Algal Research 70 (2023) 102995

Contents lists available at ScienceDirect

Algal Research
journal homepage: www.elsevier.com/locate/algal

Outdoor annual algae productivity improvements at the pre-pilot scale

through crop rotation and pond operational management strategies
John McGowen a, *, Eric P. Knoshaug b, Lieve M.L. Laurens b, Jessica Forrester a
a
Arizona Center for Algal Technology and Innovation, Arizona State University, Mesa, AZ, USA
b
Biosciences Center, National Renewable Energy Laboratory, Golden, CO, USA

A R T I C L E I N F O A B S T R A C T

Keywords: The Development of Integrated Screening, Cultivar Optimization, and Verification Research (DISCOVR)
Algae cultivation collaborative consortium operated pre-pilot scale outdoor ponds to deliver much-needed multi-year, long-term
Outdoor raceway ponds and consistent, algae cultivation data relevant to understanding the current state of technology in terms of ex­
Long-term
pected seasonal algae biomass productivity. Over the course of four years from 2018 to 2021, twelve identical
Crop rotation
Productivity
4.2 m2 mini-ponds were run in triplicate sets to test strains and operational strategies demonstrated in small-,
indoor photobioreactors, in pursuit of increasing overall algae areal productivity and projected farm yield.
Fourteen different cultivars derived from a strain screening pipeline were tested. Through deliberate seasonal
crop rotation and improvements in operational strategies, annual biomass productivity increased from 11.6 to
17.6 g m− 2 day− 1, a > 50 % increase over the 2018 baseline. Both brackish and marine strains were included and
four out of the fourteen strains consistently yielded high productivity across multiple years; brackish strains
Monoraphidium minutum (26BAM) and Scenedesmus obliquus (UTEX393), and marine strains Tetraselmis striata
(LANL1001) and Picochlorum celeri (TG2). These freely available datasets, which represent nearly complete
annual daily coverage of cultivation metrics including weather, pond temperature and pH, nutrients, and pro­
ductivity, are unique in the public domain and seek to fill agronomic and operational knowledge gaps to help in
the eventual commercialization of algal biofuels and bioproducts.

1. Introduction when growing filamentous Oscillatoria strains. These cultures would

however be replaced frequently by native Micractinium and Scenedesmus
Algae-based biofuels have the potential to substantially contribute to strains. For these studies, 1/3 of the pond volume was typically har­
renewable fuel needs however further research and development are vested daily with a 40 % recycle of the harvested biomass. Productivity
needed to improve algal productivity and outdoor pond performance, of S. quadricuada grown in sequential-batch mode was harvested every
reduce risk and uncertainty in large-scale deployment, and inform the few days and averaged 15 g m− 2 day− 1 over an 8-month period from
data gap between assumed and actual long-term agronomic values March to October with continuously diluted (during the day) cultures
[1–4]. There is a dearth of published research on long-term (spanning increasing by 20 %. At Roswell, productivities of 3.5 to 30 g m− 2 day− 1
several years), multi-season, outdoor algal cultivation in shallow were observed with various species in ponds having 3 m2 in surface area
(10–35 cm depth), paddlewheel driven ponds also known as high-rate from winter to summer respectively. While the productivities were very
algal ponds (HRAP). Dedicated research into algae-based biofuels low during the winter, it is interesting to note that even though the
began decades ago with the U.S. Department of Energy's Aquatic Species ponds froze over occasionally, there was still some biomass produced.
Program (ASP). The closeout report [5] presents results of several large- Productivities were typically lower in the large ponds (1000 m2) fluc­
scale open ponds studies summarized here; Outdoor, parallel ponds in tuating from 3 to 18 g m− 2 day− 1 from winter to summer, respectively.
California and Hawaii and ponds of up to 1000 m2 surface area in Single-day productivities of 50 g m− 2 day− 1 were observed but long-
Roswell, New Mexico, were operated non-continuously for 6 years. Ex­ term productivity was approximately 10 g m− 2 day− 1. In other early
periments at the University of California's Richmond Field Station, studies, Cyclotella gracilis reached a maximum of 40 g m− 2 day− 1 for the
generated productivities of 8.5 to 15 g m− 2 day− 1 for a week at a time period of June, July, and August in 0.35 m2 microponds and Isochyris

* Corresponding author.
E-mail address: [email protected] (J. McGowen).

https://doi.org/10.1016/j.algal.2023.102995
Received 9 May 2022; Received in revised form 6 January 2023; Accepted 26 January 2023
Available online 31 January 2023
2211-9264/© 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
J. McGowen et al. Algal Research 70 (2023) 102995

galbana reached a productivity of 23.6 g m− 2 day− 1 after dilutions were on year-round outdoor algal biomass production that could be directly
started for a period of 2 weeks in a 100 m2 pond. compared between one site and another and thus represent a conser­
More recently, productivities for two algal strains in a hybrid PBR vative baseline of non-optimized algal growth that could be expected at
(25 m3 volume)/open pond (400 m2 surface area) configuration in Kona, these sites [13–15]. Throughout the course of 1.75 years (October 2013
Hawaii were reported [6]. Harvest-based ash-free dry weight (AFDW) – July 2015) of monthly cultivation across three algae strains, pro­
productivities of 15.1 and 12.7 g m− 2 day− 1 were reported for Staurosira ductivities of 1.6 to 14 g m− 2 day− 1 were observed during winter to
sp. and 11.9 and 12.8 g m− 2 day− 1 were reported for Desmodesmus sp. summer respectively [13].
C046 in a 2-day or 3-day batch harvest cycles respectively in a low ni­ Despite the above described results, with a few exceptions, achieving
trogen fertilization case. In a high nitrogen fertilization case, AFDW consistent and high biomass productivity remains a limiting factor in the
productivities of 23.9 and 19.7 g m− 2 day− 1 were reported for Staurosira commercial success of algae cultivation for the purpose of biofuels and
sp. and 22.6 and 18.6 g m− 2 day− 1 were reported for Desmodesmus sp. bioproducts. Publicly available, long-term, outdoor algae cultivation
C046 in a 2-day or 3-day batch harvest cycle respectively. These ex­ data at a suitable scale is needed to inform techno-economic analysis
periments were performed over a 4-month period starting in April and (TEA), life cycle (LCA), and resource assessments (RA), predictive
ending in July. The open ponds were 400 m2 of lighted surface area and growth modeling, and crop protection strategies to guide and de-risk the
operated at a depth of 15 cm though it was noted that there was no development of algae agronomic practices. Identifying suitable algae
change in areal productivity with a depth increase to 30 cm. Sapphire species for a given application, understanding their optimal growth
Energy at the Columbus Algal Biomass farm, reported in-situ pro­ conditions and dependence on weather (e. g., temperature and solar
ductivities of approximately 10, 14, and 8 g m− 2 day− 1 and 13, 20, and insolation) and water chemistry (e. g., salinity, pH), susceptibility to
9 g m− 2 day− 1 for 2012 and 2014 respectively during the primary biotic failure-inducing organisms, and response to operational strategies
growing season starting in April, peaking in July, and ending in October (e. g., culture depth, harvest frequency) are all critical to successful
[7]. A 50 % year over year increase was noted in harvest yield pro­ outdoor cultivation. Recently, computational process simulations and
ductivity culminating in the 2014 growing season however detailed modeling efforts (e.g. TEA, LCA, and RA) to predict metrics (e.g.,
harvest metrics were not given. Open algae culturing ponds were either biomass productivity, evaporative losses, pond temperatures, biofuel
1.1 or 2.2 acres in surface area and operated as semi-continuous where productivity, and baseline cost) have been important to better under­
up to 40 % volume was removed and replaced daily with water recycled standing the risks and opportunities associated with outdoor open pond
from the harvesting process. A wastewater treatment demonstration was algae cultivation have been reported [16–19]. Two recent studies
conducted in 5-ha (14,000 m2 including central berm), 35 cm deep, open developed and validated algae growth models using geospatial data (e.g.
ponds over 15 months. Continuous daily harvesting reached 4.4 to 11.5 local weather, evaporation rates), reactor geometry (open pond, PBR)
g m− 2 day− 1 seasonally of colonial algal species that naturally developed inputs, and strain specific parameters (e. g., temperature and light in­
and were dominated by Micractinium and Desmodesmus sp. [8]. Over one tensity tolerance, nightly respiration rate) to predict metrics such as
year of continuous cultivation in 3.5 m2 raceway ponds fed by reclaimed biomass and biofuel productivity. Using publicly available datasets from
municipal wastewater with a novel strain Tribonema minus, cultures the ATP3, such models showed a 0.9 ± 2.35 % and − 4.59 ± 8.13 %
were shown to be more productive than a native algal polyculture, and relative accuracy in respect to productivity thus validating this approach
achieved an annual average productivity of 15.9 ± 0.3 versus 13.4 g [18,19]. Though the data for these modeling efforts is largely provided
m− 2 day− 1 for the polyculture [9]. During a 16-month outdoor open by mid-scale ponds, through the use of both mid-scale (800 L) and large
raceway pond study of Nannochloropsis oculata in 2.5 m3 (12.5 m2) scale (23,000L) ponds, the NAABB concluded that results generated at
ponds of 20 cm depth operated using batch harvests at early stages of mid-scale translated to large scale thus validating that mid-scale pro­
culturing followed by semi-continuous mode to maintain a constant duction systems can act as appropriate research tools to predict expected
biomass concentration, productivities of 0.72 to 3.61 g m− 2 day− 1 were results at larger scales [11].
measured in winter and summer respectively with an overall maximum The U.S. DOE-BETO uses an annual State of Technology (SOT)
of 14.0 g m− 2 day− 1 [10]. analysis to quantify improvements in productivity and help project
Between 2012 and 2017, the U.S. Department of Energy (DOE) future trends in algae cultivation [20–22]. From 2014 to 2017 data for
Bioenergy Technologies Office (BETO) has funded several large collab­ the SOT was generated from seasonal cultivation data by ATP3 with the
orative efforts having as a primary goal, long-term outdoor algae culti­ initial SOT baseline established in 2015 using Nannochloropsis oceanica
vation. One objective of the National Alliance for Advanced Biofuels and KA32 achieving an annual average productivity of 8.5 g m− 2 day− 1 and
Bioproducts (NAABB) was to screen algae strains and develop low-cost increasing to 10.3 g m− 2 day− 1 primarily through the addition of new
culture media with validation in outdoor ponds [11]. NAABB investi­ cultivars and a crop rotation strategy for both cool and warm seasons.
gated 9 strains outdoors at mid- and large- scale (800–23,000 L) Here we describe the continuation of year-over-year outdoor algae
generating productivities of 11 to 25.2 g m− 2 day− 1. Similarly, the cultivation trials with the goal of improving algae productivity first
Regional Algal Feedstock Testbed (RAFT) project was tasked with started under the guidance of the ATP3 and now as part of the Devel­
identifying new algae strains for outdoor pond cultivation and demon­ opment of Integrated Screening, Cultivar Optimization, and Verification
strate high seasonal and annual biomass productivities via crop rotation. Research (DISCOVR) consortium ([23] this issue). These year over year
RAFT achieved 0.9 to 35.2 g m− 2 day− 1 seasonal productivities across 7 data from January 2018 through December 2021 were obtained under
strains in 600–1000 L raceway ponds with production runs of 66–106 mainly semi-continuous cultivation conditions in 4.2 m2 open raceway
days [12]. The objective of a third consortium, the Algae Testbed Public- ponds operated simultaneously year-round under various operational
Private Partnership (ATP3) was to generate a robust dataset of algal conditions. This continuing data pipeline for the an annual SOT analysis
growth metrics in outdoor open ponds with a focus on the comparison of reports, is used as a measure of progress relative to an established target.
harvested biomass productivity in identical ponds under different sea­ t The same data also supports year over year algae agronomics useful to
sonal, climatic, and operational conditions at a small-scale (1025 L, 4.2 the broader algae research community. We explored the primary drivers
m2) by controlling the non-geographical variables of inoculum seed of phototrophic biomass production, light and temperature, and
growth, biomass production systems, processes and protocols, system assessed seasonal effects on productivity for top-performing strains
scale, and algae strain. The Unified Field Studies (UFS) were set up as the identified in DISCOVR's three-tiered strain screening pipeline ([24–27]
baseline upon which later experiments would build upon and as such, no this issue). A strain rotation strategy to maximize annual productivity
attempts at optimization of growth or lipid accumulation were made. through cultivation of the best seasonal strains was utilized, though we
Rather, the primary focus of the UFS were to cultivate algal biomass have also identified cultivars that perform well year-round.
under consistent conditions and harvesting operations to provide data

2
J. McGowen et al. Algal Research 70 (2023) 102995

2. Materials and methods was performed following rigorous methods and sample quality control
protocols as previously described [14,31–34].
2.1. Standard cultivation operations and algae biomass compositional
analysis 2.2. Strain selection, sourcing, and cultivation media

The establishment of the Arizona Center for Algae Technology and The primary source of strains for the SOT cultivation trials performed
Innovation (AzCATI) testbed and standard operational framework for at AzCATI since the summer season of 2018 were from the DISCOVR
conducting SOT cultivation trials under ATP3 was previously described consortium's strain screening and down-selection pipeline which iden­
[13–15,28] and this same framework was utilized for DISCOVR. Briefly, tified candidate cultivars for outdoor evaluation ([25–27] this issue).
outdoor fiberglass ponds (Commercial Algae Professionals, http://www. The primary strains run between 2018 and 2021 under DISCOVR as part
commercialalgae.com) of 820 L nominal volume at a depth of 20 cm and of the annual fiscal year (FY) SOT trials are shown in Table 1. The strains
a total surface area of 4.2 m2 are operated in triplicate for each exper­ that performed best within a given month/season were included in
imental condition (e.g., strain, media, dilution rate, etc.). Pond mixing is formal calculations of the seasonal and annual average productivity and
with a stainless-steel paddlewheel driven by a 1/3 hp. motor (Leeson are highlighted. The media used for a given cultivar is indicated in
model # 191201) and gearbox (IPTS model IBLCS050, 80:1 gear ratio) Table 1. The source of each strain is listed and all strains cultivated since
controlled by a variable-frequency drive (KB Genesis, Model KBDA-24D) 2018 are publicly available or available under material transfer agree­
operated at 20 Hz. Ponds were inoculated after seed scale-up in indoor ments. The majority of outdoor cultivation was performed using either a
flat panel reactors and operated as described previously [15]. The pri­ modified BG-11 media adjusted to 5 ppt salinity for brackish strains or a
mary mode of operation for the majority of the cultivation trials was to modified f/2 media adjusted to 35–50 ppt salinity for marine strains. All
operate the ponds in a semi-continuous fashion, with experiments media used for the SOT were either brackish (5 ppt) or full marine
beginning with an inoculation target density ≥ 0.05 g AFDW L− 1 and salinity (35 ppt) up to 1.5× full marine salinity (50 ppt salinity). No
subsequent grow out to 0.3–0.5 g AFDW L− 1 to trigger harvesting op­ freshwater cultivation was performed for the SOT trials. The modified f/
erations. Ponds were then harvested one to three times a week 2 media is as previously reported [37]. The modified BG-11 media was
depending on the season. Higher productivity led to more frequent composed of: 5 mM NH4HCO3, 0.31 mM K2HPO4, 46.3 μM H3BO3, 0.77
harvests with corresponding higher dilution rates in the summer while μM ZnSO4⋅7H2O, 9.15 μM MnCl2⋅4H2O, 0.17 μM Co(NO3)2⋅6H2O, 1.78
the opposite was true in the winter. On harvest days, ponds are sampled μM NaMoO4⋅2H2O, 0.316 μM CuSO4⋅5H2O, 0.304 mM MgSO4⋅7H2O,
to determine biomass density (g AFDW L− 1), the target percent of pond 0.245 mM CaCl2 ⋅ 2H2O, 22.8 μM C6H9FeNO7, 0.189 mM Na2CO3, 31.2
volume was removed, ponds were re-filled with fresh media, allowed to μM C6H8O7, 3.42 μM C10H16N2O8. To adjust salinity in the brackish
mix, and sampled again to quantify the change in various metrics due to media, 5.2 g of Instant Ocean Sea Salt (www.instantocean.com) was
the harvest (e. g., decrease in OD750 and AFDW, added macronutrients of added to 1 L of media. Additional media formulations utilized included
nitrogen and phosphorous). An estimated daily dilution rate was brackish and marine versions of DISCOVR media ([26] this issue) as well
calculated as the percentage of harvested pond volume multiplied by the as a modified artificial seawater media (MASM) [34]. Water source for
number of harvests in a week divided by 7 days. The majority of pond outdoor cultivation was municipal, potable water (City of Mesa, AZ)
operations followed this standard semicontinuous protocol of 1×–3× which was used through May 2019. A reverse osmosis (RO) system with
weekly morning harvests under nutrient replete conditions, but other in-line ultraviolet (UV) sterilization was installed and has been in use
operational modes were explored. Changes in operational mode since June 2019.
including pond depth, differing harvest/dilution rates, different pH set
points, and for select strains, different crop protection strategies, were 2.3. Data analysis and calculations
compared. While some limited work on nutrient source impact (e.g.,
ammonia versus nitrate) and overall media formulation were conducted, Data were quality checked and compiled as described previously
primarily involving salinity levels for marine strains, media optimiza­ [13]. Briefly, primary cultivation and composition data were collected
tion and media recycling were not the focus for the SOT cultivation trials in Excel spreadsheets which were then compiled based on fiscal year
to date. into one comprehensive file using R scripts [36]. The files contained
Routine daily samples were taken for optical density, AFDW, nutri­ either cultivation and composition data, water chemistry data, or
ents (N and P), pH, salinity, and microscopy. Water quality monitoring weather data. All primary datasets are freely available (https://apps.
and pH control was through the use of a YSI 5200A-DC (YSI Inc., Yellow openei.org/DISCOVR/) and the relevant summarized datasets used in
Springs, OH, USA) water quality monitoring system simultaneously this manuscript are provided as Supplemental File 1. These compiled
measuring pH, pond water temperature (◦ C), dissolved oxygen satura­ yearly data files were then used for analysis. Areal harvest yield pro­
tion (%), and salinity (g L− 1) recorded at 15-minute intervals. The pH ductivities (AHYP) in g algal biomass produced m− 2 pond surface area
was maintained by on-demand sparging with CO2 through a ceramic day− 1 were calculated based the amount of harvested algal biomass that
micro-bubbler diffuser (Sweetwater® Model# DYPFP4, www.pentaira was removed from the ponds as determined by the volume harvested
es.com) triggered by the YSI pH probe. The CO2 supply was turned off and the AFDW at the time of harvest thus representing the actual amount
at night. Weather data was collected on site using a HOBO RX3000 of biomass that would be available for downstream processing. In order
Weather Station (Onset Computer Corporation, MA), including sensors to determine the seasonal and annual averages, we first calculated AHYP
for air temperature (◦ C), relative humidity (%), rainfall (mm), photo­ on a month-by-month basis for each strain/condition by calculating the
synthetically active radiation (PAR), solar radiation (W m− 2), wind sum total biomass harvested during the month, divided by the surface
speed (m s− 1) and direction (degrees), and was collected at 5-minute area of the ponds and total days within that month over which the
intervals. In addition, samples were collected and preserved for pond biomass was harvested (harvested biomass (g)/4.2 m2 * days). The
metagenomic analysis and identification of algal pests, as well as for the following conventions were used for the seasons: Fall = September,
establishment of indoor failure assays for crop protection research as October, November; Winter = December, January, February; Spring =
new pests arise ([29,30] this issue). Finally, samples were also collected March, April, May; and Summer = June, July, August. The seasonal
at harvest points for proximate analysis of biomass composition. For average for a given strain/condition reporting are the average of the
biochemical biomass sampling, approximately 1 L of culture volume was three months within that season. The annual average is calculated based
collected and centrifuged at 4200 rpm for 10 min to collect pelleted on the average of the four seasons within a fiscal year. For any given
biomass, which was preserved by freezing at − 20 ◦ C prior to bulk month or seasonal value used to represent the best performing data for
lyophilization. Proximate composition of the harvested algae biomass that month or season, there is no overlap between strains or conditions

3
J. McGowen et al. Algal Research 70 (2023) 102995

Table 1
Species and strain (Strain ID) grown during the FY SOT trails at AzCATI from 2018 to 2021. Strains
in bold font contributed at least one month of productivity in a given season and FY year to the
formal SOT. The particular season and year a strain was included as part of the SOT calculation is
highlighted in green.

Notes: *Cultivation media indicated with superscript: a) AzCATI modified f/2, b) modified artificial
seawater media, c) AzCATI modified BG-11, d) DISCOVR medium. See Materials and Methods for
specific formulations. The strains N. oceanica KA32, Desmodesmus sp. C046, M. minutum 26BAM
(2018 winter/spring), and S. acutus LRB0401 (2018 winter/spring) were run under the Algae
Testbed Public-Private Partnership (ATP3) [15,28], D. armatus SE00107 (Summer 2018) was run
under Rewiring Algal Carbon Energetics for Renewables (RACER) [35] and S. obliquus UTEX393
(Fall 2020 thru 2021) was run under Decision Model Supported Algae Cultivation Process En­
hancements (DMSACPE, DOE funding award number DE-0008906).

within the month or season when calculating the total days for month or improve AHYP were begun under the ATP3 and generated productivities
season, the overall biomass harvest yield, and thus the AHYP. We for N. oceanica KA32 during all four seasons across a single year ranging
calculate this metric starting in the Fall season of the previous calendar from 2 to 14 g m− 2 day− 1 for winter and summer respectively with an
year and running through the Summer season of the current year to annual average of 8.5 g m− 2 day− 1 as the baseline AHYP for the first year
allow for reporting of an annual productivity number that aligns with of the SOT in 2015 [13–15,28]. Over the course of the next two years,
the federal fiscal year reporting requirements. Thus, formal SOT annual this was improved to 10.3 g m− 2 day− 1 in 2017, an increase of 21 %
averages cross calendar years (CY). When making any seasonal/annual through strain rotation. More productive cultivars relative to N. oceanica
summary calculations it is indicated if it is a function of FY running from KA32 were found for both seasons, Desmodesmus sp. C046 in the summer
September of the previous year through August of the current year, or and M. minutum 26BAM in the winter [13,20,28]. This performance
CY, from January through December of the same year. baseline and experimental framework was continued as part of the
DISCOVR consortium starting in the year 2018 with outdoor cultivation
3. Results and discussion at the AzCATI testbed site.
Using the DISCOVR strain screening pipeline ([25–27] this issue),
3.1. Outdoor pond cultivation outdoor algae cultivation at AzCATI was used as a final testing arena for
strains showing promise across multiple stages for a given selection
Outdoor algae cultivation experiments with the goal of setting a strategy. Different species were run in a given season and focused on the
baseline for algal AHYP and developing a framework within which to optimal climate conditions for a given strain to determine the best

4
J. McGowen et al. Algal Research 70 (2023) 102995

performing strains and to develop a crop rotation strategy for maximal moving forward as evidenced by the drop in 2021, but fortunately a
annual productivity. The actual feasibility or need for a crop rotation more modest 4–5 % improvement year over year is all that must be
strategy in a commercial algae cultivation facility is not yet fully clear demonstrated to achieve both the 2025 and 2030 targets. By way of
but as with terrestrial crops, which are grown in specific climates based comparison, agricultural crop yield improvements for soy and corn have
on optimal light and temperature, it will likely be similar for many shown an average of 2 % increase year over year for the last 30 years
commercial scale algae farms depending on their location and seasonal (https://www.nass.usda.gov).
climates. In our case at AzCATI, given the northern geographic location Fig. 1 shows the minimum and maximum water temperature (◦ C) for
relative to the equator and thus experiencing yearly seasons, crop 20 cm deep ponds along with the daily light integral (DLI, mol m− 2
rotation was necessary because there was no single strain that effectively day− 1, calculated as the number of photosynthetically active photons
covered the range of temperatures experienced at the site across the year (PAR, photons in the 400–700 nm wavelength range) accumulated in a
([26] this issue). In contrast, a gulf coast site or sites further south closer square meter over the course of a day) from January 1, 2018 through
to the equator, may not require crop rotation from a seasonal temper­ December 31, 2021. The Mesa, AZ AzCATI site offers a unique and
ature tolerance aspect. effective testbed location for exploring the effects of natural light and
A primary goal of the SOT trials under DISCOVR is the selection of temperature. Though there are routinely over 300 days of sunshine per
best seasonally performing strains. The main objective was to cultivate year, there is significant seasonality in solar insolation and temperature.
those strains under real-world, varying weather conditions, and work to That light and temperature are the primary drivers is evident in Fig. 2
improve their outdoor performance while simultaneously verifying if showing the monthly AHYP for CY 2018–2021. Seasonal transitions
newly identified cultivars or operational concepts in the DISCOVR from fall to winter and again from winter to spring show the fastest
pipeline which demonstrate improved performance indoors, hold up changes in both light and temperature and correlates to rapid increases
once taken out to the field. A summary of the seasonal and overall and decreases in productivity for spring and fall, respectively. This is
annual average productivities for the top performing strains for the FY reflected in the higher variability in the seasonal average AHYP for fall
SOT from 2018 to 2021 are shown in Table 2. Under ATP3, annual and spring seasons relative to winter and summer across the short three-
average AHYP were 8.5, 9.1, and 10.3 g m− 2 day− 1 for 2015, 2016, and month transitional seasons where harvest yields can almost double from
2017, respectively [22]. Under DISCOVR (2018–2021), annual AHYP March to May and decrease by a similar amount from September to
was advanced from 11.6 g m− 2 day− 1 in 2018 to 17.6 g m− 2 day− 1 in November (Fig. 2, Table 3).
2021, a 52 % increase. Since the establishment of the U. S. DOE SOT When the SOT was first established under the ATP3, the trials were
benchmark for AHYP FY cultivation metrics in 2015, annual AHYP has conducted on a seasonal basis but would often include multiple strains
more than doubled, improving on the 2015 baseline of 8.5 g m− 2 day− 1 and or operational conditions being tested with only six replicate ponds
to 17.6 g m− 2 day− 1 in 2021, a 108 % increase. In the first 6 years since available for a given testbed site. Thus, cultivation trials within a season
the inception of the DOE SOT cultivation trials in FY 2015, productivity were typically on the order of 30–45 days allowing for two rounds of
improvements of 7 %, 13 %, 14 %, 36 %, and 16 % (2016–2020) relative experimentation in a season usually with different strains, and resulted
to each preceding year were achieved. However, in 2021 there was 4 % in gaps in seasonal coverage for a given strain/condition [13]. To limit
reduction in annual AHYP relative to 2020. The decrease in productivity gaps in monthly/seasonal data which can increase the uncertainty in
in 2021 was driven primarily by a significant decrease in summer pro­ extrapolation of productivity estimates across a full season or year,
ductivity relative to 2020 of 25 %. The reasons for that decrease remain DISCOVR set an explicit goal to maximize seasonal and thus annual
unclear, though they are actively being investigated by DISCOVR and coverage throughout the full calendar year, limiting strain turnovers
will be briefly discussed in Section 3.3. during any given month or season. For DISCOVR, the cultivation trials
A benchmark, technical and economical, performance target was set were expanded to twelve ponds to allow for more conditions to be tested
for demonstrating an average productivity of 25 g m− 2 day− 1 by 2030 side-by-side (e.g., strains, operational set points, etc.) with an explicit
with an intermediate goal of 20 g m− 2 day− 1 by 2025, against which the operational target of maximizing the overall number of days of experi­
annual SOT will be compared and reported [37]. It is unlikely the sub­ mental uptime across the year. This resulted in three benefits for the
stantial year over year improvements observed from 2015 to 2020, even annual SOT cultivation trials; 1) better temporal resolution in particular
at the smaller, pre-pilot scale being run for DISCOVR, can be sustained across the rapid change in light and temperature for spring and fall

Table 2
FY SOT seasonal and annual average AHYP for 2018 through 2021.
FY CY Season No. months AHYP (g/m2 day) Total days season Annual average AHYP Total days year Percent annual increase
a
2018 2016 Fall 2 9.0 ± 2.1 42
2018 2018 Winter 2b 7.7 ± 1.6 46
2018 2018 Spring 3 14.8 ± 3.1 78
2018 2018 Summer 3 14.9 ± 1.3 56 11.6 ± 3.8 222 14%c
2019 2018 Fall 3 11.3 ± 1.6 66
2019 2019 Winter 3 6.4 ± 0.08 94
2019 2019 Spring 3 18.6 ± 6.0 88
2019 2019 Summer 3 27.1 ± 2.8 87 15.9 ± 9.0 335 37 %
2020 2019 Fall 3 15.0 ± 3.6 87
2020 2020 Winter 3 8.4 ± 1.2 93
2020 2020 Spring 3 18.4 ± 4.5 93
2020 2020 Summer 3 31.6 ± 3.9 81 18.4 ± 9.8 354 16 %
2021 2020 Fall 3 19.3 ± 8.3 92
2021 2021 Winter 3 8.3 ± 1.2 90
2021 2021 Spring 3 19.4 ± 3.7 88
2021 2021 Summer 3 23.8 ± 0.9 90 17.6 ± 6.6 360 ¡4 %
a
Fall data for FY 2018 was carried forward from fall CY 2016 as no cultivation work was conducted for the SOT in fall CY 2017.
b
The Fall and Winter FY2018 seasons included data from only 2 months, October/November CY 2016 and January/February CY 2018, respectively. The remaining
SOT years had cultivation from all 12 months of the year. AHYP is the average ± 1 standard deviation of the mean. Total days represents the total number of cultivation
days within the season that contributed to the seasonal average.
c
Percent increase over FY 2017 SOT under the ATP3.

5
J. McGowen et al. Algal Research 70 (2023) 102995

Fig. 1. Daily pond water temperatures (◦ C) (maximum (red), minimum (blue)) and DLI (mol m− 2 day− 1 (green)) from January 1, 2018, through December 31, 2021.
Note that no ponds were running on site from December 18, 2018, to January 7th, 2019, producing a minor gap in the water temperature data. (For interpretation of
the references to colour in this figure legend, the reader is referred to the web version of this article.)

seasons improving certainty for calculations of seasonal and annual 2019 to 2021 showing its best productivity in CY 2019 (annual AHYP of
average productivities, 2) the agronomic comparisons became more 15.5 g m− 2 day− 1). However, cool season performance for S. obliquus
robust with monthly resolution as we generate year over year compar­ UTEX393 considerably lags behind that of M. minutum 26BAM, with a
isons for the best performing strains within the SOT and, 3) the ability to three-year average productivity of 5.8 and 15.9 g m− 2 day− 1 for winter
extrapolate data (e. g., productivity) to other geographic locations as and spring seasons, respectively, versus 7.8 and 17.4 g m− 2 day− 1 for
different parts of the year (e. g., winter and transitional seasons) may M. minutum 26BAM. Comparing two-year averages for 2020 and 2021,
simulate closely other locations (e. g., more Northern locations) as well T. striata LANL1001 had an average AHYP of 7.9 and 15.6 g m− 2 day− 1
as allow for more robust, validated models of biomass productivity for winter and spring seasons, respectively, versus 8.2 and 18 g m− 2
within the algae research and development community with some use of day− 1 for M. minutum 26BAM. From 2018 to 2021 we improved on the
this data for such modeling already demonstrated [18,19]. baseline productivity of M. minutum 26BAM year over year in winter,
spring, and fall seasons of 7.5 %, 30.9 %, and 10.2 %, respectively
3.2. Strain selection and crop rotation relative to 2018.
T. striata LANL1001 (marine, 35 ppt), close in outdoor performance
Between 2018 and 2021 fourteen different strains vetted in the to M. minutum 26BAM, was selected as the second-best cool weather
DISCOVR strain selection pipeline were tested outdoors at AzCATI strain and the best cool weather marine strain tested, outperforming two
(Table 1, Supplemental Fig. 1). DISCOVR's overall strategy for other cool weather marine strains M. reisseri 14F2 and P. tricornutum
improving AHYP on an annual basis implies a crop rotation strategy and UTEX646, all showing comparable performance to M. minutum 26BAM
incudes cultivars shown to perform better in cool weather (e.g., in the DISCOVR indoor testing pipeline ([26,27] this issue). While no
M. minutum 26BAM, T. striata LANL1001, M. reisseri 14F2, Chlamydo­ cool weather strains matched or exceeded the productivities for
monas sp. PATC1, P. tricornutum UTEX646) and those that perform M. minutum 26BAM outdoors, T. striata LANL1001 demonstrated supe­
better in warm weather (e.g., S. obliquus UTEX393, P. celeri TG2, rior robustness with no culture crashes in 2020 or 2021, and no need for
P. renovo 39A8, P. soloecismus DOE101, P. cruentum CCMP675). The any active crop protection measures. We have not yet optimized nor
monthly and annual average AHYP for the top performing SOT strains expended a similar amount of effort on T. striata LANL1001 as with
on a CY basis (2018–2021) are shown in Fig. 2 and summarized in M. minutum 26BAM, but expect to be able to improve productivity
Table 3. Since the start of DISCOVR, assuming crop rotation and using through additional cultivation optimization in the future, in particular
the single best performing strain in a given month, an increase from 12 g as it relates to dilution rate (see Section 3.4).
m− 2 day− 1 in 2018 to 17.0 g m− 2 day− 1 was achieved, an improvement Broad, seasonal gains in productivity have been achieved from 2018
of 41.6 % on a CY basis. Of all the strains tested to date, four have to 2021, with the largest gains occurring in the warmer seasons. With
consistently been the top performers across multiple years; brackish the introduction of P. celeri TG2, first cultivated in August and
strains M. minutum 26BAM and S. obliquus UTEX393, and marine strains September of 2019, we obtained the highest productivities at AzCATI for
T. striata LANL1001 and P. celeri TG2 (Fig. 3). summer and early fall in 2020 with a four-month average (June –
The top performing cool weather strain to date has been M. minutum September) of 31 g m− 2 day− 1, and a single month high of 36 g m− 2
26BAM and while we have observed better performing cool weather day− 1 for August. As reported, P. celeri TG2 has significant potential for
strains indoors ([26] this issue), we have not as of yet found another biofuel production, with high productivity, and demonstrated genetic
strain that once taken outside to AzCATI has outperformed M. minutum engineering toolkits available [38,39]. It is now the current benchmark
26BAM. S. obliquus UTEX393 has proven to be a versatile, all-season, warm weather marine strain for the DISCOVR SOT exhibiting high
brackish strain which has been tested year-round for three years from temperature and salinity tolerance (majority of data in 2020 and 2021

6
J. McGowen et al. Algal Research 70 (2023) 102995

Fig. 2. Monthly and seasonal AHYP using best performing strains for annual SOT by calendar year (January 2018 through December 2021), solid line through
discrete points represents a spline fit to the data. Estimated monthly and average seasonal daily dilution rates (top graphs, left and right respectively) and monthly
and seasonal AHYP (bottom graphs, left and right respectively). For the monthly AHYP values, each bar graph represents the single best performing strain for the
month/year (n = 3 ponds). For seasonal summary, each bar graph represents the average values for three months within a season (n = 9). Error bars are ±1 standard
deviation from the mean.

was at 50 ppt salinity). It also was quite robust with no pond crashes in scale, no fresh water use for make-up, and thus there is a need to
2019–2021, including running over 143 days from mid-June thru the minimize environmental and economic impacts of blow down and
end of October 2020 without a pond crash or need to re-inoculate ponds strains that can tolerate higher salinity are favored [16–19]. Media
due to a drop in productivity. However, one issue observed with P. celeri recycle or allowing for increasing salinity during cultivation due to
TG2 was contamination by diatoms, which was common in all three evaporation has not been a part of routine SOT cultivation trials to date.
years, in particular for late summer and early fall of 2021 leading to In addition, for the formal reported productivity calculations shown in
several restarts once the diatom population exceed roughly 10 % (by cell Table 2 and Table 3, the calculation of the seasonal values is not
count). The presence of diatoms can lead to an increase in auto- segregated by media type as the current intent of the formal annual
flocculation and buildup of biomass at the water's edge and on the reporting is to provide the best monthly/seasonal productivities,
paddlewheels and cause increased settling of biomass. The effect of regardless of media type. However, the formal calculation of the key
diatom contamination on primary productivity for P. celeri TG2 is as of metric, minimum biomass selling price (MBSP), does account for the
yet undetermined. actual salinity for a given cultivar and the expected costs of blow-down
to maintain a strain at its optimal salinity target [21,22].
A given algae cultivation facility will more likely only have one type
3.3. Media selection: brackish versus marine strains of water available of a narrow salinity range and so it is more
commercially relevant to look at trends in AHYP based on media type.
DISCOVR has looked at both brackish and marine strains typically Fig. 4 shows the monthly AHYP values for 2019, 2020, and 2021 based
assuming a nominal 5 ppt or 35 (and higher) ppt salinity, respectively on either a brackish strain rotation or a marine strain rotation scenario
(see “target salinity’ in Table 1). Strains are evaluated for salinity with the top four DISCOVR strains. Seasonal and annual averages based
tolerance under the Tier 1 screening for DISCOVR ([26] this issue). on a brackish or marine media are summarized in Table 4. The most
Higher salinity tolerance as a function of absolute value as well as productive year for brackish strains was 2019 with an annual average
tolerance to changing salinity, is a key performance metric due to ex­ productivity of 16.9 ± 8.4 g m− 2 day− 1 and 2020 for marine strains with
pected increases in salinity over time during continuous/semi- an annual average productivity of 18.6 ± 10.1 g m− 2 day− 1. In both
continuous cultivation due to evaporation. The expectation is that at

7
J. McGowen et al. Algal Research 70 (2023) 102995

Table 3 the reasons for this large drop in productivity including exploring a
Seasonal and annual average CY AHYP for the best performing strains for 2018 combination of; 1) retrospective biomass productivity modeling via the
through 2021. PNNL Biomass Assessment Tool (BAT) [41] using the actual temperature
Calendar Season AHYP Total Annual % Annual Total and light observations for 2020 and 2021, 2) biotic differences through
year (g m− 2 days/ average increase days/ evaluation of preserved pond samples for pond metagenomic analysis,
day− 1) season AHYP AHYP year and 3) indoor lab experiments with climate-simulation photobioreactors
2018 Winter 7.1 ± 84 running water temperature and DLI scripts for specific seasons. Under­
1.7 standing year over year fluctuations in biomass productivity and the
2018 Spring 14.8 ± 78
drivers of those fluctuations are a key aspect to developing best agro­
3.1
2018 Summer 14.9 ± 56 nomic practices and gaining the experience to manage long term algae
1.3 cultivation at scale.
2018 Fall 11.3 ± 66 12.0 ± N/A 284
1.6 3.7
3.4. Operational strategies improve AHYP
2019 Winter 6.9 ± 90
0.6
2019 Spring 18.6 ± 88 While strain selection has driven a significant portion of the gains in
6.0 productivity, successfully cultivating algae at commercial scales will
2019 Summer 27.1 ± 87 require the development of best agronomic practices in order to over­
2.8
come the yield gap between indoor lab and outdoor field performance
2019 Fall 15.0 ± 87 16.9 ± 41 % 352
3.6 8.4 and requires identifying optimal operating conditions for a given strain/
2020 Winter 8.3 ± 88 season to yield improvements in productivity and robustness [3,5,7].
1.2 There are many parameters that need to be managed in a rapidly
2020 Spring 18.4 ± 93
changing environment where the key drivers of productivity, light and
4.5
2020 Summer 31.6 ± 81
temperature, remain largely uncontrollable. The DISCOVR SOT trials
3.9 have focused primarily on understanding the performance of different
2020 Fall 19.3 ± 92 19.4 ± 15 % 354 strains when run outdoors in varying environmental conditions under
8.3 9.5 non-limiting conditions for nutrients and CO2. The annual cultivation
2021 Winter 8.7 ± 89
trials operate under a recurring cycle of comparing a benchmark strain
0.8
2021 Spring 19.4 ± 88 for a given season (i.e., the best performing strain in the previous year
3.7 for that season) side-by-side with any new strain that has made it
2021 Summer 23.8 ± 90 through the DISCOVR pipeline while also adjusting operational pa­
0.9 rameters that can improve performance (e.g., depth, pH, dilution rate),
2021 Fall 16.2 ± 91 17.0 ± − 12 % 358
3.7 6.4
and as needed, the implementation of crop protection strategies. This
year over year experience with a given strain allows valuable experience
to be gained in developing best agronomic practices and improving
cases significant improvement in seasonal and thus annual pro­ productivity.
ductivities was achieved with the largest % increase in summer and fall It has been shown that culture depth and dilution rate are two factors
seasons for both media types relative to 2018. that need to be optimized in algal biomass production. Depth and
While steady progress was made over the last 4 years, periodic de­ dilution rate can be a means to regulate light availability and pond
clines were experienced year over year for particular seasons. In summer temperature providing a mechanism to keep a particular strain as close
and fall of 2020, a significant decline in productivity of 23 % was to its optimal production rate as a function of seasonal changes
observed for S. obliquus UTEX393 relative to 2019. This decline was [5–8,42–44]. Fig. 6 shows examples of comparisons for some of the key
attributed to a new bacterial pest that appeared on site in the spring of control parameters that are routinely optimized under the SOT for a
2020 and significantly decreased productivity for this cultivar (Fig. 4) given strain/season. Culture depth comparisons, such as shallower
and also significantly decreased robustness with a sharp decrease in depths in the winter to improve light availability, were evaluated for
mean time to failure (MTTF) relative to 2019 (see Section 3.5). In 2021, M. minutum 26BAM in winter as well as spring seasons with significantly
we also observed a year over year drop in the summer and early fall for higher biomass concentration at harvest observed at shallower culture
P. celeri TG2 relative to its peak in summer of 2020 with a decline of depths (1.8-2× higher for 10 cm vs 20 cm). However, there was no
almost 25 % for the summer season in 2021. The reasons for this year statistically significant impact on areal productivities (i.e., AHYP g m− 2
over year drop are less clear than the year over year decline observed for d− 1, Fig. 6a). In warmer seasons with more available light, this rela­
S. obliquus UTEX393 as we did not see any significant difference in tionship held (i.e., higher biomass concentration at harvest, but similar
contamination and little to no grazing or other weedy algae in P. celeri overall harvest yields), but as the risk of culture overheating can in­
TG2 cultures. The monsoon season of 2021 was more active relative to crease at shallower culture depths relative to a deeper water column
2020 with higher dust levels and more clouds leading to decreased solar [42], these comparisons were not explored in summer months with
insolation especially for July 2021. Fig. 5 shows the maximum and warm season strains. Given that the temperature profiles of small, pre-
minimum water temperatures and DLI for June thru September for pilot raceways do not mimic well the temperature profiles at large
2019–2021 as a function of calendar day and month. The monthly scale (i.e., >1 acre ponds) [18,19], a standardized depth of 20 cm was
average DLI for July 2021 was 10–15 % below that for 2020 and set for the SOT runs in 2020 for experimental simplification. Culture
morning water temperatures were slightly higher in the morning with a depth may continue to be explored in future SOT trials.
tighter range indicating overall warmer overnight temperatures relative Another major operational parameter evaluated was pH. That
to 2020. The combination of lower overall available light and warmer biomass productivity for algae are highly dependent on pH is well
nighttime temperatures may be expected to decrease biomass produc­ known and has been studied in natural systems as well as in open ponds
tivity through lower daytime productivity, and has the potential for and photobioreactors [45–47]. When the SOT cultivation trials were
increased nighttime biomass loss due to increased respiration [40]. first begun, the majority of cultivation in 2018 through Spring of 2019
However, August and September 2021 had the largest year over year were conducted at a pH setpoint of 7.9. In spring of 2019, DISCOVR
decreases in productivity (~34 %) but light and temperature differences began screening for effects of pH on productivity and showed for both
were much more modest relative to 2020. DISCOVR continues to explore M. minutum 26BAM and S. obliquus UTEX393 that a lower pH setpoint of

8
J. McGowen et al. Algal Research 70 (2023) 102995

Fig. 3. Monthly average AHYP (g m− 2 day− 1) by seasonality of strain (top graph) and media salinity (bottom graph). Monthly averages are the average across three
years for M. minutum 26BAM and S. obliquus UTEX393 (2019–2021) and the average across two years for T. striata LANL1001 and P. celeri TG2 (2020–2021, except
June for LANL1001 which is CY 2020 only). Error bars are ±1 standard deviation from the mean (n = 9 for M. minutum 26BAM and T. obliquus UTEX393, and n = 6
for T. striata LANL1001 and P. celeri TG2.

Fig. 4. Monthly AHYP from January 2019 through December 2021 for brackish (blue) and marine (red) top four cultivars. Warm weather strains indicated by bars
with cross hatching. For the monthly AHYP values, each bar graph represents the single best performing strain for the month/year (n = 3 ponds) and for a given
media condition (either brackish, 5 ppt salinity, or marine, 35–50 ppt salinity). Error bars are ±1 standard deviation from the mean. (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of this article.)

7.0 had a significant effect improving AHYP 20–30 % (Fig. 6b). In lower pH set points, significant outgassing of CO2 will occur [46] and
addition to higher growth rates, greater robustness for many strains was thus strategies to maintain higher productivity at conditions with lower
observed at the lower pH setpoints. While lower pH did greatly improve CO2 losses to increase carbon utilization are important to develop.
productivity for top performing DISCOVR strains, running at lower pH Daily dilution rate is another operational parameter that affects
has limitations, in particular for carbon utilization efficiency as at these productivity. As shown previously in Fig. 2, the estimated daily dilution

9
J. McGowen et al. Algal Research 70 (2023) 102995

Table 4
Seasonal and annual calendar year average AHYP for the best performing strains used in annual performance improvement calculation for 2018 through 2021.
Media CY Season No. Avg. AHYP (g Avg. AFDW at Avg. est. daily Harvests / Total days/ Annual avg. (g Total
type months m− 2 day− 1) harvest (g L− 1) dilution week season m− 2 day− 1) days/year

Brackish 2019 Winter 3 6.9 ± 0.6 0.426 ± 0.09 0.12 ± 0.02 1×-2× 90
Brackish 2019 Spring 3 18.6 ± 6.0 0.602 ± 0.12 0.25 ± 0.07 2×-3× 88
Brackish 2019 Summer 3 27.1 ± 2.8 0.430 ± 0.05 0.32 ± 0.02 3× 87
Brackish 2019 Fall 3 15.0 ± 3.6 0.355 ± 0.00 0.22 ± 0.07 2×-3× 87 16.9 ± 8.4 352
Brackish 2020 Winter 3 8.3 ± 1.2 0.536 ± 0.22 0.13 ± 0.01 1×-2× 88
Brackish 2020 Spring 3 18.4 ± 4.5 0.431 ± 0.17 0.28 ± 0.06 2×-3× 93
Brackish 2020 Summer 3 20.7 ± 3.5 0.377 ± 0.02 0.28 ± 0.04 3× 93
Brackish 2020 Fall 3 15.8 ± 5.7 0.330 ± 0.06 0.23 ± 0.06 2×-3× 77 15.8 ± 5.4 351
Brackish 2021 Winter 3 8.1 ± 1.6 0.330 ± 0.05 0.12 ± 0.02 1×-2× 89
Brackish 2021 Spring 3 19.4 ± 3.7 0.408 ± 0.04 0.24 ± 0.04 2×-3× 88
Brackish 2021 Summer 3 20.2 ± 2.4 0.349 ± 0.10 0.22 ± 0.00 2×-3× 28
Brackish 2021 Fall 3 14.5 ± 2.5 0.347 ± 0.07 0.20 ± 0.01 2×-3× 63 15.6 ± 5.6 268
Marine 2020 Winter 3 7.2 ± 1.6 0.376 ± 0.08 0.10 ± 0.04 1×-2× 84
Marine 2020 Spring 3 16.5 ± 2.9 0.324 ± 0.02 0.26 ± 0.07 2×-3× 90
Marine 2020 Summer 3 31.6 ± 3.9 0.497 ± 0.08 0.34 ± 0.00 3× 81
Marine 2020 Fall 3 19.1 ± 8.4 0.354 ± 0.10 0.27 ± 0.08 2×-3× 92 18.6 ± 10.1 347
Marine 2021 Winter 3 8.7 ± 0.9 0.318 ± 0.06 0.14 ± 0.02 1×-2× 89
Marine 2021 Spring 3 17.1 ± 3.9 0.354 ± 0.02 0.24 ± 0.05 2×-3× 84
Marine 2021 Summer 3 23.8 ± 0.9 0.378 ± 0.02 0.32 ± 0.02 3× 90
Marine 2021 Fall 3 13.2 ± 5.2 0.306 ± 0.01 0.21 ± 0.08 2×-3× 74 15.7 ± 6.4 337

Fig. 5. Daily maximum (red) and minimum (blue) water temperature (◦ C) and DLI (green) (mol m− 2 day− 1) by calendar day (left plot) and month (right plot). (For
interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

rate has a strong correlation with productivity with higher dilution rates 3.5. Crop protection and pest management
during higher productivity seasons and lower dilution rates during
lower productivity seasons. For the DISCOVR SOT trials, dilution rate is One of the most challenging aspects of achieving and maintaining
a parameter that is primarily managed empirically with minor adjust­ high algae biomass productivities at scale is controlling or avoiding
ments based on the observed growth rate and culture density at harvest. completely weedy algae, microzooplanktonic grazers, and fungal and
Our biomass density targets were to keep harvest density in the range of bacterial parasitoids, all which can have a devastating effect on biomass
0.3 to 0.5 g L− 1 at harvest and ≥0.1 g L− 1 post-harvest. On a seasonal quality and quantity. Thus, successful large-scale algae cultivation at
basis, optimization of dilution rate for a given strain is determined commercial scales requires the development of effective crop protection
directly by comparing higher or lower dilution rates side by side. Fig. 6c and integrated pest management [3,5,7,11,12,49–52]. Under ATP3,
shows examples for three of the top performing strains showing the ef­ frequent culture crashes with both marine and freshwater strains were
fect of dilution rate on AHYP for T. striata LANL1001 in the winter, reported including from grazers, weedy algae strains, and fungal para­
P. celeri TG2 in early fall, and S. obliquus UTEX393 in the spring. All sitoids, with contamination and thus the risk of pond crashes highest in
showed an increase in AHYP at higher dilution rates. More optimization the warmer seasons. Metrics were established to better quantify pond
is certainly needed, in particular the ability to move away from an failure and thus reliability, and allow for quantitative tracking to pro­
empirical, trial and error driven approach to managing dilution rate vide insight into potential pond management and contaminant mitiga­
more dynamically [5–8,11,48]. This remains a primary focus of the tion [53]. Beginning in late Spring through the Summer and early Fall of
DISCOVR annual performance, in particular for the warmer season CY 2018, significant culture crashes with brackish strains S. acutus
strains. LRB0401, M. minutum 26BAM, S. obliquus UTEX393, and D. armatus
SE00107 (18 ppt), and the marine strain Desmodesmus sp. CO46 were

10
J. McGowen et al. Algal Research 70 (2023) 102995

Fig. 6. Average harvest yield productivity (AHYP) comparisons for different operational setpoints (n = 3 ponds for each bar graph/condition) a.) M. minutum 26BAM
cultivated side by side at two culture depths of 10 cm or 20 cm from January 7, 2019 thru February 28th, 2019 and again from December 31, 2019 thru January 31,
2020 (n = 3 ponds per depth). Error bars represent ±1 standard deviation from the mean, the average AFDW at harvest is listed on each bar graph. b.) M. minutum
26BAM and S. obliquus UTEX393 cultivated at two different pH setpoints (SP) in April 2020 for 14 days (M. minutum 26BAM) and May 2019 for 18 days (S. obliquus
UTEX393). Error bars represent ±1 standard deviation from the mean with the percent improvement of lower pH relative to higher pH setpoints. c.) three strains
cultivated under two different dilution rate regimes, high dilution rate (HD), low dilution rate (LD). Harvest frequency was 3× per week for all conditions changing
only the volume of harvested culture at each set point. Average daily dilution is listed in each bar graph along with percent improvement of HD versus LD. Cultivation
dates for each strain are shown at top of each graph). Error bars represent ±1 standard deviation from the mean.

observed which limited overall annual productivity. In all cases of pond reducing fungal infections allowing for improved and sustained pro­
crashes in CY 2018, the main contaminant appeared to be algal para­ ductivity with S. dimorphous [52] and D. armatus SE00107 [54]. The
sitoids which were assumed to be fungal or fungal-like based on the active agent in the broad-spectrum fungicide used for D. armatus
morphology observed via microscopy. These algal parasitoids caused a SE00107 was fluazinam (3-chloro-N-(3-chloro-2,6-dinitro-4-tri­
rapid decline in productivity with cultures turning brown within a few fluoromethylphenyl)-5-trifluoromethyl-2-pyridinamine). A version of
days of first observations of infection. The morphology of the crashed this fungicide, Secure® (manufactured by Syngenta) has been utilized at
cultures looked different depending on strain, but specifically for AzCATI since 2018, beginning with D. armatus SE00107 and then
S. acutus LRB0401, S. obliquus UTEX393, and M. minutum 26BAM, crash expanding to both S. obliquus UTEX393 and M. minutum 26BAM in 2019.
morphology was similar to examples of algal parasitoids isolated and Fig. 7 shows the initial trials with fluazinam outdoors with S. obliquus
identified from ponds in New Mexico which infected S. obliquus and UTEX393 where an active fungal parasitoid infection observed via mi­
D. armatus SE00107 cultures and reported to be in the phylum Aphelida croscopy early in a cultivation trial led to a sharp decline in biomass
[50,51]. productivity. Two out of three replicate ponds were treated with
Chemical treatment protocols using commercially available fungi­ Secure® at an application rate of 1 ppm of the active agent every seven
cides has been reported in the literature and shown to be effective at days (4 applications across 3 weeks). The third pond was left untreated.

Fig. 7. AHYP (top) and AFDW (bottom) for

S. obliquus UTEX393 spring 2019 cultivation trial.
Parasitic fungal infection was first observed via mi­
croscopy on 3/30/2019. Ponds treated with (SPW10
and SPW12) and without (SPW14) 1 ppm of fluazi­
nam. Fungicide application began on 4/8/2019 (post-
harvest) and continued weekly for 3 weeks (4 appli­
cations total). The untreated pond crashed on 4/24/
2019. SPW14 was restarted from SPW12 on 4/26/
2019. The treated ponds fully recovered and did not
experience any further infection for >30 days from
last fungicide application. Decline in productivity in
late May was due to cloudy/cool weather.

11
J. McGowen et al. Algal Research 70 (2023) 102995

As shown in Fig. 8, S. obliquus UTEX393 ponds were heavily contami­ untreated ponds, respectively.
nated by early April with classic signs of fungal parasitoid infection (e.g., Outdoor cultures were tested using the polymerase chain reaction
ghost cells and residual bodies) and the infection had progressed (PCR) with primer sequences of known parasitoid strains in the litera­
significantly by time of first application. Within 2 weeks treated ponds ture, in particular those identified and isolated at Sapphire Energy,
fully recovered and the untreated pond turned brown and crashed. including parasitoid strains A. protococcarum FD95, A. occidentale FD01,
Dosing was stopped at the end of April and cultivation continued with and A. desmodesmi FD104 [50,51]. The progression of infection and life
the recovered culture through the end of May 2019. Productivity for cycle for the fungal parasitoids we observed in S. obliquus UTEX393 and
May 2019 was the highest sustained productivity ever observed at M. minutum 26BAM were very similar to those observed at Sapphire
AzCATI to date achieving >20 g m− 2 d− 1. Ponds were restarted with Energy and thus these pests were our first target for identification. An
fresh inoculum for the summer 2019 season with implementation of a example of a fungal parasitoid lifecycle in M. minutum 26BAM is shown
fungicide-based pest management routine initiated at the first sign of in Fig. 9b-g. This is typical for what has been observed in M. minutum
infection in ponds. With this implementation, the highest summer pro­ 26BAM and S. obliquus UTEX393 with zoospores that appear to be
ductivities at AzCATI to date were achieved with S. obliquus UTEX393 amoeboid (as opposed to flagellated). Through early 2020, outdoor
with a summer seasonal average of 27.1 g m− 2 d− 1, an 82 % increase cultures of both S. obliquus UTEX393 and M. minutum 26BAM routinely
year over year 2018 to 2019 with no culture crashes from May through tested positive for FD01, but negative for FD95 and FD104. In addition,
October demonstrating the significant impact of pest mitigation. Ponds D. armatus SE00107, the original host for FD104 isolation from outdoor
were restarted with fresh culture at the start of summer season and then ponds in New Mexico and which routinely crashed when cultivated at
again in August due to a cyanobacterial contamination (when contam­ AzCATI, were negative for FD104 (as well as FD01 and FD95) and thus
ination exceeded ~10 % of cells), but no decline in productivity was the algal parasitoid infecting in D. armatus SE00107 cultures at AzCATI,
observed. while controllable by fungicide, as of yet remains unidentified. Since
S. obliquus UTEX393 showed significant tolerance to fluazinam with spring of 2020, cultures that are positive for FD95, FD104, or FD01 have
no apparent trade-off in productivity (at doses up to 2 ppm) unlike not been observed via PCR, yet cultures continued to crash due to
D. armatus SE00107, Desmodesmus sp. C046, and M. minutum 26BAM apparent fungal parasitoids and are controllable through the use of
which all show significant productivity declines upon application of fungicides which indicates one (or more) as yet unidentified strains of
0.5–1.0 ppm. Despite the lower tolerance to fungicide, we were suc­ fungal parasitoids may be present. Work to isolate and identify these
cessful at developing dosing regimes for both D. armatus SE00107 and, fungal parasitoids is ongoing.
in particular M. minutum 26BAM, where protection from fungal para­ An active pest management program for algal cultivation needs to
sitoid infections for 1–2 months would be observed but much more care include a robust program of surveillance for known pests, active miti­
had to be used, including lower application amounts and frequency, in gation based on established thresholds for action, and surveillance for
order to minimize productivity losses. An example of the successful new threats [7,52–54]. An example of the process of developing a
implementation of fluazinam with M. minutum 26BAM is shown in mitigation strategy for one pest/pest type, only to be challenged with the
Fig. 9a. A dosing regime was established to minimize productivity loss appearance of a new pest threat, can be seen in the year over year results
and maintain protection against fungal parasitoids. Replicate sets of for S. obliquus UTEX393. We identified a fungal parasitoid issue in 2019,
ponds with and without treatment were cultivated side-by-side in April developed a mitigation strategy (e.g., fungicide treatment) that essen­
and May 2021. The ponds that were treated with fluazinam had an tially eliminated the threat of that pest only to be faced with a new pest
average harvest productivity of 21.1 g m− 2 d− 1 versus 15.6 g m− 2 d− 1 for threat in 2020 that was different in nature (i.e., non-responsive to
untreated ponds, a 29 % improvement. In addition, treated ponds lasted fungicide). This lead to significant year over year declines in summer
almost 3× longer with a MTTF of 41 versus 14 days, for treated and and fall productivity for 2020 and 2021 relative to 2019 (Fig. 4). More

Fig. 8. Photographs and optical microscopy of S. obliquus UTEX393 during Spring 2019 cultivation. Top panel shows pictures of untreated ponds and corresponding
optical microscopy (100×, measurement bar 10 μm) showing badly infected culture on 4/8/2019 and proceeding to full culture collapse on 4/24/2019. Bottom panel
shows pictures of treated ponds (as described in Fig. 6) and corresponding optical microscopy (100×, measurement bar 10 μm) after first dose showing a similar level
of infection as the untreated pond on 4/8/2019 and full culture recovery by 4/24/2019.

12
J. McGowen et al. Algal Research 70 (2023) 102995

Fig. 9. Comparison of fungicide application with M. minutum 26BAM cultures in April and May 2021. a.) AHYP comparison for triplicate ponds with (+ fungicide)
and without (− fungicide). Error bars are ±1 standard deviation from the mean. Typical lifecycle of fungal parasitoid contaminant in M. minutum 26BAM cultures
showing b.) healthy M. minutum 26BAM cell, c.) zoospore attachment encysted on cell surface, d.) zoospore replication and segmentation inside the host cell, e.)
amoeboid zoospores leaving the host cell, f.) empty host cell with residual body, and g.) completely crashed culture of M. minutum 26BAM (measurement bars on
panels b-g is 5 μm).

Fig. 10. Optical microscopy of two different pest types that routinely infect S. obliquus UTEX393 cultures at the AzCATI field site. Fungal parasitoids which were the
main contaminant observed in 2019 for S. obliquus UTEX393 showing typical early-stage (a) and late-stage(b) phenotypes and a new bacterial parasitoid (endobiotic)
that started infecting S. obliquus UTEX393 cultures in summer 2020 showing early-stage (c) and late-stage infection (d). Overall summer season productivity for CY
2019, 2020, and 2021 showing AHYP g m− 2 d− 1 (e., upper right panel) and MTTF in days (e., lower right panel). Error bars are ±1 standard deviation from the mean.
Total days of cultivation for each summer season were 87, 72, and 40 for 2019, 2020 and 2021, respectively. Summer 2019 had only one reseeding event at beginning
of August due to contamination by cyanobacteria (Stanieria sp.) but was counted as a crash event for the MTTF analysis. Summer 2020 and 2021 were restarted from
seed 5 times and 3 times, respectively.

13
J. McGowen et al. Algal Research 70 (2023) 102995

specifically, what was observed for S. obliquus UTEX393 going into that described for FD111 based on sodium hypochlorite addition how­
summer of 2020 was a new pest with a culture crash morphology very ever S. obliquus UTEX393 was too sensitive to chlorine addition and
different from that observed for fungal based crashes. Fig. 10a-b shows immediately bleached out even at concentrations 4-fold lower than
the microscopy from 2019 S. obliquus UTEX393 ponds infected with a those described (0.5 versus 2 mg L− 1) for Nannochloropsis that showed
fungal parasitoid and its progression to complete pond failure with efficacy for controlling FD111 infection [49]. One mitigation step that
typical morphology showing parasitoid attachment, penetration tubes, proved marginally successful but did not fully eliminate pond infection
and eventually empty ghost cells with residual bodies (Fig. 10a-b). This and culture collapse was increasing salinity from 5 ppt to 10–15 ppt
is in contrast to the infection and ultimate crash morphology of which improved MTTF from 14 to 24 days but only partially restored
S. obliquus UTEX393 beginning in summer of 2020 where increased productivity. This illustrates the significant knowledge gap around algae
bacterial presence was observed along with cells going chlorotic, yel­ crop pest susceptibility and the challenges for large-scale deployment of
lowing slightly with loss in pigmentation, partial or complete disap­ algae cultivation including the development and implementation of
pearance of cell contents, and most importantly, no residual bodies as robust crop protection methodologies.
seen in all our confirmed fungal infections (Fig. 10c-d). Rapid decline in
productivities were observed within 7–10 days of inoculating ponds
outside in 2020 and decreased to <7 days in 2021. Even with repeated 3.6. Biomass composition
restarts from fresh indoor seed culture, and enhanced sterilization pro­
tocols in ponds, high harvest productivities could not be maintained. Biomass samples from a subset of harvests (between 4 and 66 points
Fig. 10e shows the AHYP for the summer seasons for 2019–2021 with for a given strain, year, and season) were collected for proximate anal­
2020 and 2021 both showing a 25 % decline in AHYP relative to 2019 as ysis for each of the reported outdoor production runs. Compositional
well as a significant decline in MTTF from 62 days in 2019 dropping to analysis for carbohydrate (measured as monomers after acid hydrolysis),
an average of 14.8 and 13.3 in 2020 and 2021 respectively. While the lipid (measured as total fatty acid methyl ester (FAME) after whole
specific bacterial pathogen has not yet been isolated or identified from biomass transesterification), protein (via nitrogen-to-protein conver­
S. obliquus UTEX393, it has been confirmed that it is likely a predatory sion), and ash content determination [14], as well as elemental carbon
bacterium. Using aliquots of crashed field cultures (at 1 % v/v), they and nitrogen of the biomass, were determined. The data shown in
were shown to re-infect indoor cultures relative to controls. The crash Table 5 illustrates the overall compositional profiles, separated by sea­
phenotype could be prevented by filtering crashed pond samples son, year, and strain. A total of 768 individual samples were collected
through a filter pore size of 0.45 μm or smaller prior to introduction into and analyzed over the course of the outdoor production runs, creating an
clean cultures. In addition, cultures treated with antibiotic (25 μg/mL unprecedented depth of algae composition data, unique in the current
ampicillin) also showed no signs of infection when challenged with 1 % literature. Cultivation experiments were focused on maximizing biomass
aliquot of infected outdoor culture. productivity, and thus did not target a biochemical shift in the biomass
The overall phenotype of crashed S. obliquus UTEX393 culture to a more attractive biorefinery-ready composition (e. g., higher car­
looked similar to that reported for Nannochlorposis sp. caused by a novel bohydrate and lipid content), which is reflected in the >40 % protein
bacterial pest (designated as FD111) which was shown to have similarity content for almost all samples analyzed. However, when looking at the
to bacteria in the order Bdellovibrionales [49]. However, using the PCR data in aggregate (Table 5, Fig. 11), we can observe trends between
sequences reported, we did not get a match for FD111 and additional deployed strains that point to an inherent capacity of some of the strains
work is ongoing to isolate and or identify this extremely harmful bac­ to accumulate higher storage carbon content (primarily reflected in total
terial pest of S. obliquus UTEX393. Mitigation was attempted similar to carbohydrates) during active growth without the need for a purposeful
shift. The strains S. obliquus UTEX393, D. armatus SE000107, and

Table 5
Proximate biomass composition for strains grown during the 2018–2021 SOT outdoor cultivation experimental trials. All data shown as the mean ± 1 standard de­
viation from the mean of triplicate ponds cultivated over the course of multiple harvests throughout the season, on either a dry weight (% DW) or on an ash-free dry
weight basis (% AFDW) for the number of samples collected and analyzed for that season/strain/year combination (N = number of samples analyzed for a given
season/strain).
Season Year Strain N Ash (%DW) Carbohydrates (% AFDW) FAME (% AFDW) Protein (% AFDW) C (% AFDW) N (% AFDW)

Fall 2018 26BAM 30 12.2 ± 7.1 14 ± 2.2 7.9 ± 1.7 43 ± 4.6 52.4 ± 2.8 9±1
Fall 2018 C046 11 17.5 ± 2.3 13.3 ± 1.7 7.8 ± 1.5 44.8 ± 1.7 49.2 ± 0.5 9.4 ± 0.4
Fall 2018 SE000107 80 33.7 ± 14 13.9 ± 4 7 ± 2.1 44 ± 3.9 49.3 ± 3.8 9.2 ± 0.8
Fall 2018 UTEX393 29 11.2 ± 4.7 12.2 ± 1.7 8 ± 1.2 47.7 ± 3.1 50.9 ± 2.5 10 ± 0.6
Winter 2018 SE000107 4 17 ± 6.4 21.2 ± 14.9 9.5 ± 2.5 39 ± 12.4 51.3 ± 2.7 8.2 ± 2.6
Fall 2019 UTEX393 66 10.5 ± 2.8 12.7 ± 3.4 9.8 ± 0.9 49 ± 3.3 52.3 ± 1.2 10.3 ± 0.7
Spring 2019 26BAM 32 8.3 ± 1.1 14.3 ± 1.3 9.7 ± 0.8 46.2 ± 1.3 54.2 ± 1.1 9.7 ± 0.3
Spring 2019 SE000107 21 18.1 ± 13.7 14.7 ± 1.8 7.5 ± 1.6 45 ± 2 51.5 ± 1.5 9.4 ± 0.4
Spring 2019 UTEX393 35 8.7 ± 2.4 12.7 ± 1.9 8.8 ± 0.4 49.4 ± 1.8 52.8 ± 0.8 10.3 ± 0.4
Summer 2019 SE000107 3 10.6 ± 0.3 14.3 ± 0.9 9.2 ± 0.2 47.9 ± 4.1 52.6 ± 1.3 10 ± 0.8
Summer 2019 UTEX393 6 8.4 ± 0.3 11.6 ± 0.6 10.1 ± 0.2 45.5 ± 10.7 48.3 ± 11.2 9.5 ± 2.2
Winter 2019 26BAM 21 11.4 ± 6.4 14.5 ± 2.7 10.5 ± 1.1 43.4 ± 2.4 53.2 ± 1.6 9.1 ± 0.5
Winter 2019 UTEX393 21 14.5 ± 7.5 13.2 ± 1.7 8.8 ± 1 47.3 ± 2.5 50.9 ± 1.9 9.9 ± 0.5
Fall 2020 TG2 65 20.5 ± 8.8 5.2 ± 0.6 10.1 ± 1.3 49.6 ± 3.8 48.5 ± 4 10.4 ± 0.8
Fall 2020 LANL1001 21 18.5 ± 1.1 7 ± 5.9 10.4 ± 0.8 42.6 ± 4.7 48.2 ± 1.1 8.9 ± 1
Spring 2020 26BAM 24 6.4 ± 0.5 12.2 ± 4.2 9.2 ± 1.2 48.5 ± 5.9 53.2 ± 0.7 10.1 ± 1.2
Spring 2020 UTEX393 32 9.2 ± 0.6 10 ± 0.7 10 ± 0.8 52.7 ± 1.9 52.2 ± 0.5 11 ± 0.4
Summer 2020 TG2 4 17.5 ± 2.5 5.5 ± 0.1 11.7 ± 0.7 51.5 ± 0.8 52.4 ± 1 10.8 ± 0.2
Winter 2020 26BAM 14 6.2 ± 0.8 9.7 ± 1.4 10 ± 0.8 47.8 ± 1.1 52.8 ± 0.7 10 ± 0.2
Winter 2020 LANL1001 18 19.7 ± 1.6 8.3 ± 3.3 9 ± 0.4 43.7 ± 1.8 47.8 ± 0.7 9.1 ± 0.4
Winter 2020 UTEX393 14 9 ± 0.9 11.7 ± 1.8 10.2 ± 0.5 49.4 ± 2.1 51.3 ± 0.6 10.3 ± 0.4
Spring 2021 26BAM 58 6.8 ± 1.2 12.5 ± 2 7.9 ± 0.9 46.4 ± 2.5 51.5 ± 1.4 9.7 ± 0.5
Spring 2021 LANL1001 3 21.4 ± 0.2 3.3 ± 0.2 9.4 ± 0 47.3 ± 0.5 49.4 ± 0.3 9.9 ± 0.1
Summer 2021 TG2 62 22 ± 9.3 4.9 ± 0.5 10.8 ± 1.1 52.8 ± 2.5 51 ± 2.5 11 ± 0.5
Winter 2021 LANL1001 24 20.7 ± 0.8 7±2 9.8 ± 0.9 44.8 ± 1.7 48.5 ± 1.3 9.4 ± 0.4

14
J. McGowen et al. Algal Research 70 (2023) 102995

Figs. 11. Proximate biomass composition for strains grown during the 2018–2021 SOT outdoor cultivation experimental trials.

M. minutum 26BAM all consistently present higher carbohydrate content inducing organisms. These organisms can be broad host range as well as
of 15–25 % on an ash-free dry weight basis while P. celeri TG2 and strain specific and can change overtime for a given cultivar. Thus,
T. striata LANL1001 consistently across years and seasons do not accu­ developing effective crop protection strategies effective against a variety
mulate much storage carbon (<10 % carbohydrates or lipids). When of organisms is crucial. Other minor drivers of productivity are pH and
looking at the lipid content, across all samples, there are no statistically salinity but these drivers will largely be determined by choice of culti­
significant differences between strain and or seasons and remains con­ vation site and associated water supply. These drivers can be manipu­
stant and modest around 10 % of the biomass. It remains to be tested lated at an additional cost, though algae cultivation for biofuels will be
whether there is a relationship between the environmental conditions required to use water that is not useful for other purposes (e. g., high
(e. g., light, temperature, etc.) that were particularly conducive to the salinity, wastewater, produced water, brackish). In addition, the oper­
higher storage carbon conditions in some of the outlier points. ational strategies of dilution rate and pond depth have potential in
increasing AHYP though further study is needed and is strain and season
4. Conclusions dependent. Finally, though our studies show multi-season year over year
improvements in outdoor algae cultivation productivity, these studies
Though there has been significant effort with both public and private were done at a pre-pilot scale of 800 L. Though other studies have shown
research programs in the last several decades, cultivation of algae for the that experiments at this scale are in fact relevant to larger scales, much
purpose of conversion to biofuels remains challenging. The cost of work remains to be done to validate our productivity and pond man­
production is mainly driven by biomass productivity demanding that agement strategies on scales relevant to commercialization.
improvements be made in year over year outdoor productivity to In summary, the last 4 years have seen a maximum increase of 62 %
consistently deliver low-cost algae biomass. Our objective as part of the for CY 2018–2020, but dropping to 42 % for CY 2018–2021 due to
DISCOVR consortium was to validate algae strains vetted in the 3-tiered summer 2021 weather and other as of yet unidentified impacts. With
strain screening pipeline showing promise for highly productive outdoor these successes, nominal 4 % annual increases year over year for the
cultivation. To this end, in aggregate, we examined a total of 14 algae next several years will need to be realized to achieve the 2030 goal of 25
species/strains, through 1348 days of triplicate pond cultivation, in all g m− 2 day− 1. Through future work in the DISCOVR consortium, our
seasons across a period of 4 years. Though of no surprise, we observed successful strategy of crop rotation using seasonally-selected best strains
that for algae cultivation sites in geographic areas that experience sig­ combined with optimal operational strategies can continue to be
nificant seasonal climatic variations, such as the Phoenix, AZ, area, the improved and the publicly available datasets can be utilized for biomass
weather (e. g., temperature, daylength, and daily solar insolation) is the productivity modeling and concomitant TEA, LCA, and RA modeling to
main driver for algae productivity over the year. Maximum AHYP during enable future commercialization of renewable, sustainable, algae-based
the summer is >5× that in the winter and experiences very rapid in­ biofuels and bioproducts and position innovative algae agronomics on
creases/decreases during the transition seasons. Given these large and par with terrestrial crop production.
relatively rapid changes in temperature and insolation leading to such Supplementary data to this article can be found online at https://doi.
large changes in AHYP, and typically relatively narrow temperature org/10.1016/j.algal.2023.102995.
tolerances of algae, we have shown that a strategy of cultivating
different seasonally optimized strains (i.e., crop rotation) will be Informed consent, human/animal rights
necessary for maximizing annual productivity. Crop rotation notwith­
standing, fluctuations in the weather in a given season will still influence No conflicts, informed consent, or human or animal rights are
productivity. To wit, after three years of ever-increasing AHYP during applicable to this project.
the summer, minor fluctuations in the weather in the summer of 2021
brought about a precipitous decline (− 7.8 g m− 2 day− 1 (− 25 %)) in
CRediT authorship contribution statement
AHYP for the summer season.
The next major productivity driver is susceptibility to pond failure
John McGowen: Conceptualization, Writing – Review & Editing,

15
J. McGowen et al. Algal Research 70 (2023) 102995

Data Curation, Formal Analysis, Project Supervision; Eric Knoshaug: [4] J.C. Quinn, R. Davis, The potentials and challenges of algae based biofuels: a
review of the techno-economic, life cycle, and resource assessment modeling,
Conceptualization, Data Curation, Formal Analysis, Writing – Original
Bioresour. Technol. 184 (2015) 444–452.
Draft; Lieve Laurens: Writing – Review & Editing, Data Curation, Formal [5] J. Sheehan, T. Dunahay, J. Benemann, P. Roessler, A Look Back at the U.S.
Analysis; Jessica Forrester: Data Curation. Departments of Energy's Aquatic Species Program–Biodiesel From Algae, NREL/
TP-580-24190, 1998.
[6] M.E. Huntley, Z.I. Johnson, S.L. Brown, D.I. Sills, L. Gerber, I. Archibald, S.
Declaration of competing interest C. Machesky, J. Granados, C. Beal, C.H. Greene, Demonstrated large-scale
production of marine microalgae for fuels and feed, Algal Res. 10 (2015) 249–265.
The authors declare that they have no known competing financial [7] R.L. White, R.A. Ryan, Long-term cultivation of algae in open-raceway ponds:
lessons from the field, Ind. Biotechnol. 11 (2015) 213–220.
interests or personal relationships that could have appeared to influence [8] R. Craggs, D. Sutherland, H. Campbell, Hectare-scale demonstration of high rate
the work reported in this paper. algal ponds for enhanced wastewater treatment and biofuel production, J. Appl.
Phycol. 24 (2012) 329–337.
[9] A.K. Davis, R.S. Anderson, R. Spierling, S. Leader, C. Lesne, K. Mahan, T. Lundquist,
Data availability J.R. Benemann, T. Lane, J.E.W. Polle, Characterization of a novel strain of
Tribonema minus demonstrating high biomass productivity in outdoor raceway
Data will be made available on request. ponds, Bioresour. Technol. 331 (2021), 125007.
[10] A. Parsy, L. Bidoire, M. Saadouni, M. Bahuaud, T. Elan, F. Perie, C. Sambusiti,
Impact of seasonal variations on nannochloropsis oculata phototrophic
Acknowledgments productivity in an outdoor pilot scale raceway, Algal Res. 58 (2021), 102375.
[11] P.J. Lammers, M. Huesemann, W. Boeing, D.B. Anderson, R.G. Arnold, X. Bai,
M. Bhole, Y. Brhanavan, L. Brown, J. Brown, et al., Review of the cultivation
The authors would like to acknowledge the efforts of all of the people
program within the National Alliance for advanced biofuels and bioproducts, Algal
involved in running the ponds and gathering data during the cultivation Res. 22 (2017) 166–186.
campaigns at the AzCATI site especially, Henri Gerken and Mark Seger [12] K. Ogden, D. Anderson, S. Simpson, W.Van Voorhies, J. Brown, M. Huesemann,
for strain QA/QC development, pest isolations, and crash assay devel­ M. Kacira, R.J. Skaggs, P. Waller, Regional Algal Feedstock Testbed (RAFT) Final
Report, DOE-UAZ-0006269, 2019.
opment, Madison Clar and Clara Missman for culture scale up and strain [13] E.P. Knoshaug, E. Wolfrum, L.M.L. Laurens, V.L. Harmon, T.A. Dempster,
QA/QC, Richard Malloy for outdoor testbed site supervision, Kate J. McGowen, Unified field studies of the algae testbed public-private partnership as
Cuprak and Jason Potts for outdoor cultivation operations, Mauricio the benchmark for algae agronomics, Sci. Data 5 (2018), 180267.
[14] L.M.L. Laurens, S. Van Wychen, P.T. Pienkos, V.L. Harmon, J. McGowen,
Gonzales for crash assay execution, and Aaron Geels for strain QA/QC Harmonization of experimental approach and data collection to streamline analysis
and microscopy. The authors would also like to acknowledge Paige of biomass composition from algae in an inter-laboratory setting, Algal Res. 25
Norris for help in data compilation and providing input to the intro­ (2017) 549–557.
[15] J. McGowen, E.P. Knoshaug, L.M.L. Laurens, T.A. Dempster, P.T. Pienkos,
duction, and the Algae Analytical Team at NREL; Stefanie Van Wychen, E. Wolfrum, V.L. Harmon, The algae testbed public-private partnership (ATP3)
Bonnie Panzcak, Hannah Alt, and Alicia Sowell for the biomass framework; establishment of a national network of testbed sites to support
compositional analyses. sustainable algae production, Algal Res. 25 (2017) 168–177.
[16] Q. Bechet, N. Coulombier, C. Vasseur, T. Lasserre, L. Le Dean, O. Bernard, Full-
The DISCOVR consortium is sponsored by the U.S. Department of scale validation of an algal productivity model including nitrogen limitation, Algal
Energy (DOE) Office of Energy Efficiency and Renewable Energy Res. 31 (2018) 377–386.
(EERE), Bioenergy Technologies Office. This work was authored in part [17] J.R. Cruce, A. Beattie, P. Chen, D. Quiroz, M. Somers, S. Compton, K. DeRose,
B. Beckstrom, J.C. Quinn, Driving towards sustainable algal fuels: a harmonization
by the National Renewable Energy Laboratory and financially supported
of techno-economic and life cycle assessments, Algal Res. 54 (2021), 102169.
by the U.S. Department of Energy under Contract No. DE-AC36- [18] J.M. Greene, D. Quiroz, S. Compton, P.J. Lammers, J.C. Quinn, A validated thermal
08GO28308 with the National Renewable Energy Laboratory, as part of and biological model for predicting algal productivity in large scale outdoor
the DOE Office of Energy Efficiency and Renewable Energy, Bioenergy cultivation systems, Algal Res. 54 (2021), 102224.
[19] D. Quiroz, J.M. Greene, J. McGowen, J.C. Quinn, Geographical assessment of open
Technologies Office, under agreement number NL0032175. pond algal productivity and evaporation losses across the United States, Algal Res.
Additional funding was provided as part of the Algae Testbed Public- 60 (2021), 102483.
Private Partnership (ATP3) project (Award No. DE-EE0005996) and [20] R. Davis, L.M.L. Laurens, Algal Biomass Production Via Open Pond Algae Farm
Cultivation: 2019 State of Technology and Future Research, NRE:/TP-5100-76569,
Rewiring Algal Carbon Energetics for Renewables (RACER) project, 2020.
WBS 1.3.5.270 (Agreement No. NL32697). The U.S. Government retains [21] R. Davis, B. Klein, Algal Biomass Production via Open Pond Algae Farm
and the publisher, by accepting the article for publication, acknowledges Cultivation: 2020 State of Technology and Future Research, NREL/TP-5100-79931,
2021.
that the U.S. Government retains a nonexclusive, paid-up, irrevocable, [22] B. Klein, R. Davis, Algal Biomass Production via Open Pond Algae Farm
world-wide license to publish or reproduce the published form of this Cultivation: 2021 State of Technology and Future Research, NREL/TP-5100-82417,
work, or allow others to do so, for U.S. Government purposes. The views 2022.
[23] M.H. Huesemann, E.P. Knoshaug, L.M.L. Laurens, T. Dale, T.W. Lane, J. McGowen,
and opinions of the authors expressed herein do not necessarily state or Development of Integrated Screening, Cultivar Optimization, and Verification
reflect those of the U.S. Government or any agency thereof. Neither the Research (DISCOVR): a coordinated research-driven approach to improve
U.S. Government nor any agency thereof, nor any of their employees, microalgal productivity, composition, and culture stability for commercially viable
biofuels production, Algal Res. (2022) this issue.
makes any warranty, expressed or implied, or assumes any legal liability
[24] N. Beirne, S. Edmundson, S. Gao, J. Freeman, M. Huesemann, A streamlined
or responsibility for the accuracy, completeness, or usefulness of any approach to characterize microalgae strains for biomass productivity under
information, apparatus, product, or process disclosed, or represents that dynamic climate simulation conditions, Algal Res. (2022) this issue.
its use would not infringe privately owned rights. [25] S. Gao, S. Edmundson, M. Huesemann, A. Gutknecht, K. Pittman, M. Greer, L.M.
L. Laurens, S.Van Wychen, B. Panczak, H. Alt, DISCOVR strain pipeline screening -
Part III: strain evaluation in outdoor raceway ponds, Algal Res. (2022) this issue.
References [26] M. Huesemann, S. Edmundson, S. Gao, S. Negi, T. Dale, A. Gutknecht, H.
E. Daligault, C.K. Carr, J. Freeman, S.R. Starkenburg, DISCOVR strain pipeline
[1] R. Davis, D. Fishman, E.D. Frank, M.S. Wigmosta, A. Aden, A.M. Coleman, P. screening - Part I: specific growth rate as a function of temperature and salinity for
T. Pienkos, R.J. Skaggs, E.R. Venteris, M.Q. Wang, Renewable Diesel From Aglal 38 candidate microalgae strains for biofuels production, Algal Res. (2022) this
Lipids: An Integrated Baseline for Cost, Emissions, and Resource Potential From a issue.
Harmonized Model, NREL/TP-5100-554341, 2012. [27] M. Huesemann, S. Gao, S. Edmundson, L.M.L. Laurens, S.Van Wychen, N. Beirne,
[2] R. Davis, J. Markham, C. Kinchin, N. Grundl, E.C.D. Tan, D. Humbird, Process A. Gutknecht, R. Kruk, K. Pittman, M. Greer, DISCOVR strain pipeline screening -
Design and Economics for the Production of Algal Biomass: Algal Biomass Part II: winter and summer season areal productivities and biomass compositional
Production in Open Pond Systems and Processing Through Dewatering for shifts in climate-simulation photobioreactor cultures, Algal Res. (2022) this issue.
Downstream Conversion, NREL/TP-5100-64772, 2016. [28] J. McGowen, Final Closeout Report the Algae Testbed Public Private Partnership -
[3] EERE-BETO, Bioenergy technologies office national algal biofuels technology ATP3, DE-EE0005996, 2019.
review. https://energygov/sites/prod/files/2016/06/f33/national_algal_biofuel [29] P.D. Lane, A. Hagerstrand, T.W. Lane, Resistance of DISCOVR algae strains to
s_technology_reviewpdf, 2016. deleterious species, Algal Res. (2022) this issue.

16
J. McGowen et al. Algal Research 70 (2023) 102995

[30] K. Poorey, P. Rai, H. Kolla, D.J. Curtis, J. McGowen, P.D. Lane, T.W. Lane, pond depth and hydraulic retention time, Environ. Sci. Technol. 50 (2016)
Characterizing industrial pond ecology timeline in DISCOVR cultivation trials for 4102–4110.
early detection of pond crashes, Algal Res. (2022) this issue. [43] B.H. Kim, J.E. Coi, K. Cho, Z. Kang, R. Ramanan, D.G. Moon, H.S. Kim, Influence of
[31] M. Lane, S. Van Wychen, A. Politis, L.M.L. Laurens, A data-driven comparison of water depth on microalgal production, biomass harvest, and energy consumption
commericially available tesing methods for algae characterization, Algal Res. 53 in high rate algal pond using municipal wastewater, J. Microbiol. Biotechnol. 28
(2021), 102134. (2018) 630–637.
[32] S.Van Wychen, L.M.L. Laurens, Determination of total solids and ash in algal [44] S. Yadala, S. Cremaschi, A dynamic optimization model for designing open-channel
biomass - Laboratory Analytical Procedure (LAP). http://wwwnrelgov/doc raceway ponds for batch production of algal biomass, Processes 4 (2016) 10.
s/fy16osti/60956pdf, 2015. [45] J.C. Goldman, Y. Azov, C.B. Riley, M.R. Dennett, The effect of pH in intensive
[33] S.Van Wychen, K. Ramirez, L.M.L. Laurens, Determination of total lipids as fatty microalgal cultures. I. Biomass regulation, J. Exp. Mar. Biol. Ecol. 57 (1982) 1–13.
acid methyl esters (FAME) by in situ transesterification - Laboratory Analytical [46] K.R. Hinga, Effects of pH on coastal marine phytoplankton, Mar. Ecol. Prog. Ser.
Procedure (LAP). http://wwwnrelgov/docs/fy16osti/60958pdf, 2015. 238 (2002) 281–300.
[34] S.Van Wychen, L.M.L. Laurens, Determination of total carbohydrates in algal [47] N.R. Moheimani, Inorganic carbon and pH effect on growth and lipid productivity
biomass - Laboratory Analytical Procedure (LAP). http://wwwnrelgov/doc of Tetraselmis suecia and chlorella sp (Chlorophyta) grown outdoors in bag
s/fy16osti/60957pdf, 2015. photobioreactors, J. Appl. Phycol. 25 (2013) 387–398.
[35] D. Douchi, M. Mosey, D.P. Astling, E.P. Knoshaug, A. Nag, J. McGowen, L.M. [48] A. Malek, L.C. Zullo, P. Daoutidis, Modeling and dynamic optimization of
L. Laurens, Nuclear and chloroplast genome engineering of a productive non-model microalgae cultivation in outdoor open ponds, Ind. Eng. Chem. Res. 55 (2016)
alga desmodesmus armatus: insights into unusual and selective acquistion 3327–3337.
mechanisms for foreign DNA, Algal Res. 53 (2021), 102152. [49] P.A. Lee, K.J.L. Martinez, P.M. Letcher, A.A. Corcoran, R.A. Ryan, A novel
[36] R. C. Team, R: A Language and Environment for Statistical Computing, R predatory bacterium infecting the eukaryotic alga nannochloloropsis, Algal Res. 32
Foundation for Statistical Computing, Vienna, Austria, 2021. (2018) 314–320.
[37] B. T. Office, 2019 R&D State of Technology, DOE/EE-2082, 2020. [50] P.M. Letcher, S. Lopez, R. Schmieder, P.A. Lee, C. Behnke, M.J. Powell, R.
[38] A. Krishnan, M. Likhogrud, M. Cano, S. Edmundson, J.B. Melanson, C. McBride, Characterization of Amoeboaphelidium protococcarum, an algal
M. Huesemann, J. McGowen, J.C. Weissman, M.C. Posewitz, Picochlorum celeri as parasite new to the Cryptomycota isolated from an outdoor algal pond used for the
a model system for robust outdoor algal growth in seawater, Sci. Rep. 11 (2021) production of biofuel, PLoS ONE 8 (2013), e56232.
11649. [51] P.M. Letcher, M.J. Powell, S. Lopez, P.A. Lee, R.C. McBride, A new isolate of
[39] A. Krishnan, M. Cano, T.A. Burch, J.C. Weissman, M.C. Posewitz, Genome editing Amoeboaphelidium protococcarum, and Amoeboaphelidium occidentale, a new
using Cas9-RNA ribonucleoprotein complexes in the high-productivity marine alga species in phylum Aphelida (Opisthosporidia), Mycologia 107 (2015) 522–531.
picochlorum celeri, Algal Res. 49 (2020), 101944. [52] R.C. McBride, S. Lopez, C. Meenach, M. Burnett, P.A. Lee, F. Nohilly, C. Behnke,
[40] S. Edmundson, M. Huesemann, The dark side of algae cultivation: characterizing Contamination management in low cost open algae ponds for biofuels production,
night biomass loss in three photosynthetic algae, Chlorella sorokiniana, Ind. Biotechnol. 10 (2014) 221–227.
nannochloropsis Salina, and picochlorum sp, Algal Res. 12 (2015) 470–476. [53] V.L. Harmon, E. Wolfrum, E.P. Knoshaug, R. Davis, L.M.L. Laurens, P.T. Pienkos,
[41] M. Huesemann, B. Crowe, P. Waller, A. Chavis, S. Hobbs, S. Edmundson, M. J. McGowen, Reliability metrics and their management implications for open pond
S. Wigmosta, A validated model to predict microalgae growth in outdoor pond algae cultivation, Algal Res. 55 (2021), 102249.
cultures subjected to fluctuating light intensities and water temperatures, Algal [54] A.A. Corcoran, M.A. Saunders, A.P. Hanley, P.A. Lee, S. Lopez, R. Ryan, C.B. Yohn,
Res. 13 (2016) 196–206. Iterative screening of an evolutionary engineered desmodesmus generates robust
[42] Q. Bechet, A. Shilton, B. Guieysse, Maximizing productivity and reducing field strains with pesticide tolerance, Algal Res. 31 (2018) 443–453.
environmental impacts of full-scale algal production through optimization of open

17