, taxon Lel_cl123
corresponds to Leuciscus Leuciscus (Lel) fish with sizes < 30 cm. We defined two fish guilds
for which we expected different influence of past hydraulics on microhabitat selection. The
‘bank’ guild grouped together taxa selecting shallow and slow-flowing habitats, thus incurring
high dewatering risk. Conversely, the ‘midstream’ guild grouped together species selecting
deep and fast-flowing habitats, typically found in the center channel, which should better
sustain high flow velocity variations. Observed graylings (Thymallus thymallus) were also
assigned to the midstream guild (Mallet et al. 2000). Plichard et al. (2020) summarized the
microhabitat selection by fish taxa for microhabitat velocity V using the statistic AGV_V,
which in short corresponds to the average microhabitat velocity that would be used by the taxa
if velocities between 0 and 0.94 m s-1 were uniformly available in the river. Similarly, AVG_D
summarizes selection for depth D over the depth range 0-3 m. Our midstream guild comprised
the nine taxa with the highest value for the product AVG_V*AVG_D, and our bank guild the
eight taxa with the lowest value. The number of taxa per guild was a compromise between guild
abundance and magnitude of habitat selection.

For macroinvertebrates, we defined two guilds according to selection for bottom shear
stress reported in Forcellini et al. (in press), where shear stress was measured using hemispheres
(FST, Statzner & Müller, 1989) numbered from 0 to 19 (low to high shear stress). As done by
Plichard et al. (2020) for fish, Forcellini et al. (in press) summarized microhabitat selection by
macroinvertebrates, using AVG_V and AVG_FST, which is the equivalent of AVG_V but over
the FST range 0-19. The ‘limnophilic’ guild grouped together taxa selecting low shear stress

110
(AVG_FST < 5), and thus expected to be negatively affected by high flow velocity variations.
Conversely, the ‘rheophilic’ guild corresponded to taxa selecting high shear stress (AVG_FST
> 8).

Present and past microhabitat hydraulics

In each electrofishing and macroinvertebrates sampling microhabitat, we measured the

present water depth (D in m), and present flow velocity (V40 in m s-1; measured at 40% of water
depth from the bottom with an electromagnetic 30 Hz current meter and averaged over 10
seconds, using a Marsh McBirney FLO MATE 2000).

We described past hydraulic conditions using the 2D hydraulic model simulations for
flows observed during the 15 days preceding sampling. The duration of 15 days was consistent
with previous experiments on fish individual learning (Green 1971; Roy et al. 2018) and
previous field studies on hydropeaking (Capra et al., 2017). This 15 days duration also allowed
to account for typical discharge variations observed in the Ain River (Figure II.2.2.).
Nevertheless, we tested the sensitivity of results to our choice of a duration of 15 days, by
repeating our analyses for durations of 5 and 10 days before sampling.

For estimating past hydraulics at a given microhabitat (electrofishing point or

macroinvertebrate sample), we associated the sampled microhabitat to a node of the 2D model
that (1) had comparable water depth, and (2) was as close as possible from the sampled
microhabitat. We judged the primary use of a “depth similarity criteria” as essential for an
optimum estimation of the dewatering frequency of the microhabitat. Indeed, for a given
measurement discharge, water depth and past dewatering frequency are expected to be strongly
correlated. In addition, even if the morphology of the river had changed locally between the 2D
model calibration and the biological survey, we expected that velocity variations for a given
water depth was well represented by the 2D model. In the Ain River, mesohabitats (e.g., riffles,
run, pools) have surface areas of several hundred m2, and the 2D model is expected to well
reflect hydraulic behavior of these mesohabitats, event it can be locally imprecise (Guay et al.
2000). In practice, we selected the closest 2D node that had a modeled depth differing from the
measured depth at the microhabitat by less than 10 cm. For snorkeling observations, depth
measurement was not available and we assigned snorkeling microhabitats to the closest 2D
node.

111
 The past hydraulic conditions during the 15 days preceding sampling were described
with two variables coded as boolean categories:
(1) Duration of dewatering: we considered the microhabitat as “dewatering” if the dewatering
time (depth = 0 cm) over the 15-days period before sampling was >10 hours, and as “non-
dewatering” if ≤10 hours. We tested the sensitivity of results to our choice of a duration of 10
hours, by repeating our analyses for a duration of 1 hour.
(2) Duration of high flow velocity conditions (>1.3 m s-1): We considered that microhabitats had
“high past velocities” if the total duration with velocity >1.3 m s-1 was >10 hours, and “low past-
velocities” if ≤10 hours. We chose a threshold of 1.3 m.s-1 because this value is above the velocity
used by most aquatic organisms (Plichard et al., 2020; Forcellini et al., in press), and is also
frequently reached in many microhabitats of the Ain River. Here again, we tested the sensitivity
of results to our choice of a threshold of 1.3 m.s-1, by repeating our analyses for a threshold of
0.7 m.s-1.

112
Microhabitat selection models by guild

For each dataset separately (electrofishing, snorkeling, benthic and hyporheic

macroinvertebrates), we related guild abundance to past and present microhabitat
characteristics (measured when available, modeled otherwise). Following Plichard et al. (2020)
and Forcellini et al. (in press), we used GLMs with B-spline transformations of the present
hydraulic variable (enabling non-linear responses) and assuming a negative binomial
distribution of abundance (accounting for abundance overdispersion). We used splines with two
degrees of freedom, with a single knot positioned at the median value (Plichard et al., 2020).
To avoid overparameterization and due to limited statistical power, we had to consider models
involving combinations of a single present hydraulic variable (PresHyd = V40 or D) and a single
past hydraulic variable (PastHyd= Dew or HV). Consistently, we looked at the inter-
correlations between present and past hydraulic variables to interpret the results.

For each PresHyd × PastHyd combination, the abundance Yi of a guild in microhabitat i was
assumed to follow a negative binomial distribution of mean μi and dispersion parameter Θ:

Yi~NB(μi,Θ)

Three GLMs of increasing complexity were compared to explain μ_i as a function of

hydraulics:

M0 (no microhabitat selection): log(μi )=β0

M1 (habitat selection with present hydraulic conditions only): log(μi )= β0+ f(PresHydi)

M2 (habitat selection with present and past hydraulic conditions): log(μi)= β0+
f(PresHydi)+ β1 PastHydi

where β0 is the intercept, f() is a spline transformation of the present hydraulic variable
PresHydi, and β1 the coefficient of the past hydraulics effect. In M0, microhabitat variables
have no influence on abundance, in M1 present hydraulic conditions have an influence, and in
M2 there is an additive effect of past hydraulics.

113
 We fitted all models using the R software (R Development Core Team 2018) and the
glm.nb function of the ‘MASS’ package (Venables et al. 2002), combined with spline functions
of the ‘splines’ package. We used likelihood ratio tests comparing M0 vs. M1 to test the
influence of present hydraulics on microhabitat selection, and tests comparing M1 vs. M2 to
assess for an additive effect of past hydraulics. Following Plichard et al. (2020), we used non-
parametric Spearman Rho rank correlation (Spearman 1904) to assess fits; this is an appropriate
statistic for overdispersed data such as our abundance data, based on the correlation between
the ranks of predicted and observed abundance.

Influence of hydropeaking on taxa selection for present hydraulics

For fish and macroinvertebrate taxa (benthic and hyporheic), we computed an average
preferred velocity AGV_V and depth AVG_D as defined by Plichard et al. (2020), but
corresponding to the fits of our M1 model by taxa. We then compared our values with those of
Plichard (2020) for fish and Forcellini et al. (in press) for macroinvertebrates, to estimate how
hydropeaking modifies the selection of present microhabitat hydraulics generally observed in
rivers without hydropeaking. We used only electrofishing data for the fish comparison, because
fish size was not available in snorkeling data, preventing comparison with the size classes of
Plichard et al. (2020).

114
2.4. Results

Biological assemblages

We sampled 3,642 fish of 23 species by electrofishing. The bank guild contained 2,344
individuals and was composed mainly of Phoxinus phoxinus cl12 (N=1,785). The midstream
guild contained 73 individuals, mainly Barbus barbus cl34 (N=27) (Table II.2.1.). We sampled
1,610 fish of 12 species by snorkeling (mainly Thymallus thymallus, N=444; Barbus barbus,
N=437 and Chondrostoma nasus, N=238). In the snorkeling dataset, large Barbus barbus, large
Chondrostoma nasus and all Thymallus thymallus belonged to the midstream guild, with many
more individuals sampled than with electrofishing (N =356) (Table II.2.1.).

We sampled 48,275 macroinvertebrates: 38,196 in benthic samples and 10,079 in

hyporheic samples. The full list of taxa of both datasets are available as Appendix (Table
II.2.A1). The benthic dataset contained 149 taxa and the hyporheic dataset 97 taxa. For the
benthic dataset, the limnophilic guild contained 687 individuals (principally composed of
Chironomini, N=612) and the rheophilic guild contained 2988 individuals (principally
composed of Esolus, N=2,021) (Table II.2.2.). For the hyporheic dataset, the limnophilic guild
contained 170 individuals (principally composed of Chironomini, N=140) and the rheophilic
guild contained 961 individuals (principally composed of Esolus, N=754) (Table II.2.A2.).

115
Table II.2.1. Fish guilds, species, taxa codes, with total abundance and occurrence in microhabitats.

Common
Dataset Guild Family Scientific name Taxa code Abundance Occurence
name
Electrofishing Bank Cyprinidae Phoxinus phoxinus Minnow PhP_cl12 1785 134
N=2344 Cyprinidae Squalius cephalus Chub SqC_cl1 217 47
Cyprinidae Rutilus rutilus Roach RuR_cl1 122 15
Barbatula
Balitoridae Stone loach BaBa_cl12 116 57
barbatula
Cyprinidae Telestes soufia Blageon TeS_cl1 57 17
Cyprinidae Gobio gobio Gudgeon GoG_cl1 30 16
Percidae Perca fluviatilis Perch PER_cl12 11 11
Centrarchidae Lepomis gibbosus Pumpkinseed LeG_cl123 6 5
Midstream Cyprinidae Barbus barbus Barbel BaBu_cl34 27 17
N=73 Thymallus
Salmonidae Grayling ThT_cl2 15 10
thymallus
Chondrostoma
Cyprinidae Nase ChN_cl34 12 7
nasus
Cyprinidae Barbus barbus Barbel BaBu_cl2 8 8
Cyprinidae Alburnus alburnus Bleak AlA_cl23 4 5
Thymallus
Salmonidae Grayling ThT_cl4 3 2
thymallus
Thymallus
Salmonidae Grayling ThT_cl3 2 3
thymallus
Cyprinidae Rutilus rutilus Roach RuR_cl34 2 2
Snorkeling Midstream Thymallus
Salmonidae Grayling ThT_cl234 444 347
N= 1119 thymallus
Cyprinidae Barbus barbus Barbel BaBu_cl34 437 164
Chondrostoma Common
Cyprinidae ChN_cl34 238 43
nasus nase

116
Table II.2.2. Benthic macroinvertebrate guilds, family, genus, taxa codes, with total abundance and
occurrence in microhabitats.

Guild Family Tribe Genus Taxa Code taxa Abundance Occurrence

Limnophilic Chironomidae Chironomini CHIR 612 23
N=687 Sphaeriidae Pisidium spp. PISI 70 15
Tabanidae TABA 3 3
Baetidae Procloeon
Procloeon spp. PBIF 2 1
(non Baetis) bifidum
Rheophilic Elmidae Esolus spp. ESOL 2021 34
N=2988 Cheumatopsyche Cheumatopsyche
Hydropsychidae CLEP 393 14
spp. lepida
Psychomyia
Psychomyiidae Psychomyia spp. PPUS 214 14
pusilla
Elmidae Elmis spp. ELMI 192 17
Baetidae Baetis spp. Baetis fuscatus BFUS 102 14
Hydropsyche Hydropsyche
Hydropsychidae HINC 48 10
spp. incognita
Psychodidae PSYC 7 4
Heptagenia spp. Heptagenia
Heptageniidae HSUL 6 3
sulphurea
Baetis
Baetidae Baetis spp. BVAR 3 1
vardarensis
Hydropsyche Hydropsyche
Hydropsychidae HEXO 1 1
spp. exocellata
Baetidae Baetis spp. Baetis rhodani BRHO 1 1

Present and past hydraulic variables and their intercorrelation (Figure II.2.3)

Present depth and velocity typically ranged between 0-4 m and 0-2 m s-1 in the fish
datasets (Figure II.2.3A.). By contrast, depth was generally shallower than 0.8 m and velocity
lower than 1 m s-1 in the macroinvertebrate dataset, due to sampling limits. Water depth and
flow velocity were poorly correlated (and with different directions) for the fish data
(electrofishing: r=0.13, P<0.05; snorkeling: r=-0.48, P<0.05) (Figure II.2.3.A). The correlation
was greater in the macroinvertebrate dataset (r=0.57; P<0.05).

When calculating past hydraulics using the 2D hydrodynamic model, 76% of the
electrofishing microhabitats and 75% of the macroinvertebrate microhabitats were associated
with a model node within a radius of 5.5 m around the sampled microhabitat. In the
electrofishing dataset, 102/507 microhabitats were dewatering habitats and 255/507 had high
past velocities (Figure II.2.3B.). The snorkeling dataset contained no dewatering microhabitats
and 342/673 microhabitats with high past velocities. A little less than half of the

117
macroinvertebrate samples (32/72) were dewatering and a higher proportion (44/72) had past
low velocities. For fish (electrofishing) and macroinvertebrates, all combinations of dewatering
× high past velocity habitats were sampled (Figure II.2.3B.). Nevertheless, dewatering habitats
were generally those with low past velocity for both fish and macroinvertebrates (Xhi2 test,
P<0.05).

Correlations between present and past hydraulics were limited (r2 < 0.30 in all cases).
The strongest correlation was between flow velocity and high past velocities (Figure II.2.3C.),
especially for the electrofishing and the macroinvertebrate datasets. Dewatering habitats also
often had lower flow velocities and depths (Figure II.2.3D.).

118
Figure II.2.3. Relationship (A) among present hydraulic variables, (B) among past
hydraulic variables and (C,D) between present and past hydraulic variables for
macroinvertebrates ( ), fish sampled by electrofishing ( ), and fish sampled by
snorkeling ( ). Points represent microhabitats.

119
Microhabitat selection models by guilds

As expected from the fish guild definitions, the midstream guild used higher present
velocities and depths than the bank guild, according to both electrofishing and snorkeling data.
The bank guild significantly avoided fast-flowing and deep microhabitats (Table II.2.3., Figure
II.2.4.). Differences in response to flow velocity between guilds, however, were less clear for
macroinvertebrates than for fish (Figure II.2.4A.). Rheophilic macroinvertebrate guild (benthic
and hyporheic) significantly avoided shallow microhabitats (Table II.2.3. and Figure II.2.4A.).
Results for hyporheic macroinvertebrates are presented in Appendix (Table II.2.A3. and Figure
II.2.A1).

Several significant additive effects of past hydraulics were observed (indicated by

asterisks in Figure II.2.4.). Macroinvertebrates responded more significantly to past hydraulics,
all macroinvertebrate guilds being 3.5-15.3 times less abundant in dewatering habitats (Figure
II.2.4A.). Consistently, Spearman Rho values for macroinvertebrate M2 models involving
dewatering (between 0.27 and 0.64, Table II.2.3. and Table II.2.A3.) were much higher than
the corresponding values for M1 models (between 0.04 and 0.35, Table II.2.4. and Table
II.2.A3.). Secondarily, benthic rheophilic macroinvertebrates selected microhabitats with high
past velocities but not hyporheic macroinvertebrates. For fish (electrofishing), responses were
weaker and clearly different, with the bank guild selecting dewatering microhabitats (Figure
II.2.4A.) and, less evidently, the midstream guild avoiding dewatering microhabitats.
Midstream fish of the snorkeling dataset selected microhabitats with high past velocities.

The sensitivity tests concerning the choice of a 15 days duration for calculating past
hydraulics and a 10 hours duration for defining dewatering microhabitats did not modify any
of our results (significance tests in Table II.2.3, Figure II.2.4. and Table II.2.A3.).

By contrast, a threshold of 0.7 m.s-1 instead of 1.3 m.s-1 for defining microhabitats with
high past velocities and, a number of 20 instead of 10 pseudo-absence when building model on
the snorkeling dataset modified some of our (secondary) results on the effects of high past
velocities. With a lower 0.7 m.s-1 threshold, the bank fish guild significantly selected high past
velocities (M2 models) and the midstream fish (electrofishing) guild avoided them (M2 model
with velocity). In addition, the selection of high past velocities by midstream fish (snorkeling)

120
was no longer significant with a 0.7 m.s-1 threshold to define past high velocity or a number of
20 instead of 10 pseudo-absence when building models.

Table II.2.5. Fitting statistics for models M0 (no microhabitat selection), M1 (effects of present
hydraulics) and M2 (additional effect of past hydraulics), including AIC and Spearman Rho
values. M2 models correspond to those shown in Figure II.2.4. Asterisks indicate significance
of likelihood ratio tests comparing M1 vs. M0 or M2 vs. M1, with (*) for P<0.05 and (**) for
P<0.01. In bold: abundance is significantly higher in past high velocity or dewatering habitat;
Underlined: abundance is lower.

Model Hydraulic
Dataset
type variables involved

Electrofishing data Snorkeling data Benthic macroinvertebrate data

Midstream Limnophilic
Bank guild Midstream guild Rheophilic guild
guild guild
Spea. Spea. Spea. Spea. Spea.
AIC AIC AIC AIC AIC
Rho Rho Rho Rho Rho
M0 1780** 392** 4108** 256** 273**
M1 Flow velocity 1670*** 0.41 395** 0.11 4847** 0.16 258 0.09 370** 0.13
Flow velocity +
1650** 0.44 392** 0.14 256* 0.27 353** 0.61
Dewatering
M2
Flow velocity+
1672** 0.40 395** 0.11 4843* 0.15 259 0.09 369* 0.31
Past velocity
M1 Water depth 1734** 0.43 385** 0.10 4933** 0.04 259 0.04 370** 0.35
Water depth+
1725** 0.43 386** 0.13 258* 0.44 358** 0.53
Dewatering
M2
Water depth+
1734** 0.44 384** 0.12 4931** 0.03 259 0.15 366** 0.46
Past velocity

121
Figure II.2.4. Fits of M2 models relating guild abundance to present hydraulics and
including an additive effect of dewatering (A) or past velocity (B). Red and green curves
reflect the effect of past hydraulic variables, when significant. Graphs are shown for the
electrofishing ( ) snorkeling ( ) and benthic macroinvertebrate ( ) data sets. Stars show
the significance of the past hydraulics effect (**) P<0.01; (*) P<0.05.

122
Influence of hydropeaking on taxa selection for present hydraulics

Microhabitat selection for present hydraulics could be compared between the present
study and the literature for nine fish taxa and ten macroinvertebrate taxa. Present velocity and
water depth selection by fish taxa corresponded to those reported in the literature (Figure
II.2.5A.) and had comparable Spearman Rho values (Figure II.2.5B.). By contrast, present
velocity and water depth selection by macroinvertebrate taxa did not match values reported in
the literature (Figure II.2.5C.) and had lower Spearman Rho values here (Figure II.2.5D.).

Figure II.2.5. Comparison of the average preferred velocity (AVG_V, corresponding to

model M1) obtained in this study with data from Plichard et al. (2020) for fish and from
Forcellini et al. (in press) for macroinvertebrates. Axis regressions (full lines) are shown as
well as the y = x lines (dotted lines). For Plichard et al. (2020), AVG_V: y= -0.76x-0.09, R2
=0.76, P<0.01; Spearman Rho: -0.48x+0.33, R2 =0.13, P=0.34. AVG_D: y=0.54x+0.08, R2
=0.21, P=0.22; Spearman Rho: -0.13x+0.26, R2 =0.03, P=0.66. For Forcellini et al. (in press),
AVG_V: y= 0.08x+0.59, R2 =0.02, p-value=0.62; Spearman Rho: 0.56x+0.091, R2 =0.03,
P=0.63, AVG_D: y= -0.10x+0.52, R2 =0.01, P=0.80; Spearman Rho: -0.24x+0.33, R2 =0.12,
P=0.39. GPUL: Gammarus pulex; HYDROP: Hydropsyche contubernalis; TANYT:
Tanytarsini (tribe); LIMN: Limnomysis benedeni; OVIL: Orectochilus villosus; VCRI: Valvata
cristata; PISI: Pisidium (genus); TIPUL: Tipula (genus). See table II.2.1 and II.2.2 for other
taxa names.

123
2.5. Discussion

By studying the combined influence of present and past hydraulics on microhabitat

selection, we improved our mechanistic understanding of the impact of hydropeaking on fish
and macroinvertebrate guilds. We found that both limnophilic and rheophilic
macroinvertebrates were 3.5-15.3 times less abundant in dewatering microhabitats. This
confirms findings such as those of Blinn et al. (1995), who reported a four-fold lower
macroinvertebrate biomass in dewatering habitats of the Colorado River. Consistently with this
result, we also showed that the average velocity and water depth preferences of some
macroinvertebrate taxa differed from those previously established in mainly non-hydropeaking
rivers (Figure II.2.5.). This suggests that macroinvertebrates are not able to maintain habitat
preferences when flow changes too rapidly or frequently: they likely stay at the same location,
undergoing different hydraulic conditions according to flow conditions.

We found no major difference in response to dewatering or to past high velocities

between benthic and hyporheic macroinvertebrates. Thus, in the Ain River the hyporheic areas
does not seem to offer shelter under hydropeaking conditions. We sampled hyporheic
macroinvertebrates at a depth of 30 cm depth in the Ain River it is certainly not deep enough to
reach the permanently water-saturated area.

In contrast to macroinvertebrates, fish followed their preferred habitat preferences

observed in other rivers and were less influenced by past hydraulics. Nevertheless, fish from
the “bank” guild probably stayed along the bank and moved to dewatering microhabitats when
flow increased, in order to remain within their strict range of preferred hydraulic conditions.
This result contrasts with many studies concerning salmonid juveniles, also living near the
banks, and indicating that only a fraction of individuals relocate when flow changes (Shirvell
1994; Kemp et al. 2003; Korman et al. 2009). High velocities in the Ain River and the different
species considered here (mostly cyprinids) may explain these differences. In the Ain River, fish
from the bank guild are probably exposed to a high risk of stranding, particularly because cobble
bars have very low slopes (Tuhtan et al. 2012). It would be interesting to quantify the influence
of horizontal down-ramping rate (water surface retreat rate) on fish stranding near these cobble
bars. The fish “midstream” guild also followed their preferred range of habitat, and tended to

124
avoid dewatering and stressful microhabitats (Reebs 1996; Capra et al. 2017). Although it
concerned other species (large barbel, catfish and chub), the telemetry experiment by Capra et
al. (2017) also reported that fish could avoid dewatering microhabitats. However, due to the
small number (N=73) of fish involved in this present result, this result deserves generalization.

Observed microhabitat selection by fish and macroinvertebrates was not negatively

affected by high past velocities, suggesting a weaker influence of high past velocities compared
to dewatering. In addition, the statistical significance of the weak responses to high past
velocities depended on the threshold chosen for defining high past velocities, indicating
unstable results. However, considering that hydropeaking has occurred since the 1930s in the
Ain River, environmental filtering may have selected only species that can cope with rapid and
frequent variations in hydraulic conditions. Furthermore, our results suggest that high velocities
force fish to move (sticking to their general hydraulic preference) and macroinvertebrates to
find flow-shelter or attach to the substrate (e.g. with silk draglines, tarsal claws, or a muscular
foot; Holomuzki et al. 2000). Moving (fish) or staying in sub-optimal habitat conditions
(macroinvertebrates) may create energetic costs that influence community structure on the long
term and explain reach-scale effects of hydropeaking (Judes et al., 2021). More detailed
observations of behavior at high flows (e.g. by snorkeling or with cameras),
experimental/physiologic experiments in flumes, or additional comparisons between reaches
subjected to different degrees of hydropeaking (Schmutz et al., 2015; Judes et al., 2021) will be
useful to confirm this interpretation.

We identify four major limits and ways of improvement in our study. First, we pooled
microhabitat samples under low and high flow conditions due to a limited statistical power.
However, habitat availability strongly changes according to flow conditions. For example, in
the Ain, 77% of the wetted surface is less than 1 m deep at 14 m3 s-1 versus 33% at 214 m3 s-1.
Such differences in habitat availability could influence species microhabitat selection (Schooley
1994; Arthur et al. 1996). Further sampling could target a more detailed analysis of the effects
of habitat availability. Similarly, improved analyses should also consider the effects of woody
debris, macrophyte cover, tributary proximity and/or thermal refugia. Secondly, discharge
before and during our sampling period combined hydropeaks and artificial releases of
intermediate flows (around 100 m3 s-1, Figure II.2.2.) to compensate for low discharge in the
Rhône River. The duration of intermediate flow may influence observed fish and

125
macroinvertebrates microhabitat selection. Future sampling will need to be implemented during
different hydropeaking regimes, especially for macroinvertebrates sampling, which was
conducted at only three flow conditions here. Thirdly, because our results were little sensitive
to a number of our methodological choices (duration considered for past hydraulics, threshold
used for defining dewatering habitats and high past velocities), additional experiments/data
would be useful to specify the key quantitative drivers of biological responses. Finally, although
the 2D hydrodynamic model allowed us to estimate past hydraulics at each microhabitat, its
accuracy and the methods used for associating field observations with model nodes would
deserve further investigation (see Guay et al., 2000).

In conclusion, our microhabitat-scale study shows a stronger effect of dewatering than

high past velocities on macroinvertebrates, and secondarily fish. Although it failed to identify
critical past velocity thresholds for hydropeaking management, it suggests that fish are forced
to move and macroinvertebrate to hide due to these high past velocities. Comparing our results
with data collected in other rivers and at a larger variety of flows, combined with experimental
studies, should help further identifying the key hydropeaking characteristics to consider in
mitigation policies.

2.6. Acknowledgments

We are grateful for the financial support of EDF– Hydro Alpes (Gérald Ramos) for the
LiDAR and water line surveys. We thank EDF-CIH (Le Bourget du Lac, France) and Dimitri
Lague (Géosciences Rennes, UMR 6118, France) for collaboration in developing the Digital
Elevation Model derived from LiDAR surveys. We gratefully acknowledge all the people who
contributed to the field work.

2.7. Bibliography

Arthur SM, Manly BF, McDonald LL, Garner GW. 1996. Assessing habitat selection when
availability changes. Ecology 77:215–227.

Auer S, Zeiringer B, Führer S, Tonolla D, Schmutz S. 2017. Effects of river bank heterogeneity
and time of day on drift and stranding of juvenile European grayling (Thymallus thymallus L.)

126
caused by hydropeaking. Science of the Total Environment 575:1515–1521.

Bazin, P., Mignot, E., Paquier, A., Bazin, P., Mignot, E., & Paquier, A. 2017. Computing
flooding of crossroads with obstacles using a 2D numerical model. Journal of Hydraulic
Research, 55.

Blinn DW, Shannon JP, Stevens LE, Carder JP. 1995. Consequences of fluctuating discharge
for lotic communities. Journal of the North American Benthological Society 14:233–248.

Bou C, Rouch R. 1967. Un nouveau champ de recherches sur la faune aquatique souterraine.
C. R. Academie hebdomadaires des sciences 265:369–370.

Bruno MC, Siviglia A, Carolli M, Maiolini B. 2013. Multiple drift responses of benthic
invertebrates to interacting hydropeaking and thermopeaking waves. Ecohydrology 6:511–522.

Capra H, Plichard L, Bergé J, Pella H, Ovidio M, Mcneil E, Lamouroux N. 2017. Fish habitat
selection in a large hydropeaking river: strong individual and temporal variations revealed by
telemetry. Science of the Total Environment 578:109–120.

Chamberland JM, Lanthier G, Boisclair D. 2014. Comparison between electrofishing and

snorkeling surveys to describe fish assemblages in Laurentian streams. Environmental
Monitoring and Assessment 186:1837–1846.

Dole-Olivier MJ, Marmonier P, Beffy JL. 1997. Response of invertebrates to lotic disturbance:
Is the hyporheic zone a patchy refugium? Freshwater Biology 37:257–276.

Dole-Olivier MJ, Maazouzi C, Cellot B, Fiers F, Galassi DMP, Claret C, Martin D, Mérigoux
S, Marmonier P. 2014. Assessing invertebrate assemblages in the subsurface zone of stream
sediments (0-15 cm deep) using a hyporheic sampler. Water Resources Research 50:453–465.

Forcellini M, Plichard L, Dolédec S, Mérigoux S, Olivier JM, Cauvy-Fraunié S, Lamouroux N.

in press. Microhabitat selection by macroinvertebrates: generality among rivers and functional
interpretation. Journal of Ecohydraulics.

Green, J. 1971. High Tide Movements and Homing Behaviour of the Tidepool Sculpin
Oligocottus maculosus. Journal of the Fisheries Board of Canada, 28.

Guay JC, Boisclair D, Rioux D, Leclerc M, Lapointe M, & Legendre P 2000. Development and

127
validation of numerical habitat models for juveniles of Atlantic salmon (Salmo salar). Canadian
Journal of Fisheries and Aquatic Sciences 5:2065–2075.

Halleraker JH, Saltveit SJ, Harby A, Arnekleiv J V, Fjeldstad HP, Kohler B. 2003. Factors
influencing stranding of wild juvenile brown trout (Salmo trutta) during rapid and frequent flow
decreases in an artificial stream. River Research and Applications 19:589–603.

Holomuzki J, Biggs BJF. 2000. Taxon-specific responses to high-flow disturbance in streams.

Journal of the North American Benthological Society 19:670–679.

Judes C, Gouraud V, Capra H, Maire A, Barillier A, Lamouroux N. (2021) Consistent but

secondary influence of hydropeaking on stream fish assemblages in space and time. Journal of
Ecohydraulics.

Kemp PS, Gilvear DJ, Armstrong JD. 2003. Do juvenile Atlantic salmon parr track local
changes in water velocity? River Research and Applications 19:569–575.

Kjaerstad G, Arnekleiv J V., Speed JDM, Herland AK. 2018. Effects of hydropeaking on
benthic invertebrate community composition in two central Norwegian rivers. River Research
and Applications 34:218–231.

Korman J, Campana SE. 2009. Effects of Hydropeaking on Nearshore Habitat Use and Growth
of Age-0 Rainbow Trout in a Large Regulated River. Transactions of the American Fisheries
Society 138:78–87.

Lamouroux N, Capra H, Pouilly M, Souchon Y. 1999. Fish habitat preferences in large streams
of southern France. Freshwater Biology 42:673–687.

Manly BF, McDonald L, Thomas D, McDonald TL, Erickson WP. 2002. Resource Selection
by Animals: Statistical Design and Analysis for Field Studies. Springer Netherlands.

Patton BW, Braithwaite VA. 2015. Changing tides: Ecological and historical perspectives on
fish cognition. Wiley Interdisciplinary Reviews: Cognitive Science 6:159–176.

Pert EJ, Erman D. 1994. Habitat use by adult rainbow trout under moderate artificial
fluctuations in flow. Transactions of the American Fisheries Society 123:913–923.

128
Plichard L, Capra H, Mons R, Pella H, Lamouroux N. 2017. Comparing electrofishing and
snorkelling for characterizing fish assemblages over time and space. Canadian Journal of
Fisheries and Aquatic Sciences 74:75–86.

Plichard L, Forcellini M, Le Coarer Y, Capra H, Carrel G, Ecochard R, Lamouroux N. 2020.

Predictive models of fish microhabitat selection in multiple sites accounting for abundance
overdispersion. River Research and Applications 36:1056–1075.

Reebs SG. 1996. Time-place learning in golden shiners (Pisces, Cyprinidae). Behavioural
Processes 36:253–262.

Regis J, Pattee E, Lebreton J. 1981. A new method for evaluating the efficiency of electric
fishing. Archiv für Hydrobiologie 93:68–82.

Roy T, Bhat A. 2018. Divergences in learning and memory among wild zebrafish : Do sex and
body size play a role ? Learn Behavior 46: 124–133.

Saltveit SJ, Halleraker JH, Arnekleiv J V., Harby A. 2001. Field experiments on stranding in
juvenile Atlantic Salmon (Salmo salar) and Brown Trout (Salmo trutta) during rapid flow
decreases caused by hydropeaking. Regulated Rivers: Research & Management 17:609–622.

Schmutz S, Bakken TH, Friedrich T, Greimel F, Harby A, Jungwirth M, Melcher A, Unfer G,

Zeiringer B. 2015. Response of fish communities to hydrological and morphological alterations
in hydropeaking rivers of Austria. River research and applications 31:919–930.

Schooley. 1994. Annual Variation in Habitat Selection : Patterns Concealed by Pooled Data.
The Journal of Wildlife Management 58:367–374.

Shirvell CS. 1994. Effect of changes in streamflow on the microhabitat use and movements of
sympatric juvenile coho salmon (Oncorhynchus kisutch) and chinook salmon (O.tshawytscha)
in a natural stream. Canadian Journal of Fisheries and Aquatic Sciences 51:1644–1652.

Spearman C. 1904. The proof and measurement of association between two things.
International Journal of Epidemiology 15:72–101.

Stubbington R, Dole-Olivier MJ, Galassi DMP, Hogan JP, Wood PJ. 2016. Characterization of
macroinvertebrate communities in the hyporheic zone of river ecosystems reflects the pump-
sampling technique used. PLOS One 11.

129
Tachet P, Richoux P, Bournaud M, Usseglio-Polatera P. 2002. Invertébrés d’eau douce:
systématique, biologique et écologie. CNRS Editions.

Tuhtan JA, Noack M, Wieprecht S. 2012. Estimating stranding risk due to hydropeaking for
juvenile European grayling considering river morphology. KSCE Journal of Civil Engineering
16:197–206.

Venables WN, Ripley BD. 2002. Modern Applied Statistics with S. Springer, fourth edition.
http://www.stats.ox.ac.uk/pub/MASS4

Vila-Martínez N, Caiola N, Ibáñez C, Benejam L, Brucet S. 2019. Normalized abundance

spectra of fish community reflect hydro-peaking on a Mediterranean large river. Ecological
Indicators 97:280–289.

Zolezzi G, Siviglia A, Toffolon M, Maiolini B. 2011. Thermopeaking in Alpine streams: event

characterization and time scales. Ecohydrology 4:564–576

2.8. Appendix

List II.2.A1. List of references used for macroinvertebrate identification.

Ephemeroptera:
Bauernfeind, E., & Humpesch, U. H. (2001). Die Eintagsfliegen Zentraleuropas (Insecta:
Ephemeroptera): Bestimmung und Ökologie. Wien, Verlag des Naturhistorischen
Museums Wien.
Bauernfeind, E., & Soldán, T. (2012). The Mayflies of Europe (Ephemeroptera). Vester
Skerninge, Apollo Books.
Degrange, C., & Sowa, R. (1988). Note complémentaire sur deux espèces de Rhithrogena du
groupe alpestris (Ephemeroptera, Heptageniidae) des Alpes françaises. Polskie Pismo
Entomologiczne, 58, 711–714.
Eiseler, B. (2005). Bildbestimmungsshlüssel für die Eintagsfliegenlarven der deutschen
Mittelgebirge und des Tieflandes. Lauterbornia, 53, 1–112.

130
Soldán, T., & Landa, V. (1999). A key to the Central European species of the genus Rhithrogena
(Ephemeroptera: Heptageniidae). Klapalekiana, 38, 25–37.
Sowa, R. (1984). Contribution à la connaissance des espèces Européennes de Rhithrogena
Eaton (Ephemeroptera, Heptageniidae) avec le rapport particulier aux espèces des Alpes
et des Carpates. International conference on Ephemeroptera, 4, 37–52.
Sowa, R., & Degrange, C. (1987). Taxinomie et répartition des Rhithrogena Eaton du groupe
alpestris (Ephemeroptera: Heptageniidae) des Alpes et des Carpates. Polskie Pismo
Entomologiczne, 57(3), 475–493.
Studemann, D., Landolt, P., Sartori, M., Hefti, D., & Tomka, I. (1992). Ephemeroptera, Insecta
Helvetica, Fauna 9. Lausanne, Schweizerische Entomologische Gesellschaft.
Tomka, I., & Rasch, P. (1993). Beitrag zur Kenntnis der europäischen Rhithrogena-Arten
(Ephemeroptera, Heptageniidae): R. intermedia Metzler, Tomka & Zurwerra, 1987 eine
Art der alpestris-Gruppe sowie ergänzende Beschreibungen zu fünf weiteren
Rhithrogena-Arten. Mitteilungen der Schweizerischen Entomologischen Gesellschaft,
66, 255–281.

Plecoptera:
Aubert, J. (1946). Les Plécoptères de la Suisse Romande. Bulletin de la Société Entomologique Suisse,
20(1), 7–128.
Aubert, J. (1949). Plécoptères helvétiques - Notes morphologiques et systématiques. Bulletin de la
Société Entomologique Suisse, 22(2), 217–236.
Aubert, J. (1959). Plecoptera, Insecta Helvetica, Fauna 1. Genève, Société entomologique Suisse.
Consiglio, C. (1980). Plecotteri (Plecoptera). Guide per il riconoscimento delle specie animali delle
acque interne italiane. Consiglio Nazionale delle Ricerche AQ/1/77, 9, 1.
Lillehammer, A. (1986). Nymphs of the Fennoscandian Nemoura Latreille (Plecoptera: Nemouridae),
with a key to species. Insect Systematics & Evolution, 17(4), 511–519.
https://doi.org/10.1163/187631286X00080
Lillehammer, A. (1988). Stoneflies (Plecoptera) of Fennoscandia and Denmark. Fauna entomologica
scandinavica, 21.
Lubini, V., Knispel, S., & Vincon, G. (2012). Die Steinfliegen der Schweiz: Bestimmung und
Verbreitung / Les plécoptères de Suisse: identification et distribution. Fauna Helvetica 27.
Neuchâtel, CSCF & SEG.

131
Reding, J. P. G. (2020). Les Plécoptères du Massif du Jura : Guide d’identification des larves au genre
et à l’espèce (Identification guide (generic and specific level) to the Plecoptera larvae of the
Jura Mountains). Neuchâtel, independently published, 4–502.
Reding, J. P. G., Launay, B., Le Doaré, J., Ruffoni, A., & Vinçon, G. (2019). Two new species of
Dictyogenus klapálek, 1904 (Plecoptera: Perlodidae) from the Jura mountains of France and
Switzerland, and from the French Vercors and Chartreuse massifs. Illiesia, 15(2), 27–64.
Zwick, P. (2004). Key to the West Palaearctic genera of stoneflies (Plecoptera) in the larval stage.
Limnologica, 34, 315–348. https://doi.org/10.1016/S0075-9511(04)80004-5
Trichoptera:
Coppa, G., Graf, W., & Tachet, H. (2012). A revised description of the larvae of three species of the
Rhyacophila tristis group: Rhyacophila aquitanica, Rhyacophila pubescens and Rhyacophila
tristis (Trichoptera: Rhyacophilidae). Annales de Limnologie - International Journal of
Limnology, 48(2), 215–223. https://doi.org/10.1051/limn/2012014
Eiseler, B. (2010). Taxonomie für die Praxis: Bestimmungshilfen-Makrozoobenthos. LANUV-
Arbeitsblatt, 14(1), 1–184.
Neu, P. J., & Tobias, W. (2004). Die Bestimmung der in Deutschland vorkommenden Hydropsychidae
(Insecta: Trichoptera). Lauterbornia, 51, 1–68.
Urbanic, G., & Waringer, J. (2002). The larva of Beraea dira McLachlan, 1875 (Trichoptera,
Beraeidae). Aquatic Insects, 24(3), 213–217. https://doi.org/10.1076/aqin.24.3.213.8116
Waringer, J., & Graf, W. (2011). Atlas of Central European Trichoptera Larvae. Dinkelscherben, Erik
Mauch Verlag.
Waringer, J., & Graf, W. (2013). Key and bibliography of the genera of European Trichoptera larvae.
Zootaxa, 3640(2), 101–151. https://doi.org/10.11646/zootaxa.3640.2.1
Waringer, J., Vitecek, S., & Graf, W. (2016). Larval morphology and identification of Rhyacophila
meyeri McLachlan 1879 (Trichoptera: Rhyacophilidae). Zootaxa, 4093(4), 559–565.
https://doi.org/10.11646/zootaxa.4093.4.7
Weinzierl, A., Malicky, H., & Waringer, J. (2021). The larva of Rhyacophila albardana McLachlan
1879, including a discriminatory matrix to the Rhyacophila larvae with comb-shaped gills of
Austria, Germany and Switzerland (Rhyacophilidae, Trichoptera). Zootaxa, 4908(1), 133–140.
https://doi.org/10.11646/zootaxa.4908.1.9
Wiberg-Larsen, P. (1979). Revised key to larvae of Beraeidae in NW Europe (Trichoptera). Insect
Systematics & Evolution, 10(2), 112–118.
Diptera:
Mauch, E. (2017). Aquatische Diptera-Larven in Mittel-, Nordwest- und Nordeuropa. Übersicht über
die Formen und ihre Identifikation. Lauterbornia, 83.

132
Nilsson, A. N. (éd.). (1996). Aquatic Insects of North Europe: Ephemeroptera, Plecoptera,
Heteroptera, Neuroptera, Megaloptera, Coleoptera, Trichoptera, Lepidoptera. Stenstrup,
Apollo Books.
Others:
Tachet, H., Richoux, P., Bournaud, M., Dessaix, P., & Pattée, E. (2009). Initiation aux
invertébrés des eaux douces (familles, quelques genres et espèces). Association
Française de Limnologie.
Tachet, H., Richoux, P., Bournaud, M., Usseglio-Polatera, P. (2010). Invertébrés d’eau douce :
systématique, biologie, écologie. Paris : CNRS éditions.
Websites:
Freshwater ecology: http://www.freshwaterecology.info
GBIF: https://www.gbif.org
Opie benthos: http://www.opie-benthos.fr
Perla: http://www.perla.developpement-durable.gouv.fr

133
Table II.2.A1. Abundance per taxa in the benthic and hyporheic area.

Abundance Abundance
on the on the
Taxonomic group Taxa
benthic hyporheic
area area
BASOMMATOPHORA Gyraulus spp. 1 0
BIVALVIA Pisidium spp. 70 28
Sphaeriidae 1 0
BRYOZOA BRYOZOA 1 0
CLADOCERA Cladocera 769 188
COLEOPTERA Esolus spp. 2021 754
Dryopidae 687 46
Elmis spp. 192 19
Oulimnius spp. 192 32
Limnius spp. 168 32
Dryops spp. 167 82
Riolus spp. 158 6
Stenelmis canaliculata 44 66
Bidessus minutissimus 24 18
Hydroporinae 18 8
Bidessus spp. 12 27
Elmidae 12 131
Haliplus spp. 10 1
Hydrophilinae 9 0
Orectochilus villosus 8 1
Brychius elevatus 3 0
Haliplidae 2 0
Laccobius spp. 2 0
Stenelmis spp. 2 0
Peltodytes rotundatus 1 0
Platambus maculatus 1 0
Pomatinus substriatus 0 1
COLLEMBOLA Collembola 16 6
CRUSTACES Malacostraca Gammarus pulex 979 136
Asellidae 411 113
Amphipoda 403 126
Gammarus spp. 306 73
Asellus aquaticus 203 94
Niphargidae 159 399
Crangonyx pseudogracilis 96 80
Gammaridae 63 14
CRUSTACEA Others Cyclopoida 1309 1494
Calanoida 41 0
Harpacticoida 1 3
DIPTERA Tanytarsini 3632 150
Orthocladinae 3093 90

134
 Chironomidae 1500 158
Tanypodinae 1067 668
Chironomini 612 140
Hemerodromiinae 239 8
Antocha vitripennis 221 1
Ceratopogoninae 117 44
Simulium spp. 57 1
Tipulidae 45 2
Rhagionidae 8 0
Psychodidae 7 1
Hexatomini 4 1
Atherix spp. 3 0
Atrichops 3 1
Simuliidae 3 0
Tabanidae 3 0
Clinocerinae 2 0
Dasyheleinae 2 0
Empididae 1 0
Forcipomyinae 1 0
Hexatoma spp. 1 0
Limoniini 0 1
Diamesinae 4 0
Dolichopodidae 1 0
EPHEMEROPTERA Caenis spp. 574 135
Rhithrogena spp. 264 14
Potamanthus luteus 176 46
Baetis spp. 175 3
Heptageniidae 173 14
Baetis fuscatus 102 1
Habroleptoides confusa 72 16
Baetidae 65 1
Heptagenia spp. 17 2
Cloeon spp. 6 0
Heptagenia sulphurea 6 0
Ecdyonurus spp. 5 0
Ecdyonurus aurantiacus 5 0
Ephemera spp. 4 5
Baetis vardarensis 3 0
Procloeon spp. 2 0
Procloeon bifidum 2 2
Alainites muticus 1 0
Baetis rhodani 1 0
Cloeon dipterum 1 0
Choroterpes picteti 1 7
Ephemera danica 1 1
Serratella ignita 1 0
Caenis luctuosa 0 1

135
 Leptophlebiidae 0 2
GASTEROPODA Planorbidae 6 0
Valvata cristata 5 10
Valvata spp. 4 146
Bathyomphalus contortus 2 0
Bithynia spp. 2 0
Planorbarius corneus 1 0
Potamopyrgus antipodarum 1 0
Planorbarius spp. 0 2
HETEROPTERA Micronecta spp. 857 87
Corixinae 2 0
HIRUDINEA Erpobdella spp. 19 1
Erpobdella octoculata 8 1
Glossiphoniidae 2 1
Helobdella spp. 0 1
HYDROZOA Hydrozoa 64 15
NEMATHELMINTHA Nemathelmintha 127 279
NEMATODA Nematoda 13 9
NEMERTEA Prostoma spp. 92 22
ODONATA Onychogomphus spp. 21 3
Orthetrum spp. 2 0
Sympetrinae 2 0
Calopteryx spp. 1 0
Coenagrionidae 1 0
Crocothemis erythraea 1 0
Gomphidae 1 1
Onychogomphus forcipatus 1 0
OLIGOCHAETA Oligochaeta 5061 1033
OSTRACODA Ostracoda 4523 983
PLECOPTERA Leuctra fusca 18 1
Leuctra spp. 14 6
Siphonoperla spp. 7 8
Leuctridae 0 1
PLUMATELLIDA Cristatella mucedo 4 1
TRICHOPTERA Hydroptila spp. 1486 31
Athripsodes spp. 827 195
Setodes argentipunctellus 732 206
Cheumatopsyche lepida 393 183
Hydropsyche spp. 386 36
Hydroptilidae 346 11
Psychomyia pusilla 214 0
Leptoceridae 76 2
Hydroptila vectis 54 1
Hydroptila angulata 49 2
Hydropsyche incognita 48 3
Oecetis spp. 13 1

136
 Psychomyidae 13 0
Rhyacophila spp. 13 0
Hydropsychidae 12 7
Tinodes spp. 9 0
Athripsodes cinereus 7 0
Mystacides spp. 7 0
Polycentropodidae 4 1
TRICHOPTERA 4 1
Mystacides azurea 3 0
Tinodes waeneri 2 0
Ceraclea spp. 1 0
Ecnomus spp. 1 0
Hydropsyche exocellata 1 0
Lepidostoma spp. 1 0
Lepidostoma hirtum 1 0
Lepidostomatidae 1 0
TRICLADIDA Polycelis spp. 77 34
Tricladida 54 34
Polycelis nigra tenuis 23 15
Planaria torva 0 2
TROMBIDIFORMES Hydracarina 1403 1165
TURBELLARIA Dugesia spp. 208 32
Dendrocoelum lacteum 25 12
Planariidae 1 0

Table II.2.A2. Benthic macroinvertebrate guilds, family, genus, taxa codes, with total abundance
and occurrence in microhabitats.

Guild Family Tribe Genus Taxa Code taxa Abundance Occurrence

Limnophilic Chironomidae Chironomini CHIR 140 17
N=170 Sphaeriidae Pisidium spp. PISI 28 10
Baetidae Procloeon
Procloeon spp. PBIF 2 2
(non Baetis) bifidum
Rheophilic Elmidae Esolus spp. ESOL 754 31
N=961 Cheumatopsyc Cheumatopsyche
Hydropsychidae CLEP 183 6
he spp. lepida
Elmidae Elmis spp. ELMI 19 8
Hydropsyche Hydropsyche
Hydropsychidae HINC 3 2
spp. incognita
Psychodidae PSYC 1 1
Baetidae Baetis spp. Baetis fuscatus BFUS 1 1

137
Figure II.2.A1. Fits of M2 models relating guild abundance for hyporheic
macroinvertebrates to present hydraulics and including an additive effect of dewatering
(A) or past velocity (B) for the hyporheic macroinvertebrates data. Red and green curves
reflect the effect of the past hydraulic variables, when significant. Stars show the significance
of the past hydraulics effect (**) P<0.01; (*) P<0.05.

138
Table II.2.A3. Fitting statistics for models M0 (no microhabitat selection), M1 (effects of
present hydraulics) and M2 (additional effect of past hydraulics), including AIC and
Spearman Rho values. M2 models correspond to those shown in Figure 4. Asterisks indicate
significance of likelihood ratio tests comparing M1 vs. M0 or M2 vs. M1, with (*) for P<0.05
and (**) for P<0.01. In bold: abundance is significantly higher in past high velocity or
dewatering habitat; Underlined: abundance is lower.

Model type Hydraulic variables involved Hyporheic macroinvertebrate data

Limnophilic guild Rheophilic guild
Spea.
AIC Spea. Rho AIC
Rho
M0 179** 281**
M1 Flow velocity 180 0.13 277** 0.18

Flow velocity + Dewatering 168** 0.62 261** 0.52

M2
Flow velocity+ Past velocity 181 0.11 281 0.15

M1 Water depth 181 0.14 280 0.19

Water depth+ Dewatering 172** 0.64 267** 0.49

M2
Water depth+ Past velocity 180 0.30 282 0.18

139
3. BATZ ET AL. (SOUMIS) :
Habitat dynamics in hydropeaking rivers: the impact of flow
frequency and duration on patch-scale habitat dynamics.
Bätz N*, Judes C*, Weber C.
*= Authors contributed equally

140
3.1. Abstract

Alterations in the riverine flow regime lead to biodiversity loss and ecosystem degradation
worldwide. However, our mechanistic understanding of how frequently recurring hydropeaks,
as caused by intermittent hydropower production, affect ecological processes is limited. Here,
we link hydropeaking research and disturbance ecology by reviewing the impacts of altered
flow frequency and duration on aquatic habitats across different spatial scales. We characterize
habitats by their temporal persistency and apply novel metrics to quantify flow regime
alterations. We conclude that (i) a time series approach is needed to fully capture ecological
effects due to the multi-event nature of hydropeaking, (ii) additional metrics are required to
quantify the temporal shifts of habitats on the spatio-temporal scale that is relevant for the
organisms under consideration (i.e. patch scale) and that (iii) strategies aiming at enhancing the
ecological resilience of regulated rivers require managers to actively account for habitat
heterogeneity, patchiness and their dynamics.

Keywords : Flow alteration; heterogeneity; shifting habitat mosaic; organism perception;

landscape metrics

141
3.2. Introduction

The flow regime is a fundamental control modulating and maintaining fluvial

hydromorphological and biological processes through space and time (e.g. Kennedy et al. 2016,
Palmer and Ruhi 2019, Poff et al. 1997, Polvi et al. 2020). Flow regime components (see
definitions in Box 1) such as magnitude, frequency, relative duration, timing and rate of change
(Poff et al. 1997) affect the abiotic and biotic structure and functioning of fluvial ecosystems.
Habitat properties as well as their diversity, arrangement and dynamics are largely driven by
the interaction of the flow regime with river reach morphology (e.g. Frissell et al. 1986, Stanford
et al. 2005, Tockner et al. 2000). Accordingly, habitats can be regarded as having a spatial and
a temporal dimension, with temporal dynamics mainly being driven by flow variations and
spatial heterogeneity resulting from the geomorphic diversity of a river reach (e.g. fluvial
landform arrangement). Organisms have evolved and adapted their life history to the local
habitat dynamics over millennia (e.g. Miller et al. 2020, Poff et al. 1997) and developed
strategies to deal with natural disturbances and seasonal changes in spatio-temporal habitat
arrangement and availability (e.g. Hitchman et al. 2018, Winemiller et al. 2010).

Over the past century flow regimes have been increasingly affected by direct human
water exploitation such as hydropower production (e.g. Kennedy et al. 2016, Lange et al. 2019,
Poff and Schmidt 2016, Schmitt et al. 2018). Nilsson et al. (2005) calculated that dams regulate
the flow regime of 83% of the 292 world’s largest rivers (31% strong regulation; 52%
moderate). In addition, a predicted increase in global hydropower capacity of 73% over the next
few decades is expected to decrease the number of free-flowing large rivers from 120 to 25
(Zarfl et al. 2015). Given the rapid development of volatile renewable energy sources (wind
and solar), on-demand or intermittent hydropower production (as opposed to continuous run-
of-river production) is expected to increase to assure power system stability (e.g. Kougias et al.
2019).

When water stored in a reservoir is returned to the river for on-demand hydropower
production, a hydropeaking reach is formed that is characterized by rapid and frequent sub-
daily flow fluctuations (e.g. Bruder et al. 2016). Peak flows can easily be ten times higher than
the daily minimum flow (e.g. Premstaller et al. 2017, Tonolla et al. 2017) and peak frequency
two orders of magnitude larger than in natural systems (Archer and Newson 2002, Greimel et
al. 2016). Depending on river reach morphology, changes in flow and water levels due to
hydropeaking alter habitat dynamics by creating artificial dewatering zones or inducing rapid

142
changes of local habitat conditions (Holzapfel et al. 2017, Parasiewicz 2008, Person 2013,
Schneider et al. 2017, Vanzo et al. 2016). Most native organism are not capable of adapting to
such pronounced dynamics (Palmer and Ruhi 2019), and impacts have been shown to affect the
structure, abundance and diversity of aquatic invertebrates (Kennedy et al. 2016), fish
(Wegscheider et al. 2020) and riparian vegetation (Bejarano et al. 2020b), as well as the overall
food web structure (Holzapfel et al. 2017). In addition, hydropeaking can impact sediment
distribution and mobility (Vericat et al. 2020), water turbidity (Hauer et al. 2019), water
temperature (Vanzo et al. 2016), water gas saturation (Pulg et al. 2016) and river soundscape
(Lumsdon et al. 2018).

Understanding the spatio-temporal link between flow-induced alterations on habitat

dynamics and the structuring and functioning of ecological communities is fundamental to
developing sustainable and effective management strategies for the conservation and restoration
of fluvial systems worldwide (Frissell et al. 1986, Newson and Newson 2000, Palmer and Ruhi
2019, Stanford et al. 2005). Several approaches have therefore been proposed to assess sub-
daily flow alterations (for a recent review, see Bejarano et al. 2020a) and modelling has been
used to derive reach habitat distribution at base and peak flow (i.e. flow magnitude) in order to
assess impacts of hydropeaking on habitat availability (e.g. Holzapfel et al. 2017, Person 2013,
Schneider et al. 2017). These approaches have found their way into application and are
progressively used to develop mitigation measures as required by several legal frameworks
worldwide (for a recent review, see Moreira et al. 2019).

Hydropeaking, however, not only changes flow magnitude but also significantly affects
flow frequency and relative duration (see example in Figure II.3.1). In natural systems, a
correlation between flow frequency and magnitude can be observed (Biggs et al. 2005).
Following flow regulation, flood magnitude can, however, become almost decoupled from
frequency as examples for the Maggia River in Switzerland (Perona et al. 2009), the Ume River
in Sweden (Widén et al. 2021) and the Coalburn catchment in the UK (Archer and Newson
2002) show. In addition, repeated sub-daily hydropeaking can increase flow frequency and
decrease relative duration (Archer and Newson 2002, Clausen and Biggs 1997).

Evidence from disturbance ecology shows that flow frequency and relative duration
significantly matter for the abiotic and biotic integrity of fluvial ecosystems (Chanut et al. 2019,
Halleraker 2001, Robinson 2012), especially when the alterations propagate to the habitats at
reach (Freeman et al. 2001, Parasiewicz 2008) and patch scale (Beisel et al. 2000, Bowen et al.

143
1998, Miller et al. 2020, Palmer et al. 2000, Winemiller et al. 2010). Consequently, approaches
comparing reach-scale habitat availability at base and peak flow neglect (i) the importance of
the temporal dynamics within a single event, i.e. between the two flow extremes; (ii) the
hierarchical temporal complexity inherent in hydropeaking regimes, with single events being
nested within a time series (or trajectory) of multiple events (sensu Ryo et al. 2019); and (iii)
the scale-dependency of effects on, for instance, habitats ranging from reach- to patch-scale.

In this article, we review how alterations in temporal flow regime components (i.e.
frequency and relative duration) propagate down to the habitat dynamics on different spatial
scales (i.e. reach- to patch scale) and suggest new avenues for such quantification for research
and application, with special emphasis on the patch-scale. We focus on hydropeaking since this
has a major anthropogenic impact on the two temporal components of the flow regime (Harby
and Noack 2013). However, the concepts discussed here can easily be transferred or adapted to
other types of flow regime alterations (e.g. residual flow), their management (e.g. e-flows) as
well as to other kind of regimes (e.g. temperature, sediment transport, ice).

144
Figure II.3.1. Comparison of two neighboring mountain rivers in Switzerland, with the
Aare being affected by hydropeaking (station 2019; FOEN 2020), whereas hydrological
alterations for the Lütschine can be neglected (station 2109; FOEN 2020). Both catchments
are comparable in glacier surface, altitudinal gradient and climatic/weather conditions (see
Tonolla et al. 2017). For a more direct comparison, flows of the Lütschine have been multiplied
by a factor of 1.46, corresponding to the size ratio between the two catchments. A shows the
hydrographs for Aare and Lütschine for a few exemplary days in October 2012 and the regime
components (table) calculated for a 24h window (dashed line) for five flow ranges, each
equaling 10m3/s (see y-axis). Frequency represents the number of time flow switches into a
given flow range. Relative duration reflects the average duration of the flow in a specific flow
range (see also definitions in Box 1). B, C and D compare three regime components (frequency,
rate of change, relative duration) averaged over five successive winters (Dec.-Feb. for the years
2012-2017). The whiskers indicate the maximum and minimum, respectively, during the period
analyzed. Note the logarithmic scale of the y-axis. Hydropeaking frequency is up to two orders
of magnitude higher than flow frequency in the natural regime (B). Relative duration (D) is
negatively related to frequency, with low flows showing high relative durations under natural
regime as compared to hydropeaking. Note also the high interannual variation for the relative
duration for flow class <10 m3/s, ranging from stable flows over 90 consecutive days (= the
entire winter) for the natural regime to 1,8 days in the hydropeaking regime. Moreover, the rate
of change appears to be two orders of magnitude higher under hydropeaking than for the natural
regime.

145
3.3. Habitat shifts at the reach scale

Near-natural river reaches are characterized by a high diversity of hydraulic and

geomorphic conditions forming a diverse mosaic of habitats within the reach (Newson and
Newson 2000, Stanford et al. 2005; see definition in Box 1). Morphological structures such as
boulders or instream wood can locally buffer and/or intensify hydraulic conditions (e.g. Yarnell
et al. 2006). A diverse environment offers a high diversity of niches for different organism
groups, species and life stages (Tews et al. 2004) and has been shown to affect demographic
processes such as birth, death, immigration or emigration (Lowe et al. 2006) due to spatio-
temporal variation in ambient characteristics, resource availability or biotic interactions (e.g.
competition, predation; Hitchman et al. 2018).

The habitat mosaic in rivers shifts (Stanford et al. 2005) meaning that the spatial
distribution of habitats is changing over time. Traditionally, the shifting habitat mosaic concept
has been used to represent absolute shifts in the spatial positions of habitat following flood-
induced geomorphic changes caused by erosion and deposition processes. For instance,
sediment movement frequency proved to be a major determinant in benthic community
development (ASCE 1992). The flood-induced shifting habitat mosaic also contributes to
resilience and diversity within a river ecosystem. For instance, the provision of diverse refugia
for differential use by various species and functional groups is a major factor for survival and
recolonization following disturbance (Lake 2000, Winemiller et al. 2010).

However, shifts can also represent temporary fluctuations as they occur during smaller
non-morphogenic floods (or “flow pulses” sensu, Tockner et al. 2000), when the hydraulic
conditions fluctuate back and forth without major geomorphic adjustments. These temporary
shifts are habitat specific, with different habitat types responding differently to change (Palmer
et al. 2000, Weber et al. 2013, Winemiller et al. 2010) as a function of their lateral and vertical
connectivity (Chanut et al. 2019, Tockner et al. 2000). Lancaster and Hildrew (1993), for
instance, identified patches exhibiting low shear stress and current velocity independently of
flow variation, while other habitat patches with the same initial hydraulic conditions
experienced increased shear stress with increasing flow. Consequently, a river reach can be seen
as a continuously shifting mosaic of habitats that follows fluctuations in flow, be they floods or
flow pulses. Hydropeaking has influenced the shifting habitat mosaic by significantly
accelerating shift frequency and reducing the relative duration between the shifts (Archer and
Newson 2002, Clausen and Biggs 1997).

146
 Each shift in the habitat mosaic caused by flow fluctuation is associated with a shift in
habitat availability at the reach scale (see definitions in Box 1). Since each species or life stage
requires a particular set of habitat conditions for e.g. resting, feeding or reproduction, the
availability of suitable habitats directly affects individual fitness (e.g. reproductive success;
Hitchman et al. 2018). For instance, the temporal availability of suitable habitats at reach scale
was correlated with fish species abundance at a given life stage (e.g. juvenile or adult; Stalnaker
et al. 1996). Flow managers therefore often use the relationships between habitat availability
and flow to maximize reach scale habitat availability for a given target fish species (Boavida et
al. 2015, Capra et al. 1995, Parasiewicz 2008). Freeman et al. (2001) showed, however, that the
temporal persistency in habitat availability for a reach of the hydropeaking-impacted
Tallapoosa River (US) was more important for ensuring an abundance of juvenile fish (all
species) than the median habitat availability. Similarly, Hedger et al. (2018) showed via
modelling that stranding-induced mortality (i.e. rapidly dewatered habitats) of juvenile Atlantic
salmon during a single hydropeaking event can be buffered by natural populations dynamics, a
process that is ineffective in the case of frequent hydropeaking (i.e. frequent mosaic shifts).
Similarly, the persistency of spawning grounds and shallow habitats for the early life stages of
fish at the reach scale have been shown to have a positive effect on the population size (Bowen
et al. 1998). For invertebrates, the first hydropeaking event after a longer period without
hydropeaking (i.e. period with high persistency) caused pronounced drift rates whereas the
impacts of subsequent daily hydropeaking events were lower (Bruno et al. 2016, Irvine 1985).

To summarize, the high frequency and low relative duration of hydropeaking events
affects the number of shifts and the related persistency of available habitats at the reach scale,
particularly by creating potential habitat bottlenecks (i.e. limited temporal availability of key
habitats; Bowen et al. 1998). Long lasting habitat bottlenecks can affect population size
(Stalnaker et al. 1996). It is therefore expected that a small number of large daily flow shifts
and associated habitat shifts at the reach scale are less detrimental to the fish community than a
large number of smaller shifts (Senay et al. 2017). However, the cumulative indirect effects on
survival and reproduction and their propagation to the population level are difficult to capture
(Judes et al. 2021).

147
Box 1: Glossary

Flow regime components are driven by river size and catchment characteristics such as
climate, geology, topography and land cover (Poff et al. 1997):
Magnitude: represents a simple measure (e.g. min, max, difference) of flow for a given
time interval (Poff et al. 1997).
Rate of change or flashiness: defines how quickly flow changes over time (Poff et al.
1997).
Frequency: can be expressed as the flood return period (Poff et al. 1997). For flow
regulation like hydropeaking, a definition based on the number of times flow changes
into a given flow magnitude class (e.g. Archer and Newson 2002) seems more
appropriate.
Relative duration: is defined relative to a particular flow event (e.g. “a floodplain may
be inundated for a specific number of days by a ten-year flood” Poff et al. 1997) and is
opposed to the more commonly used cumulative duration which aggregates flows over a
defined time interval (e.g. “the number of days in a year when flow exceeds some
value” Poff et al. 1997).
Timing or predictability: refers to the regularity with which flows of a given magnitude
occur on different time scales (Poff et al. 1997).
Habitat: corresponds to particular ambient conditions typically defined and aggregated on
various spatial scales by means of dominant physico-chemical environmental
characteristics (e.g. Newson and Newson 2000). Each species and life cycle stage requires
a certain array of habitats to satisfy needs such as nutrition, reproduction, resting or refuge
(e.g. Hitchman et al. 2018).
Habit mosaic: comprises a spatial arrangement of habitat types and is commonly defined at
the reach scale (Stanford et al. 2005). The diversity of habitats found within a mosaic is
river-type specific – different river types and watersheds lead to different habitat mosaics
(Brennan et al. 2019) depending on the various drivers operating on different scales (Polvi
et al. 2020).
Reach: defines a river section that is delimited by breaks in channel slope, valley floor
width and side-slopes, riparian vegetation and bank material (Frissell et al. 1986).
Patch: is defined as any geographically fixed area (Pickett and Rogers 1997); we consider
patch sizes ranging between 0-100m2, with the appropriate scale depending on the species

148
 and age class studied and their respective mobility (Pringle et al. 1988, Winemiller et al.
2010).
Shift: a shift occurs when environmental characteristics change at different scale in space
and/or time (Stanford et al. 2005). Different types of shifts can be distinguished:
Shift in habitat availability at the reach scale: changes of total or relative habitat area
within a reach.
Spatial shifts of habitats: spatial movement of particular habitat types in time.
Habitat shifts within patches: changes in habitat type within a patch in time.
Persistency: as opposed to shifts, persistency defines stable habitat characteristics in space
and time on different scales:
Persistency of habitat availability at the reach scale: refers to the temporal persistency
of a sufficient quantity of a habitat type within a reach (see Bovee 1982, Bowen et al.
1998). The higher the persistency at the reach scale, the lower the number of shifts in
habitat availability.
Patch persistency: refers to the temporal persistency of a habitat type within a patch.
The higher the patch persistency, the lower the number of habitat shifts within the patch.

3.4. Habitat shifts at the patch scale

Flow variation not only causes shifts in habitat availability at the reach scale, but also
influences the spatial arrangement of single habitats at the patch scale. As organisms perceive
and respond to the habitat conditions immediately surrounding them, habitat dynamics at the
patch scale can have a strong filtering effect on ecological processes (Hitchman et al. 2018,
Pringle et al. 1988). In a river with a natural flow regime, habitats and related patches are
typically defined based on dominant ambient conditions (e.g. dominant current velocity) since
conditions stay fairly constant over a day, week or season, with floods often regarded as a
punctual disturbance (e.g. Lake 2000, Tockner et al. 2000). In contrast, the flow in
hydropeaking rivers varies so frequently and rapidly (Figure II.3.1) that the assumption of a
patch matching typical or dominant ambient conditions (e.g. Newson and Newson 2000) does
not hold anymore, therefore a decoupling of the geographically defined patch from physically
defined dominant habitat properties must be presumed. It follows that shifts in hydropeaking
rivers can be looked at either from the perspective of a single patch (habitat shifts within
patches) or from the perspective of a given habitat type (spatial shifts of habitats; see also
definitions in Box 1). Figure II.3.2 conceptualizes both types of shifts. The decision as to which

149
perspective to take depends on the species studied and the traits of the related life cycle stage
(Turner et al. 2015, Winemiller et al. 2010).

Habitat shifts within patches represent the conditions that an organism in its respective
life cycle stage would perceive if persisting in a patch (see also example in Box 2). This
perspective is therefore strictly connected to patch persistency (see definitions in Box 1). Egg
laying organisms such as macroinvertebrates and fish require specific conditions at
reproduction sites for successful hatching and survival in the early life stages. For instance, to
minimize the risk of egg-detachment from the substrate, bottom current velocity of patches used
for fish spawning should not exceed 0.7 m/s (Bartoň et al. 2021). Similarly Kennedy et al.
(2016) and Miller et al. (2020) showed that egg viability of macroinvertebrates is reduced by
80-98% if patches are dewatered for more than one consecutive hour (shown for Baetis,
Brachycentrus occidentalis, Hydropsyche occidentalis and Chironomidae). The authors
therefore conclude that the egg life stage represents a bottleneck in the life cycle that requires
particular attention in hydropeaking mitigation. Moreover, the feeding efficiency of sessile
filter-feeding macroinvertebrates (e.g. Simuliidae and Hydropsychidae) can be reduced by
changes in flow conditions since filtering is optimized for a particular range of current velocities
(Brooks and Haeusler 2016). For Chinook salmon larvae (Oncorhynchus tshawytsacha), it has
been shown that dewatering of patches should not last for more than 1 hour to avoid mortal
stranding (Becker et al. 1982). Drift of salmonid larvae can be prevented when current velocities
do not exceed 0.10-0.25 m/s (Heggenes and Traaen 1988). The frequency of inundation of a
patch also influences plant response. Particularly the duration of peaking appears to impact the
germination of sensitive plant species (Bejarano et al. 2020b).

A focus on the spatial shifts of habitats allows the perspective of a mobile organism that
can follow its preferences for particular ambient conditions to be taken (see also example in
Box 2). Some species and life stages have a limited range of mobility (i.e. most invertebrates
and juvenile fish), but are able to follow habitat shifts when individual moving capacity equals
or exceeds the rate of change in habitat conditions. If this requirement is not fulfilled, suitable
habitats, although available at the reach scale, are not accessible. Kjærstad et al. (2018) found
in the River Bævra (Norway) that it took 48 days for Chironomids and the mayfly Baetis
rhodani to recolonize the dewatered zone after hydropeaking. Similarly, Gore (1985) calculated
a recolonization lag of at least 75 days for every 200m from the upstream invertebrate drift
source after disturbance in the Tongue River (US). Blinn et al. (1995) found that in the Colorado

150
River (US) the macroinvertebrate biomass in the permanently submerged area was four times
larger than in the dewatering zone. Winemiller et al. (2010) state that in a resilient system the
dispersal rate of species between patches should be significantly higher than the rate of patch
disturbance, giving some indications of how to identify potential thresholds of anthropogenic
disturbance. The authors stress, however, the importance of disturbance on the scale of the
single patch as this allows competitively inferior species to colonize patches that would
otherwise be dominated by more competitive species (i.e. intermediate disturbance concept).

The spatio-temporal habitat dynamics at patch scale may also affect highly mobile
organisms such as adult fish, for which habitat availability is usually calculated at the reach
scale. When facing changing habitat conditions, highly mobile organisms can choose to stay in
a given patch or to seek preferred habitat conditions, immediately or at some point in time,
when for instance a critical threshold is exceeded. The choice of moving or staying is associated
with costs, such as the risk of predation, loss of energy, the abandonment of feeding territory or
physical damage requiring some macroinvertebrate species to repair or rebuild their cases
(Korman and Campana 2009). There are differences in the willingness to move, with some fish
individuals being reluctant to change position when flow varies (e.g. Kemp et al. 2003, Korman
and Campana 2009). Staying in a non-ideal patch may cause energy losses due to withstanding
the current or due to a decrease in energy intake resulting from reduced feeding efficiency
(Lagarrigue et al. 2002). The energetic costs of moving may as well be difficult to cover for
some organisms during certain seasons as, for instance, fish metabolism slows down during
winter and swimming capacity is considerably reduced to save energy (Cunjak et al. 1998). For
fish, the decision to move therefore depends on the (1) frequency of habitat shift, (2) how much
habitat conditions deviate from the optimal conditions, (3) the proximity to alternative habitats
and (4) the relative duration of the unsuitable habitat condition (Capra et al. 2017). This lead
Capra et al. (2017) to the assumption that fish may opt for a “least constraining” habitat
selection by preferentially using patches with the lowest temporal habitat variation. This
certainly explains why Geist et al. (2008) found that sturgeons (Acipenser transmontamus)
occupy habitats where current velocities remain relatively constant independent of the flow. In
hydropeaking reaches, such habitats are most probably found between the dewatering zone and
the midstream where current velocities are generally too high. The spatio-temporal habitat
dynamics also affect intra- and interspecific relationships (Puffer et al. 2017) and may propagate
to the entire food web. For instance, patches with high habitat shifts may show low

151
macroinvertebrate availability, therefore affecting the food availability for fish using these
patches (Holzapfel et al. 2017).

Summarizing, the patch scale allows the mechanisms behind the shifting habitat mosaic
to be captured by assessing how the interactions between patches affect the behavior of
organisms and their life cycle strategies (Hitchman et al. 2018).

152
Figure II.3.2. (A) shows the habitat dynamics at the patch scale considering a typical day
or week for a river section with a natural versus hydropeaking flow regime and (B) the
related frequency and relative duration of dewatering. Invertebrates ( ) have a low range
of mobility. Under hydropeaking, invertebrates are affected by habitat shifts within patches,
leading to e.g. dewatering and desiccation. The high frequency and low relative duration of
dewatering strongly influences the spatial presence of low-mobile organisms. Fish ( ) are able
to follow changing habitat conditions. Under hydropeaking, frequent spatial shifts of habitats
force fish to change their position (i.e. the patch). Potential stranding may, among other impacts
be controlled by the rate of change of dewatering at the patch scale, which is defined as spatial
shifts of habitats over time.

153
Box 2: Example of habitat dynamics at the patch scale

The example presented here is based on a channelized reach of the Aare River in
Switzerland (for details on the study site, see Tonolla et al. 2017 and Person 2013). The
methodology consists of an adaptation of the analysis by Capra et al. (1995) to the patch-
scale using hydraulic modelling to develop relationships between habitat metrics and flow
characteristics. These metric-flow relationships are then applied to a flow time-series (same
as in Figure II.3.1), which results in a time-series for each individual metric applied. The
metric-specific time-series are used to assess how hydropeaking-induced alteration in flow
frequency and relative duration affect habitat dynamics at the patch scale.

Metrics for habitat shifts within patches (see Box 1 for definition)
 Spatio-temporal habitat dominance is equal to the sum of the sizes of each habitat type
in the patch over time. The metric expresses the average absolute size or average
relative proportion of each habitat type in the patch (in our example reference size at
base flow Q=5m3/s) and represents the habitat conditions an organism would perceive
most frequently.
 Shift index is inspired by the flashiness index of Baker et al. (2004) and represents the
sum of size-changes for each habitat type divided by the patch size at base flow (in our
example Q=5m3/s) and by the length of the study period. The metric captures how
often an organism perceives changes in habitat conditions.
Metrics for spatial shifts of habitats (see Box 1 for definition)
 Habitat expansion-contraction represents the size of a given habitat type at the patch
scale for a given flow over time. The metric is expressed in terms of absolute surface
and reflects the available space an organism has in a particular habitat at time t.
 Habitat displacement quantifies the distance between the centroid of the habitat of a
given type at base flow and the centroid of the most proximate habitat of the same type
at a given flow. The metric represents the distance an organism would have to travel
when following particular habitat conditions.

Figure II.3.3 shows the modelling results, marking the positions of two exemplary patches
(nr. 10 and 20) and the displacement of the associated habitat conditions (habitat type 1 and
2, respectively) when increasing the flow from base to peak-flow.

154
Figure II.3.4 illustrates the habitat shifts within patches for a few exemplary days in
October 2012. Results for six consecutive winters (2012-2017) are then summarized in
Table II.3.1. Both patches 10 and 20 show a wide range of spatio-temporal habitat
dominance under hydropeaking flow while a clear dominance for habitat type 1 and 2,
respectively, can be observed for the natural regime. The shift index under hydropeaking
scores highest for patch 10 and reaches intermediate values for patch 20, while under
natural flow both patches achieve comparatively low values.
Figure II.3.5 illustrates the spatial shifts of habitats for the same period in October as shown
in Figure II.3.4. Summarizing the results from six consecutive winters in Table II.3.2, the
difference between the two regimes is striking; for instance, the habitat is displaced with a
rate of change of two orders of magnitude higher as compared to the natural regime.
Similar patterns can be observed for the expansion-contraction of habitat type 1 in patch 10.
The example highlights the importance of considering the patch scale in hydropeaking
analysis and assessment. Natural patch dynamics may, however, change in season. During
snow/glacier melt or thunderstorms, the natural regime might show higher dynamics.
Although strong evidence exists (see text), the ecological implications of altered patch scale
dynamics on e.g. ecological thresholds, tipping points and resilience when compared to
natural regime dynamics have not yet been well defined and further investigations are
critical for the development of more effective mitigation measures (see conclusions).

155
Figure II.3.3. Results from a 2D model and classification in five simple habitat types
(current velocity classes) for a channelized, hydropeaking-affected reach with artificial
groynes in the Aare River close to Innertkirchen in Switzerland. A: Habitat distribution at
low flow (5 m3/s), with two reference patches marked with the white line. Patch 10 is dominated
by habitat type 1 and patch 20 is dominated by habitat type 2. B: Habitat distribution at peak
flow (45 m3/s), with the most proximate positions of habitat types 1 (for patch 10) and 2 (for
patch 20) marked with a black and white dotted line and their trajectory from the original
patches (white line) at base flow with a fine dotted white line. C: Shows the intermediate steps
between A and B. Note that with increasing flow habitat type 1 contracts within patch 10, while
habitat type 2 is displaced from patch 20 towards the margins.

156
Figure II.3.4. Comparisons of habitat shifts within patches between hydropeaking and
natural regime during a few exemplary days in autumn 2012 for patch 10 (A-B) and patch
20 (C-D). The coloring of the habitat types corresponds to the one described in Figure II.3.3.
The grey line shows the flow over the defined period. The arrows indicate the timing of the
situation as shown in Figure II.3.3. Although the proportions (i.e. magnitudes) of the different
habitat types over the entire week are comparable between the two regimes, shifts under
hydropeaking are much more frequent.

157
Figure II.3. 5. Comparisons of spatial shifts of habitats between hydropeaking and natural
regime during a few exemplary days in autumn 2012. A-B habitat expansion-contraction of
habitat type 1 (patch 10 in Figure II.3.3A). C-D displacement of the habitat type 2 relative to
the position at Q5 (patch 20 in Figure II.3.3A). The grey line shows the flow over the defined
period. The arrows indicate the timing of the situation as shown in Figure II.3.3. Note how
much both metrics fluctuate under hydropeaking as compared to the natural regime.

158
Table II.3.1. Summary statistics for the habitat shifts within patches for hydropeaking
and natural regime over the entire winter season (Dec.-Feb.) for six consecutive years
(2012-2017; see also Figure II.3.1 and Figure II.3.4). The number indicates the average, with
shading intensity reflecting increasing dominance or shift indices, respectively. In brackets, the
minimum and maximum for the period are plotted to give an indication of interannual variation.

Habitat shifts for patch nr. 10 Habitat shifts for patch nr. 20
Dominance Shift index Dominance Shift index
(%) (%/h) (%) (%/h)
Habitat
Hydrop. Natural Hydrop. Natural Hydrop. Natural Hydrop. Natural
type
53.1% 89.4% 14.6% 0.4% 0.3% 0.7% 0.0% 0.0%
1
(+27.6/-22.4) (+9.7/-16.7) (+6.3/-4.0) (+0.2/-0.3) (+0.3/-0.2) (+0.1/-0.2) (+0.0/-0.0) (+0.0/-0.0)
12.4% 3.4% 4.6% 0.1% 41.5% 85.8% 1.1% 0.0%
2
(+5.4/-6.1) (+3.9/-2.5) (+1.5/-1.7) (+0.1/-0.1) (+33.1/-27.4) (+12.0/-20.2) (+0.8/-0.6) (+0.1/-0.0)
19.6% 4.3% 8.0% 0.2% 13.3% 4.1% 0.5% 0.0%
3
(+9.0/-10.4) (+6.7/-4.3) (+2.5/-3.1) (+0.2/-0.2) (+4.9/-5.5) (+3.9/-2.7) (+0.3/-0.2) (+0.0/-0.0)
14.6% 2.8% 6.3% 0.1% 42.7% 9.0% 1.2% 0.0%
4
(+8.0/-10.8) (+5.9/-2.8) (+3.6/-4.4) (+0.2/-0.1) (+22.6/-25.7) (+15.0/-9.0) (+0.5/-0.4) (+0.1/-0.0)
0.3% 0.1% 0.2% 0.0% 2.2% 0.4% 0.2% 0.0%
5
(+0.2/-0.3) (+0.2/-0.0) (+0.1/-0.2) (+0.0/-0.0) (+2.0/-2.1) (+1.5/-0.4) (+0.3/-0.2) (+0.0/-0.0)

Table II.3.2. Summary statistics for the comparisons of spatial shifts of habitats between
hydropeaking and natural regime for the entire winter season (Dec.-Feb.) for six
consecutive years (2012-2017; see also Figure II.3.1 and Figure II.3.3). The number
indicates the average, whereas the minimum and maximum for the period are given in brackets
as an indication of interannual variation.

Expansion-contraction of habitat type 1 Displacement of habitat type 2

Hydropeaking Natural Hydropeaking Natural
294.8 m2/day 8.4 m2/day 15.5 m/day 0.4 m/day
Rate of change
(+127.7/-81.2) (+5.7/-7.5) (+7.0/-4.8) (+0.2/-0.4)
8.1 nr./day 0.2 nr./day 4.7 nr./day 0.1 nr./day
Frequency
(+2.3/-4.6) (+0.3/-0.2) (+2.2/-1.6) (+0.1/-0.1)
Relative 1.0 h 6.1 days 1.5 h 15.7 days
duration (+0.3/-0.2) (+29.9/-6) (+0.9/-0.5) (+74.3/-15.4)

159
3.5. Challenges for patch-scale metrics

Most studies use the magnitude of a single hydropeaking event to quantify ecological
impacts (e.g. Boavida et al. 2015, García et al. 2011, Holzapfel et al. 2017, Person 2013,
Premstaller et al. 2017, Schneider et al. 2017, Tonolla et al. 2017). As ecological impacts of
hydropeaking are also to be expected from repeated hydropeaking (see Figure II.3.1), it is
fundamental to expand the assessment to multiple peaking events (Harby and Noack 2013, Ryo
et al. 2019, Wolkovich et al. 2014). Time series analysis, as suggested for the evaluation of
habitat availability at the reach scale (see Bovee 1982, Capra et al. 1995, Freeman et al. 2001,
Parasiewicz 2008, Spurgeon et al. 2019, Boavida et al. 2015), provides a powerful tool to
evaluate alteration in frequency and relative duration at the patch scale (see example in Box 2).
As indicated above, it is the patch scale that best represents the habitat conditions that are
perceived and exploited by the organism, but its importance is widely ignored in science and
application (Pringle et al. 1988).

As proposed in our example in Box 2, habitat types can be predefined and their spatially
explicit arrangement (i.e. the habitat mosaic) modelled for different flow stages. The habitat
distribution for different flow stages can then be used to assess the dynamics of single patches
under different flow scenarios (i.e. time series of altered versus natural flow), such as the
relative duration and the frequency of habitat shifts within patches (e.g. dewatering) or the drift
risk for invertebrates and fish associated with high current velocities (e.g. > 1 m/s). A time
series approach also allows the patch-specific rates of change to be evaluated and hence to
account for the species and their life stage specific moving capacity.

Methods are needed that integrate results retrieved from single patches into ecologically
meaningful metrics at higher scales given that practical management often operates at the reach
scale (Winemiller et al. 2010). Generally, metrics should be able to quantify the expected effects
resulting from different management interventions and ideally be transparent, easy to calculate
and simple to understand by scientists and non-scientists (Bruder et al. 2016, Pickett and Rogers
1997, Turner et al. 2015). Specifically, the chosen metrics should (i) capture the effects of
altered flow regime components on ecological structures and functions of interest (e.g.
particular species, life stages), (ii) consider the quality and accuracy of the available data (e.g.
spatio-temporal resolution), (iii) account for the comparability of results (e.g. between different
reach morphologies, catchments), (iv) avoid correlations between metrics and (v) maximize
variance and range (Pringle et al. 1988, Turner et al. 2015). Moreover, summary statistics for

160
patch dynamics at the reach scale should not only consider the average of a metric, but also its
variance (e.g. coefficient of variation, percentiles, range) given that variance is the emergent
property of patch dynamics (Palmer et al. 1997, Pringle et al. 1988, Winemiller et al. 2010).

Over the past two decades landscape ecologists have explored how to quantify
landscape patterns on different scales and how to relate these patterns to ecological processes
(Pickett and Rogers 1997, Turner et al. 2015). This has led to a myriad of landscape metrics
that can broadly be classified in addressing patch shape, size, configuration, juxtaposition,
diversity and heterogeneity, density, connectivity, duration and formation (Pickett and Rogers
1997, Pringle et al. 1988, Turner et al. 2015). In aquatic ecology, landscape metrics have mostly
been used for the assessment of habitat patch heterogeneity (Cooper et al. 1997, Winemiller et
al. 2010). Cooper et al. (1997), for instance, experimentally proved that algae patch size and
arrangement (fractal dimension) affect the density of invertebrate grazers. Beisel et al. (2000)
showed that patch heterogeneity in terms of composition (number and proportion of patches)
and configuration (spatial arrangement, shape, and contrasts between neighboring patches)
affect invertebrate community structure in the Mortagne River, France. Similarly, Palmer et al.
(2000) revealed that predictive power in the regression model for Chironomid and Copepod
abundance could be significantly increased if the spatial distribution of patches (mean and
standard deviation of patch size, modal patch size, perimeter-area ratio, patch shape,
arrangement) was added. A field study by Hitchman et al. (2018) on the Upper Neosho River,
US, indicated a relationship between fish species richness and patch type, size, number,
diversity and abundance. Some further examples exist (e.g. Brown 2003, Newton et al. 2008,
Yarnell et al. 2006), but the integration of landscape metrics into practical river management is
still rare (e.g. Jacobson et al. 2009, Remo et al. 2013, Rohde et al. 2005).

3.6. Conclusions

In this article, we showed how hydropeaking induced alterations in the temporal flow
regime components (i.e. frequency and relative duration) translate to impacts on the habitat
dynamics at different spatial scales (i.e. reach to patch scale; see also Figure II.3.6). Four major
conclusions can be drawn:

1. Neglected hydrological alterations: Most studies and methodological approaches have

focused on the magnitude of single events, i.e. on individual hydropeaking events.
Accordingly, important aspects of regime alteration emerging from frequent peaking

161
 events, such as fundamental changes in flow frequency and relative duration, have been
widely neglected in both science and practical management.

2. Mismatch in habitat scales: Most studies assessing the effects of hydropeaking assess
habitat availability at the reach scale and rarely address alterations in habitat dynamics
(i.e. shifts). We argue that it is the temporal dynamics at the patch scale that matters,
this being the scale best representing the perception of an organism living in and moving
between aquatic habitats and patches.

3. Underestimation of habitat dynamics: Habitat dynamics at the patch scale considering

multiple hydropeaking events (i.e. time series) cannot be described by average habitat
characteristics only. Particularly, frequent peaking causes habitat shifts within patches
(= shifts of habitat type in a geographically fixed location) and spatial shifts of habitats
(= spatial movement of particular ambient condition).

4. Metrics are needed: the extent to which habitat shifts within patches and/or spatial shifts
of habitats need to be evaluated for impact assessment depends on the required
persistency and degree of mobility of a species and its life stage. Preliminary metrics,
as presented in Box 2 for both types of shifts, show differences exceeding an order of
magnitude between unregulated and hydropeaking rivers. However, additional sensitive
metrics need to be defined that are able to capture altered patch dynamics.

162
Figure II.3.6. Conceptual model summarizing the effect of altered flow regime
components on the habitats at reach and patch scale.

3.7. Perspectives

The major conclusions listed above suggest several needs for action for both research and
practice, which we have structured around four strongly interrelated points (see also Table II.3.3
for detailed questions):

 Assess ecological effects: In the example (Box 2), we show that shifts at the patch scale
in hydropeaking rivers can exceed those in unregulated rivers by an order of magnitude.
The literature reviewed suggests that altered habitat dynamics may propagate down to
the biotic level by affecting the survival of more sessile species and life stages (e.g.
invertebrate eggs; Miller et al. 2020) as well as the energy budget of mobile species and
life stages (e.g. adult fish; Capra et al. 2017). However, studies are needed to test how
organisms react to altered dynamics (i.e. increase in shifts) at the patch scale due to
anthropogenic changes in flow frequency and relative duration (i.e. multi event).
Particularly empirical approaches, be they experimental or observational, are needed to
understand the causal relationships of short, medium and long-term ecological
responses to such alterations (see also Winemiller et al. 2010). This implies the
identification of ecologically critical responses to habitat dynamics at the patch scale
(e.g. persistency) and the definition of how these responses translate to the reach scale
and beyond.

163
 Account for spatio-temporal variation: When looking at time series, the quantification
of variance becomes essential (Palmer et al. 1997). Different species and life stages
require different degrees of patch persistency, with requirements varying over days,
weeks to seasons (i.e. low number of habitat shifts within patches during a sensitive
period). On the other hand, natural disturbance at the scale of a single patch is essential
to rejuvenate habitat characteristics (e.g. avoid clogging; Hauer et al. 2019) and/or to
allow competitively inferior species to (re-)colonize patches otherwise dominated by
more competitive species (i.e. intermediate disturbance concept; Winemiller et al.
2010). It follows that different patches (as a function of river morphology) show
different habitat dynamics (e.g. persistency; Lancaster and Hildrew 1993) that may vary
in time as a function of the flow regime. Approaches are needed that allow such
variability to be captured and shifts for natural and hydrologically altered river systems
to be compared.

 Develop management options: The patch scale allows a process-oriented approach that
represents what organisms perceive when living and interacting with their environment
(Pringle et al. 1988). Managers taking such a patch perspective therefore have the
advantage of i) making the management goals explicit by allowing a spatio-temporal
evaluation, ii) finding trade-offs for habitat requirements at the patch scale for different
target species (e.g. persistency) and iii) planning of spatio-temporally explicit
management schemes affecting the dynamics and structure of the habitat mosaic
(Hitchman et al. 2018, Pickett and Rogers 1997, Pringle et al. 1988). Moreover, a patch
scale approach will facilitate a quantitative comparison within and between river
systems (Pringle et al. 1988). Although landscape ecologists have, for instance,
extensive experience quantifying the shapes, sizes, configurations, diversity, density
and connectivity of patches, a set of ecologically meaningful assessment metrics that
can be used by river managers is missing. Research and collaborative learning (Weber
et al. 2017) between research and practice is needed to develop management approaches
that account for the patch scale.

 Explore transferability: We focused our article on the effects of hydropeaking since this
an anthropogenic impact with significant influence on the temporal flow regime
components (i.e. frequency and relative duration). However, we consider our findings
to be transferable to other types of flow regime alterations. For instance, rivers affected

164
 by residual flow may experience the opposite hydrological extremes if compared to
hydropeaking (very stable versus very dynamic daily flows). The concepts presented
here could therefore be used by mangers to design more effective mitigation measures
(e.g. E-flows). In addition, other regimes (e.g. temperature, sediment transport, ice) may
influence habitat quality, and research is needed to assess to which extent the concepts
presented here can be applied.

We are convinced that addressing the questions presented here will help promote a more
comprehensive view of hydropeaking effects and lead to a management approach targeting on
the resilience of impacted river systems.

Table II.3.3. Open questions for future research and practice to better account for the
effects of temporal flow regime components (i.e. frequency and relative duration) on
ecological processes on different spatial scales (i.e. reach to patch scale) in hydropeaking
rivers.

Assess ecological effects:

 How much do spatial shifts of habitats affect the fitness of mobile species? How do
habitat shifts within patches affect the fitness of sessile species?
 Do organisms adjust their behavior or physiology to compensate for altered habitat
dynamics?
 Which factors trigger the movement of organisms during pulse and ramp disturbances
(e.g. refuge seeking)?
 How do altered habitat dynamics affect biological interactions (e.g. intra- or
interspecific competition)?
 How do arrangement and connectivity between patches affect the composition and
diversity of aquatic communities?
 What functional characteristics (traits) of organisms should be considered for
assessing the ecological effects of alterations in flow frequency and relative duration
resulting from hydropeaking?

Account for spatio-temporal variation:

 How heterogeneous is patch persistency within a reach (i.e. patches that are naturally
not persistent versus patches that are persistent)?

165
  Which spatio-temporal factors drive patch persistency (e.g. season, morphology,
regime)?How much disturbance (with and without sediment transport) is necessary
to ensure patch species abundance and diversity?

Develop management options:

 Which metrics from landscape ecology best capture the patch scale habitat dynamics
for river systems? How can these metrics be used for practical river management?
 How can habitat dynamics at the patch scale be summarized and made available for
reach-scale management?
 What are the minimum requirements (e.g. resolution, precision, frequency) for
hydrological and morphological data in hydropeaking rivers to run time series
analyses at the patch-scale?
 What measurements should be taken and what metrics be quantified for monitoring
and evaluation of hydropeaking mitigation?
 From a management perspective, is it possible to increase the persistent habitat
availability through small-scale morphological measures?

Explore transferability:
 Are the dynamics observed in residual flow reaches the exact inverse of those for
hydropeaking rivers?
 To what extent can the concepts presented here be applied or adapted to turbidity,
temperature and sound peaking caused by hydropeaking?
 Can we use the concepts presented here to assess the ecological impacts of
morphogenic flows (i.e. flows causing sediment movement and turnover)?
 How are the temporal components of the hydrological regime (i.e. flow frequency,
relative duration) affected by land-use alteration in the catchment or by changes
resulting from climate warming?

3.8. Acknowledgments

The authors particularly thank Lorenzo Gorla, Herve Capra, Maria Alp, Davide Vanzo,
Jan Baumgartner and Esther Leitgeb for early discussions on habitat dynamics in hydropeaking
rivers. Feedback from Nathalie Friese on an earlier version of the manuscript was greatly
appreciated. We are grateful for financial support from the Swiss Federal Office for the

166
Environment (FOEN) for Nico Bätz and from Électricité de France (EDF) for Clarisse Judes.
We thank Jan Baumgartner, Benjamin Berger and Steffen Schweizer (Kraftwerke Oberhasli
AG) for providing data for the example.

3.9. Bibliography

Archer D, Newson M. 2002. The use of indices of flow variability in assessing the hydrological
and instream habitat impacts of upland afforestation and drainage. Journal of Hydrology 268:
244-258.

ASCE (Task Committee on Sediment Transport and Aquatic Habitats) 1992. Sediment and
Aquatic Habitat in River Systems. Journal of Hydraulic Engineering 118: 669-687.

Baker DB, Richards RP, Loftus TT, Kramer JW. 2004. A new flashiness index: characteristics
and applications to midwestern rivers and streams. JAWRA Journal of the American Water
Resources Association 40: 503-522.

Bartoň D, Bretón F, Blabolil P, Souza AT, Vejřík L, Sajdlová Z, Kolařík T, Kubečka J, Šmejkal
M. 2021. Effects of hydropeaking on the attached eggs of a rheophilic cyprinid species.
Ecohydrology: e2280.

Becker CD, Neitzel DA, Fickeisen DH. 1982. Effects of dewatering on chinook salmon redds:
tolerance of four developmental phases to daily dewaterings. Transactions of the American
Fisheries Society 111: 624-637.

Beisel J-N, Usseglio-Polatera P, Moreteau J-C. 2000. The spatial heterogeneity of a river
bottom: a key factor determining macroinvertebrate communities. Hydrobiologia 422: 163-171.

Bejarano MD, García-Palacios JH, Sordo-Ward A, Garrote L, Nilsson C. 2020a. A new tool for
assessing environmental impacts of altering short-term flow and water level regimes. Water 12:
2913.

Bejarano MD, Sordo-Ward Á, Alonso C, Jansson R, Nilsson C. 2020b. Hydropeaking affects

germination and establishment of riverbank vegetation. Ecological Applications 30: e02076.

167
Biggs BJ, Nikora VI, Snelder TH. 2005. Linking scales of flow variability to lotic ecosystem
structure and function. River Research and Applications 21: 283-298.

Blinn W, Shannon JP, Stevens LE, Carder JP. 1995. Consequences of fluctuating discharge for
lotic communities. Journal of the North American Benthological Society 14: 233-248.

Boavida I, Santos JM, Ferreira T, Pinheiro A. 2015. Barbel habitat alterations due to
hydropeaking. Journal of Hydro-environment Research 9: 237-247.

Bovee KD. 1982. A guide to stream habitat analysis using the Instream Flow Incremental
Methodology. Instream Flow Information Paper 12. Fish and Wildlife Service Environmental
Protection Agency / Soil Conservation Service, Geological Survey, Office of Biological
Services. FWS/OBS - 82/26. 248 pp.

Bowen ZH, Freeman MC, Bovee KD. 1998. Evaluation of generalized habitat criteria for
assessing impacts of altered flow regimes on warmwater fishes. Transactions of the American
Fisheries Society 127: 455-468.

Brennan SR, Schindler DE, Cline TJ, Walsworth TE, Buck G, Fernandez DP. 2019. Shifting
habitat mosaics and fish production across river basins. Science 364: 783-786.

Brooks AJ, Haeusler T. 2016. Invertebrate responses to flow: trait-velocity relationships during
low and moderate flows. Hydrobiologia 773: 23-34.

Brown BL. 2003. Spatial heterogeneity reduces temporal variability in stream insect
communities. Ecology Letters 6: 316-325.

Bruder A, Tonolla D, Schweizer SP, Vollenweider S, Langhans SD, Wüest A. 2016. A

conceptual framework for hydropeaking mitigation. Science of The Total Environment 568:
1204-1212.

Bruno MC, Cashman MJ, Maiolini B, Biffi S, Zolezzi G. 2016. Responses of benthic
invertebrates to repeated hydropeaking in semi‐natural flume simulations. Ecohydrology 9: 68-
82.

Capra H, Breil P, Souchon Y. 1995. A new tool to interpret magnitude and duration of fish
habitat variations. Regulated Rivers: Research & Management 10: 281-289.

168
Capra H, Plichard L, Berge J, Pella H, Ovidio M, McNeil E, Lamouroux N. 2017. Fish habitat
selection in a large hydropeaking river: strong individual and temporal variations revealed by
telemetry. Science of The Total Environment 578: 109-120.

Chanut PCM, Datry T, Gabbud C, Robinson CT. 2019. Direct and indirect effects of flood
regime on macroinvertebrate assemblages in a floodplain riverscape. Ecohydrology 12: e2095.

Clausen B, Biggs B. 1997. Relationships between benthic biota and hydrological indices in
New Zealand streams. Freshwater Biology 38: 327-342.

Cooper SD, Barmuta L, Sarnelle O, Kratz K, Diehl S. 1997. Quantifying Spatial Heterogeneity
in Streams. Journal of the North American Benthological Society 16: 174-188.

Cunjak RA, Prowse TD, Parrish DL. 1998. Atlantic salmon (Salmo salar) in winter: "the season
of parr discontent"? Canadian Journal of Fisheries and Aquatic Sciences 55: 161-180.

FOEN (Federal Office for the Environment) 2020. Environmental data available under
www.hydrodaten.admin.ch.

Freeman MC, Bowen ZH, Bovee KD, Irwin ER. 2001. Flow and habitat effects on juvenile fish
abundance in natural and altered flow regimes. Ecological Applications 11: 179-190.

Frissell CA, Liss WJ, Warren CE, Hurley MD. 1986. A hierarchical framework for stream
habitat classification: viewing streams in a watershed context. Environmental Management 10:
199-214.

García A, Jorde K, Habit E, Caamaño D, Parra O. 2011. Downstream environmental effects of

dam operations: changes in habitat quality for native fish species. River Research and
Applications 27: 312-327.

Geist DR, Arntzen EV, Murray CJ, McGrath KE, Bott Y-J, Hanrahan TP. 2008. Influence of
rver level on temperature and hydraulic gradients in chum and fall chinook salmon spawning
areas downstream of Bonneville Dam, Columbia River. North American Journal of Fisheries
Management 28: 30-41.

Gore JA. 1985. Mechanisms of colonization and habitat enhancement for benthic
macroinvertebrates in restored river channels. Pages 81-101 in Gore JA, ed. The Restoration of
Rivers and Streams. Theories and Experience.

169
Greimel F, Zeiringer B, Höller N, Grün B, Godina R, Schmutz S. 2016. A method to detect and
characterize sub-daily flow fluctuations. Hydrological Processes 30: 2063-2078.

Halleraker JH. 2001. The environmental impacts of peaking at hydropower plants. Xergi
(Trondheim) 41: 5.

Harby A, Noack M. 2013. Rapid flow fluctuations and impacts on fish and the aquatic
ecosystem. Pages 323-335 in Maddock I, Harby A, Kemp P, Wood P, eds. Ecohydraulics: an
Integrated Approach.

Hauer C, Holzapfel P, Tonolla D, Habersack H, Zolezzi G. 2019. In situ measurements of fine

sediment infiltration (FSI) in gravel-bed rivers with a hydropeaking flow regime. Earth Surface
Processes and Landforms 44: 433-448.

Hedger RD, Sauterleute J, Sundt‐Hansen LE, Forseth T, Ugedal O, Diserud OH, Bakken TH.
2018. Modelling the effect of hydropeaking‐induced stranding mortality on Atlantic salmon
population abundance. Ecohydrology 11: e1960.

Heggenes J, Traaen T. 1988. Downstream migration and critical water velocities in stream
channels for fry of four salmonid species. Journal of Fish Biology 32: 717-727.

Hitchman SM, Mather ME, Smith JM, Fencl JS. 2018. Identifying keystone habitats with a
mosaic approach can improve biodiversity conservation in disturbed ecosystems. Global
Change Biology 24: 308-321.

Holzapfel P, Leitner P, Habersack H, Graf W, Hauer C. 2017. Evaluation of hydropeaking

impacts on the food web in alpine streams based on modelling of fish- and macroinvertebrate
habitats. Science of The Total Environment 575: 1489-1502.

Irvine J. 1985. Effects of successive flow perturbations on stream invertebrates. Canadian

Journal of Fisheries and Aquatic Sciences 42: 1922-1927.

Jacobson RB, Johnson HEI, Dietsch BJ. 2009. Hydrodynamic Simulations of Physical Aquatic
Habitat Availability for Pallid Sturgeon in the Lower Missouri River, at Yankton, South Dakota,
Kenslers Bend, Nebraska, Little Sioux, Iowa, and Miami, Missouri, 2006–07. US Geological
Survey. Report no.

170
Judes C, Gouraud V, Capra H, Maire A, Barillier A, Lamouroux N. 2021. Consistent but
secondary influence of hydropeaking on stream fish assemblages in space and time. Journal of
Ecohydraulics.

Kemp PS, Gilvear DJ, Armstrong JD. 2003. Do juvenile Atlantic salmon parr track local
changes in water velocity? River Research and Applications 19: 569-575.

Kennedy TA, Muehlbauer JD, Yackulic CB, Lytle DA, Miller SW, Dibble KL, Kortenhoeven
EW, Metcalfe AN, Baxter CV. 2016. Flow management for hydropower extirpates aquatic
insects, undermining river food webs. BioScience 66: 561-575.

Kjærstad G, Arnekleiv JV, Speed JDM, Herland AK. 2018. Effects of hydropeaking on benthic
invertebrate community composition in two central Norwegian rivers. River Research and
Applications 34: 218-231.

Korman J, Campana SE. 2009. Effects of hydropeaking on nearshore habitat use and growth of
age-0 rainbow trout in a large regulated river. Transactions of the American Fisheries Society
138: 76-87.

Kougias I, et al. 2019. Analysis of emerging technologies in the hydropower sector. Renewable
and Sustainable Energy Reviews 113: 109257.

Lagarrigue T, Céréghino R, Lim P, Reyes-Marchant P, Chappaz R, Lavandier P, Belaud A.

2002. Diel and seasonal variations in brown trout (Salmo trutta) feeding patterns and
relationship with invertebrate drift under natural and hydropeaking conditions in a mountain
stream. Aquat. Living Resour. 15: 129-137.

Lake PS. 2000. Disturbance, patchiness, and diversity in streams. Journal of the North
American Benthological Society 19: 573-592.

Lancaster J, Hildrew AG. 1993. Flow Refugia and the Microdistribution of Lotic
Macroinvertebrates. Journal of the North American Benthological Society 12: 385-393.

Lange K, et al. 2019. Small hydropower goes unchecked. Frontiers in Ecology and the
Environment 17: 256-258.

Lowe WH, Likens GE, Power ME. 2006. Linking scales in stream ecology. BioScience 56:
591-597.

171
Lumsdon AE, Artamonov I, Bruno MC, Righetti M, Tockner K, Tonolla D, Zarfl C. 2018.
Soundpeaking – hydropeaking induced changes in river soundscapes. River Research and
Applications 34: 3-12.

Miller SW, Schroer M, Fleri JR, Kennedy TA. 2020. Macroinvertebrate oviposition habitat
selectivity and egg-mass desiccation tolerances: Implications for population dynamics in large
regulated rivers. Freshwater Science 39: 584-599.

Moreira M, Hayes DS, Boavida I, Schletterer M, Schmutz S, Pinheiro A. 2019. Ecologically-

based criteria for hydropeaking mitigation: a review. Science of The Total Environment 657:
1508-1522.

Newson MD, Newson CL. 2000. Geomorphology, ecology and river channel habitat: mesoscale
approaches to basin-scale challenges. Progress in Physical Geography: Earth and Environment
24: 195-217.

Newton TJ, Woolnough DA, Strayer DL. 2008. Using landscape ecology to understand and
manage freshwater mussel populations. Journal of the North American Benthological Society
27: 424-439.

Nilsson C, Reidy CA, Dynesius M, Revenga C. 2005. Fragmentation and flow regulation of the
world's large river systems. Science 308: 405-408.

Palmer MA, Hakenkamp CC, Nelson-Baker K. 1997. Ecological heterogeneity in streams: why
variance matters. Journal of the North American Benthological Society 16: 189-202.

Palmer MA, Ruhi A. 2019. Linkages between flow regime, biota, and ecosystem processes:
Implications for river restoration. Science 365: eaaw2087.

Palmer MA, Swan CM, Nelson K, Silver P, Alvestad R. 2000. Streambed landscapes: evidence
that stream invertebrates respond to the type and spatial arrangement of patches. Landscape
Ecology 15: 563-576.

Parasiewicz P. 2008. Habitat time series analysis to define flow augmentation strategy for the
Quinebaug River, Connecticut and Massachusetts, USA. River Research and Applications 24:
439-452.

172
Perona P, Molnar P, Savina M, Burlando P. 2009. An observation‐based stochastic model for
sediment and vegetation dynamics in the floodplain of an Alpine braided river. Water Resources
Research 45.

Person E. 2013. Impact of Hydropeaking on Fish and their Habitat. PhD thesis. École
Politechnique Fédérale de Lausanne (EPFL).

Pickett STA, Rogers KH. 1997. Patch Dynamics: The Transformation of Landscape Structure
and Function Pages 101-127 in Bissonette JA, ed. Wildlife and Landscape Ecology: Effects of
Pattern and Scale

Poff NL, Allan JD, Bain MB, Karr JR, Prestegaard KL, Richter BD, Sparks RE, Stromberg JC.
1997. The natural flow regime. BioScience 47: 769-784.

Poff NL, Schmidt JC. 2016. How dams can go with the flow. Science 353: 1099-1100.

Polvi LE, Lind L, Persson H, Miranda-Melo A, Pilotto F, Su X, Nilsson C. 2020. Facets and
scales in river restoration: nestedness and interdependence of hydrological, geomorphic,
ecological, and biogeochemical processes. Journal of Environmental Management 265:
110288.

Premstaller G, Cavedon V, Pisaturo GR, Schweizer S, Adami V, Righetti M. 2017.

Hydropeaking mitigation project on a multi-purpose hydro-scheme on Valsura River in South
Tyrol/Italy. Science of The Total Environment 574: 642-653.

Pringle CM, Naiman RJ, Bretschko G, Karr JR, Oswood MW, Webster JR, Welcomme RL,
Winterbourn MJ. 1988. Patch dynamics in lotic systems: the stream as a mosaic. Journal of the
North American Benthological Society 7: 503-524.

Puffer M, Berg OK, Huusko A, Vehanen T, Einum S. 2017. Effects of intra- and interspecific
competition and hydropeaking on growth of juvenile Atlantic salmon (Salmo salar). Ecology
of Freshwater Fish 26: 99-107.

Pulg U, Vollset KW, Velle G, Stranzl S. 2016. First observations of saturopeaking:

Characteristics and implications. Science of The Total Environment 573: 1615-1621.

173
Remo JWF, Khanal A, Pinter N. 2013. Assessment of chevron dikes for the enhancement of
physical-aquatic habitat within the Middle Mississippi River, USA. Journal of Hydrology 501:
146-162.

Robinson CT. 2012. Long-term changes in community assembly, resistance, and resilience
following experimental floods. Ecological Applications 22: 1949-1961.

Rohde S, Schütz M, Kienast F, Englmaier P. 2005. River widening: an approach to restoring

riparian habitats and plant species. River Research and Applications 21: 1075-1094.

Ryo M, Aguilar-Trigueros CA, Pinek L, Muller LAH, Rillig MC. 2019. Basic principles of
temporal dynamics. Trends in Ecology & Evolution 34: 723-733.

Schmitt RJP, Bizzi S, Castelletti A, Kondolf GM. 2018. Improved trade-offs of hydropower
and sand connectivity by strategic dam planning in the Mekong. Nature Sustainability 1: 96-
104.

Schneider M, Kopecki I, Tuhtan J, Sauterleute JF, Zinke P, Bakken TH, Zakowski T, Merigoux
S. 2017. A fuzzy rule-based model for the assessment of macrobenthic habitats under
hydropeaking impact. River Research and Applications 33: 377-387.

Senay C, Taranu ZE, Bourque G, Macnaughton CJ, Lanthier G, Harvey-Lavoie S, Boisclair D.

2017. Effects of river scale flow regimes and local scale habitat properties on fish community
attributes. Aquatic Sciences 79: 13-26.

Spurgeon J, Pegg M, Parasiewicz P, Rogers J. 2019. River-wide habitat availability for fish
habitat guilds: implications for in-stream flow protection. Water 11.

Stalnaker CB, Bovee KD, Waddle TJ. 1996. Importance of the temporal aspect of habitat
hydraulics to fish population studies. Regulated Rivers: Research & Management 12: 145-153.

Stanford JA, Lorang MS, Hauer FR. 2005. The shifting habitat mosaic of river ecosystems. SIL
Proceedings, 1922-2010 29: 123-136.

Tews J, Brose U, Grimm V, Tielbörger K, Wichmann MC, Schwager M, Jeltsch F. 2004.

Animal species diversity driven by habitat heterogeneity/diversity: the importance of keystone
structures. Journal of Biogeography 31: 79-92.

174
Tockner K, Malard F, Ward JV. 2000. An extension of the flood pulse concept. Hydrological
Processes 14: 2861-2883.

Tonolla D, Bruder A, Schweizer S. 2017. Evaluation of mitigation measures to reduce

hydropeaking impacts on river ecosystems – a case study from the Swiss Alps. Science of The
Total Environment 574: 594-604.

Turner M, Gardner R, (eds.). 2015. Landscape Ecology in Theory and Practice.

Vanzo D, Zolezzi G, Siviglia A. 2016. Eco-hydraulic modelling of the interactions between

hydropeaking and river morphology. Ecohydrology 9: 421-437.

Vericat D, Ville F, Palau-Ibars A, Batalla RJ. 2020. Effects of hydropeaking on bed mobility:
evidence from a Pyrenean river. Water 12.

Weber C, Åberg U, Buijse AD, Hughes FMR, McKie BG, Piégay H, Roni P, Vollenweider S,
Haertel-Borer S. 2017. Goals and principles for programmatic river restoration monitoring and
evaluation: collaborative learning across multiple projects. Wiley Interdisciplinary Reviews:
Water: e1257.

Weber C, Nilsson C, Lind L, Alfredsen KT, Polvi L. 2013. Winter disturbances and riverine
fish in temperate and cold regions. BioScience 63: 199-210.

Wegscheider B, Linnansaari T, Monk WA, Curry RA. 2020. Linking fish assemblages to hydro-
morphological units in a large regulated river. Ecohydrology: e2233.

Widén Å, Renöfält BM, Degerman E, Wisaeus D, Jansson R. 2021. Let it flow: Modeling
ecological benefits and hydropower production impacts of banning zero-flow events in a large
regulated river system. Science of The Total Environment 783: 147010.

Winemiller KO, Flecker AS, Hoeinghaus DJ. 2010. Patch dynamics and environmental
heterogeneity in lotic ecosystems. Journal of the North American Benthological Society 29: 84-
99.

Wolkovich EM, Cook BI, McLauchlan KK, Davies TJ. 2014. Temporal ecology in the
Anthropocene. Ecology Letters 17: 1365-1379.

175
Yarnell SM, Mount JF, Larsen EW. 2006. The influence of relative sediment supply on riverine
habitat heterogeneity. Geomorphology 80: 310-324.

Zarfl C, Lumsdon AE, Berlekamp J, Tydecks L, Tockner K. 2015. A global boom in

hydropower dam construction. Aquatic Sciences 77: 161-170.

Geist DR, Arntzen EV, Murray CJ, McGrath KE, Bott Y-J, Hanrahan TP. 2008. Influence of
rver level on temperature and hydraulic gradients in chum and fall chinook salmon spawning
areas downstream of Bonneville Dam, Columbia River. North American Journal of Fisheries
Management 28: 30-41.

Gore JA. 1985. Mechanisms of colonization and habitat enhancement for benthic
macroinvertebrates in restored river channels. Pages 81-101 in Gore JA, ed. The Restoration of
Rivers and Streams. Theories and Experience.

Greimel F, Zeiringer B, Höller N, Grün B, Godina R, Schmutz S. 2016. A method to detect and
characterize sub-daily flow fluctuations. Hydrological Processes 30: 2063-2078.

Halleraker JH. 2001. The environmental impacts of peaking at hydropower plants. Xergi
(Trondheim) 41: 5.

Harby A, Noack M. 2013. Rapid flow fluctuations and impacts on fish and the aquatic
ecosystem. Pages 323-335 in Maddock I, Harby A, Kemp P, Wood P, eds. Ecohydraulics: an
Integrated Approach.

Hauer C, Holzapfel P, Tonolla D, Habersack H, Zolezzi G. 2019. In situ measurements of fine

sediment infiltration (FSI) in gravel-bed rivers with a hydropeaking flow regime. Earth Surface
Processes and Landforms 44: 433-448.

Hedger RD, Sauterleute J, Sundt‐Hansen LE, Forseth T, Ugedal O, Diserud OH, Bakken TH.
2018. Modelling the effect of hydropeaking‐induced stranding mortality on Atlantic salmon
population abundance. Ecohydrology 11: e1960.

Heggenes J, Traaen T. 1988. Downstream migration and critical water velocities in stream
channels for fry of four salmonid species. Journal of Fish Biology 32: 717-727.

176
Hitchman SM, Mather ME, Smith JM, Fencl JS. 2018. Identifying keystone habitats with a
mosaic approach can improve biodiversity conservation in disturbed ecosystems. Global
Change Biology 24: 308-321.

Holzapfel P, Leitner P, Habersack H, Graf W, Hauer C. 2017. Evaluation of hydropeaking

impacts on the food web in alpine streams based on modelling of fish- and macroinvertebrate
habitats. Science of The Total Environment 575: 1489-1502.

Irvine J. 1985. Effects of successive flow perturbations on stream invertebrates. Canadian

Journal of Fisheries and Aquatic Sciences 42: 1922-1927.

Jacobson RB, Johnson HEI, Dietsch BJ. 2009. Hydrodynamic Simulations of Physical Aquatic
Habitat Availability for Pallid Sturgeon in the Lower Missouri River, at Yankton, South Dakota,
Kenslers Bend, Nebraska, Little Sioux, Iowa, and Miami, Missouri, 2006–07. US Geological
Survey. Report no.

Judes C, Gouraud V, Capra H, Maire A, Barillier A, Lamouroux N. 2021. Consistent but

secondary influence of hydropeaking on stream fish assemblages in space and time. Journal of
Ecohydraulics.

Kemp PS, Gilvear DJ, Armstrong JD. 2003. Do juvenile Atlantic salmon parr track local
changes in water velocity? River Research and Applications 19: 569-575.

Kennedy TA, Muehlbauer JD, Yackulic CB, Lytle DA, Miller SW, Dibble KL, Kortenhoeven
EW, Metcalfe AN, Baxter CV. 2016. Flow management for hydropower extirpates aquatic
insects, undermining river food webs. BioScience 66: 561-575.

Kjærstad G, Arnekleiv JV, Speed JDM, Herland AK. 2018. Effects of hydropeaking on benthic
invertebrate community composition in two central Norwegian rivers. River Research and
Applications 34: 218-231.

Korman J, Campana SE. 2009. Effects of hydropeaking on nearshore habitat use and growth of
age-0 rainbow trout in a large regulated river. Transactions of the American Fisheries Society
138: 76-87.

Kougias I, et al. 2019. Analysis of emerging technologies in the hydropower sector. Renewable
and Sustainable Energy Reviews 113: 109257.

177
Lagarrigue T, Céréghino R, Lim P, Reyes-Marchant P, Chappaz R, Lavandier P, Belaud A.
2002. Diel and seasonal variations in brown trout (Salmo trutta) feeding patterns and
relationship with invertebrate drift under natural and hydropeaking conditions in a mountain
stream. Aquat. Living Resour. 15: 129-137.

Lake PS. 2000. Disturbance, patchiness, and diversity in streams. Journal of the North
American Benthological Society 19: 573-592.

Lancaster J, Hildrew AG. 1993. Flow Refugia and the Microdistribution of Lotic
Macroinvertebrates. Journal of the North American Benthological Society 12: 385-393.

Lange K, et al. 2019. Small hydropower goes unchecked. Frontiers in Ecology and the
Environment 17: 256-258.

Lowe WH, Likens GE, Power ME. 2006. Linking scales in stream ecology. BioScience 56:
591-597.

Lumsdon AE, Artamonov I, Bruno MC, Righetti M, Tockner K, Tonolla D, Zarfl C. 2018.
Soundpeaking – hydropeaking induced changes in river soundscapes. River Research and
Applications 34: 3-12.

Miller SW, Schroer M, Fleri JR, Kennedy TA. 2020. Macroinvertebrate oviposition habitat
selectivity and egg-mass desiccation tolerances: Implications for population dynamics in large
regulated rivers. Freshwater Science 39: 584-599.

Moreira M, Hayes DS, Boavida I, Schletterer M, Schmutz S, Pinheiro A. 2019. Ecologically-

based criteria for hydropeaking mitigation: a review. Science of The Total Environment 657:
1508-1522.

Newson MD, Newson CL. 2000. Geomorphology, ecology and river channel habitat: mesoscale
approaches to basin-scale challenges. Progress in Physical Geography: Earth and Environment
24: 195-217.

Newton TJ, Woolnough DA, Strayer DL. 2008. Using landscape ecology to understand and
manage freshwater mussel populations. Journal of the North American Benthological Society
27: 424-439.

178
Nilsson C, Reidy CA, Dynesius M, Revenga C. 2005. Fragmentation and flow regulation of the
world's large river systems. Science 308: 405-408.

Palmer MA, Hakenkamp CC, Nelson-Baker K. 1997. Ecological heterogeneity in streams: why
variance matters. Journal of the North American Benthological Society 16: 189-202.

Palmer MA, Ruhi A. 2019. Linkages between flow regime, biota, and ecosystem processes:
Implications for river restoration. Science 365: eaaw2087.

Palmer MA, Swan CM, Nelson K, Silver P, Alvestad R. 2000. Streambed landscapes: evidence
that stream invertebrates respond to the type and spatial arrangement of patches. Landscape
Ecology 15: 563-576.

Parasiewicz P. 2008. Habitat time series analysis to define flow augmentation strategy for the
Quinebaug River, Connecticut and Massachusetts, USA. River Research and Applications 24:
439-452.

Perona P, Molnar P, Savina M, Burlando P. 2009. An observation‐based stochastic model for

sediment and vegetation dynamics in the floodplain of an Alpine braided river. Water Resources
Research 45.

Person E. 2013. Impact of Hydropeaking on Fish and their Habitat. PhD thesis. École
Politechnique Fédérale de Lausanne (EPFL).

Pickett STA, Rogers KH. 1997. Patch Dynamics: The Transformation of Landscape Structure
and Function Pages 101-127 in Bissonette JA, ed. Wildlife and Landscape Ecology: Effects of
Pattern and Scale

Poff NL, Allan JD, Bain MB, Karr JR, Prestegaard KL, Richter BD, Sparks RE, Stromberg JC.
1997. The natural flow regime. BioScience 47: 769-784.

Poff NL, Schmidt JC. 2016. How dams can go with the flow. Science 353: 1099-1100.

Polvi LE, Lind L, Persson H, Miranda-Melo A, Pilotto F, Su X, Nilsson C. 2020. Facets and
scales in river restoration: nestedness and interdependence of hydrological, geomorphic,
ecological, and biogeochemical processes. Journal of Environmental Management 265:
110288.

179
Premstaller G, Cavedon V, Pisaturo GR, Schweizer S, Adami V, Righetti M. 2017.
Hydropeaking mitigation project on a multi-purpose hydro-scheme on Valsura River in South
Tyrol/Italy. Science of The Total Environment 574: 642-653.

Pringle CM, Naiman RJ, Bretschko G, Karr JR, Oswood MW, Webster JR, Welcomme RL,
Winterbourn MJ. 1988. Patch dynamics in lotic systems: the stream as a mosaic. Journal of the
North American Benthological Society 7: 503-524.

Puffer M, Berg OK, Huusko A, Vehanen T, Einum S. 2017. Effects of intra- and interspecific
competition and hydropeaking on growth of juvenile Atlantic salmon (Salmo salar). Ecology
of Freshwater Fish 26: 99-107.

Pulg U, Vollset KW, Velle G, Stranzl S. 2016. First observations of saturopeaking:

Characteristics and implications. Science of The Total Environment 573: 1615-1621.

Remo JWF, Khanal A, Pinter N. 2013. Assessment of chevron dikes for the enhancement of
physical-aquatic habitat within the Middle Mississippi River, USA. Journal of Hydrology 501:
146-162.

Robinson CT. 2012. Long-term changes in community assembly, resistance, and resilience
following experimental floods. Ecological Applications 22: 1949-1961.

Rohde S, Schütz M, Kienast F, Englmaier P. 2005. River widening: an approach to restoring

riparian habitats and plant species. River Research and Applications 21: 1075-1094.

Ryo M, Aguilar-Trigueros CA, Pinek L, Muller LAH, Rillig MC. 2019. Basic principles of
temporal dynamics. Trends in Ecology & Evolution 34: 723-733.

Schmitt RJP, Bizzi S, Castelletti A, Kondolf GM. 2018. Improved trade-offs of hydropower
and sand connectivity by strategic dam planning in the Mekong. Nature Sustainability 1: 96-
104.

Schneider M, Kopecki I, Tuhtan J, Sauterleute JF, Zinke P, Bakken TH, Zakowski T, Merigoux
S. 2017. A fuzzy rule-based model for the assessment of macrobenthic habitats under
hydropeaking impact. River Research and Applications 33: 377-387.

180
Senay C, Taranu ZE, Bourque G, Macnaughton CJ, Lanthier G, Harvey-Lavoie S, Boisclair D.
2017. Effects of river scale flow regimes and local scale habitat properties on fish community
attributes. Aquatic Sciences 79: 13-26.

Spurgeon J, Pegg M, Parasiewicz P, Rogers J. 2019. River-wide habitat availability for fish
habitat guilds: implications for in-stream flow protection. Water 11.

Stalnaker CB, Bovee KD, Waddle TJ. 1996. Importance of the temporal aspect of habitat
hydraulics to fish population studies. Regulated Rivers: Research & Management 12: 145-153.

Stanford JA, Lorang MS, Hauer FR. 2005. The shifting habitat mosaic of river ecosystems. SIL
Proceedings, 1922-2010 29: 123-136.

Tews J, Brose U, Grimm V, Tielbörger K, Wichmann MC, Schwager M, Jeltsch F. 2004.

Animal species diversity driven by habitat heterogeneity/diversity: the importance of keystone
structures. Journal of Biogeography 31: 79-92.

Tockner K, Malard F, Ward JV. 2000. An extension of the flood pulse concept. Hydrological
Processes 14: 2861-2883.

Tonolla D, Bruder A, Schweizer S. 2017. Evaluation of mitigation measures to reduce

hydropeaking impacts on river ecosystems – a case study from the Swiss Alps. Science of The
Total Environment 574: 594-604.

Turner M, Gardner R, (eds.). 2015. Landscape Ecology in Theory and Practice.

Vanzo D, Zolezzi G, Siviglia A. 2016. Eco-hydraulic modelling of the interactions between

hydropeaking and river morphology. Ecohydrology 9: 421-437.

Vericat D, Ville F, Palau-Ibars A, Batalla RJ. 2020. Effects of hydropeaking on bed mobility:
evidence from a Pyrenean river. Water 12.

Weber C, Åberg U, Buijse AD, Hughes FMR, McKie BG, Piégay H, Roni P, Vollenweider S,
Haertel-Borer S. 2017. Goals and principles for programmatic river restoration monitoring and
evaluation: collaborative learning across multiple projects. Wiley Interdisciplinary Reviews:
Water 5: e1257.

181
Weber C, Nilsson C, Lind L, Alfredsen KT, Polvi L. 2013. Winter disturbances and riverine
fish in temperate and cold regions. BioScience 63: 199-210.

Wegscheider B, Linnansaari T, Monk WA, Curry RA. 2020. Linking fish assemblages to hydro-
morphological units in a large regulated river. Ecohydrology 7: e2233.

Widén Å, Renöfält BM, Degerman E, Wisaeus D, Jansson R. 2021. Let it flow: Modeling ecological
benefits and hydropower production impacts of banning zero-flow events in a large regulated river
system. Science of The Total Environment 783: 147010.

Winemiller KO, Flecker AS, Hoeinghaus DJ. 2010. Patch dynamics and environmental
heterogeneity in lotic ecosystems. Journal of the North American Benthological Society 29: 84-
99.

Wolkovich EM, Cook BI, McLauchlan KK, Davies TJ. 2014. Temporal ecology in the
Anthropocene. Ecology Letters 17: 1365-1379.

Yarnell SM, Mount JF, Larsen EW. 2006. The influence of relative sediment supply on riverine
habitat heterogeneity. Geomorphology 80: 310-324.

Zarfl C, Lumsdon AE, Berlekamp J, Tydecks L, Tockner K. 2015. A global boom in

hydropower dam construction. Aquatic Sciences 77: 161-170.

182