Functional and biological diversity jointly shape growth and recovery of Synechococcus communities under stressors

combined stressors, ecological stability, functional diversity, phytoplankton 12 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint

13 Biodiversity is expected to enhance the stability of ecological communities under 14 environmental stress, but the relative roles of functional, interspecific, and intraspecific 15 diversity remain poorly resolved, particularly under multiple concurrent stressors. We 16 tested how these diversity dimensions shaped the growth and recovery of marine 17 Synechococcus communities in a microcosm experiment manipulating strain composition 18 across four strain-richness levels and two interspecific diversity levels under control, 19 atrazine, warming, and combined atrazine-plus-warming treatments. Functional diversity 20 was quantified from flow-cytometric trait data and analyzed as initial functional diversity 21 during the stress phase and assembled functional diversity during recovery. Contrary to 22 our expectations, higher initial functional diversity was associated with lower community 23 growth during stress, while higher assembled functional diversity was generally 24 associated with weaker recovery. However, these relationships depended on stressor 25 identity and interspecific diversity: in two-species communities, the negative effects of 26 functional diversity were reduced, and under combined stress, higher assembled 27 functional diversity was associated with improved recovery. In contrast, intraspecific 28 diversity consistently enhanced community growth and recovery, while interspecific 29 diversity primarily promoted functional recovery. Together, our results show that 30 functional, interspecific, and intraspecific diversity can influence stress responses through 31 distinct pathways. 32 1. Introduction 33 The stability of biological communities is increasingly threatened by environmental 34 stressors whose intensity and duration are exacerbated by anthropogenic global change 35 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint (IPCC 2023; Pecl et al., 2017). In natural systems, however, stressors rarely act in 36 isolation. Instead, communities are often exposed to simultaneous chemical, thermal, and 37 biological stressors whose combined effects are difficult to predict (Nguyen et al., 2021). 38 Such stressors can alter community structure and ecosystem functioning in ways that are 39 difficult to predict from single-stressor responses (Jackson et al., 2021; Polazzo & Rico 40 2021; Orr et al., 2024). Because community stability underpins critical ecosystem 41 functions such as primary production and nutrient cycling, identifying the mechanisms 42 that buffer communities under multiple concurrent stressors has become a central 43 challenge in ecology (Oliver et al., 2015). 44 Ecological stability is commonly characterized by two complementary components: the 45 ability to maintain performance under stress and the ability to recover once stress is 46 removed (Pimm 1984; Hillebrand et al. 2018). These components need not covary: 47 communities may maintain functioning during stress but recover slowly, or conversely, 48 decline rapidly yet recover completely (Van Meerbeek et al. 2021; Donohue 2016, 2013). 49 Distinguishing the ecological drivers of these stability dimensions is therefore essential 50 for predicting how communities respond to increasingly complex stress regimes. 51 One mechanism that may explain performance under stress acts through functional 52 diversity (FD), defined using the distribution of functional traits within a community 53 (Schleuter et al. 2010; Cadotte 2011; de Bello et al. 2021, Barabás et al. 2022). 54 Communities with higher FD are often expected to better maintain ecosystem functioning 55 under stress because trait differences can increase the likelihood that some taxa continue 56 to perform under changing conditions (Loreau & de Mazancourt 2013). Community 57 functioning may also depend on biodiversity dimensions other than FD. Interspecific 58 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint diversity (species richness) can influence community performance through effects that 59 are not necessarily captured by trait-space metrics alone (Tilman et al. 1997; Violle et al. 60 2012; Holzwart et al., 2015; MacArthur & Levins 1967; Spaak & De Laender 2021; 61 Barabás et al. 2022). A community composed of more species, or more genotypes per 62 species, can shape responses to stress by increasing the likelihood that some individuals 63 persist under altered conditions (Bolnick et al. 2011; Siefert et al. 2015; Barabás et al. 64 2022). For example, the insurance effect occurs when taxa performing similar ecological 65 roles respond differently to environmental change and thence improve performance 66 (Yachi & Loreau 1999; McCann 2000; Baert et al., 2018). Similar mechanisms may 67 operate within species at the intraspecific level and genotypic diversity may also improve 68 stability through a similar insurance effect. This variation in responses, often termed as 69 response diversity, can buffer ecosystem functioning when environmental conditions 70 fluctuate because some taxa maintain performance while others decline (Elmqvist et al. 71 2003; Ross and Petchey 2023). Such effects may stabilize communities through 72 demographic buffering rather than through changes in functional trait space. Interspecific 73 and intraspecific diversity can therefore contribute to community performance through 74 distinct mechanisms than FD, yet how these biodiversity dimensions impact recovery has 75 not yet been jointly evaluated. 76 Recovery implies a potential restructuring due to historical conditions. Stress exposure 77 may selectively eliminate trait variants and thereby compress or reorganize the functional 78 trait space of a community (Mouillot et al. 2013; Díaz et al. 2016; Fischer et al. 2016). As 79 a result, the FD present after stress may differ substantially from the FD present before it. 80 This distinction is important because the FD initially present in a community may foster 81 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint performance during stress, whereas FD after stress - here termed assembled FD - may 82 hamper recovery (Baert et al. 2017, Pennekamp et al., 2018). Distinguishing between 83 initial FD and assembled FD is therefore needed to understand how biodiversity 84 influences both growth and recovery. 85 Despite increasing interest in biodiversity-stability relationships (Campbell et al., 2011, 86 Craven et al., 2018; Eisenhauer 2024), there are currently insufficient studies that 87 simultaneously test for effects of interspecific, intraspecific and functional diversity on 88 stability under combined stress (Palacio et al., 2025). To address this, we experimentally 89 manipulated species diversity and strain diversity in phytoplankton communities, 90 specifically Synechococcus sp., a globally distributed primary producer in marine 91 ecosystems, exposed to two environmental stressors in combination as well as in 92 isolation. We quantified how intraspecific, interspecific and initial functional diversity 93 (FD) influenced community performance and FD change during stress exposure. We then 94 examined how intraspecific, interspecific, diversity and assembled functional diversity 95 influenced the recovery of growth and function following stress removal. 96 We tested the following hypotheses. First, we hypothesized that initial FD increases 97 community growth during stress exposure (H1). Second, we hypothesized that assembled 98 FD improves recovery of growth and FD once stressors are removed (H2). Third, 99 because of response diversity, we expected inter- and intraspecific diversity to have a 100 positive effect on community performance and recovery. (H3). Finally, we expected the 101 effects of biodiversity and functional diversity on community performance and recovery 102 to vary with stressor identity, producing different outcomes under thermal, chemical, and 103 combined stress (H4). 104 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint 2. Materials and Methods 105 We conducted a microcosm experiment to test the effects of interspecific and 106 intraspecific diversity and functional diversity (FD) on community responses to stress. 107 We used four strains of marine Synechococcus sp. (Table 1), selected to represent two 108 clades (hereafter “species”) and two pigmentation types (Six et al. 2007, Farrant et al. 109 2016, Grebert et al. 2018, Sliwinska-Wilczewska et al. 2020). All strains were obtained 110 from the Roscoff Culture Collection. Prior to the experiment, strains were maintained 111 under control conditions in 100 mL Erlenmeyer flasks. 112 Table 1 : information on the four strains of Synechococcus sp. used for this study. All 113 information was retrieved from the Roscoff Culture Collection website. 114 Roscoff ID Strain ID Clade (Species) Pigment type RCC 2383 Synechococcus sp. RS9909 VIII 1 RCC 2434 Synechococcus sp. RS9906 VIII 1 RCC 2375 Synechococcus sp. A15-46 V 2 RCC 2524 Synechococcus sp. BL_10 V 2 115 We manipulated inter- and intraspecific diversity by manipulating strain richness and 116 composition within each strain richness level (Table 2): there were four mono-cultures 117 (one strain, interspecific diversity of one), six duo-cultures (two strains, interspecific 118 diversity of one to two), four tri-cultures (three strains, interspecific diversity of two), and 119 one tetra-culture (four strains, interspecific diversity of two). Each community was 120 subjected to one of four stressor treatments, which was replicated three times, yielding 121 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint 180 experimental units total (48 monoculture, 72 duo-cultures, 48 tricultures, and 12 122 tetra-culture replicates). 123 Table 2 : Information on the composition of communities used in this study. Strains refer 124 to the RCC ID as indicated in Table 1. 125 Strains Species Strain richness Species richness 2383, 2524 VIII,V 2 2 2434, 2524 VIII,V 2 2 2383, 2434 VIII,VIII 2 1 2383, 2434, 2524 VIII,VIII,V 3 2 2375, 2524 V,V 2 1 2375, 2383 V,VIII 2 2 2375, 2434 V,VIII 2 2 2375, 2383, 2524 V,VIII,V 3 2 2375, 2434, 2524 V,VIII,V 3 2 2375, 2383, 2434 V,VIII,VIII 3 2 2375, 2383, 2434, 2524 V,VIII,VIII,V 4 2 126 We did not manipulate FD directly but instead calculated it for each microcosm (section 127 Functional diversity computation) and evaluated the effect of its initial value (initial FD) 128 and assembled values (assembled FD) on community responses in a regression analysis 129 (section Data processing and statistical analysis). 130 2.1. Treatments and microcosms 131 We tested four treatments: (1) control (ambient temperature 20°C , 0 mg L-1 atrazine), (2) 132 atrazine (ambient temperature, 0.1 mg L -1 atrazine), (3) elevated temperature (+2°C 133 above ambient, 0 mg L-1 atrazine), and (4) atrazine-temperature (0.1 mg L-1 atrazine and 134 +2°C above ambient). The temperature increase was selected to avoid nonlinear threshold 135 effects on growth, and the atrazine concentration was chosen to reduce intrinsic growth 136 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint rates by approximately 50% (Holmes et al. 2025). Communities were cultured in 6 mL of 137 PCRS-11 Red Sea medium (Rippka et al. 2000) in six-well plates (Sarstedt standard flat 138 6-well plate, 83.3920) housed in temperature-controlled incubators (Lovibond 139 Thermostatic Cabinet). Cultures were maintained under white light (50 μ mol photons 140 m/i2 ² s /i2 ¹ at 6500 K; LedAquaristik Sky Bar) on a 12 h:12 h light:dark cycle and 141 continuously mixed at 150 rpm using an orbital shaker (VWR mini shaker). 142 2.2. Experimental protocol and sampling 143 The 21-day experiment comprised two phases: a stress phase (days 0–10) and a recovery 144 phase (days 11–20). Communities were inoculated at approximately 5000 cells/mL on 145 day 0, with strains added in equal proportions in multi-strain communities. Every 24 h, 146 microcosms were sampled under a laminar flow hood using a standardized procedure: (1) 147 57 μ L of pure water was added to each well to compensate for evaporation; (2) 200 μ L 148 was removed from each well into a 96-well plate for flow cytometric analysis; (3) 200 μ L 149 of fresh medium containing the appropriate atrazine concentration was added to replace 150 the removed volume; and (4) the collected sample was serially diluted to a density of 151 500–5000 cells/μ L prior to measurement to be suitable for flow cytometry analysis. On 152 day 10, the microcosms were subjected to a perturbation (dilution) and the stress 153 treatments were reset. This consisted of setting the temperature to ambient levels (20°C) 154 and replacing 50% (3 mL) of the medium of each microcosm with fresh medium (0 mg L-155 1 atrazine), reducing the atrazine concentration by 50% and acting as a perturbation to the 156 communities (dilution). 157 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint 2.3 Flow cytometry and data processing 158 Population density and cell traits were quantified using a Guava easyCyte 12HT flow 159 cytometer. Five optical channels were used to distinguish live cells from non-living 160 debris and to characterize proxies of cell traits (Rutten et al. 2005): forward scatter (cell 161 size), side scatter (cell complexity), red-blue fluorescence (chlorophyll-a concentration), 162 yellow-blue fluorescence (phycoerythrin concentration), and red-red fluorescence 163 (phycocyanin concentration). Cell densities were computed after filtering doublets and 164 debris from the raw flow cytometry data in R (v. 4.3.3) using the peacocq, cyanoFilter, 165 flowCore, and flowDensity packages (R Core Team 2024, Olusoji et al. 2021, Ellis et al. 166 2024, Malek et al. 2023, Emmaneel et al. 2022). 167 2.4 Functional diversity computation 168 We computed FD from flow cytometric trait data following the approach of Olusoji et al. 169 (2023) and Barabás et al. (2022). Only three of the five measured traits were used during 170 the computation of functional diversity: cell size, chlorophyll, and phycoerythrin. These 171 traits were the least redundant (i.e. lowest correlation with other traits) which reduced the 172 impact of highly correlated traits on the computed FD. Additionally, this reduction of 173 dimensionality served to reduce computational load and allowed for higher grid 174 resolution. The multidimensional trait space was discretized into a grid (50 divisions per 175 channel), and the kernel density of occupied trait space was estimated using the vine and 176 dvine functions from the rvinecopulib R package (Nagler et al. 2025). Hill diversity (q = 177 1) was then calculated from the resulting kernel density estimate. To ensure 178 comparability across microcosms, the grid boundaries were defined using the entire 179 pooled dataset. 180 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint 2.5. Data processing and statistical analysis 181 Four response metrics were calculated for each microcosm: 182 Growth of community density = 183 ln /g2888/g2915/g2924/g2929/g2919/g2930/g2935 /g2930/g2869/g2868 /g2888/g2915/g2924/g2929/g2919/g2930/g2935 /g2930/g2868 184 Recovery of community density = 185 ln Density t20 Density t10 Change in functional diversity = 186 ln FD t10 FD t0 Functional recovery = 187 ln FD t20 FD t10 To assess how interspecific, intraspecific, and functional diversity, and stressor identity 188 influenced community performance and FD change, we fitted four generalized least 189 squares (GLS) models using the “nlme” package (v. 3.1-168; Pinheiro et al. 2022) in R. 190 We used a GLS model over an ordinary least squares model to better accommodate 191 heterogeneity in residual variance across treatment groups, modeled using a variance 192 structure allowing residual spread to differ by treatment (varIdent, form = ~1 | treatment). 193 All four models shared the same predictor structure: a focal FD variable (diversity at t /i2 194 or t/i2/i2 ) in a three-way interaction with species diversity and stressor treatment (control, 195 atrazine, thermal, combined), with strain diversity included as an additive covariate. The 196

level for all models was set to species diversity (1), strain diversity (1), and 197 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint treatment (control). Residual diagnostic plots were inspected for all models to evaluate 198 adequacy of fit (Fig. S1). 199 Model coefficients and associated p-values were extracted using the “broom.mixed” 200 package (v. 1.0.9). To characterize the direction and significance of the FD effect within 201 each species diversity-by-treatment combination, we used the emtrends() function from 202 the “emmeans” package (v. 1.11.2; Lenth 2023). Pairwise contrasts between slopes at the 203 two species diversity levels were computed within each treatment using the contrast 204 function grouped by treatment, to determine whether species diversity modulated the FD 205 effect. This procedure was applied identically to all four models. Statistical significance 206 was assessed at α = 0.05, with significance thresholds reported as p < 0.05 (*), p < 0.01 207 (**), and p < 0.001 (***). All analyses were conducted in R v. 4.5.2. 208 Figure Captions 209 Figure 1 : The effects of functional diversity, strain and species diversity, and 210 stressor treatment, on community growth and recovery . Relationships are shown for 211 exposure to control (C), atrazine (A), thermal stress (T), and combined atrazine + thermal 212 stress (AT) treatments at two species diversity levels (1 and 2 species) and three strain 213 diversity levels (blue = 1 strain, green = 2 strains, olive = 3 strains, and orange = 4 214 strains). The left panels depict how initial functional diversity relates to community 215 growth during the stress phase, while the right panels show how assembled functional 216 diversity (measured at the end of the stress phase) relates to density recovery after stress 217 removal. Points represent individual microcosms and lines are fitted relationships within 218 groups, with the black line representing the overall trend within each panel. 219 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint Figure 2: The effects of functional diversity, strain and species diversity, and 220 stressor treatment, on change and recovery of functional diversity . Relationships are 221 shown for exposure to control (C), atrazine (A), thermal stress (T), and combined atrazine 222 + thermal stress (AT) treatments at two species diversity levels (1 and 2 species) and 223 three strain diversity levels (blue = 1 strain, green = 2 strains, olive = 3 strains, and 224 orange = 4 strains). The left panels depict how initial functional diversity relates to 225 change of functional diversity during the stress phase, while the right panels show how 226 assembled functional diversity (measured at the end of the stress phase) relates to 227 recovery of functional diversity after stress removal. Points represent individual 228 microcosms and lines are fitted relationships within groups, with the black line 229 representing the overall trend within each panel. 230 Figure 3: Effect of initial and assembled functional diversity (FD) on community 231 responses across treatments and species diversity levels. Each point represents the 232 estimated slope of the relationship between functional diversity and the response variable 233 (community density or FD) within a given treatment and species diversity level, derived 234 from generalized least squares models. Horizontal lines indicate 95% confidence 235 intervals. Negative slopes indicate that higher functional diversity is associated with 236 reduced community performance; slopes not significantly different from zero indicate no 237 detectable FD effect. Results are shown separately for one-species (top row) and two-238 species communities (bottom row), and for community density and functional diversity 239 during the stress phase (days 0–10) and recovery phase (days 11–20). Point colors denote 240 treatment: control (C, black), atrazine (A, blue), thermal stress (T, yellow), and combined 241 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint stressor (AT, pink). Asterisks indicate slopes significantly different from zero (* p < 0.05, 242 ** p < 0.01, *** p < 0.001). 243 3. Results 244 3.1 Effects of stressors on growth and recovery of community density 245 Atrazine significantly reduced community growth relative to the control ( β = −2.4 ± 0.1, 246 p < 0.001, Table S1). A similar reduction was observed under the combined atrazine and 247 thermal treatment ( β = −2.4 ± 0.1, p < 0.001). Thermal stress alone had no significant 248 effect on growth (p = 0.54). 249 During the recovery phase, communities previously exposed to atrazine recovered 250 significantly better than control communities ( β = 1.6 ± 0.5, p = 0.002, Table S1). 251 Communities recovering from the combined treatment also recovered more relative to 252 control (β = 1.3 ± 0.6, p = 0.045), whereas communities exposed to thermal stress alone 253 did not differ significantly from the control (p = 0.13). 254 3.2 Effects of diversity on community growth 255 Functional diversity at the start of the 10-day stress phase (initial FD) had a significant 256 overall negative effect on community density during the stress phase ( β = −4957.6 ± 257 1089.1, p < 0.001, Table S1). However, this relationship depended on interspecific 258 diversity and stressor identity. In single-species communities under control conditions, 259 higher initial FD strongly reduced community growth. In two-species communities, this 260 effect was reversed, resulting in a positive relationship between FD and growth (Fig 1). 261 This difference in interspecific diversity levels was significant under atrazine (p = 0.002) 262 and combined stress (p = 0.025), but not under control or thermal treatments. 263 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint Interspecific diversity itself had a negative main effect on growth ( β = −0.8 ± 0.4, p = 264 0.04, Table S1). Intraspecific diversity (the number of strains per community) positively 265 influenced density across treatments. Duo-cultures (β = 0.8 ± 0.2, p < 0.001), tri-cultures 266 (β = 0.7 ± 0.2, p < 0.001) and tetra-cultures ( β = 0.8 ± 0.2, p < 0.001) had better growth 267 than monocultures. These effects were consistent across treatments and did not 268 significantly interact with stressor identity. 269 270 Figure 1 271 3.3 Effects of diversity on community recovery 272 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint FD after the 10-day stress phase (assembled FD) had a significant overall negative effect 273 on community recovery ( β = −1829.9 ± 711.2, p = 0.01, Table S2). (Fig 1). However, 274 treatment type and interspecific diversity modified this trend. 275 Pairwise contrasts indicated that the effect of assembled FD differed significantly 276 between single and two-species communities following exposure to the combined 277 stressor (p = 0.003, Table S5), but not after atrazine or thermal stress alone (Fig. 3). This 278 pattern was reflected in the significant three-way interaction in the GLS model, where 279 two-species communities recovering from combined stress exhibited a reversal of the 280 overall negative relationship between assembled FD and community recovery ( β = 3851, 281 p = 0.036, Table S2). Interspecific diversity alone did not significantly affect community 282 recovery (p = 0.34, Table S2). Intraspecific diversity had a positive effect on recovery, 283 with tri-cultures ( β = 0.4 ± 0.2, p = 0.04) and tetra-cultures ( β = 0.5 ± 0.2, p = 0.01) 284 growing better, while duo-cultures did not differ significantly from monocultures. 285 3.4 Effects of stressors on functional diversity change and functional recovery 286 Treatments significantly influenced changes in FD during the stress phase (Table S3). 287 Atrazine negatively influenced FD change relative to the control ( β = −0.3 ± 0.1, p < 288 0.001). In contrast, thermal stress positively influenced FD change ( β = 0.3 ± 0.1, p < 289 0.001), and the combined atrazine and thermal treatment also resulted in a small but 290 significant positive effect on FD change (β = 0.2 ± 0.1, p = 0.03). 291 During the recovery phase, none of the treatments significantly affected functional 292 recovery (p > 0.45, Table S4). 293 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint 294 Figure 2 295 3.5 Effects of diversity on functional diversity change 296 Initial FD negatively affected FD change during the stress phase (Table S3). 297 Communities with higher initial FD experienced larger declines ( β = −4178.6 ± 808.5, p 298 < 0.001). This relationship differed within interspecific diversity levels and treatments 299 (Fig. 2). Initial FD negatively affected FD change in both single and two-species 300 communities but the magnitude of this effect was reduced in the latter. The interaction 301 between interspecific diversity and initial FD was positive ( β = 2830.1 ± 827.0, p < 302 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint 0.001), indicating that species diversity partially buffered the loss of FD in initially 303 diverse communities. 304 Slope contrasts between interspecific diversity levels showed that this buffering effect 305 was significant under the combined stress treatment (p = 0.022, Table S5), but not under 306 the other treatments. Intraspecific diversity did not significantly influence FD change 307 during the stress phase. 308 3.6 Effects of diversity on functional recovery 309 Assembled FD had an overall non-significant, slightly negative effect on functional 310 recovery (β = -152.3, p = 0.88, Table S4). However, interspecific diversity significantly 311 influenced functional recovery overall ( β = 1.5 ± 0.5, p = 0.01, Table S4). Specifically, 312 two-species communities exhibited better functional recovery than single species (Fig 3). 313 The relationship between assembled FD and functional recovery further varied among 314 treatments. In single-species communities, assembled FD did not significantly influence 315 recovery under any treatment. In two-species communities, assembled FD negatively 316 affected recovery under control conditions and following atrazine exposure. However, 317 assembled FD did not affect functional recovery under combined and thermal stress. 318 Under these treatments, a higher assembled FD was associated with greater functional 319 recovery relative to control (combined: β = 3529, p = 0.02; thermal: β = 2959, p = 0.03, 320 Table S4). Intraspecific diversity had no detectable effect on functional recovery. 321 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint 322 Figure 3 323 4. Discussion 324 Functional diversity (FD) influenced community responses to stress, but its effects 325 depended strongly on stressor identity and species diversity. Contrary to our expectation 326 that a higher initial FD would promote community growth under stress (H1), we found 327 that initial FD reduced community growth during stress exposure. During recovery, 328 assembled FD also had a negative effect on community growth and no effect on 329 functional recovery reflecting a limitation on functional organization after stress removal. 330 (H2). These negative relationships with initial and assembled FD were buffered by 331 interspecific and intraspecific diversity consistently for both community growth and 332 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint recovery (H3). Finally, as expected, these outcomes differed based on stressor identity, 333 with stressors in isolation having a different effect than combined stressors (H4). 334 4.1 Stressor identity determines community growth and recovery 335 Community growth differed based on treatments, indicating that stressor identity shaped 336 community performance under stress. Atrazine suppressed growth during the stress 337 phase, consistent with its mode of action as a photosystem II inhibitor that constrains 338 photosynthetic energy acquisition and phytoplankton productivity (Yang et al., 2021). In 339 contrast, warming alone had little effect on growth relative to the control (Table S1). 340 Temperature changes can alter phytoplankton physiology and community dynamics 341 without necessarily imposing strong reductions in growth (Anderson et al., 2022). 342 Communities previously exposed to atrazine exhibited greater recovery than control 343 communities (Table S2). One possible explanation is that chemical stress altered trait 344 distributions in ways that changed which traits contributed most to growth after stress 345 removal. Chemical stressors can modify response-effect relationships such that strains 346 conferring tolerance under stress differ from those maximizing growth under control 347 conditions (Mensens et al., 2017). More generally, trade-offs between stress tolerance 348 and growth are widely observed across organisms (Schucht & Blasius 2025). Under this 349 scenario, atrazine-tolerant strains may maintain limited growth during stress, whereas 350 faster-growing strains experience stronger inhibition because photosynthesis is disrupted. 351 Once the stressor is removed, these fast growers may rapidly resume growth and drive 352 community recovery. Because strain identities were not tracked through time, this 353 mechanism remains a hypothesis rather than a demonstrated process. 354 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint 4.2 Negative effects of initial functional diversity on community growth 355 A higher initial functional diversity (FD) was associated with reduced community growth 356 during stress (Table S3), suggesting that trait differentiation imposed competitive costs 357 rather than complementarity benefits, contrary to H1. Trait differences can generate both 358 niche differences and fitness differences, although not necessarily in conjunction 359 (Mayfield and Levine, 2010; Spaak and De Laender 2021). Under atrazine, which 360 constrains photosynthetic energy availability, the most functionally specialized strains 361 were also likely the most metabolically vulnerable, contributing disproportionately to 362 performance loss (Violle et al., 2017; Mouillot et al., 2013). Trait-based biodiversity 363 theory similarly shows that ecosystem functioning can be shaped by opposing 364 mechanisms, where selection driven by fitness differences can override complementarity 365 effects (Cadotte, 2017). 366 This perspective explains why the negative relationship between FD and growth was 367 strongest in single-species but weaker in two-species communities (Table S1 and S2). 368 Within single-species communities, trait variation among strains may primarily reflect 369 differences in competitive performance or stress tolerance, potentially increasing 370 dominance by specific phenotypes under stress. By contrast, differences among species 371 may be more likely to correspond to niche differentiation, allowing some of the effects of 372 fitness differences to be offset by partial complementarity. Although our experiment 373 cannot directly separate niche from fitness differences, the pattern is consistent with 374 theoretical work showing that increasing diversity does not necessarily increase niche 375 differences and may instead increase fitness differences (Spaak et al., 2021). 376 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint Intraspecific diversity consistently promoted community growth without affecting FD 377 change and recovery (Table S1-S4). This pattern suggests that intraspecific diversity 378 enhanced growth primarily through demographic buffering: increasing the number of 379 strains increases the likelihood that some strains maintain growth under stress and resume 380 growth once conditions improve. 381 4.3 Assembled functional diversity constrained community recovery 382 Communities with higher assembled FD at the end of the stress phase showed weaker 383 recovery across most conditions (Table S2). This reduction in community recovery likely 384 shows that communities retaining broader functional variation after stress had less scope 385 for further trait reorganization and therefore showed less growth once stressors were 386 removed (H2). Stress can reorganize communities by altering trait distributions and 387 thereby changing which ecological strategies remain available during recovery (Mouillot 388 et al., 2013; de Bello et al., 2021). 389 Under chemical stress this asymmetry was most evident: communities exposed to 390 atrazine, though functionally impoverished after stress, contained fast-growing, tolerant 391 strains able to rapidly recover after stressor removal, leading to greater recovery despite 392 low assembled FD. This effect was reversed under specific conditions: two-species 393 communities recovering from the combined stressor, higher assembled FD instead 394 enhanced recovery ( β = 3851, p = 0.036; Table S2), suggesting a possible interaction 395 between atrazine's filtering effect and warming's diversifying effect produced assembled 396 communities with both greater tolerance and greater FD than either stressor would 397 generate independently. Intraspecific diversity continued to promote density recovery 398 (strain diversity 4: p = 0.012, Table S2), suggesting demographic buffering as 399 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint communities containing more strains are more likely to retain strains capable of 400 sustaining growth post-stress (Bolnick et al., 2011; Siefert et al., 2015). 401 4.4 Functional diversity change during stress depended on stressor identity 402 FD changed during the stress phase even in control communities, indicating that 403 community trait structure was inherently dynamic over time (Fig S2). Relative to this 404 baseline change, atrazine shifted FD in a more negative direction whereas warming 405 shifted it in a more positive direction (Table S3). Chemical stress may have constrained 406 the range of physiological strategies capable of sustaining growth by limiting energy 407 acquisition through photosynthesis (Yang et al., 2021). Thermal stress, by contrast, can 408 modify metabolic processes and physiological traits, potentially increasing trait variation 409 among individuals (Anderson et al., 2022; Huang et al., 2025; Armin et al., 2025). The 410 combined treatment produced intermediate patterns of FD change (Table S3), consistent 411 with an interaction between two opposing mechanisms that moderated the FD outcomes 412 of either stressor acting alone (Jackson et al., 2021; Fig. 2). These contrasting outcomes 413 demonstrate that the mode of action of a stressor, governed whether stress contracted or 414 expanded community functional trait space. 415 4.5 Negative effect of initial functional diversity on functional diversity change during stress 416 Communities with higher initial FD experienced more negative FD change over time 417 (Table S4, Fig. 2). Because FD change was also negative in control communities, this 418 pattern reflects temporal changes in trait distributions that occur even in the absence of 419 stress. Trait convergence through time can arise from competitive exclusion or 420 environmental filtering that favor specific trait combinations as communities develop 421 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint (Mayfield and Levine, 2010; Kraft et al., 2015). In this context, stress may modify the 422 trajectory of this convergence rather than initiate it. 423 Interspecific diversity was associated with more negative FD change during the stress 424 phase (Fig S1). One possible explanation is that communities starting with larger trait 425 space may experience greater convergence as competitive interactions and environmental 426 constraints favor a narrower subset of trait combinations over time. Thus, the magnitude 427 of FD loss may scale with the amount of FD initially present. Because strain identities 428 were not tracked through time, the specific competitive dynamics driving this 429 convergence (whether through selective elimination of specific strains or through 430 phenotypic plasticity within surviving strains), remain a hypothesis rather than a 431 demonstrated process, and represent a target for future work. 432 4.6 Functional recovery primarily shaped by interspecific diversity 433 Interspecific diversity promoted functional recovery once stress was removed (species 434 diversity 2: p = 0.006; Table S4), reversing its negative stress-phase effect: strains 435 suppressed or reduced below detection thresholds during stress, may have re-established 436 once selective pressure was relaxed, restoring functional diversity to levels unattainable 437 by single-species communities (Elmqvist et al., 2003; Loreau et al., 2021; Loreau & de 438 Mazancourt, 2013). The contribution of assembled FD to functional recovery was 439 stressor-dependent and composition-specific. In two-species communities recovering 440 from the combined stress, higher assembled FD enhanced functional recovery ( β = 3529, 441 p = 0.024; thermal: β = 2959, p = 0.030; Table S4), suggesting that the trait combinations 442 retained under combined stress were particularly well positioned to diversify further once 443 stressors were removed. Rather than implying a universal positive effect of assembled FD 444 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint on recovery, this result indicates that recovery trajectories depend on which trait 445 combinations remain after stress and how those combinations interact with the post-stress 446 environment. 447 4.7 Perspectives 448 This study advances mechanistic understanding of diversity-stability relationships by 449 explicitly considering how multiple diversity types interact under stress, thereby 450 revealing the link between biodiversity, functional diversity, and community 451 performance. Different kinds of diversity had different effects: net negative effect of both 452 initial and assembled FD; net positive effects of inter and intraspecific diversity; and 453 case-specific effects of stressor identity. Taken together, these findings indicate that 454 community performance and functional responses to stress can become decoupled, with 455 communities recovering more rapidly than they regain functional trait diversity. These 456 findings carry direct relevance for predicting ecosystem trajectories under global change, 457 where chemical and thermal stressors increasingly co-occur and where recovery in 458 community performance may depend on how functional diversity is reorganized during 459 stress. Future work should characterize intraspecific trait distributions to identify which 460 strain-level properties confer vulnerability or resilience and extend temporal scope to 461 determine whether limitations on functional recovery are transient or reflect enduring 462 reorganization of community trait structure. 463 Author Contributions: Mark Holmes (data curation : lead, conceptualization: equal, analysis : 464 equal, writing: equal) ; Arunima Sikder (conceptualization: equal, analysis: equal, writing: lead); 465 Pauline Witsel (data curation: equal) ; Frederik de Laender (conceptualization: equal, analysis: 466 equal, writing: equal, supervision: lead). 467 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint Conflict of Interest Statement: the authors declare no conflict of interests. 468

469 1. Anderson, S. I., G. Franzè, J. D. Kling, P. Wilburn, C. T. Kremer, S. Menden-Deuer, E. 470 Litchman, D. A. Hutchins, and T. A. Rynearson. 2022. "The Interactive Effects of 471 Temperature and Nutrients on a Spring Phytoplankton Community." Limnology and 472 Oceanography 67 (3): 634–645. https://doi.org/10.1002/lno.12023 473 2. Armin, G., G. Boros, M. Kis, M. Burányi, H. Horváth, K. Krassován, T. Masuda, G. 474 Bernát, and K. Inomura. 2025. "The Effect of Temperature on Phytoplankton Physiology: 475 A Mesocosm and Modeling Study." Microbiology Spectrum 13 (10): e00457-25. 476 https://doi.org/10.1128/spectrum.00457-25 477 3. Baert, J. M., N. Eisenhauer, C. R. Janssen, and F. De Laender. 2018. "Biodiversity 478 Effects on Ecosystem Functioning Respond Unimodally to Environmental Stress." 479 Ecology Letters 21 (8): 1191–1199. https://doi.org/10.1111/ele.13088 480 4. Barabás, G., C. Parent, A. Kraemer, F. Van De Perre, and F. De Laender. 2022. "The 481 Evolution of Trait Variance Creates a Tension Between Species Diversity and Functional 482 Diversity." Nature Communications 13 (1): 2521. https://doi.org/10.1038/s41467-022-483 30090-4 484 5. Bolnick, D. I., P. Amarasekare, M. S. Araújo, R. Bürger, J. M. Levine, M. Novak, V. H. 485 W. Rudolf, S. J. Schreiber, M. C. Urban, and D. A. Vasseur. 2011. "Why Intraspecific 486 Trait Variation Matters in Community Ecology." Trends in Ecology and Evolution 26: 487 183–192. 488 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint 6. Cadotte, M. W. 2011. "The New Diversity: Management Gains Through Insights into the 489 Functional Diversity of Communities." Journal of Applied Ecology 48 (5): 1067–1069. 490 https://doi.org/10.1111/j.1365-2664.2011.02056.x 491 7. Cadotte, M. W. 2017. "Functional Traits Explain Ecosystem Function Through Opposing 492 Mechanisms." Ecology Letters 20: 989–996. 493 8. Campbell, V., G. Murphy, and T. N. Romanuk. 2011. "Experimental Design and the 494 Outcome and Interpretation of Diversity–Stability Relations." Oikos 120 (3): 399–408. 495 9. Craven, D., N. Eisenhauer, W. D. Pearse, Y. Hautier, F. Isbell, C. Roscher, M. Bahn, et 496 al. 2018. "Multiple Facets of Biodiversity Drive the Diversity–Stability Relationship." 497 Nature Ecology & Evolution 2 (10): 1579–1587. https://doi.org/10.1038/s41559-018-498 0647-7 499 10. de Bello, F., S. Lavorel, L. M. Hallett, E. Valencia, E. Garnier, C. Roscher, L. Conti, T. 500 Galland, et al. 2021. "Functional Trait Effects on Ecosystem Stability: Assembling the 501 Jigsaw Puzzle." Trends in Ecology and Evolution 36: 822–836. 502 11. Díaz, S., J. Kattge, J. H. C. Cornelissen, I. J. Wright, S. Lavorel, S. Dray, B. Reu, et al. 503 2016. "The Global Spectrum of Plant Form and Function." Nature 529 (7585): 167–171. 504 https://doi.org/10.1038/nature16489 505 12. Donohue, I., H. Hillebrand, J. M. Montoya, O. L. Petchey, S. L. Pimm, M. S. Fowler, K. 506 Healy, et al. 2016. "Navigating the Complexity of Ecological Stability." Ecology Letters 507 19 (9): 1172–1185. https://doi.org/10.1111/ele.12648 508 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint 13. Donohue, I., O. L. Petchey, J. M. Montoya, A. L. Jackson, L. McNally, M. Viana, K. 509 Healy, M. Lurgi, N. E. O'Connor, and M. C. Emmerson. 2013. "On the Dimensionality of 510 Ecological Stability." Ecology Letters 16 (4): 421–429. https://doi.org/10.1111/ele.12086 511 14. Eisenhauer, N., K. Mueller, A. Ebeling, G. Gleixner, Y. Huang, A.-M. Madaj, C. 512 Roscher, et al. 2024. "The Multiple-Mechanisms Hypothesis of Biodiversity–Stability 513 Relationships." Basic and Applied Ecology 79: 153–166. 514 https://doi.org/10.1016/j.baae.2024.07.004 515 15. Ellis, B., P. Haaland, F. Hahne, N. Le Meur, N. Gopalakrishnan, J. Spidlen, M. Jiang, and 516 G. Finak. 2024. flowCore: Basic Structures for Flow Cytometry Data. R package manual. 517 16. Elmqvist, T., C. Folke, M. Nyström, G. Peterson, J. Bengtsson, B. Walker, and J. 518 Norberg. 2003. "Response Diversity, Ecosystem Change, and Resilience." Frontiers in 519 Ecology and the Environment 1: 488–494. 520 17. Emmaneel, A. 2022. PeacoQC: Peak-Based Selection of High Quality Cytometry Data . 521 R package manual. https://doi.org/10.18129/B9.bioc.PeacoQC 522 18. Farrant, G. K., H. Doré, F. M. Cornejo-Castillo, F. Partensky, M. Ratin, M. Ostrowski, F. 523 D. Pitt, et al. 2016. "Delineating Ecologically Significant Taxonomic Units from Global 524 Patterns of Marine Picocyanobacteria." Proceedings of the National Academy of Sciences 525 113 (24): E3365–E3374. https://doi.org/10.1073/pnas.1524865113 526 19. F i s c h e r , F . M . , A . J . W r i g h t , N . E i s e n h a u e r , A . E b e l i n g , C . R o s c h e r , C . W a g g , A . 527 Weigelt, W. W. Weisser, and V. D. Pillar. 2016. "Plant Species Richness and Functional 528 Traits Affect Community Stability After a Flood Event." Philosophical Transactions of 529 the Royal Society B 371 (1694): 20150276. https://doi.org/10.1098/rstb.2015.0276 530 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint 20. Fontana, S., M. K. Thomas, M. Moldoveanu, P. Spaak, and F. Pomati. 2018. "Individual-531 Level Trait Diversity Predicts Phytoplankton Community Properties Better Than Species 532 Richness or Evenness." The ISME Journal 12 (2): 356–366. 533 https://doi.org/10.1038/ismej.2017.160 534 21. Grébert, T., H. Doré, F. Partensky, G. K. Farrant, E. S. Boss, M. Picheral, L. Guidi, et al. 535 2018. "Light Color Acclimation Is a Key Process in the Global Ocean Distribution of 536 Synechococcus Cyanobacteria." Proceedings of the National Academy of Sciences 115 537 (9): E2010–E2019. https://doi.org/10.1073/pnas.1717069115 538 22. Hillebrand, H., S. Langenheder, K. Lebret, E. Lindström, Ö. Östman, and M. Striebel. 539 2018. "Decomposing Multiple Dimensions of Stability in Global Change Experiments." 540 Ecology Letters 21 (1): 21–30. https://doi.org/10.1111/ele.12867 541 23. Holmes, M., P. Witsel, and F. De Laender. 2025. "Effects of Atrazine Concentration on 542 Growth Rates of Marine Synechococcus Strains." Dataset. Zenodo. 543 https://doi.org/10.5281/zenodo.17522383 544 24. Holzwarth, F., N. Rüger, and C. Wirth. 2015. "Taking a Closer Look: Disentangling 545 Effects of Functional Diversity on Ecosystem Functions with a Trait-Based Model 546 Across Hierarchy and Time." Royal Society Open Science 2 (3): 140541. 547 https://doi.org/10.1098/rsos.140541 548 25. Huang, D., C.-Q. Cheng, H.-Y. Zhang, Y. Huang, S.-Y. Li, Y.-T. Huang, X.-L. Huang, et 549 al. 2025. "Heat Shock Transcription Factor-Mediated Thermal Tolerance and Cell Size 550 Plasticity in Marine Diatoms." Nature Communications 16 (1): 3404. 551 https://doi.org/10.1038/s41467-025-58547-2 552 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint 26. Intergovernmental Panel on Climate Change (IPCC). 2023. Climate Change 2021 – The 553 Physical Science Basis: Working Group I Contribution to the Sixth Assessment Report . 554 Cambridge: Cambridge University Press. https://doi.org/10.1017/9781009157896 555 27. Jackson, M. C., S. Pawar, and G. Woodward. 2021. "The Temporal Dynamics of 556 Multiple Stressor Effects: From Individuals to Ecosystems." Trends in Ecology and 557 Evolution 36: 402–410. 558 28. Kraft, N. J. B., P. B. Adler, O. Godoy, et al. 2015. "Community Assembly, Coexistence 559 and the Environmental Filtering Metaphor." Functional Ecology 29: 592–599. 560 29. Loreau, M., and C. de Mazancourt. 2013. "Biodiversity and Ecosystem Stability: A 561 Synthesis of Underlying Mechanisms." Ecology Letters 16 (s1): 106–115. 562 https://doi.org/10.1111/ele.12073 563 30. Loreau, M., M. Barbier, E. Filotas, D. Gravel, F. Isbell, S. J. Miller, J. M. Montoya, et al. 564 2021. "Biodiversity as Insurance: From Concept to Measurement and Application." 565 Biological Reviews 96 (5): 2333–2354. https://doi.org/10.1111/brv.12756 566 31. Ma, Z., H. Liu, Z. Mi, Z. Zhang, Y. Wang, W. Xu, L. Jiang, and J.-S. He. 2017. "Climate 567 Warming Reduces the Temporal Stability of Plant Community Density Production." 568 Nature Communications 8 (1): 15378. https://doi.org/10.1038/ncomms15378 569 32. MacArthur, R., and R. Levins. 1967. "The Limiting Similarity, Convergence, and 570 Divergence of Coexisting Species." The American Naturalist 101 (921): 377–385. 571 https://www.jstor.org/stable/2459090 572 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint 33. Malek, M. J. T. 2023. flowDensity: Sequential Flow Cytometry Data Gating . R package 573 manual. https://doi.org/10.18129/B9.bioc.flowDensity 574 34. Mayfield, M., and J. M. Levine. 2010. "Opposing Effects of Competitive Exclusion on 575 the Phylogenetic Structure of Communities." Ecology Letters 13 (9): 1085–1093. 576 https://doi.org/10.1111/j.1461-0248.2010.01509.x 577 35. McCann, K. S. 2000. "The Diversity-Stability Debate." Nature 405 (6783): 228–233. 578 https://doi.org/10.1038/35012234 579 36. Mensens, C., F. De Laender, C. R. Janssen, K. Sabbe, and M. De Troch. 2017. "Different 580 Response–Effect Trait Relationships Underlie Contrasting Responses to Two Chemical 581 Stressors." Journal of Ecology 105: 1571–1582. 582 37. Mouillot, D., N. A. J. Graham, S. Villéger, N. W. H. Mason, and D. R. Bellwood. 2013. 583 "A Functional Approach Reveals Community Responses to Disturbance." Trends in 584 Ecology & Evolution 28 (3): 167–177. https://doi.org/10.1016/j.tree.2012.10.004 585 38. Nagler, T., and T. Vatter. 2025. Rvinecopulib: High Performance Algorithms for Vine 586 Copula Modeling. R package manual. https://CRAN.R-project.org/package=rvinecopulib 587 39. Nguyen, J., J. Lara-Gutiérrez, and R. Stocker. 2021. "Environmental Fluctuations and 588 Their Effects on Microbial Communities, Populations and Individuals." FEMS 589 Microbiology Reviews 45 (4): fuaa068. https://doi.org/10.1093/femsre/fuaa068 590 40. Oliver, T. H., N. J. B. Isaac, T. A. August, B. A. Woodcock, D. B. Roy, and J. M. 591 Bullock. 2015. "Declining Resilience of Ecosystem Functions Under Biodiversity Loss." 592 Nature Communications 6 (1): 10122. https://doi.org/10.1038/ncomms10122 593 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint 41. Olusoji, O. D., G. Barabás, J. W. Spaak, S. Fontana, T. Neyens, F. De Laender, and M. 594 Aerts. 2022. "Measuring Individual-Level Trait Diversity: A Critical Assessment of 595 Methods." Oikos 2023 (4). https://doi.org/10.1111/oik.09178 596 42. Olusoji, O. D., J. W. Spaak, M. Holmes, T. Neyens, M. Aerts, and F. De Laender. 2021. 597 "cyanoFilter: An R Package to Identify Phytoplankton Populations from Flow Cytometry 598 Data Using Cell Pigmentation and Granularity." Ecological Modelling 460: 109743. 599 https://doi.org/10.1016/j.ecolmodel.2021.109743 600 43. Orr, J. A., S. J. Macaulay, A. Mordente, B. Burgess, D. Albini, J. G. Hunn, K. Restrepo-601 Sulez, et al. 2024. "Studying Interactions Among Anthropogenic Stressors in Freshwater 602 Ecosystems: A Systematic Review of 2396 Multiple-Stressor Experiments." Ecology 603 Letters 27 (6): e14463. https://doi.org/10.1111/ele.14463 604 44. P a l a c i o , F . X . , G . O t t a v i a n i , S . M a m m o l a , C . G r a c o - R o z a , F . d e B e l l o , a n d C . P . 605 Carmona. 2025. "Integrating Intraspecific Trait Variability in Functional Diversity: An 606 Overview of Methods and a Guide for Ecologists." Ecological Monographs 95 (2): 607 e70024. https://doi.org/10.1002/ecm.70024 608 45. Pecl, G. T., M. B. Araújo, J. D. Bell, J. Blanchard, T. C. Bonebrake, I.-C. Chen, T. D. 609 Clark, et al. 2017. "Biodiversity Redistribution Under Climate Change: Impacts on 610 Ecosystems and Human Well-Being." Science 355 (6332): eaai9214. 611 https://doi.org/10.1126/science.aai9214 612 46. Pennekamp, F., M. Pontarp, A. Tabi, F. Altermatt, R. Alther, Y. Choffat, E. A. Fronhofer, 613 et al. 2018. "Biodiversity Increases and Decreases Ecosystem Stability." Nature 563 614 (7729): 109–112. https://doi.org/10.1038/s41586-018-0627-8 615 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint 47. Pimm, S. 1984. "The Complexity and Stability of Ecosystems." Nature 307: 321–326. 616 https://doi.org/10.1038/307321a0 617 48. Polazzo, F., and A. Rico. 2021. "Effects of Multiple Stressors on the Dimensionality of 618 Ecological Stability." Ecology Letters 24 (8): 1594–1606. 619 https://doi.org/10.1111/ele.13770 620 49. R Core Team. 2024. R: A Language and Environment for Statistical Computing . Vienna: 621 R Foundation for Statistical Computing. https://www.R-project.org/ 622 50. Rippka, R., T. Coursin, W. Hess, C. Lichtle, D. J. Scanlan, K. A. Palinska, I. Iteman, F. 623 Partensky, J. Houmard, and M. Herdman. 2000. "Prochlorococcus marinus Chisholm et 624 al. 1992 subsp. Pastoris subsp. Nov. Strain PCC 9511, the First Axenic Chlorophyll 625 A2/B2-Containing Cyanobacterium." International Journal of Systematic and 626 Evolutionary Microbiology 50 (5): 1833–1847. https://doi.org/10.1099/00207713-50-5-627 1833 628 51. Ross, S. R. P.-J., O. L. Petchey, T. Sasaki, and D. W. Armitage. 2023. "How to Measure 629 Response Diversity." Methods in Ecology and Evolution 14 (5): 1150–1167. 630 https://doi.org/10.1111/2041-210X.14087 631 52. Rutten, T. P. A., B. Sandee, and A. R. T. Hofman. 2005. "Phytoplankton Monitoring by 632 High Performance Flow Cytometry: A Successful Approach?" Cytometry Part A 64A (1): 633 16–26. https://doi.org/10.1002/cyto.a.20106 634 53. Schleuter, D., M. Daufresne, F. Massol, and C. Argillier. 2010. "A User's Guide to 635 Functional Diversity Indices." Ecological Monographs 80 (3): 469–484. 636 https://doi.org/10.1890/08-2225.1 637 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint 54. Schucht, T., and B. Blasius. 2025. "Shared Preferences Along Stress Gradients: How a 638 Growth-Tolerance Trade-Off Drives Unimodal Diversity and Trait Lumping." 639 Theoretical Ecology 18 (1): 13. https://doi.org/10.1007/s12080-025-00607-w 640 55. Siefert, A., C. Violle, L. Chalmandrier, et al. 2015. "A Global Meta-Analysis of the 641 Relative Extent of Intraspecific Trait Variation in Plant Communities." Ecology Letters 642 18: 1406–1419. 643 56. Six, C., J. C. Thomas, L. Garczarek, M. Ostrowski, A. Dufresne, N. Blot, D. J. Scanlan, 644 and F. Partensky. 2007. "Diversity and Evolution of Phycobilisomes in Marine 645 Synechococcus spp.: A Comparative Genomics Study." Genome Biology 8 (12). 646 https://doi.org/10.1186/gb-2007-8-12-r259 647 57. Ś liwiń ska-Wilczewska, S., Z. Konarzewska, K. Wi ś niewska, and M. Konik. 2020. 648 "Photosynthetic Pigments Changes of Three Phenotypes of Picocyanobacteria 649 Synechococcus sp. Under Different Light and Temperature Conditions." Cells 9 (9): 650 2030. https://doi.org/10.3390/cells9092030 651 58. Spaak, J. W., and F. De Laender. 2021. "Effects of Pigment Richness and Size Variation 652 on Coexistence, Richness and Function in Light-Limited Phytoplankton." Journal of 653 Ecology 109 (6): 2385–2394. https://doi.org/10.1111/1365-2745.13645 654 59. Spaak, J. W., C. Carpentier, and F. De Laender. 2021. "Species Richness Increases 655 Fitness Differences, but Does Not Affect Niche Differences." Ecology Letters 24 (12): 656 2611–2623. https://doi.org/10.1111/ele.13877 657 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint 60. Tilman, D., J. Knops, D. Wedin, P. Reich, M. Ritchie, and E. Siemann. 1997. "The 658 Influence of Functional Diversity and Composition on Ecosystem Processes." Science 659 277 (5330): 1300–1302. https://doi.org/10.1126/science.277.5330.1300 660 61. Van Meerbeek, K., T. Jucker, and J.-C. Svenning. 2021. "Unifying the Concepts of 661 Stability and Resilience in Ecology." Journal of Ecology 109 (9): 3114–3132. 662 https://doi.org/10.1111/1365-2745.13651 663 62. Violle, C., B. J. Enquist, B. J. McGill, L. Jiang, C. H. Albert, C. M. Hulshof, V. Jung, and 664 J. Messier. 2012. "The Return of the Variance: Intraspecific Variability in Community 665 Ecology." Trends in Ecology and Evolution 27 (4): 244–252. 666 https://doi.org/10.1016/j.tree.2011.11.014 667 63. Yachi, S., and M. Loreau. 1999. "Biodiversity and Ecosystem Productivity in a 668 Fluctuating Environment: The Insurance Hypothesis." Proceedings of the National 669 Academy of Sciences 96 (4): 1463–1468. https://doi.org/10.1073/pnas.96.4.1463 670 64. Yang, L., S. Mou, H. Li, Z. Zhang, N. Jiao, and Y. Zhang. 2021. "Terrestrial Input of 671 Herbicides Has Significant Impacts on Phytoplankton and Bacterioplankton Communities 672 in Coastal Waters." Limnology and Oceanography 66 (11): 4028–4045. 673 https://doi.org/10.1002/lno.11940 674 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted April 23, 2026. ; https://doi.org/10.64898/2026.04.21.719904doi: bioRxiv preprint