Received: 20 July 2017 | Revised: 19 December 2017 | Accepted: 22 December 2017

DOI: 10.1111/geb.12719

RESEARCH PAPER

Ecological drivers of spatial community dissimilarity, species

replacement and species nestedness across temperate forests

Xugao Wang1 | Thorsten Wiegand2,3 | Kristina J. Anderson-Teixeira4,5 |

Norman A. Bourg6 | Zhanqing Hao1 | Robert Howe7 | Guangze Jin8 |

David A. Orwig9 | Marko J. Spasojevic10 | Shunzhong Wang11 | Amy Wolf7 |

Jonathan A. Myers12

1
Key Laboratory of Forest Ecology and
Management, Institute of Applied Ecology,
Abstract
Chinese Academy of Sciences, Shenyang,
Aims: Patterns of spatial community dissimilarity have inspired a large body of theory in ecology
P. R. China
2
and biogeography. Yet key gaps remain in our understanding of the local-scale ecological processes
Department of Ecological Modelling,
Helmholtz Centre for Environmental underlying species replacement and species nestedness, the two fundamental components of spa-
Research-UFZ, Leipzig, Germany tial community dissimilarity. Here, we examined the relative influence of dispersal limitation,
3
German Centre for Integrative Biodiversity habitat filtering and interspecific species interactions on local-scale patterns of the replacement
Research (iDiv) Halle-Jena-Leipzig, Leipzig, and nestedness components in eight stem-mapped temperate forest mega-plots at different
Germany
4
ontogenetic stages (large versus small trees).
Conservation Ecology Center, Smithsonian
Conservation Biology Institute, National Location: Eight large (20–35 ha), fully mapped temperate forest plots in northern China and
Zoological Park, Front Royal, Virginia
northern U.S.A.
5
Center for Tropical Forest Science–Forest
Global Earth Observatory, Smithsonian Time period: 2004–2016.
Tropical Research Institute, Panama,
Major taxa studied: Woody plants.
Republic of Panama
6
U.S. Geological Survey, National Research Methods: We combined decomposition of community dissimilarity (based on the Ru
zička
Program – Eastern Branch, Reston, Virginia
index) and spatial point-pattern analysis to compare the spatial (i.e., distance-dependent)
7
Department of Natural and Applied
replacement and nestedness components of each plot with that expected under five
Sciences, University of Wisconsin-Green
Bay, Green Bay, Wisconsin
spatially explicit null models representing different hypotheses on community-assembly
8
Center for Ecological Research, Northeast
mechanisms.
Forestry University, Harbin, China
Results: Our analyses revealed complex results. In all eight forests, spatial community dissimi-
9
Harvard Forest, Harvard University,
larity was best explained by species replacement among local tree assemblages and by a null
Petersham, Massachusetts
10
model based on dispersal limitation. In contrast, spatial nestedness for large and small trees
Department of Biology, University of
California Riverside, Riverside, California was best explained by random placement and habitat filtering, respectively, in addition to
11
State Key Laboratory of Vegetation and dispersal limitation. However, interspecific interactions did not contribute to local replacement
Environmental Change, Institute of Botany, and nestedness.
Chinese Academy of Sciences, Xiangshan,
Beijing, China Main conclusions: Species replacement is the predominant process accounting for spatial commu-
12
Department of Biology & Tyson Research nity dissimilarity in these temperate forests and caused largely by local-scale species clustering
Center, Washington University in St Louis, associated with dispersal limitation. Nestedness, in contrast, is less prevalent and primarily associ-
St Louis, Missouri ated with larger variation in local species richness as caused by spatial richness gradients or

Correspondence ‘hotspots’ of local species richness. The novel use of replacement and nestedness measures in
Xugao Wang, Key Laboratory of Forest point pattern analysis is a promising approach to assess local-scale biodiversity patterns and to
Ecology and Management, Institute of explore their causes.
Applied Ecology, Chinese Academy of
Sciences, Shenyang 110016, P. R. China.
Email: [email protected]

Global Ecol Biogeogr. 2018;1–12. wileyonlinelibrary.com/journal/geb V

C 2018 John Wiley & Sons Ltd | 1
2 | WANG ET AL.

Funding information KEYWORDS

National Natural Science Foundation of
beta diversity, biotic interactions, community assembly, dispersal limitation, habitat filtering, nest-
China, Grant/Award Number: 31722010
and 31770666; Chinese Academy of edness, pattern reconstruction, point pattern analysis, species replacement
Sciences, Grant/Award Number:
XDPB0203; National Key Research and
Development Program of China, Grant/
Award Number: 2016YFC0500300;
Smithsonian Global Earth Observatory
Initiative; HSBC Climate Partnership;
International Center for Advanced
Renewable Energy and Sustainability;
National Science Foundation, Grant/Award
Number: DEB 1557094; Tyson Research
Center; The 1923 Fund; Cofrin Center for
Biodiversity; European Research Council
(ERC), Grant/Award Number: 233066

Editor: Andres Baselga

1 | INTRODUCTION Miller, 2003; Mouquet & Loreau, 2003). Habitat filtering, dispersal limi-
tation and interspecific interactions may also interact in complex ways
Disentangling the relative importance of mechanisms that create spatial to influence spatial community dissimilarity. For example, whereas
variation in community composition is a central challenge in ecology increased dispersal may homogenize community composition under
and biogeography (Anderson et al., 2011; Whittaker, 1960). For exam- weak habitat filtering, it may increase community dissimilarity under
ple, increases in community dissimilarity (beta diversity) with spatial dis- strong habitat filtering (Myers & LaManna, 2016). Thus, patterns of
tance have been extensively described for diverse organisms and spatial community dissimilarity may reflect the outcome and interplay
comprise one of the most widely recognized and important spatial of multiple community assembly processes.
biodiversity patterns (Nekola & White, 1999; Soininen, McDonald, & Given the complexity of potential interactions among processes, it
Hillebrand, 2007). This generalization, also called distance-decay of is not surprising that the relative importance of processes in driving
community similarity, is widely used in community ecology to under- patterns of spatial community dissimilarity remains largely unresolved
stand processes of community assembly (Condit et al., 2002; Morlon (Svenning, Fløjgaard, & Baselga, 2011). However, these difficulties may
et al., 2008; Wang et al., 2011). At biogeographical scales, patterns of also be grounded in an overly simplistic characterization of community
spatial species turnover provide important insights into historical and dissimilarity. For example, researchers have recognized for some time
regional processes underlying the composition and dynamics of that the compositional dissimilarity of two local communities may
regional biotas (Nekola & White, 1999; Qian & Ricklefs, 2012; reflect two different phenomena: spatial species turnover (also known
Tuomisto, Ruokolainen, & Yli-Halla, 2003). as species replacement) and nestedness (Baselga, 2010; Baselga &
The processes that determine increasing community dissimilarity Leprieur, 2015; Legendre, 2014; Podani & Schmera, 2011). At one
with distance (hereafter ‘spatial community dissimilarity’) have impor- extreme (perfect species replacement), local communities contain the
tant implications for community assembly and the maintenance of bio- same number of species, but no species are shared among them. In this
diversity in ecological communities. At local scales, spatial community case, their dissimilarity is purely driven by species replacement. At the
dissimilarity is driven by several non-mutually exclusive processes, other extreme (perfect nestedness), the species composition of one
including dispersal limitation, habitat filtering and species interactions local community is a completely nested subset of the other. In this
(Morlon et al., 2008; Wang et al., 2011, 2015). Dispersal limitation (e.g., case, the replacement component is zero, and the dissimilarity is
local seed dispersal) is predicted to increase spatial community dissimi- entirely driven by species loss that causes nestedness. Therefore,
larity by increasing spatial aggregation of conspecific individuals. In nestedness and replacement must be disentangled at different spatial
contrast, increased dispersal should decrease spatial community dissim- scales in order to identify the underlying, possibly antithetic, processes
ilarity and homogenize community composition (Catano, Dickson, responsible for observed patterns of community dissimilarity (Baselga,
Myers, & Rejmanek, 2017; Hubbell, 2001). Likewise, habitat filtering 2010).
(where species arrive at a site but fail to persist owing to the abiotic Species nestedness and replacement patterns have been attributed
conditions; Kraft et al., 2015) can increase spatial community dissimilar- to various processes of community assembly. Previous macroecological
ities by increasing species sorting across abiotic gradients (Chase & studies showed that the nestedness component was more important in
Myers, 2011). Interspecific species interactions, such as competition or areas affected by recent glaciations (e.g., Baselga, 2010; Dobrovolski,
predation, can counteract positive effects of dispersal on diversity and Melo, Cassemiro, & Diniz-Filho, 2012; Svenning et al., 2011). At more
can homogenize community composition by removing competitively local scales, nestedness patterns can emerge from habitat filtering
inferior species or species lacking enemy-tolerant traits (Kneitel & across environmental gradients (Greve, Gremmen, Gaston, & Chown,
WANG ET AL. | 3

2005), where species that occur in harsh environments are a nested

of small
number

34,837

53,635

21,645

30,967

83,670

24,597

48,549

43,124
subset of species that occur in more benign environments (Chase,

Total

trees
2007). In contrast, species-replacement patterns at local scales are

richness,
expected from neutral dynamics under dispersal limitation, ecological

Species

treesb
small
drift, high levels of speciation and isolation of local communities, or

33

53

53

42

44

48

60

51
niche selection where different species occur under different environ-

richness, number
of large

14,021

23,901

11,242

10,329

17,197
mental conditions (e.g., habitat filtering). Studies partitioning commu-

8,269

6,521

9,558
Species Total

trees
nity dissimilarity into species replacement and nestedness components
have mainly been conducted at macroecological scales (e.g., Dobrovol-

richness individuals treesa

ski et al., 2012; Svenning et al., 2011), but little is known about the

precipitation species number of large

23

34

49

39

25

30

34

36
relative importance of processes that influence patterns of species
replacement and nestedness at local scales.

48,858

77,536

29,914

37,488

94,912

34,926

65,746

52,682
In this study, we examined the relative influence of dispersal

Total
limitation, habitat filtering and interspecific species interactions on
local-scale patterns of spatial community dissimilarity and its species

Total
replacement and nestedness components in eight stem-mapped tem-

36

55

61

46

46

51

63

51
perate forest mega-plots (20–35 ha) in the Smithsonian Forest Global
Earth Observatory (ForestGEO; Anderson-Teixeira et al., 2015; Condit,

Characteristics of the eight Forest Global Earth Observatory (ForestGEO) plots in northern China and northern U.S.A.
1998). Four of the temperate forest plots are located in northern China,

Mean

1,150

1,029
Area (ha) Latitude Longitude Elevation (m) temperature (8C) (mm)

748

992

684

700

700

650
and the other four plots in the northern U.S.A. The diversity of settings
and variation in species richness among the eight forest plots (Table 1)
allow us to explore the generality of the patterns and processes
revealed by our analyses across different types of temperate forests
that vary in their climate and evolutionary history. To test underlying
Mean

20.5
12.8

13.6
4.2

8.8

3.6

3.6

4.8
local mechanisms, we simulated spatially explicit ‘null communities’ that
maintain the observed richness and relative abundances of species in

1,290–1,509
each plot (i.e., we removed regional species-pool effects), but random-
488–514

340–368

273–338

172–233

401–492

792–810

781–802
ize tree locations within each plot following spatial point process
models that resemble different (null) hypotheses on the presence or
absence of dispersal limitation, habitat filtering and interspecific species
288.79

272.18

278.15

290.56

129.12

128.08

128.01

115.43
interactions (Wang et al., 2015; Wiegand & Moloney, 2014). We then
compared the simulated patterns with their observed counterparts
(Morlon et al., 2008; Shen et al., 2009; Wang et al., 2011; Wiegand &
45.55

42.54

38.89

38.52

48.08

42.38

42.33

39.96
Moloney, 2014). To explore the relative importance of these processes
at different ontogenetic stages, we conducted separate analyses for
large trees [tree diameter at 1.3 m aboveground (dbh)
10 cm] and
25.2

Smithsonian Conservation Biology Institute, U.S.A. (SCBI) 25.6

small trees (dbh < 10 cm). We asked the following questions. (a) What
35

20

30

25

24

20

is the relative importance of species replacement and species nested-

ness in determining spatial community dissimilarity at different spatial
scales? (b) How do dispersal limitation, habitat filtering and interspecific
species interactions influence species replacement and nestedness, and
do these processes differ among ontogenetic stages? We describe spe-
cific hypotheses and predictions for each of these mechanisms below.
Tyson Research Center, U.S.A. (TRC)

Diameter at breast height < 10 cm.

Diameter at breast height
10 cm.

2 | MATERIALS AND METHODS

Harvard Forest, U.S.A. (HF)

Changbaishan, China (CBS)

Donglingshan, China (DLS)

Wabikon, U.S.A. (WAB)

2.1 | Study sites: Temperate forest-dynamics plots in

northern China and northern U.S.A.
Fenglin, China (FL)

Baihe, China (BH)

Eight large (20–35 ha) temperate forest-dynamics plots were used in

the present study (Table 1). Four plots are located in northern China
TA BL E 1

and four in the northern U.S.A. The plots range in latitude from 38.52
Plot

to 48.088 N. Total species richness among the plots ranges from 36 to

b
a
4 | WANG ET AL.

63 (Table 1). All free-standing woody stems (excluding lianas) with dbh We followed Wang et al. (2011, 2015) and used six environmental

1 cm were mapped, tagged, measured and identified to species using variables (elevation, slope, aspect, terrain convexity, topographic wet-
standardized ForestGEO protocols (Anderson-Teixeira et al., 2015; ness index and altitude above channels) with a spatial resolution of 5 m
Bourg, McShea, Thompson, McGarvey, & Shen, 2013; Condit, 1998). 3 5 m. Elevation, slope, aspect and terrain convexity are widely used
topographical variables. The other two indices, topographic wetness
index (TWI) and vertical distance to the channel network (Chn), are
2.2 | Partitioning community dissimilarity into
used to compute topographical control on hydrological processes
abundance-based species replacement and nestedness
(Kanagaraj, Wiegand, Comita, & Huth, 2011; Punchi-Manage et al.,
For each forest plot, we were interested in the local-scale (< 250 m) 2013). TWI represents the ratio of the area upslope of each quadrat to
pattern of distance decay of community similarity and its components. the local slope for that quadrat, which can capture important informa-
Therefore, we combined the analytical framework of decomposing tion on wetness. To calculate TWI, we used Tarboton’s deterministic
community dissimilarity into nestedness and replacement components infinity method, described by Tarboton (1997) and Sørensen, Zinko,
(Baselga, 2010, 2017; Legendre, 2014) with the distance-centred and Seibert (2006), implemented in the open-source software SAGA-
framework of spatial point pattern analysis (Shen et al., 2009; Wang GIS (Conrad et al., 2015). The vertical distance from the channel net-
et al., 2015; Wiegand & Moloney, 2014; Wiegand et al., 2017). To work (i.e., drainage lines of surface rainfall) was also calculated in
obtain distance-dependent dissimilarity measures, we resampled the SAGA-GIS.
data of the fully mapped plots with the spatial grain of 20 m 3 20 m
subplots. This grain is commonly used in the analysis of ForestGEO 2.4 | Ecological processes and hypotheses
plots, and it is also the typical size of patches used in forest gap
We generated five types of spatially explicit null communities
models (e.g., Fischer et al., 2016). The number of trees and species in
representing different hypotheses on the relative roles of dispersal
these subplots is large enough for meaningful analysis (Supporting
limitation, habitat filtering and interspecific species interactions in com-
Information Table S1).
munity assembly. They are based on explicit maps of the locations of
In a first step of the estimation of distance-dependent dissimilarity
the individuals (instead of the number of individuals of species s in
measures, we randomly located within a given forest plot 500 pairs of
site i), and this additional structural realism allows us to devise null
subplots that were distance r apart. For each subplot i, we then deter-
models with more direct biological interpretation. All null communities
mined the abundance as,i of species s. Second, we followed Legendre
conserved the species richness and relative abundances of species at
(2014) and Baselga (2017) and calculated the three abundance-based
each plot and were assembled by independent superposition of the
dissimilarity indices TDi,j, Repli,j and Nesi,j for the 500 pairs of subplots i
distribution patterns of individual species simulated by specific point
and j (see Supporting Information Appendix S1 for detail). TDi,j is the
process models. This corresponds to the assumption of no species
total dissimilarity in species abundances between subplot i and j as
interactions (e.g., McGill, 2010; Wiegand et al., 2012). The algorithms
measured by the Ru
zička index (an abundance-weighted Jaccard index),
of the specific point process models have been described in detail by
Repli,j is the replacement component of TDi,j, and Nesi,j is the nestedness
Wiegand & Moloney (2014) and Wang et al. (2015). Here, we briefly
component, with TDi,j 5 Repli,j 1 Nesi,j. In the third step, we took the
summarize the basic framework of these point process models used to
mean of TDi,j, Repli,j and TDi,j over the 500 pairs of plots that were dis-
test the five hypotheses below.
tance r apart to obtain our final measures, mTD(r), mRepl(r) and mNes(r).
This procedure was repeated for all distances r of 21–250 m in steps 2.4.1 | The random-placement hypothesis
of 1 m. We used the 21–250 m distance range because subplots with This hypothesis assumes that all individuals in the study area are ran-
r  20 m show substantial spatial overlap, and because for distances domly and independently distributed. It represents the extreme case of
> 250 m focal subplots have only a few neighbour subplots at distance communities without spatial structure that do not, therefore, show dis-
r inside the plot (Wiegand & Moloney, 2014). tance dependence in spatial community dissimilarity and its compo-
nents. To implement the random-placement hypothesis, we used for
each species i a homogeneous Poisson process model with the
2.3 | Environmental variables
observed intensity ki (Illian, Penttinen, Stoyan, & Stoyan, 2008) that
Two of our five null community models considered the impact of envi- assigns each tree a random location within the given study area.
ronmental variables on species distribution. We followed an established Significant deviations from this null model indicate the existence of
approach in point process theory and modelled the spatially variable non-random spatial structures in spatial community dissimilarity,
intensity function ki(x) at location x for each species i by the log-linear species replacement and nestedness.
regression model ki(x) 5 exp[b0 1 b1 v1(x) 1 . . . 1 bn vn(x)] with envi-
ronmental variable vi(x) and coefficients bi that were fitted to the data 2.4.2 | The habitat-filtering hypothesis
using maximum likelihood estimation (Waagepetersen & Guan, 2009). This hypothesis assumes that the distribution of each species is only
For species with < 20 individuals, we used the plot-scale density ki, driven by local habitat suitability, but all further mechanisms of species
being the number of species divided by the area of the plot. patterning are removed. To test the habitat filtering hypothesis, we
WANG ET AL. | 5

used for each species i an inhomogeneous Poisson process model of ki(x) with bandwidth R (Wiegand et al., 2013) replaced the
(Shen et al., 2009), where the habitat suitability at location x is repre- parametric estimate. The nonparametric estimate basically smoothes
sented by the intensity function ki(x) (see section 2.3 ‘Environmental the observed distribution pattern and therefore faithfully reproduces
variables’ above). If a species did not show significant relationships with the observed larger-scale variation in local tree density. Significant
environmental variables, we used the constant plot-scale density ki deviations from this null model can therefore happen only at distances
instead of ki(x). The inhomogeneous Poisson process produces species r smaller than the bandwidth R, and mainly as a result of local interspe-
patterns where the local density of individuals is proportional to the cific species interactions (or imperfect pattern reconstructions or
local habitat suitability given by ki(x), but no additional mechanisms of small-scale edaphic factors). We used a bandwidth of R 5 50 m, like
species aggregation are considered. Significant deviations from this null Wiegand, Gunatilleke, Gunatilleke, and Huth (2007) and Wang et al.
model indicate that mechanisms and processes beyond habitat filtering (2015) (see Supporting Information Appendix S2).
are operating. However, departures may also be caused by missing
environmental variables or extinction–recolonization dynamics where
not all suitable areas are occupied by the species. 2.5 | Evaluating the fit of the different hypotheses

2.4.3 | The dispersal-limitation hypothesis We calculated the scale-dependent dissimilarity summary functions Si(r)
[i.e., representing total dissimilarity mTD(r), species replacement
This hypothesis assumes that the community is assembled only by
mRepl(r) or nestedness mNes(r)] for the null communities generated
the effects of intraspecific aggregation or inhibition mechanisms of
by the five point process models in the same way as for the
population dynamics (e.g., dispersal limitation or negative conspe-
observed data. To compare the observed summary functions S0(r)
cific density dependence) without consideration of the influences of
(indicated by subscript i 5 0) and that resulting from i 5 1, . . ., 100
habitat filtering or interspecific species interactions. To implement
realizations of the null community models, we first calculated the
this hypothesis, we used a homogeneous nonparametric annealing
standardized effect sizes (SES), as follows:
algorithm (Tscheschel & Stoyan, 2006; Wiegand, He, & Hubbell,
2013) that is able to create, for each species, null-distribution pat- SESi ðrÞ5½Si ðrÞ 2 
SðrÞ=rS ðrÞ; (1)
terns that closely match the spatial structure of the original pattern
where 
SðrÞ and rS(r) are the mean and the SD of the summary func-
as captured by summary functions, such as the pair correlation
tions Si(r) of the 100 null community realizations, respectively. For a
function, the K-function and the kth nearest neighbour functions
given distance r, the null community model can then be accepted
(for detail, see Wiegand et al., 2013). Note that this homogeneous
with a ‘pointwise’ significance level of a if 2za < SES0(r) < za. For
algorithm does not preserve the spatial intensity function ki(x) of
a 5 .05, we have za 5 1.96 (Wiegand, Grabarnik, & Stoyan, 2016).
species i, but it preserves the observed overall aggregation (that can
That means that we test whether the observed summary function
be co-determined by habitat filtering). Significant deviations from
S0(r) is located within the 2.5th and 97.5th percentiles of the corre-
this null model indicate that habitat filtering and/or interspecific
sponding null model distributions [i.e., the pointwise simulation enve-
species interactions contribute to the observed patterns.  
lopes S2 ðrÞ5SðrÞ2z 1
a rS ðrÞ and S ðrÞ5SðrÞ1za rS ðrÞ; black dashed

2.4.4 | The combined habitat and dispersal hypothesis lines in Figure 1a]. The standardized effects sizes therefore transform
the original summary functions in a way that the resulting pointwise
This hypothesis assumes that the community is driven by the joint
simulation envelopes are constants –za and za (Figure 1b) (Wiegand
effects of habitat filtering and dispersal limitation. We created null
et al., 2016). The standardized effect size is a measure of fit that con-
communities like those generated by the dispersal-limitation hypothe-
siders the stochasticity of the null communities. If the stochasticity is
sis, but the relocation of individuals of species i was additionally con-
large, for example owing to small sample sizes, the observed commun-
strained by the spatial intensity function ki(x) used in the habitat
filtering hypothesis (Wiegand et al., 2013). Significant deviations from ities may not be distinguishable from null communities.
To assess the significance of the observed dissimilarity functions
this null model may result from unmeasured environmental factors that
are ignored in the log-linear regression models and by interspecific spe- over a given distance interval (e.g., 21–250 m), it is important to control

cies interactions that are not considered (because the individual species for type I error that results from the b multiple tests conducted at the

patterns are independently superimposed). different distance bins r. To correct for this effect, we used the
simulation-based version of the global envelope test presented by
2.4.5 | The independent-placement hypothesis Wiegand et al. (2016) that applies the standard ‘maximal absolute
This hypothesis tests for local interspecific interactions by randomizing difference’ (MAD) test to the transformed summary functions SESi(r)
species independently of one another, while preserving the overall (Myllymäki, Mrkvicka, Grabarnik, Seijo, & Hahn, 2017). We first esti-
intraspecific aggregation and the observed larger-scale distribution [i.e., mated the maximal absolute value Simax of SESi(r) (over r 5 21, . . ., 250),
the observed intensity function ki(x)]. Thus, individuals of different spe- where the upper global envelope zb is the fifth highest value of the
cies are placed at smaller scales without regard to each other (McGill, Simax (over i 5 1, . . ., 100), and the lower global envelope is 2zb (red
2010). To test this hypothesis, we used the method of the combined dashed lines in Figure 1b). We found that zb was c. 3.4. The null
habitat and dispersal hypothesis, but a nonparametric kernel estimate hypothesis can be rejected with significance level of a 5 .05 if the
6 | WANG ET AL.

FIGURE 1 Determining significant departures form the null communities. (a) The observed summary function S0(r) (black bold line), the
mean SðrÞ of the summary functions Si(r) of the 100 null community realizations (bold grey line), the pointwise simulation envelopes being
the 2.5th and 97.5th percentiles of the distribution of the null model simulations Si(r) (dashed black line) and the global simulation
envelopes for the 21–250 m distance interval (red dashed lines) that correct for multiple testing. (b) Same as (a), but for the summary
functions transformed to standardized effect sizes (Equation 1). Note that the resulting simulation envelopes are constants. This allows us
to define the index Err that describes the mean magnitude of departure from the null communities over the 21–250 m distance interval (the
green area)

observed summary functions wander at one (or more) distance bins hypothesis occur for values of Err < 1 (Wiegand et al., 2016). For the
r outside the global envelopes. pointwise test (i.e., b 5 1), Err 5 1 means that the null hypothesis is
To obtain an index Err of the overall strength of departure of a accepted with a significance level of a 5 .05 (that results in za 5 1.96),
given null community model from the observed data over a given dis- but would be rejected with a significance level of a 5 .003 (that results
tance interval, we estimated the average of the significant component in za 5 2.96).
of SES(r) (i.e., the green area in Figure 1b) over the 21-250 m distance
interval:
8
3 | RESULTS
>
< 0 if2zb  SESðrÞ  zb
errðrÞ5
> 3.1 | Observed patterns of spatial community
:
jSESðrÞj2zb otherwise dissimilarity and its components
1 Xrmax
Err5 errðrÞ ð2Þ
b r5rmin Spatial community dissimilarity mTD(r) generally increased with spatial
distance r in all eight temperate forests (Figure 2); only in the Tyson
We have Err 5 0 if the null hypothesis is accepted over the interval
Research Center (TRC) forest did it decrease at larger distances
(rmin, . . ., rmax), and as a rule of thumb, small departures from the null
(Figure 2a,e). The three plots [Fenglin (FL), Changbaishan (CBS) and

F I G U R E 2 (a-c, e-g) The observed spatial community dissimilarities [mTD(r), mRepl(r) and mNes(r)] in the eight temperate forest plots.
Large 5 individuals with diameter at breast height (dbh)
10 cm; Small 5 individuals with dbh < 10 cm. Total 5 total spatial community
dissimilarity. (d, h) The proportion of total spatial community dissimilarities attributed to the replacement component.
WANG ET AL. | 7

TA BL E 2 The mean error Err (Equation 2) describing the fit of the different null community models for distances r 5 21–250 m for large trees
and small trees (in parentheses) in the eight forests

Hypothesis
Country Plots Random placement Habitat filtering Dispersal limitation Habitat and dispersal Independent placement

A) Total dissimilarity
China FL 7.1 (63.3) 6.8 (30.3) 0.0 (1.3) 0.0 (0.0) 0.0 (0.3)
China CBS 2.9 (28.9) 0.5 (14.0) 0.0 (0.0) 0.0 (0.4) 0.0 (0.0)
China BH 12.0 (42.3) 4.4 (20.7) 0.9 (0.5) 0.0 (0.5) 0.1 (0.9)
China DLS 20.6 (67.6) 8.4 (12.7) 0.0 (0.7) 0.9 (1.5) 0.0 (0.0)
U.S.A. WAB 8.2 (48.6) 2.7 (9.0) 0.3 (0.5) 0.0 (0.0) 0.0 (0.0)
U.S.A. HF 19.0 (71.7) 7.0 (8.1) 0.0 (1.4) 0.5 (0.3) 0.0 (0.0)
U.S.A. SCBI 5.8 (35.5) 2.1 (6.9) 0.0 (0.1) 0.0 (0.4) 0.0 (0.0)
U.S.A. TRC 11.3 (63.1) 2.0 (4.5) 0.0 (2.5) 0.2 (1.8) 0.0 (0.0)

B) Replacement component
China FL 4.4 (40.4) 4.2 (37.2) 0.0 (0.6) 0.3 (0.2) 0.5 (0.0)
China CBS 3.4 (19.7) 1.7 (15.2) 0.3 (0.4) 1.0 (0.3) 0.0 (0.0)
China BH 10.4 (26.0) 5.9 (18.3) 0.4 (0.0) 3.1 (4.7) 0.9 (0.0)
China DLS 20.3 (52.5) 8.5 (14.1) 1.0 (0.4) 1.3 (1.7) 0.0 (0.0)
U.S.A. WAB 4.5 (10.4) 4.4 (14.8) 0.0 (1.4) 0.7 (3.4) 0.3 (0.0)
U.S.A. HF 9.2 (34.3) 1.3 (13.5) 0.0 (0.3) 0.2 (3.0) 0.0 (0.0)
U.S.A. SCBI 5.4 (27.3) 3.0 (4.3) 0.3 (0.0) 0.3 (0.5) 0.4 (0.0)
U.S.A. TRC 9.9 (42.7) 2.8 (7.2) 0.7 (1.3) 0.2 (3.4) 0.0 (0.0)

C) Nestedness component
China FL 0.0 (9.3) 0.0 (3.1) 0.0 (0.0) 0.7 (0.0) 0.1 (0.0)
China CBS 0.4 (2.0) 1.0 (1.5) 0.6 (0.1) 1.6 (0.0) 0.3 (0.0)
China BH 0.1 (3.8) 0.8 (1.0) 0.4 (0.3) 3.7 (4.2) 0.4 (0.0)
China DLS 1.7 (11.7) 1.5 (1.9) 0.8 (0.0) 1.2 (1.4) 0.6 (0.0)
U.S.A. WAB 0.0 (34.2) 1.4 (0.0) 0.9 (0.8) 1.5 (0.8) 0.8 (0.0)
U.S.A. HF 5.2 (25.5) 6.3 (2.1) 0.7 (2.9) 4.1 (3.5) 0.0 (0.0)
U.S.A. SCBI 1.5 (4.8) 1.3 (0.0) 0.7 (0.1) 1.3 (0.5) 0.4 (0.0)
U.S.A. TRC 0.9 (13.6) 0.6 (1.2) 0.8 (0.7) 0.7 (3.2) 0.2 (0.0)

Note. For Err 5 0, the null hypothesis is accepted with significance level of .05 over the entire distance interval; weak departures from the null
hypothesis occur for values of Err < 1.

Baihe (BH)] in northern China showed only a weak increase of dissimi- 3.2 | Relative importance of dispersal limitation,
larity with distance, whereas the Donglingshan (DLS) plot in northern habitat filtering and interspecific interactions
China and the Harvard Forest (HF) plot in the northern U.S.A. showed
3.2.1 | The replacement component
strong increases (Figure 2a,e). In general, the communities of small
trees showed stronger increases in dissimilarity with distance than the The random placement hypothesis, which represents communities
communities of large trees. without any spatial structure, yielded the poorest fit to the observed
In all eight forests, spatial community dissimilarity was primarily replacement components for both large and small trees (Table 2B;
related to species replacement among local assemblages distance r Supporting Information Figure S2). The index Err of departures from
apart (Figure 2d,h). For large trees, we found that between 70 and the null community ranged for large trees between 3.4 and 20.3 and
96% of the dissimilarity (at distances > 20 m) could be attributed to for small trees between 10.4 and 52.5 (Table 2B). The mean strength
the species replacement component (Figure 2d). The largest contribu- of departures Err from random placement were much larger for small
tion of species replacement occurred at the Smithsonian Conservation trees, especially for the FL, DLS, HF and TRC plots. Interestingly, the
Biology Institute (SCBI) plot (c. 94%; red line in Figure 2d) and the index Err was correlated with the elevation range of the plots shown in
lowest for the HF plot (c. 74%; blue line in Figure 2d). We observed Table 1 (correlation coefficients were 0.78 for large trees and 0.77 for
similar patterns for small trees at all four plots in northern China small trees). The habitat filtering hypothesis, which accounted for the
(between 78 and 87%); however, for small trees at the plots in north- effects of topographic variables, improved the fit in most cases, but still
ern U.S.A. the nestedness component became more important at produced highly significant departures, with Err ranging between 1.3
scales between 20 and 100 m (Figure 2h), and at the Wabikon (WAB) and 8.5 for large trees and between 4.3 and 37.2 for small trees
forest it even accounted for half of the dissimilarity. The absolute (Table 2B).
nestedness values for large trees and small trees in northern China In most cases, the dispersal-limitation hypothesis fitted the
showed relatively little response to spatial scale (Figure 2c,g), but nest- observed replacement component (i.e., Err 5 0) or produced only small
edness of small trees decreased at intermediate scales at the plots in departures (i.e., Err  1; Table 2B). The more complex combined habitat
northern U.S.A. (Figure 2g). and dispersal limitation hypothesis produced similar good fits for large
8 | WANG ET AL.

trees (except the BH plot), but for smaller trees in the BH, WAB, HF association of species to topographic variables (i.e., the habitat-filtering
and TRC plots larger departures (with Err > 2). The independent- hypothesis) provided mostly poor fits. However, nestedness of large
placement hypotheses fitted the observed replacement trees was also well fitted by the random placement and habitat filtering
component in all eight plots (smaller departures for large trees in BH hypotheses, and for small trees nestedness was also well fitted by the
and SCBI, otherwise Err 5 0) for both the large and small tree commun- habitat filtering hypothesis. Third, we found that the nestedness com-
ities (Table 2B). ponents of the plots in the northern U.S.A. varied in general more than
that in northern China. This was especially true for small trees, where
3.2.2 | The nestedness component
nestedness values of the northern U.S.A. plots were substantially larger
The nestedness component of community dissimilarity was driven by than those in northern China. Overall, our results support the hypothe-
different mechanisms than the replacement component. For large sis that the two components of spatial community dissimilarity are
trees, we found that random placement produced surprisingly good driven by different mechanisms of community assembly.
approximations of the nestedness component (with Err < 1.7), except
at the HF forest, with Err 5 5.2, which showed the largest
4.1 | The relative importance of the replacement and
nestedness component for large trees among all eight forests
nestedness components
(Table 2C). Habitat filtering produced good fits, similar to random
placement (Supporting Information Figure S3). The dispersal-limitation In all eight forest plots, we found that patterns of community dissimi-
hypothesis also produced only weak departures (i.e., Err  1) for the HF larity primarily reflected species replacement among local commun-
forest. ities. Overall, we obtained similar results from forest plots in northern
For small trees, in contrast, random placement did not fit the China and northern U.S.A. However, the contributions of the nested-
observed nestedness components and produced large departures for ness components to total dissimilarities were higher in northern U.S.
some forests, especially for the WAB (Err 5 34.2) and HF (Err 5 25.5) A., especially at distances < 100 m. The generally low contribution of
forests. Habitat filtering improved the fit substantially in all cases, nestedness to overall dissimilarity of local species assemblages can be
including the WAB and SCBI forests, yielding otherwise only weak explained by the relatively small local variation in species richness
departures (Table 2C). Dispersal limitation produced only weak depar- among local communities (i.e., 20 m 3 20 m subplots). We found that
tures, except for the HF plot (Table 2C). The independent-placement the high nestedness component of small trees at the WAB and HF
hypothesis fitted the data well in all cases. forests coincided with a high local variation in species richness
(coefficient of variation 5 0.54 and 0.62) and number of individuals
3.2.3 | Total dissimilarity (coefficient of variation 5 0.95 and 1.21), a pattern inevitably leading
In general, the null model analyses yielded similar results for total dis- to ‘diversity hotspots’ (Supporting Information Table S1). This obser-
similarity and the replacement component (cf. Table 2A,B; Supporting vation parallels the high degree of nestedness found within island
Information Figure S1). Here, the dispersal-limitation hypothesis and archipelagos (Lomolino, 1996; Si, Baselga, Ding, & Machado, 2015),
the combined habitat and dispersal hypothesis produced similar fits, where selective immigration and extinction lead to ordered patterns
with Err < 1, in most cases. Notably, the combined habitat and dispersal of species assemblages. In particular, depauperate islands tend to be
hypothesis produced better fits for total dissimilarity than its replace- occupied by predictable subsets of species occurring on species-rich
ment component (Table 2A). islands. In contrast, smaller variation in local richness, as observed for
large trees and small trees at the plots in northern China, leaves little
4 | DISCUSSION room for typical nestedness configurations with strong spatial
richness gradients or ‘hotspots’ of local species richness (or tree
In this study, we evaluated the relative importance of two complemen- abundance).
tary processes (species replacement and nestedness) underlying tree The relatively large nestedness component for small trees in the
community dissimilarity measures (Baselga, 2010; Legendre, 2014; northern U.S.A. forests is probably related to their overall lower species
Podani & Schmera, 2011) in eight fully mapped ForestGEO plots in the richness that scales down to the low mean species richness of small
U.S.A. and China (Anderson-Teixeira et al., 2015). We used spatial point trees at the subplot scale (ranging between 4.9 and 6.9 in northern U.S.
pattern analysis and spatially explicit null models (Wang et al., 2015; A. and between 7.6 and 13.7 in northern China; Supporting Information
Wiegand & Moloney, 2014; Wiegand et al., 2017) to analyse Table S1). Interestingly, all plots showed for small trees a similar SD in
mechanisms underlying observed patterns in the two dissimilarity local species richness (c. 2.7; Supporting Information Table S1), which
components. Our study yielded three key findings. First, total spatial then results in higher local variation in richness of the North American
community dissimilarity was primarily driven by species replacement plots (Supporting Information Table S1). For example, the WAB and HF
among local assemblages. The species-replacement components and forest plots, with the high nestedness values for small trees, both show
total dissimilarity generally increased with spatial distances, whereas patchy local hotspots of high density (and richness) of small trees. The
the species-nestedness component showed little response to spatial generation of local hotspots of species richness at these two plots was
scale. Second, the dispersal-limitation hypothesis provided the best probably enhanced by localized logging during the 1900s at WAB and
explanation for species replacement and total dissimilarity, whereas the following a hurricane of 1938 at HF.
WANG ET AL. | 9

4.2 | Ecological mechanisms underlying replacement scales analysed here, we find that the replacement component
and nestedness accounts for spatial community dissimilarity caused by ‘homogeneous’
species clustering as described by the dispersal limitation hypothesis
Habitat filtering and dispersal limitation have been regarded as two
(i.e., species clusters of different species are independent and can
major mechanisms of community assembly, but their relative roles
appear everywhere in the plot with the same probability). This is also
across forests remains elusive. Our results showed consistently for all
the spatial pattern expected from neutral theory (Hubbell, 2001). In
forests that the dispersal-limitation hypothesis (i.e., maintaining the
contrast, the spatial nestedness component of spatial dissimilarity can
observed intraspecific aggregation) contributed more to the average
only become more important if the community shows high heterogene-
total compositional dissimilarity between local communities than habi-
ity in local species richness and/or local individual numbers. Otherwise,
tat filtering. This result is consistent with previous studies in tropical
in more homogeneous communities, nestedness remains small and
(Morlon et al., 2008) and temperate forests (Wang et al., 2011).
arises from stochasticity as captured by random placement.
Moreover, this general finding challenges the idea that dispersal assem-
bly may be relatively more important than niche assembly in high-
4.3 | Effects of interspecific species interactions
diversity (e.g., tropical) compared with low-diversity (e.g., temperate)
forest communities (Hubbell, 2001; Myers et al., 2013); our results Surprisingly, our results suggest that small-scale species interactions
suggest that it is important universally. have a relatively weak influence on overall spatial patterns of commu-
Our study advances previous work on the processes governing nity dissimilarity in the temperate forests analysed. The independent-
distance decay of similarity at local scales (e.g., Morlon et al., 2008; placement hypothesis that distributes individuals of a species locally
Wang et al., 2011, 2015) by separate analysis of the species replace- irrespective of individuals of other species (McGill, 2010; Wiegand
ment and nestedness components of spatial community dissimilarity. et al., 2012) generated excellent approximations of spatial community
We found that the species replacement component was able to dissimilarity, species replacement and nestedness in all forests at dis-
discriminate clearly among competing hypotheses. Our analysis led to tances of 21–250 m (Table 2). This result is inconsistent with previous
acceptance of the dispersal limitation hypotheses as the simplest studies that have provided evidence for the importance of pairwise
hypotheses that fitted the observations reasonably well, clearly reject- interspecific species interactions, especially for tree species in temper-
ing the random placement and habitat filtering hypotheses. However, ate forests (e.g., Canham et al., 2006; Wang et al., 2010). For instance,
the nestedness component was influenced by different processes. The Wang et al. (2010) found that c. 30% of all pairs of large tree species
random-placement hypothesis explained the low species-nestedness showed evidence of interspecific interactions at the CBS plot. Possible
component for large trees well (except for the HF forest, which explanations for this finding are that the effects of significant negative
showed the largest nestedness component for large trees), suggesting pairwise interspecific interactions on community level patterns might
that spatial distributions of large trees may be strongly influenced by be ‘diluted’ by the non-significant interactions (e.g., the 60% found by
stochastic community assembly resulting from ecological drift (Hubbell, Wang et al., 2010) or by stochasticity (Wang et al., 2016), or the effect
2001). In contrast, the habitat-filtering hypothesis provided a better of positive and negative pairwise interspecific interactions on commu-
explanation for the larger species-nestedness components of small nity level patterns might cancel one another. As a consequence,
trees (especially at the WAB forest), suggesting a stronger influence of significant pairwise interactions might leave no strong signal in
deterministic community assembly at earlier life stages or for under- community-level patterns such as the species–area relationship and
storey species relative to canopy species. spatial community dissimilarity (Wang et al., 2011).
The inability of the habitat-filtering models to explain observed An alternative but complementary explanation for our reported
spatial community dissimilarities can be explained partly by effects of patterns is that the detectable small-scale interactions in some of the
from extinction–recolonization dynamics (where not all species occupy
species pairs translated only into subtle small-scale patterns that are
all suitable areas) or from the omission of some important environmen-
not detectable with the relatively coarse 20 m spatial resolution used
tal variables. For instance, soil nutrients such as nitrogen, phosphorus
here. For example, Wang et al. (2015) detected at scales < 15 m subtle
and aluminum have been shown to influence local spatial distributions
effects of interspecific interactions in patterns of phylogenetic and
rez, & Turner,
of species in tropical (e.g., Condit, Engelbrecht, Pino, Pe
functional beta diversity (their fig. D6 in appendix D) quantified by
2013; John et al., 2007) and temperate (LaManna, Walton, Turner,
high-resolution point pattern summary functions not based on subplot
Myers, & Rejmanek, 2016; Spasojevic, Turner, Myers, & Jones, 2016)
counting. Moreover, weak interspecific interactions could result from a
forests. Also, factors such as land-use history (e.g., grazing, hunting and
relatively stronger influence of intraspecific interactions, such as intra-
logging) may have affected the spatial pattern of species, especially at
specific competition for limiting resources or negative density depend-
the WAB and SCBI plots, where historical logging is known to have
ence caused by specialized natural enemies (LaManna et al., 2016).
occurred. Given a lack of more detailed environmental data, we leave
this as an issue for forthcoming studies.
4.4 | Conclusions
In addition, we found that patterns of species nestedness and
replacement were influenced by variation in local species richness and Analyzing patterns of spatial community dissimilarity, species replace-
densities of individuals at the 20 m 3 20 m subplot scale. At the local ment and species nestedness have substantially advanced our
10 | WANG ET AL.

understanding of the processes controlling macro-scale biodiversity pat- R.H., G.J., D.A.O., M.J.S., S.W., A.W. and J.A.M. provided data and
terns, but the ways in which different community assembly processes commented on the manuscript.
create these patterns at local scales remain largely unexplored. Across
eight temperate forests in northern China and northern U.S.A., we show
that spatial community dissimilarity is primarily explained by species ORC ID

replacement among local assemblages. Moreover, we show that differ- Xugao Wang http://orcid.org/0000-0003-1207-8852
ent assembly processes influence the species replacement and species Thorsten Wiegand http://orcid.org/0000-0002-3721-2248
nestedness components of community dissimilarity. Species replacement Guangze Jin http://orcid.org/0000-0002-9852-0965
of both large and small trees was primarily explained by dispersal limita-
tion in all forests, whereas small nestedness was primarily associated R EFE R ENC E S
with stochastic processes (random spatial distributions of large trees), Anderson, M. J., Crist, T. O., Chase, J. M., Vellend, M., Inouye, B. D.,
and larger nestedness (of small trees) was primarily associated with Freestone, A. L., . . . Swenson, N. G. (2011). Navigating the multiple
larger heterogeneous variability in local species richness. In contrast, meanings of b diversity: A roadmap for the practicing ecologist.
Ecology Letters, 14, 19–28.
interspecific species interactions appeared to have no detectable influ-
Anderson-Teixeira, K. J., Davies, S. J., Bennett, A. C., Gonzalez-Akre, E.
ence on spatial community dissimilarity and its replacement and nested-
B., Muller-Landau, H. C., Joseph Wright, S., . . . Zimmerman, J. (2015).
ness components across all forests. Collectively, these results indicate CTFS-ForestGEO: A worldwide network monitoring forests in an era
that the species replacement and nestedness components of community of global change. Global Change Biology, 21, 528–549.
patterns provide key insights into the relative importance of different Baselga, A. (2010). Partitioning the turnover and nestedness components
ecological processes underlying community assembly and local of beta diversity. Global Ecology and Biogeography, 19, 134–143.

biodiversity patterns; these insights could easily be overlooked in studies Baselga, A. (2017). Partitioning abundance-based multiple-site dissimilar-
ity into components: Balance variation in abundance and abundance
that focus solely on patterns of community dissimilarity. Thus, the novel
gradients. Methods in Ecology and Evolution, 8, 799–808.
use of replacement and nestedness measures in point pattern analysis of
Baselga, A., & Leprieur, F. (2015). Comparing methods to separate
large, fully mapped forest dynamics plots is a promising approach to components of beta diversity. Methods in Ecology and Evolution, 6,
assess local-scale biodiversity patterns and to explore their causes. 1069–1079.
Bourg, N. A., McShea, W. J., Thompson, J. R., McGarvey, J. C., & Shen,
AC KNOW LEDG MENT S X. (2013). Initial census, woody seedling, seed rain, and stand struc-
ture data for the SCBI SIGEO large forest dynamics plot. Ecology, 94,
This work was supported by National Natural Science Foundation of 2111–2112.
China (31722010), the Strategic Priority Research Program of the Chi- Canham, C. D., Papaik, M. J., Uriarte, M., McWilliams, W. H., Jenkins, J.
nese Academy of Sciences (XDPB0203), National Key Research and C., & Twery, M. J. (2006). Neighborhood analyses of canopy tree
Development Program of China (2016YFC0500300) and National Nat- competition along environmental gradients in New England forests.
Ecological Applications, 16, 540–554.
ural Science Foundation of China (31770666, 31570630). Funding for
the SCBI plot was provided by the Smithsonian Global Earth Observa- Catano, C. P., Dickson, T. L., Myers, J. A., & Rejmanek, M. (2017).
Dispersal and neutral sampling mediate contingent effects of disturb-
tory initiative, the Smithsonian Institution, National Zoological Park
ance on plant beta-diversity: A meta-analysis. Ecology Letters, 20,
and the HSBC Climate Partnership. Funding for the Tyson Research 347–356.
Plot was provided by the International Center for Advanced Renew- Chase, J. M. (2007). Drought mediates the importance of stochastic com-
able Energy and Sustainability (I-CARES) at Washington University in munity assembly. Proceedings of the National Academy of Sciences
St Louis, the National Science Foundation (DEB 1557094 to J.A.M. USA, 104, 17430–17434.

and M.J.S.) and the Tyson Research Center. Research at the Wabikon Chase, J. M., & Myers, J. A. (2011). Disentangling the importance of eco-
logical niches from stochastic processes across scales. Philosophical
Forest Dynamics Plot was supported by The 1923 Fund, the Smithso-
Transactions of the Royal Society B: Biological Sciences, 366, 2351–
nian Institution, and the Cofrin Center for Biodiversity at the Univer- 2363.
sity of Wisconsin-Green Bay. T.W. was supported by the European Condit, R. (1998). Tropical forest census plots: Methods and results from
Research Council (ERC) advanced grant 233066. Barro Colorado Island, Panama and a comparison with other plots.
Berlin, Heidelburg: Springer-Verlag.

DAT A ACC ES SIBI LI TY Condit, R., Engelbrecht, B. M. J., Pino, D., Perez, R., & Turner, B. L.
(2013). Species distributions in response to individual soil nutrients
Data on these temperate forest plots are available at http://www.for- and seasonal drought across a community of tropical trees. Proceed-
estgeo.si.edu/ and accessible by contacting the principal investigator of ings of the National Academy of Sciences USA, 110, 5064–5068.
each plot. Condit, R., Pitman, N., Leigh, E. G., Chave, J., Terborgh, J., Foster, R. B.,
. . . Hubbell, S. P. (2002). Beta-diversity in tropical forest trees.
Science, 295, 666–669.
AU THOR CONT RIB UT IONS
Conrad, O., Bechtel, B., Bock, M., Dietrich, H., Fischer, E., Gerlitz, L., . . .
X.W and T.W. designed the study; X.W, T.W. and J.A.M. wrote the Bo€ hner, J. (2015). System for automated geoscientific analyses
manuscript; X.W. performed data analyses; K.J.A.-T., N.A.B., Z.H., (SAGA) v. 2.1.4. Geoscientific Model Development, 8, 1991–2007.
WANG ET AL. | 11

Dobrovolski, R., Melo, A. S., Cassemiro, F. A. S., & Diniz-Filho, J. A. F. Punchi-Manage, R., Getzin, S., Wiegand, T., Kanagaraj, R., Savitri Gunatil-
(2012). Climatic history and dispersal ability explain the relative leke, C. V., Nimal Gunatilleke, I. A. U., . . . Huth, A. (2013). Effects of
importance of turnover and nestedness components of beta diversity. topography on structuring local species assemblages in a Sri Lankan
Global Ecology and Biogeography, 21, 191–197. mixed dipterocarp forest. Journal of Ecology, 101, 149–160.
Fischer, R., Bohn, F., Dantas de Paula, M., Dislich, C., Groeneveld, J., Qian, H., & Ricklefs, R. E. (2012). Disentangling the effects of geographic
rrez, A. G., . . . Huth, A. (2016). Lessons learned from applying a
Gutie distance and environmental dissimilarity on global patterns of species
forest gap model to understand ecosystem and carbon dynamics of turnover. Global Ecology and Biogeography, 21, 341–351.
complex tropical forests. Ecological Modelling, 326, 124–133. Shen, G., Yu, M., Hu, X., Mi, X., Ren, H., Sun, I., & Ma, K. (2009).
Greve, M., Gremmen, N. J. M., Gaston, K. J., & Chown, S. L. (2005). Nest- Species–area relationships explained by the joint effects of dispersal
edness of Southern Ocean island biotas: Ecological perspectives on a limitation and habitat heterogeneity. Ecology, 90, 3033–3041.
biogeographical conundrum. Journal of Biogeography, 32, 155–168. Si, X., Baselga, A., Ding, P., & Machado, R. B. (2015). Revealing beta-
Hubbell, S. P. (2001). The unified neutral theory of biodiversity and diversity patterns of breeding bird and lizard communities on inun-
biogeography. Princeton, NJ: Princeton University Press. dated land-bridge islands by separating the turnover and nestedness
components. PLoS One, 10, e0127692.
Illian, J., Penttinen, A., Stoyan, H., & Stoyan, D. (2008). Statistical analysis
and modelling of spatial point patterns. New York: Wiley. Soininen, J., McDonald, R., & Hillebrand, H. (2007). The distance decay
of similarity in ecological communities. Ecography, 30, 3–12.
John, R., Dalling, J. W., Harms, K. E., Yavitt, J. B., Stallard, R. F.,
Mirabello, M., . . . Foster, R. B. (2007). Soil nutrients influence spatial Sørensen, R., Zinko, U., & Seibert, J. (2006). On the calculation of the
distributions of tropical tree species. Proceedings of the National topographic wetness index: Evaluation of different methods based on
Academy of Sciences USA, 104, 864–869. field observations. Hydrology and Earth System Sciences, 10, 101–112.

Kanagaraj, R., Wiegand, T., Comita, L. S., & Huth, A. (2011). Tropical tree Spasojevic, M. J., Turner, B. L., Myers, J. A., & Jones, R. (2016). When
species assemblages in topographical habitats change in time and does intraspecific trait variation contribute to functional beta-diver-
sity? Journal of Ecology, 104, 487–496.
with life stage. Journal of Ecology, 99, 1441–1452.
Svenning, J., Fløjgaard, C., & Baselga, A. (2011). Climate, history and neu-
Kneitel, J. M., & Miller, T. E. (2003). Dispersal rates affect species com-
trality as drivers of mammal beta diversity in Europe: Insights from
position in metacommunities of Sarracenia purpurea inquilines. The
multiscale deconstruction. Journal of Animal Ecology, 80, 393–402.
American Naturalist, 162, 165–171.
Tarboton, D. G. (1997). A new method for determination of flow direc-
Kraft, N. J. B., Adler, P. B., Godoy, O., James, E. C., Fuller, S., Levine, J.
tions and upslope areas in grid digital elevation models. Water
M., & Fox, J. (2015). Community assembly, coexistence, and the envi-
Resource Research, 33, 309–319.
ronmental filtering metaphor. Functional Ecology, 29, 592–599.
Tscheschel, A., & Stoyan, D. (2006). Statistical reconstruction of random
LaManna, J. A., Walton, M. L., Turner, B. L., Myers, J. A., & Rejmanek, M.
point patterns. Computational Statistics and Data Analysis, 51, 859–871.
(2016). Negative density dependence is stronger in resource-rich
environments and diversifies communities when stronger for com- Tuomisto, H., Ruokolainen, K., & Yli-Halla, M. (2003). Dispersal, environ-
mon but not rare species. Ecology Letters, 19, 657–667. ment, and floristic variation of western Amazonian forests. Science,
299, 241–244.
Legendre, P. (2014). Interpreting the replacement and richness difference
components of beta diversity. Global Ecology and Biogeography, 23, Waagepetersen, R., & Guan, Y. (2009). Two-step estimation for inhomo-
1324–1334. geneous spatial point processes. Journal of the Royal Statistical Society
B, 71, 685–702.
Lomolino, M. (1996). Investigating causality of nestedness of insular
Wang, X., Wiegand, T., Hao, Z., Li, B., Ye, J., & Lin, F. (2010). Species
communities: Selective immigrations or extinctions? Journal of
associations in an old-growth temperate forest in north-eastern
Biogeography, 23, 699–703.
China. Journal of Ecology, 98, 674–686.
McGill, B. J. (2010). Matters of scale. Science, 328, 575–576.
Wang, X., Wiegand, T., Kraft, N. J. B., Swenson, N. G., Davies, S. J., Hao,
Morlon, H., Chuyong, G., Condit, R., Hubbell, S., Kenfack, D., Thomas, D., Z., . . . Wolf, A. (2016). Stochastic dilution effects weaken determinis-
. . . Green, J. L. (2008). A general framework for the distance-decay tic effects of niche-based processes in species rich forests. Ecology,
of similarity in ecological communities. Ecology Letters, 11, 904–917. 97, 347–360.
Mouquet, N., & Loreau, M. (2003). Community patterns in source-sink Wang, X., Wiegand, T., Swenson, N. G., Wolf, A. T., Howe, R. W., Hao,
metacommunities. The American Naturalist, 162, 544–557. Z., . . . Yuan, Z. (2015). Mechanisms underlying local functional and
Myers, J. A., & LaManna, J. A. (2016). The promise and pitfalls of beta phylogenetic beta diversity in two temperate forests. Ecology, 96,
diversity in ecology and conservation. Journal of Vegetation Science, 1062–1073.
27, 1081–1083. Wang, X., Wiegand, T., Wolf, A., Howe, R., Davies, S. J., & Hao, Z.
nez, I., Jørgensen, P. M., Araujo-Murakami,
Myers, J. A., Chase, J. M., Jime (2011). Spatial patterns of tree species richness in two temperate for-
A., Paniagua-Zambrana, N., & Seidel, R. (2013). Beta-diversity in tem- ests. Journal of Ecology, 99, 1382–1393.
perate and tropical forests reflects dissimilar mechanisms of commu- Whittaker, R. H. (1960). Vegetation of the Siskiyou mountains, Oregon
nity assembly. Ecology Letters, 16, 151–157. and California. Ecological Monographs, 30, 279–338.
Myllymäki, M., Mrkvicka, T., Grabarnik, P., Seijo, H., & Hahn, U. (2017). Wiegand, T., & Moloney, K. A. (2014). Handbook of spatial point pattern
Global envelope tests for spatial processes. Journal of the Royal analysis in ecology. Boca Raton, FL: Chapman and Hall/CRC Press.
Statistical Society Series B, 79, 381–404. Wiegand, T., Grabarnik, P., & Stoyan, D. (2016). Envelope tests for spatial
Nekola, J. C., & White, P. S. (1999). The distance decay of similarity in point patterns with and without simulation. Ecosphere, 7, e01365.
biogeography and ecology. Journal of Biogeography, 26, 867–878. Wiegand, T., Gunatilleke, C. V. S., Gunatilleke, I. A. U. N., & Huth, A.
Podani, J., & Schmera, D. (2011). A new conceptual and methodological (2007). How individual species structure diversity in tropical forests.
framework for exploring and explaining pattern in presence-absence Proceedings of the National Academy of Sciences USA, 104, 19029–
data. Oikos, 120, 1625–1638. 19033.
12 | WANG ET AL.

Wiegand, T., He, F., & Hubbell, S. P. (2013). A systematic comparison of focuses on spatial patterns, dynamics and underlying mechanisms of
summary characteristics for quantifying point patterns in ecology. plant communities in forests.
Ecography, 36, 92–103.
Wiegand, T., Huth, A., Getzin, S., Wang, X., Hao, Z., Gunatilleke, C. V. S.,
& Gunatilleke, I. A. U. N. (2012). Testing the independent species’ SU PP ORT ING INF OR MATI ON
arrangement assertion made by theories of stochastic geometry of Additional Supporting Information may be found online in the sup-
biodiversity. Proceedings of the Royal Society B: Biological Sciences, porting information tab for this article.
279, 3312–3320.
Wiegand, T., Uriarte, M., Kraft, N. J. B., Shen, G., Wang, X., & He, F. (2017).
Spatially explicit metrics of species diversity, functional diversity, and
phylogenetic diversity: Insights into plant community assembly proc- How to cite this article: Wang X, Wiegand T, Anderson-Teixeira
esses. Annual Review of Ecology, Evolution, and Systematics, 48, 329–351. KJ, et al. Ecological drivers of spatial community dissimilarity,
species replacement and species nestedness across temperate
forests. Global Ecol Biogeogr. 2018;00:1–12. https://doi.org/10.
BIOSK ET CH
1111/geb.12719
XUGAO WANG is a professor of forest ecology at Shenyang Institute of
Applied Ecology, Chinese Academy of Sciences. His current research