Competition Increases Risk of Species Extinction during Extreme Warming

Temperature and interspecific competition are fundamental drivers of community structure in natural systems and can interact to affect many measures of species performance. However, surprisingly little is known about the extent to which competition affects extinction temperatures during extreme warming. This information is important for evaluating future threats to species from extreme high-temperature events and heat waves, which are rising in frequency and severity around the world. Using experimental freshwater communities of rotifers and ciliates, this study shows that interspecific competition can lower the threshold temperature at which local extinction occurs, reducing time to extinction during periods of sustained warming by as much as 2 weeks. Competitors may lower extinction temperatures by altering biochemical characteristics of the natural environment that affect temperature tolerance (e.g., levels of dissolved oxygen, nutrients, and metabolic wastes) or by accelerating population decline through traditional effects of resource depletion on life history parameters that affect population growth rates. The results suggest that changes in community structure in space and time could drive variability in upper thermal limits.


Introduction
Periods of extreme warming are becoming more frequent and severe (Rahmstorf and Coumou 2011;Ummenhofer and Meehl 2017), exposing diverse ecosystems to a wide range of destructive effects (Allen et al. 2010;McKechnie and Wolf 2010;Bellard et al. 2012;Hughes et al. 2018;Robinson et al. 2019) and representing an increasing threat to species around the world (Cahill et al. 2013;Hoffmann et al. 2013;Williams et al. 2013;Kingsolver et al. 2013;Thompson et al. 2013;Vasseur et al. 2014;McManus et al. 2020).Global warming could eventually drive the extinction of a significant percentage of global biodiversity (Maclean and Wilson 2011;Urban 2015), so researchers and governments around the world are calling for the development of strategies to mitigate effects of increasing temperature on ecosystem stability and function (Miller et al. 2018;Prober et al. 2019;IPCC 2022).Unfortunately, even basic frameworks that can be used to identify species most at risk of extinction are complicated by a limited understanding of the effects of species interactions on temperature tolerances (Williams et al. 2008;Gilbert et al. 2014;Sinclair et al. 2016).
Interspecific competition figures prominently among interactions proposed to exacerbate risk of extinction due to global warming (Gilman 2010;Kordas et al. 2011;Urban et al. 2012).Temperature and interspecific competition are fundamental drivers of community structure in natural systems and can interact to affect many measures of species performance, including distribution (Davis et al. 1998;Elsen et al. 2017), behavior (Taniguchi and Nakano 2000;Milazzo et al. 2013), abundance (Jiang and Kulczycki 2004;Bestion et al. 2018), persistence (Jiang and Morin 2004a;Comeault andMatute 2021), growth (Alexander et al. 2015;Grainger et al. 2018), and reproduction (Møller et al. 2020).However, surprisingly little is known about whether competition affects temperature tolerances during extreme warming, information important for evaluating future threats to species from extreme high-temperature events and heat waves.
Theoretical and empirical studies treating effects of temperature on species often focus on thermal performance curves (Huey and Kingsolver 1989;Angilletta 2009), which are used to present the effects of temperature on a wide range of species traits, including but not limited to standard metabolic rates (DeLong et al. 2018;Kellermann et al. 2019), locomotion (John-Alder et al. 1988;Phillips et al. 2014), fecundity (Klepsatel et al. 2013;Clemson et al. 2016), and population growth rate (Luhring and DeLong 2017;Childress and Letcher 2017).Thermal performance curves most often assume the form of a modified Gaussian function (Angilletta 2006) in which the species' trait rises with temperature until reaching optimal performance temperature (where fitness is highest) and then drops rapidly over a short range of temperature, the lower and upper bounds of the curve identifying the low and high temperatures that correspond to death.If interspecific interactions can alter lethal temperature, then interspecific interactions should also alter the upper bound of thermal performance curves that are commonly used to compare species' temperature tolerances and extrapolate the evolutionary and ecological effects of global warming.
This study addressed two questions concerning loss of species during a period of extreme warming.First, can interspecific competition reduce the threshold temperature at which local extinction occurs?Second, if so, do effects of competition on temperature of local extirpation vary across members of the same community and guild?To make maximum use of limited equipment and space, this study was conducted simultaneously with and incorporated density data from two treatments that were shared with and first published in Walberg andGreen (2021a, 2021b; see constant-polyculture and increasing-polyculture treatments below).

System
The study system, documented in detail in Walberg and Green (2021a) and summarized here, comprised 10 study species (four rotifers [Cephalodella sp., Lecane sp., Lepadella sp., and Rotaria sp.] and six ciliates [Aspidisca sp., Coleps sp., Euplotes daidaleos, Frontonia sp., Paramecium bursaria, and Paramecium caudatum]) feeding on a naturally occurring background assemblage of algae, bacteria, and fungi.This community (the source culture for polyculture replicates) was extracted in a single sample of water from Bamboo Pond at Rutgers University and cultured for 14 years prior to commencement of the study under standardized conditions in 0.95-L sterilized glass containers with 400 mL of distilled water infused with leaves and grass (1.8 g dry weight, equal parts Quercus rubra and Microstegium vimineum).Every week, half of the water and detritus was removed and replaced to provide continuing nutrients for basal microbes and remove waste materials.
Monocultures for the three test species, E. daidaleos, Coleps, and Rotaria (the source cultures for monoculture replicates), were created one year prior to the start of the experiment by isolating organisms using a micropipette and Nikon SMZ microscope at 25# power.Approximately 20 individuals of the three test species were extracted from the 10-species community following standardized procedures (supplemental PDF) and cultured under the same long-term conditions as the 10-species community.

Experimental Design
The three-factor experiment included two temperature regimes (constant and increasing), two community compositions (monoculture and polyculture), and three test species (Coleps sp., E. daidaleos, and Rotaria sp.).Polyculture communities included all 10 species (the four rotifers and six ciliates) cultured together in a single microcosm.Monoculture communities included E. daidaleos, Coleps sp., and Rotaria sp.(the three test species of this experiment), each cultured in isolation.As this study was conducted simultaneously with Walberg andGreen (2021a, 2021b), the number of species examined for effects of competition was limited by available laboratory space, electrical power, and equipment.Test species were selected from the 10-studyspecies community following trial experiments indicating their densities were substantially higher in monoculture than polyculture, providing evidence for interspecific competition (fig.S1; figs.S1-S3 are available online).Lepadella and Lecane were excluded from consideration following preliminary data indicating they had much higher temperature tolerances than the other eight study species (reasoning that the other eight species, once extinct, would no longer have the ability to affect the extinction temperatures of Lepadella and Lecane).
All 10 study species were documented algivores, bacterivores, fungivores, or mixotrophs (fig.S2), and no predation was observed between study species over 11,000 hours of observation.All species but Aspidica were also collected individually from the same pond and variously used in previous studies where they were also identified as algivores, bacterivores, fungivores, or mixotrophs (McGrady-Steed et al. 1997;Morin and McGrady-Steed 2004;Jiang and Morin 2004b;Long et al. 2006;Carrara et al. 2012;Altermatt and Holyoak 2012;Faillace and Morin 2016;Garnier et al. 2017;Delong and Gilbert 2019).Together with data from this study and trial experiments showing that six of the 10 study species maintained significantly higher densities in monoculture than polyculture, the preponderance of evidence strongly supported attribution of effects of community composition to interspecific competition.However, because the study community used in the experiment was complex, the possibility that other indirect species interactions might have played at least some role in effects of community composition, as they do in complex communities in natural systems, could not be ruled out.
Increasing-temperature communities were first increased from room temperature to acclimation temperature (26.07C) and held there for a period of 2 weeks.Subsequent to the acclimation period, temperature was increased incrementally 10.57C week 21 until temperatures reached 42.07C, resulting in the death of all species and roughly approximating warming rates preceding documented aquatic heat waves (Wernberg et al. 2012;Kahn 2015;Ruthrof et al. 2018).Temperature was controlled using Cole-Parmer StableTemp 28-L water baths (model EW-14576-16, resolution 0.17C) in a climate-controlled room (room temperature, 23:57C50:737C) that was equipped with heating and cooling backup systems (fig.S3).Holding all communities at a shared acclimation temperature prior to ramping temperature followed the dynamic method of determining relative temperature tolerances (Terblanche et al. 2007;Kingsolver and Umbanhowar 2018).Bathwater temperature was calibrated at each 10.57C incremental increase in temperature using dual Fisher Science traceable digital thermometers (resolution 0.17C) located in standardized positions on the right and left side of the baths, and water pumps circulated water in the bath to keep within-bath variation in temperature as low as possible.All temperature control equipment was attached to a 10-kVA online double-conversion power supply (TrippLite, model SU10000RT3U2TF).
Consistent with long-term culturing conditions, constanttemperature communities were held at room temperature (23:57C50:737C) throughout the duration of the experiment.Following Jiang and Morin (2004a) and Hao et al. (2015), these communities served as controls for the effects of increasing temperature, their continuing coexistence preceding and throughout the duration of the experiment demonstrating that loss of species in warmed communities was due to experimental warming rather than temperatureindependent ecological processes.
All constant-monoculture (n p 2), constant-polyculture (n p 6), increasing-monoculture (n p 6), and increasingpolyculture (n p 6) replicate communities were created shortly prior to the start of the experiment following standardized procedures (Walberg and Green 2021a).Constant-monoculture communities had lower replication because of space constraints.

Data Collection
At the end of each 10.57C week, eight samples (0.375 mL each) were collected from each replicate using a spatially stratified sampling technique (Walberg and Green 2021a).The eight samples were then consolidated in a single petri dish (3.0 mL total) and manipulated into a C-shaped liquid corridor to facilitate counting.Densities of Coleps sp., E. daidaleos, and Rotaria sp.(the three test species) were enumerated by searching the corridor from one end to the other for individuals using a Nikon SMZ microscope (25# power searching, 75# power to confirm species identity where necessary).When densities of species were too high for accurate enumeration in 3.0-mL samples, counts were taken from subsamples.Approximately 0.25 mL (volume measured precisely by sample weight using an electronic balance) was withdrawn from the gently mixed 3.0-mL sample using a glass pipette and deposited on the surface of a second petri dish in the form of 10-15 small drops, after which individuals in each of the drops were enumerated and summed.Measuring population density at each 0.57C incremental increase in temperature allowed examination of the effects of interspecific competition on species abundance throughout the duration of the experiment.
In increasing-monoculture and increasing-polyculture communities, temperature of local extinction (T e ) was the sampling temperature at which no further organisms were observed alive (motile).In this study, T e acted as a proxy for extinction temperature synonymous with lethal temperature (LT 100 ) in the physiological literature, which is defined as the temperature causing death of 100% of the individuals in a sample (Stillman and Somero 2000;Angilletta 2009;Rezende et al. 2014).After a species was identified as extinct, its absence was also confirmed in all subsequent remaining weekly sampling procedures.Determination of the temperature of local extinction in increasing-monoculture and increasing-polyculture communities allowed examination of the effect of interspecific competition on the temperature-dependent extinction of all three species.
Days to local extinction (D e ) was the corresponding time from commencement of warming (starting at 267C) to temperature of local extinction.Because increasing-monoculture and increasing-polyculture communities were exposed to a constant rate of warming, every extinction temperature corresponded to a precise period of elapsed time (e.g., species with T e of 307C, 347C, and 387C had corresponding D e of 58, 112, and 168 days, respectively).
Presence-absence data for the remaining seven community members in increasing-polyculture replicates (Walberg and Green 2021a, 2021b) allowed tracking changes in the composition of the competitor community with increasing temperature.

Analyses
Statistical analyses were conducted using parametric and nonparametric tests conducted in R version 4.2.2 (R Core Team 2022).Effects of community composition on mean population density (across all time points) were investigated using an ANOVA (type III SS; factors: temperature regime, community composition, and species) for an experiment with unbalanced design and a post hoc Tukey test.Effects Competition Lowers Lethal Temperature 325 of community composition on temperature of extinction (T e ) were analyzed using an ANOVA (factors: species and community composition) for an experiment with balanced design and a post hoc Tukey test.Because data were in discrete 0.57C increments, results were cross-checked with Wilcoxon rank sum tests run separately for each species and with P values adjusted for the effects of multiple comparisons (Benjamini and Hochberg 1995).Days to extinction (D e ) was not separately analyzed because analyses of T e and D e produced identical results.This occurs when (and only when) the relationship between temperature and time during warming is both linear and constant.

Results
The ANOVA on mean density found a significant threeway interaction among community composition, temperature regime, and species (table S1a; tables S1-S3 are available online).The post hoc Tukey tests showed that the competitors always suppressed density, but the magnitude of the suppression depended on both the species and the temperature regime (table S1b; fig.1).
Differences between mean density in monoculture and polyculture were highest for Coleps and lowest for Euplotes in both temperature regimes, and differences between mean density in monoculture and polyculture were higher in constant-temperature regimes than increasing-temperature regimes for all species.Together with information regarding feeding mechanisms of the community members and density data from trial experiments for all 10 species (discussed in "Methods"; figs.S1, S2), these results provide strong support for attribution of effects of community composition to interspecific competition.
The ANOVA on temperature of local extinction (T e ) found significant two-way interaction effects between community composition and species (table S2a).Interaction effects were ordinal (i.e., the difference between monoculture and polyculture was positive for all three species), although the post hoc Tukey tests indicated that effects were significant for Euplotes and Coleps but not Rotaria (figs.1, 2; table S2b).The Wilcoxon rank sum tests indicated that T e was higher in monoculture than polyculture for Euplotes and Coleps, the same species identified in the ANOVA (table S2c).
Throughout the term of the experiment, all 10 species in controls for the effects of increasing temperature (constant-monoculture and constant-polyculture communities) coexisted without extinctions (fig.1), continuing the 14-year pattern of coexistence preceding the experiment and indicating that species loss in increasingtemperature communities was due to warming rather than temperature-independent ecological processes.

Discussion
The results of this study provide experimental evidence that interspecific competition can lower the threshold temperature at which local extinction occurs, reducing time to extinction during periods of sustained warming by as much as 2 weeks.In an ecosystem of constantly fluctuating temperature and using Hutchinson's theory of the niche (Hutchinson 1957) as a theoretical foundation, temperature of extinction (T e ) in monoculture and polyculture can be conceptualized as estimates of the upper limits of a species' fundamental and realized thermal niche (respectively) for a specific rate of warming.The results demonstrate the importance of considering stresses from interspecific competition when forecasting effects of extreme high-temperature events and heat waves on communities.They also suggest that differences in community structure in space and time may drive variability in lethal temperatures (i.e., variability in the upper bound of thermal performance curves).
The effects of competition varied across the three study species, likely because of variation in their traits (Morin 2011).For example, the large effect of competition on the density of Coleps relative to both Euplotes and Rotaria could indicate that Coleps had more competitors than did Euplotes and Rotaria and/or the per capita interaction effect of competition was greater.Effects of competition also depended on the temperature regime, suggesting that changing temperature altered either the interaction strength of competitors or the relative efficiencies of the metabolic pathways by which they gained and lost energy (Tilman et al. 1981(Tilman et al. , 1982;;Bestion et al. 2018;Lewington-Pearce et al. 2019;Jiang and Morin 2004a;Jiang and Morin 2007;Lang et al. 2012).Moreover, interaction between the effects of competition and temperature regime differed among the three study species, a finding consistent with studies showing that effects of temperature on competitive outcomes can depend on individual species traits-for example, interspecific differences in the temperature dependence of resource acquisition and growth (Tilman et al. 1981(Tilman et al. , 1982;;Bestion et al. 2018;Lewington-Pearce et al. 2019).
Interspecific competition reduced the temperature of local extinction for only two of three species examined, demonstrating that competition only sometimes affects the temperature of extinction.This could occur if the ability of competitors to alter lethal temperature depends on species' functional characteristics (sensu Petchy and Gaston 2002;Schneider et al. 2017)-for example, some competitors having the ability to modify the shared biological and physical environment in a manner that reduces lethal temperature, others not.Another possibility is that effects of competition depend more on the degree to which all competitors suppress density.Note that suppression of the density of Euplotes by competitors remained high throughout the period of warming that preceded Euplotes' extinction (which is not surprising considering that all nine potential competitors were still extant when Euplotes was extirpated), and the effect of competition on the temperature of extinction of Euplotes was the highest observed among all three test species.In contrast, suppression of the density of Rotaria by competitors decreased dramatically immediately preceding its extirpation (from 347C to 387C), and competition had no effect on the temperature of extinction of Rotaria.The decrease in suppression of density coincided in time with the extinction of seven more temperature-sensitive species in the community, a scenario consistent with competitive , long-term holding temperature; B) regimes.Mean density was higher in monoculture than polyculture, providing strong evidence for interspecific competition (data analyzed using ANOVA and post hoc Tukey test; all differences significant to the .001level).In constant-temperature communities, which represented controls for the effects of increasing temperature, all three species persisted throughout the 29-week duration of the experiment, continuing the 14-year pattern of coexistence preceding the experiment and demonstrating that species loss in increasing-temperature communities was due to warming rather than temperature-independent ecological processes.Days 0, 28, 56, 84, 112, 140, 168, and 196 in constant-temperature communities corresponded in time to temperatures 267C, 287C, 307C, 327C, 347C, 367C, 387C, and 407C, respectively, in increasing-temperature communities.
Competition Lowers Lethal Temperature 327 release (Davidson 1978).Together, the two scenarios suggest that differences in the effect of competition on T e among the three test species could have been due to changes in the number of extant competitors throughout the period of warming.Subsequent analysis showed that suppression of density was positively correlated with the number of extant competitors at each temperature for Rotaria (Spearman correlation coefficient, r p 0:458, P p :018) but not Coleps (r p 0:3688, P p :08325), and correlation could not be analyzed for Euplotes because it was the first species extirpated.
While effects of competition on temperature of extinction varied across the three test species, the range in effects was small, perhaps because this study examined a limited number of species in a relatively small microcosm.If effects of competition vary across taxa and natural environments, the range in larger natural systems could be substantially greater.Rising temperature can also increase the strength of interspecific competition (Jiang and Morin 2004a;Jiang and Morin 2007;Lang et al. 2012; but see Rudolf and Roman 2018), so species unaffected by competition at low temperature may nonetheless experience significant effects of competition at high temperature.As a result, quantifying the degree to which effects of competition vary across natural systems would require testing many species randomly selected from large communities, regardless of the level of competition observed at low temperature.
There are several possible ecological and evolutionary mechanisms by which interspecific competition might lower extinction temperature during periods of extreme warming.First, competition may reduce the maximum temperature at which species are able to maintain basic biochemical processes (e.g., aerobic metabolism [Schulte et al. 2011;Pörtner et al. 2017], the conformation of cellular membranes [Hazel and Williams 1990;Hazel 1995], and the function of cellular enzymes [Hochachka and Somero 2002;Somero 2020]).This could occur if competitors modify the biochemical characteristics of the shared environment in a manner detrimental to temperature tolerance.For example, nutrient availability has been shown to affect temperature sensitivity (McLeod et al. 2013;Wiedenmann et al. 2013;Thomas and Litchman 2016;Thomas et al. 2017; Huey and Kingsolver  Frontonia (e), P. caudatum (f ), Coleps (g), Rotaria (h), Lepadella (i), and Lecane (j).Test species are highlighted in yellow.
2019), and interspecific competition may reduce nutrient concentrations available to some species in the community.Dissolved oxygen can also affect sensitivity to temperature (Pörtner and Knust 2007;Stevens et al. 2010;Verberk et al. 2016;Pörtner et al. 2017), and increasing numbers of interspecific competitors could fill open heterotrophic niche space in aquatic systems, lowering levels of dissolved oxygen by increasing rates of respiration relative to photosynthesis.Finally, the introduction of chemical compounds in the environment by competing species (e.g., the release of metabolic wastes and chemical defenses associated with allelopathic competition) may also exacerbate sensitivity to extreme temperature.Examples of chemical (allelopathic) competition exist across taxa (Śliwińska-Wilczewska et al. 2021) and are documented in microbial systems with increasing frequency (Song et al. 2017;Wang et al. 2017;Poulin et al. 2018).Second, competition may accelerate population decline during warming by interacting with temperature to reduce demographic growth rates (i.e., driving additional population decline through more traditional effects of resource depletion on life history parameters affecting population growth rates).This can occur if competitive interactions that are sustainable (i.e., allowing coexistence) at lower temperatures become unsustainable (i.e., resulting in competitive exclusion) at higher temperatures (Jiang and Morin 2004a;Comeault and Matute 2021).Conceptualized using traditional multispecies resource competition theory (Tilman 1982), the outcome of competition for a species might transition from coexistence to exclusion if warming alters the efficiency of the metabolic pathways by which the species gains and loses energy so that it no longer utilizes any resource more efficiently than its competitors (i.e., as a result of the change in temperature, it no longer has the lowest R * for any resource).This scenario would not be surprising, as studies show that minimum resource requirements can be temperature dependent (Tilman et al. 1981), species can vary in the ways in which their minimum resource requirements respond to changes in temperature (Lewington-Pearce et al. 2019), and interspecies differences in minimum resource requirements can help predict the outcome of competition in single-resource environments (Bestion et al. 2018).The transition from coexistence to gradual exclusion might also occur if temperature increases the interaction strength between species sufficiently to destabilize coexistence.This has been shown for both competing species (Jiang and Morin 2004a;Jiang and Morin 2007;Lang et al. 2012) andpredators andprey (O'Connor 2009;Rall et al. 2010;Rall et al. 2012;Novich et al. 2014) and can occur if warming increases the degree of niche overlap (Barton and Schmitz 2009) or alters physiological efficiencies associated with resource uptake and assimilation (Vucic-Pestic et al. 2011).Although the effects that extreme periods of warming have on competitive interactions may be temporary, time spent at higher temperatures may be sufficient to begin the process of competitive exclusion, speeding population decline and extinction.
Third, competition might reduce the maximum temperature species are able to survive by driving evolutionary trajectories that reduce temperature tolerances.For example, if competitive ability and thermal tolerance are genetically correlated (Betancourt and Presgraves 2002;Barton 2010;Paaby and Rockman 2013;Watanabe et al. 2019), evolutionary responses to competition could cause the evolution of reduced thermal tolerance.With respect to the communities in this experiment, differential evolutionary trajectories could have begun one year preceding the start of the experiment, on the date the three source monocultures were separated from the 10-species polyculture.
This study utilized an experimental warming rate consistent with modes of interannual climate variability and approximating warming rates preceding documented aquatic heat waves (Wernberg et al. 2012;Kahn 2015;Ruthrof et al. 2018).However, rates of warming can also be much faster, especially those associated with diurnal temperature cycles in terrestrial systems (Whitman et al. 1997;Braganza et al. 2004;Sardon 2007).Differences in warming rates are important because extinction temperature and time to extinction are often highly positively and negatively correlated with warming rates (respectively) in both experimental and natural ecosystems (Becker and Genoway 1979;Elliot and Elliot 1995;Terblanche et al. 2007;Chown et al. 2009;Allen et al. 2012Allen et al. , 2016;;Faulkner et al. 2014;Bentley et al. 2016), a relationship that occurs because stress from high temperature tends to increase with exposure time (Rezende et al. 2014(Rezende et al. , 2020;;Kingsolver and Umbanhowar 2018).As warming rates have a strong effect on both upper thermal limits and time to extinction (Terblanche at al. 2007;Kovacevic et al. 2019;Rezende et al. 2020), they might also alter effects of competition on upper thermal limits.For example, slow rates of warming may allow more time for the outcome of altered competitive interactions to play out than rapid warming, in which case rapid warming might generate outcomes determined primarily by effects that competitors have on biochemical characteristics of the shared environment affecting temperature tolerances (see the first mechanism discussed above), and slow warming might generate outcomes increasingly influenced by the effect of altered competitive interactions on demographic growth rates (see the second mechanism discussed above).
Extreme high-temperature events and heat waves are expected to worsen over the next century, regardless of whether emissions are curbed (IPCC 2022), necessitating an improved understanding of their potential future effects on communities (Ma et al. 2015;Buckley and Huey Competition Lowers Lethal Temperature 329 2016).Studies examining the effects of interspecific interactions on temperature sensitivity during extreme warming can improve our understanding of the complex and interacting ecophysiological processes underpinning the relationship between climate and biodiversity (Sinclair et al. 2016;Sears et al. 2019) and contribute to the development of mechanistic frameworks used to forecast future effects of extreme high-temperature events and heat waves on ecosystems (Buckley et al. 2010;Urban et al. 2016).

Figure 1 :
Figure1: Density of Euplotes, Coleps, and Rotaria in monoculture (purple) and polyculture (blue) communities under increasing-temperature (A) and constant-temperature (i.e., long-term holding temperature; B) regimes.Mean density was higher in monoculture than polyculture, providing strong evidence for interspecific competition (data analyzed using ANOVA and post hoc Tukey test; all differences significant to the .001level).In constant-temperature communities, which represented controls for the effects of increasing temperature, all three species persisted throughout the 29-week duration of the experiment, continuing the 14-year pattern of coexistence preceding the experiment and demonstrating that species loss in increasing-temperature communities was due to warming rather than temperature-independent ecological processes.Days 0,28, 56, 84, 112,  140, 168, and 196  in constant-temperature communities corresponded in time to temperatures 267C, 287C, 307C, 327C, 347C, 367C, 387C, and 407C, respectively, in increasing-temperature communities.

Figure 2 :
Figure 2: Mean and standard deviation of temperature of local extinction (T e ; 7C) and days to local extinction (D e ) for Euplotes, Coleps, and Rotaria in monoculture (purple) and polyculture (blue).Note the different ranges of the y-axes.Data were analyzed using ANOVA and post hoc Tukey test (tables S2a, S2b).Significant differences (P !:05) are identified by two asterisks.