Support for the social buffering hypothesis, especially under unpredictable precipitation regimes

This work was supported by a NERC Pushing the Frontiers 25 grant (NE/X013766/1) to RSG. Some ideas for this paper originated in discussions with 26 colleagues during the Royal Society meeting “Understanding age and society using 27 natural populations”. 28 Conflicts of interest None 29 Author Contributions RSG conceived the ideas and designed the methodological 30 approach with input from KD and LB. RSG collected the pertinent data or accessed it 31 via open-access repositories. RSG curates COMADRE. RSG performed the analyses 32 with input from GS. RSG wrote the first draft with input from KD. All coauthors 33 contributed to subsequent versions of the manuscript. 34 Statement of inclusivity The authorship is composed of researchers of all academic 35 stages, genders, and several nationalities, including multiple languages. 36 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 3

37 Sociality is hypothesised to buffer populations against environmental stochasticity by 38 stabilising key vital rates, yet this has rarely been tested across diverse taxa. We 39 compiled demographic time series from 87 populations for 66 animal species spanning 40 12 taxonomic classes to test whether sociality promotes demographic buffering and 41 whether this relationship depends on environmental context. Using stochastic 42 elasticities and phylogenetically comparative analyses, we find broad support for the 43 social buffering hypothesis: more social species exhibit lower sensitivity of population 44 growth to vital rate variance (lower stochastic elasticity to vital rate variances, |T σ |), 45 particularly in juvenile survival and reproduction. Contrary to expectations, juvenile vital 46 rates, but not adult survival, showed the strongest buffering patterns in social taxa, 47 suggesting life-stage-specific social mechanisms. Furthermore, we show that social 48 species invest more in maximising the average performance of adult survival and 49 reproduction (higher stochastic elasticity to vital rate means, Tμ ), consistent with 50 adaptive strategies favouring canalisation of fitness-related traits. This buffering benefit 51 is especially relevant in highly unpredictable environments: in climates with erratic 52 precipitation, social species have greater demographic advantages, exhibiting even 53 lower Tσ values. Together, these findings provide support for the social buffering 54 hypothesis, especially in highly stochastic environments. Social traits confer some 55 demographic stability under moderate climatic conditions, but may become more 56 relevant under extreme stochasticity—an insight particularly important for understanding 57 population resilience in a rapidly changing world. 58 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 4

comparative demography; cooperative breeding; demographic buffering; 59 environmental stochasticity; phylogenetic analysis; sociality continuum. 60 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 5

61 Sociality is crucial for the persistence of natural populations. The way in which 62 individuals interact within social groups influences not only individual fitness (Silk, 63 2007b; Snyder-Mackler et al., 2020), but also shapes population-level processes such 64 as survival (Creel & Christianson, 2008; Salguero-Gómez, 2024), reproduction (Lukas & 65 Clutton-Brock, 2012; Salguero-Gómez, 2024), dispersal (Greenwood, 1980), and 66 resistance to environmental pressures (Oro, 2020; Rubenstein & Lovette, 2007; 67 Salguero-Gómez, 2024). Social organisations, from ephemeral aggregations to highly 68 coordinated closed social groups, have evolved across diverse taxa as adaptive 69 responses to predation (Krause & Ruxton, 2002), resource limitation (Koenig & 70 Dickinson, 2016), or reproductive constraints (Emlen, 1982). Interactions that occur 71 between members of a social organisation often mediate access to resources and 72 mates (Clutton-Brock, 2009) , affect stress physiol ogy (Young & B ennett, 2013), and 73 enable individuals to cope with environmental uncertainty (Bourke, 2011). As such, 74 sociality is a central feature of the viability of natural populations. 75 Social organisation is a multifaceted trait. However, this characteristic has been 76 historically often boiled down to a binary trait that is mostly examined in a few taxonomic 77 groups. Indeed, while much of the previous literature distinguishes species as either 78 social or non-social (Lukas & Clutton-Brock, 2012), this dichotomy is now often 79 considered as insufficient for capturing the complexity and gradation of social 80 behaviours observed across the animal kingdom. For instance, some spiders exhibit 81 facultative sociality, forming colonies under high resource availability but remaining 82 solitary otherwise (Avilés et al., 2012). Moreover, most comparative research in this 83 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 6 area has focused on birds (Jetz et al., 2012) and mammals (Lukas & Clutton-Brock, 84 2012), where social systems have been extensively studied. This narrow focus has left 85 a gap regarding the generality of how social organisations influence demography across 86 animals, including in less well-studied taxa such as reptiles, amphibians, invertebrates 87 (but see (Crump, 2015; Doody et al., 2013; Wong & Balshine, 2011)), and marine 88 organisms. A growing consensus now acknowledges the need for integrative 89 frameworks that account for the dimensionality of sociality (Albery et al., 2024; Doody et 90 al., 2013; Firth et al., 2024; Salguero-Gómez, 2024), including the kinds, intensity, and 91 frequency of said interactions. 92 There are clear implications regarding how sociality should allow populations to 93 persist via the social buffering hypothesis (Klug & Bonsall, 2014; Rubenstein & Lovette, 94 2007; Shine, 1978). This hypothesis posits that social interactions, particularly those 95 involving cooperative care, resource sharing, or communal defence, can mitigate the 96 negative demographic impacts of environmental fluctuations. By pooling resources, 97 enhancing survival, or smoothing reproductive success over time, sociality may reduce 98 demographic variance and stabilise population growth. However, the strength and 99 consistency of such buffering effects are expected to vary depending on the type of 100 social organisation, with cooperative breeders hypothesised to benefit more strongly 101 from such mechanisms than solitary, loosely gregarious, or communal species 102 (Salguero-Gómez, 2024). 103 Population ecologists have extensively examined the drivers of demographic 104 buffering (J.-M. Gaillard & Yoccoz, 2003; Morris et al., 2008; Pfister, 1998; Tuljapurkar, 105 1982), a life-history strategy whereby natural selection reduces temporal variability in 106 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 7 vital rates that most affect population fitness. This strategy is predicted to evolve under 107 environmental stochasticity, as a means of stabilising population growth in the face of 108 unpredictable ecological pressures (Bruijning et al., 2020; Hilde et al., 2020; Le Coeur et 109 al., 2022; Tuljapurkar, 1982). Traditionally, demographic buffering has been assessed, 110 among other ways (Lewontin & Cohen, 1969), by examining the coefficient of variation 111 in vital rates or their temporal variance (Hilde et al., 2020; Pfister, 1998). However, 112 recent theoretical and empirical work has emphasised the utility of stochastic elasticity 113 analysis in explicitly quantifying how changes in the mean and variance of vital rates 114 affect long-term population growth rate ( λ/i2 ) (Ezard & Coulson, 2010; S. Gascoigne et 115 al., 2024; S. J. L. Gascoigne et al., 2023; Giaimo & Traulsen, 2023; Haridas & 116 Tuljapurkar, 2005). 117 Stochastic elasticities directly quantify the buffering capacity of a population due 118 to its vital rates. In this framework, the stochastic elasticity to the mean of vital rates (Tμ ) 119 reflects the relative importance of selection for improving average demographic 120 performance, whereas the stochastic elasticity to vital rate variance ( Tσ ) captures the 121 sensitivity of λ/i2 to interannual variability in those same vital rates. Importantly, Tμ and 122 Tσ probe alternative buffering mechanisms by which populations can persist in variable 123 environments (S. J. L. Gascoigne et al., 2024). Across species, the impacts of mean 124 vital rates on population growth rate in variable environments covary with life history 125 strategies. For example, mean adult survival is especially important in species with 126 longer generation times (Oli & Dobson, 2003; Saether & Bakke, 2000). Furthermore, Tσ 127 also covaries with life history strategies. Specifically, the quantification of Tσ across taxa 128 has demonstrated the increased sensitivity of species with shorter lifespans to variation 129 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 8 in demographic rates relative to those with longer lifespans (Morris et al., 2008). 130 Together, these metrics provide a powerful and evolutionarily relevant view of how 131 organisms manage trade-offs between maximising fitness and buffering environmental 132 uncertainty. Despite their theoretical relevance, T μ and T σ have rarely been applied in 133 macroecological or behavioural studies—particularly not to test the role of sociality in 134 modulating selection on demographic stability. Few studies have explicitly linked 135 behavioural traits such as social organisation to either of these components of 136 population dynamics, despite predictions that social behaviour may act as a buffering 137 mechanism (Rubenstein & Lovette, 2007; Salguero-Gómez, 2024). Bridging this gap is 138 essential to understand whether social systems merely enhance survival and 139 reproduction, or also reduce vulnerability to environmental stochasticity—a distinction 140 with major implications for forecasting population resilience under global change. 141 Here, we collate data on 87 populations and 66 animal species spanning 12 142 taxonomic classes, thus encompassing a wide diversity of life histories and social 143 systems, to examine the demographic consequences of sociality through the lens of 144 demographic buffering. We test four key hypotheses. (H1) Social species should exhibit 145 stronger demographic buffering, operationalised as lower temporal variance vital rates, 146 relative to solitary taxa, in line with the social buffering hypothesis (Rubenstein & 147 Lovette, 2007). This buffering is predicted to be especially pronounced in cooperative 148 breeders, where group-living and alloparental care redistribute energetic costs and 149 reduce reproductive uncertainty. (H2) This buffering effect should be primarily mediated 150 via adult stages, particularly adult survival, which tends to be more canalised in long-151 lived species (J.-M. Gaillard & Yoccoz, 2003; Morris et al., 2008; Pfister, 1998), while 152 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 9 juvenile rates should contribute minimally to buffering patterns. (H3) More social species 153 should exhibit higher stochastic elasticity to the mean of adult survival and reproduction 154 than to juvenile survival and maturation, indicating stronger selection for maximising the 155 average performance of key adult vital rates in social species compared to solitary 156 ones. Finally, (H4) the hypothesised relationship between sociality and demographic 157 buffering should be modulated by environmental stochasticity, such that buffering 158 effects in social species diminish under highly variable conditions, e.g., in arid regions 159 with high rainfall variability. Indeed, recent evidence has emerged suggesting that 160 demographic buffering can be ineffective in extremely stochastic environments 161 (Rodriguez-Caro et al., 2021; Santos et al., 2024). To test these four hypotheses, we 162 develop a comparative phylogenetic framework that incorporates time series of 163 demographic information, species-specific climatic variability, and a recently introduced 164 sociality continuum, which classifies species from more solitary to fully social (Salguero-165 Gómez, 2024). 166 167

168 To test the hypotheses that sociality modulates demographic buffering in animals via 169 adult survival and its interaction with environmental stochasticity, we performed a suite 170 of phylogenetically-informed comparative analyses. Our analyses build on and extend 171 two recent methodological frameworks: one developing novel demographic metrics to 172 quantify the degree of demographic buffering in natural populations using matrix 173 population models (MPMs, hereafter) (Santos et al., 2023), and another one introducing 174 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 10 a continuum of social organisation for animal species, from more solitary to more social 175 (Salguero-Gómez, 2024). By incorporating longitudinal demographic data specifically 176 designed to capture temporal variation in survival and reproduction, and linking them to 177 the climatic patterns associated with each studied location, we also test the hypothesis 178 that the expected relationships between sociality and demography break down in 179 extreme environments. 180 181 Demographic data 182 To quantify the strength of demographic buffering across species, we first compiled 183 demographic time series data for animal populations from the COMADRE Animal Matrix 184 Database (v.4.32.3.1) (Salguero-Gomez et al., 2016), a global open-access repository 185 of age- and stage-structured matrix population models (MPMs) for animals. We 186 complemented this database with demographic data on Suricata suricatta (meerkat) 187 from (Conquet et al., 2023), which also includes an annual long-term series of MPMs. 188 To ensure the reliability, comparability, and biological realism of these MPMs, we 189 applied a stringent series of selection criteria. Indeed, in its version 4.23.3.1, 190 COMADRE contains 3,488 MPMs from 429 populations across 415 peer-reviewed 191 studies. These studies, though peer-reviewed, were not necessarily built with 192 comparative demography in mind by each single set of authors (Salguero-Gomez et al., 193 2021). As such, we imposed the following set of selection criteria using R (v. 4.3.2) and 194 the R packages Rcompadre and Rage (Jones et al., 2022), and popbio (Stubben & 195 Milligan, 2007). We retained only MPMs that satisfied the following conditions: 196 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 11 - Wild populations only: We excluded MPMs parameterised with data from 197 laboratory settings, captive populations ( e.g., zoos), or experimentally 198 manipulated populations by filtering the metadata of the COMADRE R object by 199 the variables ‘Captivity’ and ‘MatrixTreatment’. This step ensured that emergent 200 demographic traits would reflect natural ecological and social dynamics. 201 - Proper matrix decomposition: We required that each MPM be decomposed into 202 its survival-development (U ) and reproduction ( F) submatrices such that the 203 overall MPM A = U + F (Caswell 2001). This structure allowed us to calculate 204 standard life history traits and to perform matrix algebra reliably across species. 205 - Biological plausibility of vital rates: Using the ‘cdb_flag’ function in the R package 206 Rcompadre, we excluded matrices with missing values or with biologically 207 implausible values ( i.e.., survival probabilities 0 < σ ≤ 1). MPMs containing F 208 submatrices solely composed of zero-values were also excluded. 209 - Animals only: We retained only MPM associated with the kingdom Animalia by 210 filtering the ‘Kingdom’ metadata, and thus excluding any entries from Bacteria or 211 Fungi. 212 - Time series completeness: We grouped MPMs by species, study (based on the 213 ‘Authors’, ‘Source’ and ‘YearPublication’ metadata), and population 214 (‘MatrixPopulation’), and retained only those populations for which at least three 215 annual MPMs were available. This step allowed us to estimate the impact of 216 interannual variability in vital rates on population growth while minimising bias 217 due to small sample sizes. We note that the primary results we show are largely 218 insensitive to the available data duration of each study (Table S1). 219 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 12 - Population selection for multi-population species: For species represented by 220 multiple studies (3 species in this study), we retained the studies containing the 221 longest and best spatially replicated data. 222 - Humans: In COMADRE, we have access to temporally replicated MPMs for 223 humans ( Homo sapiens sapiens ) for 42 countries (Nicol-Harper et al., 2018). 224 However, to strike a compromise between having humans represented in this 225 analysis, but not overwhelm the analyses with highly unbalanced population 226 replications for any one species, and also due to the difficulties in assigning 227 buffering abilities against environmental st ochasticity to humans (Mondal et al., 228 2024), we retained only one human population: Spain. 229 - Standardising projection intervals: Because MPMs in COMADRE span different 230 projection intervals (e.g., 6 months, 5 years; Figure S1), we back-transformed all 231 MPMs to a 1-year time step using the methods by elevating each matrix element 232 aij to the power of the frequency of study, following Salguero-Gómez and 233 Gamelon (Salguero-Gomez & Gamelon, 2021) and Salguero-Gómez (Salguero-234 Gómez, 2024). 235 This set of selection criteria resulted in a subset of COMADRE for the next 236 analytical steps, comprising 87 populations from 66 animal species, and totalling 955 237 MPMs (Figure S2). 238 Estimates of demographic buffering 239 To quantify demographic buffering across the resulting subset of animal species, we 240 focused on interannual variability in stage-specific survival ( σ ), maturation ( γ ), and 241 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 13 reproduction (φ ). However, as the dimension of the MPMs varied c onsiderably (Figure242 S3), we collapsed all MPMs with dimension >2 ( i.e., representing more than two stages243 in the life cycle of the species) down to 2×2 MPMs, where t he first stage represents244 juveniles (J) and the second stage adults ( A). To do so, we implemented the collapsing245 criterion developed by Salguero-Gómez and Plotkin (2010) , which allows for the246 collapsing of any MPM while retaining it s eigenstructure. Following suggestions from247 Salguero-Gómez and Plotkin (2010) , we kept the first stage unaltered, and collapsed248 into the adult stage from the second life cycle stage onwards. As such, the overall249 structure of the resulting MPMs contain four vital rates, shown in Eq. 1: juvenile survival250 (σ J), juvenile maturation (γ ), adult survival (σ A), and reproduction (φ ). 251 A = (Eq. 1)252 Next, we quantified the stochastic elasti cities of population growth rate ( λ s) to253 both changes in the mean and in the variance of vital rates across each population’s254 time series of MPMs. Briefly, stochastic elasticities describe the propor tional change in255 long-term stochastic growth rate in response to small, proportional changes in a given256 vital rate, thus making them ideal tools to examine how natural populations respond to257 environmental stochasticity (Haridas & Tuljapurkar, 2005; Tuljapurkar et al., 2003) .258 Following Haridas and Tuljapurkar (2005) , we distinguished between two forms of259 elasticity: (i) the elasticity of t he stochastic population growth rate ( λ s) to mean of vital260 rates, Evr μ , which isolates the impact of changi ng the mean value of a given vital rate vr261 in the MPM (Eq. 1) while holding its variability constant; and (ii) the elasticity of λ s to the262 changes in the variance of a given vital rate, Evr σ , which isolates the impact of changing263 13 re es ts ng he m ed all al 1) to ’s in en to . of tal vr he ng .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 14 variability while holding the mean of said vital rate (and all others) constant. These two 264 types of stochastic elasticities are critical for evaluating demographic buffering, which is 265 defined as selection to reduce the sensitivity of λ s to variability in vital rates (Hilde et al., 266 2020; Pfister, 1998; Santos et al., 2023). 267 We calculated these elasticities using population-specific MPMs constructed from 268 each of the four annual vital rate estimates shown in Eq. 1: juvenile survival ( σ J), 269 juvenile maturation ( γ ), adult survival ( σ A), and reproduction ( φ ). For each population, 270 we implemented the approach detailed in Haridas and Tuljapurkar (Haridas & 271 Tuljapurkar, 2005), which expresses λ s as a function of the full time series of MPMs and 272 uses perturbation analysis to compute the stochastic elasticities. Specifically, we 273 calculated Evr μ as the derivative of log( λ s) with respect to the mean of the vital rate vr, 274 and Evr σ as the derivative with respect to its standard deviation, both obtained via 275 numerical approximations using finite perturbations that are relative to the value of the 276 vital rates under examination (S. J. L. Gascoigne et al., 2023; Haridas & Tuljapurkar, 277 2005). By separating effects of means and variances, this approach allows us to 278 estimate the total elasticity of λ s to changes in the mean of vital rates ( Tμ ) and the total 279 elasticity of λ s to changes in the variance of vital rates ( Tσ ) by summing across all vital 280 rate elasticities in each population: 281 Tμ = ∑ Evr μ ( E q . 2 ) 282 Tσ = ∑ Evr σ ( E q . 3 ) 283 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 15 To compute these elasticities empirically, we used the ‘stoch.elas’ function in the 284 popbio R package (Stubben & Milligan, 2007) as a baseline, which we then modified to 285 generate separate elasticities for the mean and standard deviation of each vital rate 286 using Monte Carlo simulation across MPM resamples. Specifically, for each population, 287 we calculated the stochastic growth rate λ s from 10,000 iterations of matrix multiplication 288 through the random time series of MPMs (one per year). Then, we introduced a small 289 perturbation (1%) to either the mean or the variance of each vital rate independently 290 while keeping all others fixed. We then recalculated λ s and took the finite-difference 291 approximation of each elasticity (Haridas & Tuljapurkar, 2005). The separate 292 components of the sum of mean elasticities ( Tμ ) thus measures the aggregate selective 293 importance of vital rate averages and helps us test (H3) whether more social species 294 show higher stochastic elasticity to adult survival and reproduction than to juvenile 295 rates, suggesting stronger selection on adult vital rate averages in social than solitary 296 species. In contrast, the sum of variance elasticities ( Tσ ) measures the aggregate 297 selective importance of vital rate variances , and as such is directly relevant to our test 298 (H1) of whether more social species exhibit reduced sensitivity to vital rate variance. As 299 the distributions of T μ and Tσ were right-skewed, we log 10-transformed them before 300 using them as response variables in our phylogenetic comparative analyses (below). 301 We also applied a log 10-transformation to the four vital rate stochastic elasticities to the 302 mean (Eσ J μ , Eγ μ , Eσ A μ and Eφ μ ) and to the variance (Eσ J σ , Eγ σ , Eσ A σ and Eφ σ ). 303 It is worth noting that, ecologically and evolutionarily, Tμ and Tσ (and their 304 respective underlying vital rate components: Evr μ and Evr σ ) capture complementary 305 forces. The stochastic elasticity to changes in mean vital rates reflects selection for life 306 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 16 history traits that increase average performance ( e.g., survival, reproduction), while the 307 stochastic elasticity to changes in the variance of vital rates reflects selection against 308 variability ( i.e., demographic buffering). Under the demographic buffering hypothesis, 309 we expect a higher degree of sociality to correlate with lower T σ , indicating reduced 310 exposure of λ s to environmental noise. The framework from Haridas and Tuljapurkar 311 (Haridas & Tuljapurkar, 2005) also contains a unique property associated with elasticity 312 analyses, shown in Eq. 4. Specifically, with equal proportional changes in mean and 313 variance in vital rates, the stochastic elasticity E vr is the summation of E vr μ and E vr σ . 314 Since, with all levels of variability in vital rates the sum of stochastic elasticity Evr sums 315 to one, the summations of elasticity values Tμ and T σ also equal to one. This property 316 holds for all possible life histories. In turn, just like their deterministic counterparts 317 (Takada et al., 2018), Tμ and T σ are valuable tools for comparative analysis as their 318 values are unbiased by the matrix dimensionality (i.e., the number of stages in the MPM 319 prior to downscaling to a 2×2 MPM, as done here), life history complexity ( e.g., 320 iteroparous vs. semelparous), or asymptotic properties ( e.g., declining, stable or 321 increasing populations as per λ s) of the MPMs. 322 Tμ + Tσ = 1 ( E q . 4 ) 323 Importantly, Tμ and Tσ inhabit different numeric domains. Tμ is invariably positive 324 as the summed impacts of minor increases in the mean of all vital rates yield an 325 increase to the stochastic population growth rate λ s (Haridas & Tuljapurkar, 2005). On 326 the other hand, T σ is invariably negative as the summed impacts of minor increases in 327 the variance of all vital rates lead to a decrease in λ s (Haridas & Tuljapurkar, 2005) . In turn, to 328 allow for the log 10-transformation of T σ , Eσ J σ , E γ σ , E σ A σ and E φ σ , we used absolute 329 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 17 values for comparative analysis as in Santos et al. (2023). Consequently, the absolute-330 value transformation yields values that negatively relate to the degree of demographic 331 buffering, with higher values of Tσ and |Evr σ | identifying less buffered populations. 332 333 Sociality classification 334 To classify the degree of sociality across the 66 animal species included in this study, 335 we employed a five-level sociality continuum previously introduced and justified by 336 Salguero-Gómez (2024). This continuum captures gradations in social organisation 337 across taxa by integrating spatial cohesion, temporal stability, and the frequency of 338 social interactions. The five ordinal levels are: (1) Solitary: individuals live alone except 339 during brief breeding encounters ( e.g., tigers, some wasps); (2) Gregarious: individuals 340 form temporary or fluid groups but engage in limited social coordination (e.g. , 341 wildebeest, schooling fish); (3) Communal: individuals cohabit or share nesting sites but 342 do not engage in cooperative care (e.g. , purple martins, some reef fish); (4) Colonial: 343 individuals consistently share nesting or living spaces, often in dense aggregations 344 (e.g., seabirds, coral polyps); and (5) Social: individuals form stable, cooperative groups 345 with persistent social bonds and behavioural coordination, including cooperative 346 breeding or hierarchical organisations (e.g ., meerkats, female elephants, baboons). 347 These categories were designed to be taxonomically agnostic, allowing consistent 348 application across a wide range of animal taxa. 349 We assigned sociality scores based on species-level information gathered from a 350 combination of curated databases (e.g ., Animal Diversity Web (Dewey et al., 2010), 351 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 18 FishBase (Froese & Pauly, 2002), IUCN Red List (Gearty & Chamberlain, 2025)), peer-352 reviewed literature, and expert consultation. When available, we relied on published 353 ethological reviews and species-specific studies documenting group size, social 354 behaviour, and breeding systems. Each classification was independently validated by at 355 least one taxonomic expert. In cases of ambiguity, we adopted a conservative approach 356 by assigning the species to the lower of two adjacent categories unless strong evidence 357 supported otherwise. See Salguero-Gómez (2024) and Table S2 for further details on 358 our scoring criteria. 359 360 Adult body mass 361 As life history traits, vital rates, and demographic buffering metrics often covary 362 allometrically with body size (Calder 1984; Charnov 1993; Blueweiss et al. 1978), we 363 explicitly discounted the effect of body mass in our phylogenetic comparative models. 364 For each species in our dataset, we obtained adult body mass estimates from curated 365 trait databases specific to major vertebrate and invertebrate taxa. Mammalian and avian 366 body mass data were extracted primarily from the AnAge (de Magalhães & Costa, 367 2009) and AVONET databases (Tobias et al., 2022), respectively, while data for 368 reptiles, amphibians, and fish were sourced from the Amniote Database (Myhrvold et 369 al., 2015) and FishBase (Froese & Pauly, 2002), the latter using the rfishbase R 370 package (Boettiger et al., 2012). For invertebrates and less-studied groups, we relied on 371 MOSAIC (Bernard et al., 2023), data from Healy et al. (2019), and primary literature 372 searches via Web of Science using species names and the keywords “adult body mass” 373 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 19 or “adult weight.” All mass values were converted to grams and log 10-transformed prior 374 to analysis to meet assumptions of linearity and homoscedasticity. When multiple mass 375 estimates were available for a species, we took the mean of adult female body mass 376 when reported, as this tends to better reflect demographic contributions in iteroparous 377 animals (J.-M. Gaillard et al., 2005; Isaac et al., 2007). 378 379 Phylogenetic tree 380 To account for non-independence due to shared evolutionary history, we constructed a 381 phylogenetic tree spanning all 66 species in our dataset using the Open Tree of Life as 382 implemented in the rotl R package (Michonneau et al., 2016). First, we matched species 383 names to OTL taxonomy using the ‘tnrs_match_names’ function, resolving any 384 synonyms or misspellings manually. Using the matched taxon IDs, we retrieved a 385 synthetic, ultrametric phylogeny. This resulting tree incorporates curated backbone 386 information from published phylogenies across major animal clades. Branch lengths 387 were scaled using divergence time estimates where available in the OTL backbone. For 388 compatibility with downstream comparative models, we ensured the tree was fully 389 bifurcating and resolved any polytomies using the ‘multi2di’ function from the ape R 390 package (Paradis et al., 2004). 391 For species represented by multiple populations (n = 12 species, Table S2), we 392 modified the tree above to create population-level phylogenetic tips by duplicating the 393 species-level branch and assigning each population a unique identifier (e.g., 394 Species_1_Pop_1, Species_1_Pop_2). These within-species duplicates were assigned 395 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 20 to nearly zero-length terminal branches, thereby preserving the species-level topology 396 and divergence times while accommodating population-level replication in the data. This 397 approach allowed us to test hypotheses using the full population-level dataset (n = 87 398 populations) while still accounting for phylogenetic structure at the species level. The 399 order of introduction of the populations within the tip of each species does not affect 400 assessments in macroecological studies using COMADRE (Merrien et al., 2021). The 401 resulting population-expanded tree was used for all comparative analyses, including 402 phylogenetic ANOVAs, PGLS models, and estimation of phylogenetic signal (Pagel’s λ ), 403 using the R packages caper (Orme et al., 2013), nlme (Heisterkamp et al., 2017), and 404 phytools (Revell, 2012). 405 406 Climatic data and environmental stochasticity 407 To assess whether (H4) the demographic buffering effects of sociality vary across 408 environments with differing climatic regimes, we obtained long-term climate data for 409 each population using the NASA POWER (Prediction of Worldwide Energy Resources) 410 database ( NASA Prediction Worldwide Energy Resources (POWER) Climate Data: 411 Data Access Viewer , 2020). This resource offers high resolution climatic products at 412 approx. 4 km 2 resolution from January 1958. For each of the 87 populations in our 413 dataset, for which we have GPS coordinates in COMADRE, we extracted monthly mean 414 temperature (T2M) and total precipitation (PRECTOTCORR) over the duration of the 415 corresponding demographic time series and the 30 years preceding the start of the 416 study, using the nasapower R package (Sparks, 2018). We then computed interannual 417 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 21 summary statistics for each location and variable, including mean, maximum, minimum, 418 and variance of precipitation and temperature, which formed the basis for estimating 419 environmental predictability. Here, in regards to aquatic species, it is important to note 420 that, even though species such as fish and corals are buffered on the short-term from 421 fast fluctuations in precipitation, this abiotic factor remains a strong proxy for key 422 hydrological processes—such as stream flow, turbidity, nutrient input, and water 423 temperature, which in turn directly influence vital rates in freshwater and coastal 424 environments (Buisson et al., 2008; Comte & Grenouillet, 2013; Fabricius, 2005). 425 We calculated three standardised metrics of climatic predictability following 426 Colwell’s framework (1974), adapted for time series climate data: constancy, 427 contingency, and predictability. Constancy quantifies the extent to which a climatic 428 variable remains stable over time (1 - variance/mean), while contingency reflects the 429 extent to which fluctuations are structured and recurrent. Predictability is then defined 430 as the sum of constancy and contingency. These metrics were calculated separately for 431 our records of monthly temperature and precipitation. We merged these values with 432 each population’s demographic data, allowing us to explore how environmental 433 predictability modulates the relationship between sociality and demographic buffering. 434 To construct a parsimonious and interpretable climatic PCA, we first reduced 435 collinearity among the aforementioned environmental variables. Starting with a 436 comprehensive set of temperature and precipitation-derived metrics (e.g. mean, 437 variance, predictability), we computed all pairwise Spearman correlation coefficients to 438 identify highly correlated variables. We visualised these correlations using custom 439 pairwise scatterplots (Figure S4) and used a threshold of | ρ | > 0.85 to flag and remove 440 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 22 strongly collinear variables using the function ‘findCorrelation’ in the caret R package 441 (Irizarry, 2019). This threshold balances the need to retain meaningful ecological 442 variation while minimizing redundancy and inflation of variance along principal 443 component axes (Dormann et al., 2013). This step dropped off the following variables 444 from our next steps: contingency and predictability of temperature, as well as constancy 445 and contingency of precipitation. Next, for the climatic variables that we retained (mean, 446 variance, and constancy of temperature, as well as mean, variance, and predictability of 447 precipitation) we conducted a PCA using the ‘prcomp’ function in base R, centering and 448 scaling all variables. We retained only the first two principal component (PC) axes, as 449 only they raised associated eigenvalues > 1 (Figure S5) in agreement with Kaiser’s 450 criterion (Legendre & Legendre, 2012). 451 This multivariate approach revealed how our 87 examined populations across 66 452 animal species are organised along two axes of climatic variability. The climatic 453 principal component analysis reveals two axes (Figure 2) whose associated eigenvalue 454 > 1 (Figure S5), and thus we retained these axes to test (H4) that the relationship 455 between degree of sociality and demographic buffering would break apart in more 456 stochastic environments. The first principal component (PC1), which explains 51% of 457 the variance, positions populations along a continuum of climates with more variable 458 precipitation (right) vs. more predictable precipitation (left). The second principal 459 component (PC2), explaining 18% of variance, places populations along a continuum of 460 higher mean annual precipitation (bottom) vs. climates with higher constancy in 461 temperature (top). 462 463 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 23 Comparative phylogenetic models 464 To test our four main hypotheses, we implemented a suite of phylogenetically-informed 465 comparative models that account for the shared evolutionary history of the species in 466 our dataset. As a first step, to discount the effect of adult body mass on the stochastic 467 elasticities, a standard approach in comparative demography (J. M. Gaillard et al., 1989; 468 Healy et al., 2019), we constructed a comparative data object using the 469 ‘comparative.data’ function from the caper R package (Orme et al., 2013) to obtain the 470 residuals of the phylogenetic generalised least squares (PGLS) models between each 471 of the stochastic elasticities (Total: Tμ , Tσ ; and vital rate specific: |Evr μ |, |Evr σ |) and body 472 mass. We then used these residuals for the next analytical steps. 473 To evaluate H1, which posits that more social species exhibit stronger 474 demographic buffering, we examined variation in the sum of stochastic elasticities to 475 variance ( Tσ ) and in the coefficients of variation of individual vital rates. We ran 476 phylogenetic ANOVAs with sociality (ordinal, 1–5) as the predictor and each buffering 477 metric as a response. These phylogenetic ANOVAs were implemented by fitting linear 478 models followed by post hoc Tukey HSD tests, and their outputs were interpreted within 479 a phylogenetic context based on trait alignment to the tree. To ensure results were 480 phylogenetically robust, we repeated these tests in a PGLS framework, comparing 481

across both model types. 482 To test H2, which predicts that demographic buffering is primarily driven by adult 483 vital rates, we used both phylogenetic ANOVAs and PGLS models with stochastic 484 elasticities of the stochastic population growth rate to changes in the temporal variance 485 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 24 of specific vital rates, separately. Specifically, we modelled the interannual CV of 486 juvenile survival (| Eσ σ J|), maturation (| Eσ γ |), adult survival (| Eσ σ A|), and reproduction 487 (|Eσ φ |) separately, using the degree of sociality as a predictor. Following significant 488 ANOVA results, we carried out post-hoc Tukey test comparisons to determine which 489 sociality levels differed most in their buffering patterns. By combining these models with 490 the analysis of stochastic elasticities to variance ( Eσ ) by life cycle stage, we further 491 tested whether adult rates were the primary targets of buffering in more social taxa. 492 For H3, which posits that social species exhibit greater stochastic elasticity to the 493 mean of adult vital rates, we analysed both total elasticity to the mean (T μ ) and vital-494 rate-specific (| Evr μ |) values as response variables. We used PGLS models with the 495 degree of sociality as predictor, and again followed up with phylogenetic ANOVAs and 496 Tukey tests where model residuals supported group-wise comparisons. This allowed us 497 to assess whether more social species rely more heavily on maximising the mean of 498 critical demographic parameters—particularly adult survival and reproduction—rather 499 than buffering their variance compared to less social species. 500 To test H4, that social buffering is weaker in highly stochastic environments, we 501 extended the above model. Specifically, we ran two separate models evaluating the 502 interaction between the degree of sociality and each of the two principal component 503 axes of our climatic PCA above—our proxy for environmental stochasticity. These were 504 implemented in PGLS models where response variables included either Tσ or |Evr σ |. We 505 centred and scaled both predictors and examined interaction terms to assess whether 506 the buffering benefit of sociality diminished as environments became more variable. We 507 also conducted model comparisons with and without interaction terms using AIC to 508 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 25 evaluate support for climate-modulated effects. A significant interaction term between 509 the degree of sociality and a climatic driver would indicate that any potential relationship 510 between demographic buffering and sociality is indeed modulated by the climate. A 511 positive effect of this interaction would imply that climatic extremes render the 512 demographic buffering of more social species less effective, as demographic buffering 513 is highest when its symbol is smallest. 514 Across all analyses, we tested assumptions of model normality, checked for 515 heteroscedasticity, and confirmed the presence of phylogenetic signal in the residuals. 516 When strong effects were detected, we visualised results using standardised model 517 coefficients and plotted predicted means from Tukey post-hoc tests to highlight 518 contrasts across the sociality continuum. This multi-model approach ensured that our 519

were robust to model structure and phylogenetic correction, and allowed us 520 to rigorously test how sociality interacts with demography and environment across 521 animal taxa. 522 523

524 Our results support the hypothesis (H1) that more social species exhibit enhanced 525 demographic buffering compared to more solitary species. Indeed, more social species 526 show a lower stochastic elasticity to vital rate variance ( Tσ ) (Figure 3A). We found 527 significant differences in Tσ across the sociality continuum, with social species exhibiting 528 the lowest Tσ values, followed by communal, and then colonial, gregarious, and solitary 529 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 26 (P = 0.0049). Solitary species consistently showed the highest sensitivity to changes in 530 the variance of all vital rates. 531 We found partial support for the prediction (H2) that adult stages drive 532 demographic buffering patterns in more social species. Across the full dataset, 533 stochastic elasticities to the variance of reproduction (| Eσ φ |) were consistently lower in 534 more social species (Figure 4D, P = 0.0012). However, we found no significant 535 difference across our social continuum regarding the stochastic elasticity to the variance 536 of adult survival (| Eσ σ A|; Figure 4C, P = 0.6666). Unexpectedly, we found a lower value 537 of the stochastic elasticity to variance in juvenile survival (| Eσ σ J|) in more social species 538 compared to less social ones (Figure 4A, P = 0.0128), implying that more social species 539 buffer more against stochastic environments in early stages of their development. The 540 stochastic elasticity to variance in maturation (| Eσ γ |) showed no significant differences 541 across sociality groups. 542 We found support for our hypothesis (H3) that more social species invest more in 543 maximising the mean of adult vital rates than more solitary species. Indeed, total 544 stochastic elasticity to the mean values (|Tμ |) differ significantly across sociality levels (P 545 = 0.0062; Figure 3B), with social species having the lowest values. The vital-rate-546 specific analyses also reveal that social species exhibit significantly higher stochastic 547 elasticity values to changes in the mean of adult survival (|Eμ σ A|; Figure 5C, P = 0.0017) 548 and reproduction (| Eμ φ |; Figure 5D, P = 0.0125). However, the stochastic elasticity to 549 mean juvenile survival is also lowest in more social species (| Eμ σ J|; Figure 5A, P = 550 0.0046). There is no effect for mean maturation on the stochastic population growth rate 551 (λ s) (| Eμ γ |; Figure 5B, P = 0.5481). The differences in vital-rate specific stochastic 552 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 27 elasticities are clearest between solitary and social species in post-hoc comparisons. 553 These patterns imply that more social species achieve population stability not only by 554 buffering variance, as discussed above, but also by enhancing the average 555 performance of adult and juvenile vital rates. 556 We found evidence that (H4) the benefits of social buffering are climatically-557 dependent. However, we did not find evidence to support that these potential benefits 558 break down under highly unpredictable climatic regimes, but in fact increase non-559 linearly. We found a negative, statistically significant coefficient for the interaction 560 between sociality and precipitation predictability (PC1 in Figure 2) in the PGLS model 561 predicting Tσ (P = 0.027; Table 1A). This means that, as environments become less 562 climatically predictable (left to right in PC1; e.g., more erratic rainfall), highly social 563 species diminish their stochastic elasticity to variance even further compared to highly 564 social species in more predictable environments, indicating enhanced demographic 565 buffering in highly stochastic environments. The interaction between sociality and PC2 566 (Temperature constancy) did not significantly predict Tσ (Table 1). The fact that the 567 positioning of solitary, gregarious, communal, colonial, and social species along this 568 climatic PCA is not significantly associated with either axis (Table S3) indicates that this 569 key finding is not driven by environmental filtering on each group in different regions of 570 the climatic space. 571 572

573 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 28 Understanding how sociality influences demographic performance under environmental 574 stochasticity is essential for predicting population responses to ongoing global change. 575 Although social behaviour is widely recognised as a key axis of animal life history 576 (Clutton-Brock, 2009; Lukas & Clutton-Brock, 2012; Silk, 2007a), its role in modulating 577 the demographic consequences of environmental variability remains poorly quantified 578 across taxa. This gap is especially important given the increasing recognition that social 579 traits can act as buffering mechanisms against ecological uncertainty (Albery et al., 580 2022; Doody et al., 2013; Rubenstein & Lovette, 2007). Our study provides the first 581 macroecological and phylogenetically informed test of the social buffering hypothesis 582 across the animal kingdom, using time-series demographic data from 87 populations of 583 66 animal species spanning 12 taxonomic classes. We find broad support for the idea 584 that more social species exhibit stronger demographic buffering (H1), especially in adult 585 stages (H2), and partially support the life history theory expectation that social species 586 maximise mean demographic performance in adult vital rates (H3). Crucially, we show 587 that these buffering benefits are conditional and become more important in highly 588 stochastic environments (H4), thus adding a critical environmental contingency to the 589 social buffering hypothesis. 590 Our results support the social buffering hypothesis (Rubenstein & Lovette, 2007), 591 showing that population growth in more social species is less sensitive to interannual 592 fluctuations in vital rates, as measured by lower total stochastic elasticity to variance 593 (Tσ ; (Haridas & Tuljapurkar, 2005). This finding suggests that features of systems that 594 are organised socially—such as alloparental care, food and information sharing, 595 cooperative hunting, and defence—confer demographic stability by reducing the 596 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 29 influence of environmental noise on fitness-related traits. These results echo and extend 597 previous work in specific taxonomic groups ( e.g., birds: (Jetz et al., 2012); mammals: 598 (Silk, 2007b); reptiles: (Doody et al., 2013)) and are consistent with theoretical 599 predictions that selection should favour strategies that reduce the fitness costs of 600 environmental variance (Hilde et al., 2020; Morris et al., 2008; Pfister, 1998). By 601 demonstrating this pattern across 12 taxonomic classes, our findings indicate that the 602 demographic benefits of sociality are widespread and likely not restricted to well-studied 603 animal groups. 604 The ability of social species to buffer against environmental stochasticity is not 605 uniformly distributed across their vital rates or life cycle stages. Rather, we provide 606 evidence that their vital rates can exert different levels of demographic buffering. The 607 most prominent buffering mechanism we show here in the context of sociality is related 608 to the forces of natural selection that shape adult survival and reproduction through 609 environmental canalisation. Our finding is consistent with theoretical predictions 610 (Tuljapurkar, 1982) and empirical evidence (J.-M. Gaillard & Yoccoz, 2003; Morris et al., 611 2008; Santos et al., 2023) that traits contributing most to population growth rate are 612 often the most canalised ( i.e., buffered). In this sense, it is not surprising that, in our 613 study, juvenile survival and maturation show little evidence of buffering ( i.e., greater 614 values of |Evr σ |) in more social species, as juvenile vital rates are typically under weaker 615 selection—and thus more variable —in long-lived species (McDonald et al., 2016; 616 Stearns, 1976, 1999). Social systems likely reinforce this canalisation by promoting 617 adult stability through mechanisms such as cooperative breeding, protection from 618 predation, and shared resource acquisition (Bourke, 2011; Koenig & Dickinson, 2016). 619 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 30 In addition, we also provide evidence of a second buffering mechanism that acts on 620 juvenile survival—where canalisation is less expected. Although juvenile survival is 621 generally under weak natural selection, we show that increasing sociality significantly 622 reduces the extent to which variation in this vital rate affects population growth. As 623 sociality increases, juvenile survival becomes more buffered —a pattern not detected in 624 vital rates that are already highly buffered, such as adult survival. This secondary 625 buffering mechanism reinforces the idea that demographic buffering is life cycle stage 626 dependent, but also challenges the assumption that demographic buffering occurs 627 exclusively at vital rates where stability most enhances population growth (Hilde et al., 628 2020; Morris et al., 2008; Pfister, 1998). 629 In partial support of H3, we found that social species do not exhibit higher total 630 stochastic elasticity to mean vital rates (Tμ ; (Haridas & Tuljapurkar, 2007) overall, but do 631 show higher elasticity to the mean of adult survival and reproduction specifically 632 compared to less social species. This finding suggests that, while sociality may not 633 increase the overall importance of mean performance across all vital rates, it does 634 amplify the role of key adult traits in determining population growth. This pattern is 635 consistent with life-history theory predicting that species should maximise the 636 performance of traits most important to fitness (Stearns, 1999). The elevated | Eμ σ A| and 637 |Eμ /i5 | values in more social species likely reflect adaptive strategies that enhance adult 638 demographic performance in stable or cooperative contexts. However, this pattern did 639 not hold across juvenile traits, again pointing to a stage-specific effect of sociality on 640 demographic investment. Indeed, social buffering might be expected to 641 disproportionately benefit adults over juveniles if social relationships establish over the 642 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 31 life course (Firth et al. 2024) . Specifically, if adults typically hold stronger and more 643 consistent social relationships compared to juveniles (Woodman et al. 2024) , this may 644 enhance their ability to leverage group support under environmental stress. For 645 instance, in cooperatively breeding mammalian species (such as meerkats, or 646 baboons), dominant adult breeders often hold established social relationships to other 647 group members and benefit from priority access to shared resources, reduced predation 648 risk through increased vigilance, and assistance in offspring care; these mechanisms 649 would all directly stabilise adult survival and fecundity more than they would affect 650 juvenile traits (Clutton-Brock et al. 2016; Silk et al. 2016). Similarly, adult individuals in 651 stable avian breeding groups often benefit from load-lightening (via helper contributions 652 to territory defense, and joint nest provisioning), thus potentially reducing adult mortality 653 and reproductive variability relative to juvenile birds, who are often less integrated into 654 established social roles in cooperatively breeding systems (Rubenstein & Abbot 2017). 655 These differences in social roles across the life-course suggest that demographic 656 buffering by sociality might be particularly pronounced for adult vital rates. 657 Importantly, our results reveal that the demographic buffering effects of sociality 658 are context-dependent and may be more important in highly stochastic environments 659 (H4). This finding adds a novel macroecological insight to the social buffering 660 hypothesis. Contrary to some work suggesting that demographic buffering mechanisms 661 can break down under extreme environmental conditions (Rodriguez-Caro et al., 2021; 662 Santos et al., 2024), our results indicate that sociality may become even more effective 663 in buffering demographic processes against environmental stochasticity when 664 conditions are highly unpredictable. This pattern echoes long-standing predictions about 665 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 32 the evolution of cooperative systems in harsh environments, such as the development 666 of eusociality in termites, (semi)social cockroaches, and naked mole rats, where group 667 living is thought to have evolved in response to arid, variable conditions (Alexander, 668 1974; Faulkes & Bennett, 2013; Field & Toyoizumi, 2020; Lubin & Bilde, 2007). Rather 669 than social organisation being strained or cooperative behaviours faltering under 670 extreme variability (Hayes, 2017), our findings suggest that sociality can enhance 671 demographic resilience precisely where environmental unpredictability is greatest. Thus, 672 we highlight the importance of considering environmental context in assessments of 673 social evolution and suggest that sociality may play a crucial role in mediating species’ 674 demographic responses to ongoing climate change. 675 Together, our findings suggest that sociality promotes demographic stability by 676 buffering the variance of adult vital rates, and that these benefits are environmentally 677 dependent. In doing so, we bridge theoretical work on demographic buffering (Hilde et 678 al., 2020; Pfister, 1998; Tuljapurkar, 1982) with recent calls to integrate social behaviour 679 into eco-evolutionary frameworks (Albery et al., 2022; Firth et al., 2024; Salguero-680 Gómez, 2024). Our use of time-series demographic data across a broad range of taxa, 681 combined with a structured classification of sociality and high-resolution climate 682 products, offers a general framework to assess how social and environmental factors 683 interact to shape population dynamics. In an era of increasing environmental 684 unpredictability (Bathiany et al., 2018), understanding the limits of social buffering is 685 vital for predicting which species are most vulnerable to demographic destabilisation—686 and why. 687 688 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 33 689 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 34

690 Albery, G. F., Bansal, S., & Silk, M. J. (2024). Comparative approaches in social 691 network ecology. Ecology Letters, 27(1), e14345. https://doi.org/10.1111/ele.14345 692 Albery, G. F., Clutton-Brock, T. H., Morris, A., Morris, S., Pemberton, J. M., Nussey, D. 693 H., & Firth, J. A. (2022). Ageing red deer alter their spatial behaviour and become 694 less social. Nature Ecology & Evolution , 6(8), 1231–1238. 695 https://www.nature.com/articles/s41559-022-01817-9 696 Alexander, R. D. (1974). The evolution of social behavior. Annual Review of Ecology 697 and Systematics , 5(1), 325–383. 698 https://doi.org/10.1146/annurev.es.05.110174.001545 699 Avilés, L., Harwood, G., & Koenig, W. (2012). A quantitative index of sociality and its 700 application to group-living spiders and other social organisms. Ethology: Formerly 701 Zeitschrift Für Tierpsychologie , 118(12), 1219–1229. 702 https://doi.org/10.1111/eth.12028 703 Bathiany, S., Dakos, V., Scheffer, M., & Lenton, T. M. (2018). Climate models predict 704 increasing temperature variability in poor countries. Science Advances , 4(5), 705 eaar5809. https://doi.org/10.1126/sciadv.aar5809 706 Bernard, C., Santos, G. S., Deere, J. A., Rodriguez-Caro, R., Capdevila, P., Kusch, E., 707 Gascoigne, S. J. L., Jackson, J., & Salguero-Gómez, R. (2023). MOSAIC - A 708 unified trait database to complement structured population models. Scientific Data, 709 10(1), 335. https://doi.org/10.1038/s41597-023-02070-w 710 Boettiger, C., Lang, D. T., & Wainwright, P. C. (2012). rfishbase: exploring, manipulating 711 and visualizing FishBase data from R. Journal of Fish Biology , 81(6), 2030–2039. 712 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 35 https://doi.org/10.1111/j.1095-8649.2012.03464.x 713 Bourke, A. F. G. (2011). Social group maintenance. In Principles of Social Evolution (pp. 714 129–161). Oxford University Press. 715 https://doi.org/10.1093/acprof:oso/9780199231157.003.0005 716 Bruijning, M., Metcalf, C. J. E., Jongejans, E., & Ayroles, J. F. (2020). The evolution of 717 variance control. Trends in Ecology & Evolution , 35(1), 22–33. 718 https://doi.org/10.1016/j.tree.2019.08.005 719 Buisson, L., Thuiller, W., Lek, S., Lim, P., & Grenouillet, G. (2008). Climate change 720 hastens the turnover of stream fish assemblages. Global Change Biology, 14(10), 721 2232–2248. https://doi.org/10.1111/j.1365-2486.2008.01657.x 722 Clutton-Brock, T. (2009). Cooperation between non-kin in animal societies. Nature, 723 462(7269), 51–57. https://doi.org/10.1038/nature08366 724 Colwell, R. K. (1974). Predictability, constancy, and contingency of periodic 725 phenomena. Ecology, 55(5), 1148–1153. https://doi.org/10.2307/1940366 726 Comte, L., & Grenouillet, G. (2013). Do stream fish track climate change? Assessing 727 distribution shifts in recent decades. Ecography, 36(11), 1236–1246. 728 https://doi.org/10.1111/j.1600-0587.2013.00282.x 729 Conquet, E., Ozgul, A., Blumstein, D. T., Armitage, K. B., Oli, M. K., Martin, J. G. A., 730 Clutton-Brock, T. H., & Paniw, M. (2023). Demographic consequences of changes 731 in environmental periodicity. Ecology, 104(3), e3894. 732 https://doi.org/10.1002/ecy.3894 733 Creel, S., & Christianson, D. (2008). Relationships between direct predation and risk 734 effects. Trends in Ecology & Evolution , 23(4), 194–201. 735 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 36 https://doi.org/10.1016/j.tree.2007.12.004 736 Crump, M. L. (2015). Anuran reproductive modes: Evolving perspectives. Journal of 737 Herpetology, 49(1), 1–16. https://doi.org/10.1670/14-097 738 de Magalhães, J. P., & Costa, J. (2009). A database of vertebrate longevity records and 739 their relation to other life-history traits. Journal of Evolutionary Biology, 22(8), 1770–740 1774. https://doi.org/10.1111/j.1420-9101.2009.01783.x 741 Dewey, T., Shefferly, N., & Havens, A. (2010). Animal Diversity Web. University of . 742 https://animaldiversity.ummz.umich.edu/site/accounts/information/Felis_silvestris.ht743 ml 744 Doody, J. S., Burghardt, G. M., & Dinets, V. (2013). Breaking the social–non /i2 social 745 dichotomy: A role for reptiles in vertebrate social behavior research? Ethology: 746 Formerly Zeitschrift Für Tierpsychologie , 119(2), 95–103. 747 https://doi.org/10.1111/eth.12047 748 Dormann, C. F., Elith, J., Bacher, S., Buchmann, C., Carl, G., Carré, G., Marquéz, J. R. 749 G., Gruber, B., Lafourcade, B., Leitão, P. J., Münkemüller, T., McClean, C., 750 Osborne, P. E., Reineking, B., Schröder, B., Skidmore, A. K., Zurell, D., & 751 Lautenbach, S. (2013). Collinearity: a review of methods to deal with it and a 752 simulation study evaluating their performance. Ecography , 36(1), 27–46. 753 https://doi.org/10.1111/j.1600-0587.2012.07348.x 754 Emlen, S. T. (1982). The evolution of helping. I. an ecological constraints model. The 755 American Naturalist, 119(1), 29–39. https://doi.org/10.1086/283888 756 Ezard, T. H. G., & Coulson, T. (2010). How sensitive are elasticities of long-run 757 stochastic growth to how environmental variability is modelled? Ecological 758 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 37 Modelling, 221(2), 191–200. https://doi.org/10.1016/j.ecolmodel.2009.09.017 759 Fabricius, K. E. (2005). Effects of terrestrial runoff on the ecology of corals and coral 760 reefs: review and synthesis. Marine Pollution Bulletin, 50(2), 125–146. 761 https://doi.org/10.1016/j.marpolbul.2004.11.028 762 Faulkes, C. G., & Bennett, N. C. (2013). Plasticity and constraints on social evolution in 763 African mole-rats: ultimate and proximate factors. Philosophical Transactions of the 764 Royal Society of London. Series B, Biological Sciences , 368(1618), 20120347. 765 https://doi.org/10.1098/rstb.2012.0347 766 Field, J., & Toyoizumi, H. (2020). The evolution of eusociality: no risk-return tradeoff but 767 the ecology matters. Ecology Letters , 23(3), 518–526. 768 https://doi.org/10.1111/ele.13452 769 Firth, J. A., Albery, G. F., Bouwhuis, S., Brent, L. J. N., & Salguero-Gómez, R. (2024). 770 Understanding age and society using natural populations. Philosophical 771 Transactions of the Royal Society of London. Series B, Biological Sciences , 772 379(1916), 20220469. https://doi.org/10.1098/rstb.2022.0469 773 Froese, R., & Pauly, D. (2002). FishBase: A Global Information System on Fishes. 774 World Wide Web Electronic Publication: Www.fishbase.org. , 000–000. 775 https://oceanrep.geomar.de/id/eprint/2964 776 Gaillard, J. M., Pontier, D., Allaine, D., Lebreton, J. D., Trouvilliez, J., & Clobert, J. 777 (1989). An analysis of demographic tactics in birds and mammals. Oikos , 56(1), 778 59–76. https://doi.org/10.2307/3566088 779 Gaillard, J.-M., & Yoccoz, N. G. (2003). Temporal variation in survival of mammals: A 780 case of environmental canalization? Ecology, 84(12), 3294–3306. 781 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 38 https://doi.org/10.1890/02-0409 782 Gaillard, J.-M., Yoccoz, N. G. , Lebreton, J.-D., Bonenfant, C., Devillard, S., Loison, A., 783 Pontier, D., & Allaine, D. (2005). Generation time: a reliable metric to measure life-784 history variation among mammalian populations. The American Naturalist , 166(1), 785 119–123; discussion 124–128. https://doi.org/10.1086/430330 786 Gascoigne, S. J. L., Kajin, M., & Salguero-Gómez, R. (2024). Criteria for buffering in 787 ecological modeling. Trends in Ecology & Evolution , 39(2), 116–118. 788 https://doi.org/10.1016/j.tree.2023.11.006 789 Gascoigne, S. J. L., Kajin, M., Tuljapurkar, S., Santos, G. S., Compagnoni, A., Steiner, 790 U. K., Vinton, A. C., Jaggi, H., Sepil, I., & Salguero-Gómez, R. (2023). Structured 791 demographic buffering: A framework to explore the environment drivers and 792 demographic mechanisms underlying demographic buffering. In bioRxiv (p. 793 2023.07.20.549848). https://doi.org/10.1101/2023.07.20.549848 794 Gascoigne, S., Kajin, M., Sepil, I., & Salguero-Gomez, R. (2024). Testing for efficacy in 795 four measures of demographic buffering. In EcoEvoRxiv. 796 https://doi.org/10.32942/x23911 797 Gearty, W., & Chamberlain, S. (2025). rredlist: “IUCN” Red List Client. 798 Giaimo, S., & Traulsen, A. (2023). Generation time in stage-structured populations 799 under fluctuating environments. The American Naturalist , 201(3), 404–417. 800 https://doi.org/10.1086/722608 801 Greenwood, P. J. (1980). Mating systems, philopatry and dispersal in birds and 802 mammals. Animal Behaviour , 28(4), 1140–1162. https://doi.org/10.1016/s0003-803 3472(80)80103-5 804 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 39 Haridas, C. V., & Tuljapurkar, S. (2005). Elasticities in variable environments: properties 805 and implications. The American Naturalist , 166(4), 481–495. 806 https://doi.org/10.1086/444444 807 Haridas, C. V., & Tuljapurkar, S. (2007). Time, transients and elasticity. Ecology Letters, 808 10, 1143–1153. 809 Hayes, L. D. (2017). Cooperative Breeding in Vertebrates: Studies of Ecology, 810 Evolution, and Behavior. Edited by Walter D. Koenig and Janis L. Dickinson; 811 illustrated by Stef den Ri dder. Cambridge and New York: Cambridge University 812 Press. $140.00. x + 379 p. + 12 pl.; ill.; index. ISBN: 978-1-107-04343-5. 2016. The 813 Quarterly Review of Biology, 92(1), 99–100. https://doi.org/10.1086/690874 814 Healy, K., Ezard, T. H. G., Jones, O. R., Salguero-Gomez, R., & Buckley, Y. M. (2019). 815 Animal life history is shaped by the pace of life and the distribution of age-specific 816 mortality and reproduction. Nature Ecology & Evolution , 3(8), 1217–1224. 817 https://doi.org/10.1038/s41559-019-0938-7 818 Heisterkamp, S., Willigen, E., Diderichsen, P.-M., & Maringwa, J. (2017). Update of the 819 nlme package to allow a fixed standard deviation of the residual error. The R 820 Journal, 9(1), 239. https://doi.org/10.32614/rj-2017-010 821 Hilde, C. H., Gamelon, M., Saether, B. E., Gaillard, J. M., Yoccoz, N. G., & Pelabon, C. 822 (2020). The Demographic Buffering Hypothesis: Evidence and Challenges. Trends 823 in Ecology & Evolution, 35(6), 523–538. https://doi.org/10.1016/j.tree.2020.02.004 824 Irizarry, R. A. (2019). The caret package. In Introduction to Data Science (pp. 523–528). 825 Chapman and Hall/CRC. https://doi.org/10.1201/9780429341830-30 826 Isaac, N. J. B., Turvey, S. T., Collen, B., Waterman, C., & Baillie, J. E. M. (2007). 827 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 40 Mammals on the EDGE: conservation priorities based on threat and phylogeny. 828 PloS One, 2(3), e296. https://doi.org/10.1371/journal.pone.0000296 829 Jetz, W., Thomas, G. H., Joy, J. B., Hartmann, K., & Mooers, A. O. (2012). The global 830 diversity of birds in space and time. Nature, 491(7424), 444–448. 831 https://doi.org/10.1038/nature11631 832 Jones, O. R., Barks, P., Stott, I., James, T. D., Levin, S., Petry, W. K., Capdevila, P., 833 Che-Castaldo, J., Jackson, J., Romer, G., Schuette, C., Thomas, C. C., & 834 Salguero-Gomez, R. (2022). Rcompadre and Rage-Two R packages to facilitate 835 the use of the COMPADRE and COMADRE databases and calculation of life-836 history traits from matrix population models. Methods in Ecology and Evolution / 837 British Ecological Society. https://doi.org/10.1111/2041-210x.13792 838 Keesey, M. T. (2020). PhyloPic. PhyloPic. http://phylopic.org 839 Klug, H., & Bonsall, M. B. (2014). What are the benefits of parental care? The 840 importance of parental effects on developmental rate. Ecology and Evolution, 4(12), 841 2330–2351. https://doi.org/10.1002/ece3.1083 842 Koenig, W. D., & Dickinson, J. L. (2016). Cooperative Breeding in Vertebrates: Studies 843 of Ecology, Evolution, and Behavior. Cambridge University Press. 844 Krause, J., & Ruxton, G. D. (2002). Evolutionary considerations. In Living in Groups (pp. 845 104–122). Oxford University PressOxford. 846 https://doi.org/10.1093/oso/9780198508175.003.0007 847 Le Coeur, C., Yoccoz, N. G., Salguero-Gómez, R., & Vindenes, Y. (2022). Life history 848 adaptations to fluctuating environments: Combined effects of demographic 849 buffering and lability. Ecology Letters , 25(10), 2107–2119. 850 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 41 https://doi.org/10.1111/ele.14071 851 Legendre, P., & Legendre, L. (2012). Numerical Ecology (3rd ed.). Elsevier. 852 Lewontin, R. C., & Cohen, D. (1969). On population growth in randomly varying 853 environments. Proceedings of the National Academy of Sciences of the United 854 States of America, 62(4), 1056 – &. https://doi.org/10.1073/pnas.62.4.1056 855 Lubin, Y., & Bilde, T. (2007). The evolution of sociality in spiders. In Advances in the 856 Study of Behavior (pp. 83–145). Elsevier. https://doi.org/10.1016/s0065-857 3454(07)37003-4 858 Lukas, D., & Clutton-Brock, T. (2012). Cooperative breeding and monogamy in 859 mammalian societies. Proceedings. Biological Sciences , 279(1736), 2151–2156. 860 https://doi.org/10.1098/rspb.2011.2468 861 McDonald, J. L., Bailey, T., Delahay, R. J., McDonald, R. A., Smith, G. C., & Hodgson, 862 D. J. (2016). Demographic buffering and compensatory recruitment promotes the 863 persistence of disease in a wildlife population. Ecology Letters , 19(4), 443–449. 864 https://doi.org/10.1111/ele.12578 865 Merrien, T., Davis, K., Di Marco, M., Capdevila, P., & Salguero-Gomez, R. (2021). 866 Human pressures filter out the less resilient demographic strategies in natural 867 populations of plants and animals worldwide. bioRxiv, 2021.09.29.462372. 868 Michonneau, F., Brown, J. W., & Winter, D. J. (2016). rotl: an R package to interact with 869 the Open Tree of Life data. Methods in Ecology and Evolution / British Ecological 870 Society, 7(12), 1476–1481. https://doi.org/10.1111/2041-210x.12593 871 Mondal, R., Aburto, J. M., Sear, R., Tuljapurkar, S., Mishra, U. S., & Salguero-Gómez, 872 R. (2024). Human populations with low survival at advanced ages and postponed 873 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 42 fertility reduce long-term growth in high inflation environments. In bioRxiv. 874 https://doi.org/10.1101/2024.07.31.606043 875 Morris, W. F., Pfister, C. A., Tuljapurkar, S., Haridas, C. V., Boggs, C. L., Boyce, M. S., 876 Bruna, E. M., Church, D. R., Coulson, T., Doak, D. F., Forsyth, S., Gaillard, J. M., 877 Horvitz, C. C., Kalisz, S., Kendall, B. E., Knight, T. M., Lee, C. T., & Menges, E. S. 878 (2008). Longevity can buffer plant and animal populations against changing climatic 879 variability. Ecology, 89(1), 19–25. 880 Myhrvold, N. P., Baldridge, E., Chan, B., Sivam, D., Freeman, D. L., & Ernest, S. K. M. 881 (2015). An amniote life-history database to perform comparative analyses with 882 birds, mammals, and reptiles. Ecology, 96(11), 3109–3000. 883 https://doi.org/10.1890/15-0846r.1 884 NASA Prediction Worldwide Energy Resources (POWER) Climate Data: Data Access 885 Viewer. (2020). 886 Nicol-Harper, A., Dooley, C., Packman, D., Mueller, M., Bijak, J., Hodgson, D., Townley, 887 S., & Ezard, T. (2018). Inferring transient dynamics of human populations from 888 matrix non /i2 normality. Population Ecology , 60(1-2), 185–196. 889 https://doi.org/10.1007/s10144-018-0620-y 890 Oli, M. K., & Dobson, F. S. (2003). The relative importance of life-history variables to 891 population growth rate in mammals: Cole’s prediction revisited. The American 892 Naturalist, 161(3), 422–440. https://doi.org/10.1086/367591 893 Orme, D., Freckleton, R., Thomson, G., Petzoldt, T., Fritz, S., Isaac, N., & Pearse, W. 894 (2013). The caper package: comparative analysis of phylogenetics and evolution in 895 R. http://cran.r-project.org/web/packages/caper/ 896 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 43 Oro, D. (2020). Perturbation, Behavioural Feedbacks, and Population Dynamics in 897 Social Animals: When to Leave and Where to Go . Oxford University Press. 898 https://play.google.com/store/books/details?id=f4_UDwAAQBAJ 899 Paradis, E., Claude, J., & Strimmer, K. (2004). APE: Analyses of Phylogenetics and 900 Evolution in R language. Bioinformatics , 20(2), 289–290. 901 https://doi.org/10.1093/bioinformatics/btg412 902 Pfister, C. A. (1998). Patterns of variance in stage-structured populations: evolutionary 903 predictions and ecological implications. Proceedings of the National Academy of 904 Sciences of the United States of America, 95(1), 213–218. 905 Revell, L. J. (2012). phytools: an R package for phylogenetic comparative biology (and 906 other things). Methods in Ecology and Evolution, 3(2), 217–223. 907 https://doi.org/10.1111/j.2041-210x.2011.00169.x 908 Rodriguez-Caro, R. C., Capdevila, P., Gracia, E., Barbosa, J. M., Gimenez, A., & 909 Salguero-Gomez, R. (2021). The limits of demographic buffering in coping with 910 environmental variation. Oikos , 130(8), 1346–1358. 911 https://doi.org/10.1111/oik.08343 912 Rubenstein, D. R., & Lovette, I. J. (2007). Temporal environmental variability drives the 913 evolution of cooperative breeding in birds. Current Biology: CB, 17(16), 1414–1419. 914 https://doi.org/10.1016/j.cub.2007.07.032 915 Saether, B. E., & Bakke, O. (2000). Avian life history variation and contribution of 916 demographic traits to the population growth rate. Ecology, 81(3), 642–653. 917 Salguero-Gómez, R. (2024). More social species live longer, have longer generation 918 times and longer reproductive windows. Philosophical Transactions of the Royal 919 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 44 Society of London. Series B, Biological Sciences , 379(1916), 20220459. 920 https://doi.org/10.1098/rstb.2022.0459 921 Salguero-Gomez, R., & Gamelon, M. (2021). Demographic Methods across the Tree of 922 Life. Oxford University Press. 923 https://play.google.com/store/books/details?id=wtg9EAAAQBAJ 924 Salguero-Gomez, R., Jackson, J., & Gascoigne, S. J. L. (2021). Four key challenges in 925 the open-data revolution. The Journal of Animal Ecology , 90(9), 2000–2004. 926 https://doi.org/10.1111/1365-2656.13567 927 Salguero-Gomez, R., Jones, O. R., Archer, C. R., Bein, C., de Buhr, H., Farack, C., 928 Gottschalk, F., Hartmann, A., Henning, A., Hoppe, G., Romer, G., Ruoff, T., 929 Sommer, V., Wille, J., Voigt, J., Zeh, S., Vieregg, D., Buckley, Y. M., Che-Castaldo, 930 J., … Vaupel, J. W. (2016). COMADRE: a global data base of animal demography. 931 The Journal of Animal Ecology , 85(2), 371–384. https://doi.org/10.1111/1365-932 2656.12482 933 Salguero-Gómez, R., & Plotkin, J. B. (2010). Matrix dimensions bias demographic 934 inferences: implications for comparative plant demography. The American 935 Naturalist, 176(6), 710–722. https://doi.org/10.1086/657044 936 Santos, G. S., Gascoigne, S. J. L., Dias, A. T. C., Kajin, M., & Salguero-Gómez, R. 937 (2023). A unified framework to identify demographic buffering in natural 938 populations. In bioRxiv (p. 2023.07.03.547528). 939 https://doi.org/10.1101/2023.07.03.547528 940 Santos, G. S., Yang, X., L Gascoigne, S. J., Compagnoni, A., Dias, A. T. C., 941 Tuljapurkar, S., Kajin, M., & Salguero-Gómez, R. (2024). Population responses to 942 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 45 environmental stochasticity are primarily driven by survival-reproduction trade-offs 943 and mediated by aridity. In bioRxiv. https://doi.org/10.1101/2024.07.24.604949 944 Shine, R. (1978). Propagule size and parental care: The “safe harbor” hypothesis. 945 Journal of Theoretical Biology , 75(4), 417–424. https://doi.org/10.1016/0022-946 5193(78)90353-3 947 Silk, J. B. (2007a). Social components of fitness in primate groups. Science, 317(5843), 948 1347–1351. https://doi.org/10.1126/science.1140734 949 Silk, J. B. (2007b). The adaptive value of sociality in mammalian groups. Philosophical 950 Transactions of the Royal Society of London. Series B, Biological Sciences , 951 362(1480), 539–559. https://doi.org/10.1098/rstb.2006.1994 952 Snyder-Mackler, N., Burger, J. R., Gaydosh, L., Belsky, D. W., Noppert, G. A., Campos, 953 F. A., Bartolomucci, A., Yang, Y. C., Aiello, A. E., O’Rand, A., Harris, K. M., Shively, 954 C. A., Alberts, S. C., & Tung, J. (2020). Social determinants of health and survival 955 in humans and other animals. Science (New York, N.Y.) , 368(6493), eaax9553. 956 https://doi.org/10.1126/science.aax9553 957 Sparks, A. (2018). Nasapower: A NASA POWER global meteorology, surface solar 958 energy and climatology data client for R. Journal of Open Source Software , 3(30), 959 1035. https://doi.org/10.21105/joss.01035 960 Stearns, S. C. (1976). Life-history tactics: a review of the ideas. The Quarterly Review 961 of Biology, 51(1), 3–47. https://doi.org/10.1086/409052 962 Stearns, S. C. (1999). The Evolution of Life Histories. Oxford University Press. 963 Stubben, C., & Milligan, B. (2007). Estimating and analyzing demographic models using 964 the popbio package in R. Journal of Statistical Software, 22(11), 1–23. 965 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 46 Takada, T., Kawai, Y., & Salguero-Gómez, R. (2018). A cautionary note on elasticity 966 analyses in a ternary plot using randomly generated population matrices. 967 Population Ecology, 60(1-2), 37–47. https://doi.org/10.1007/s10144-018-0619-4 968 Tobias, J. A., Sheard, C., Pigot, A. L., Devenish, A. J. M., Yang, J., Sayol, F., Neate-969 Clegg, M. H. C., Alioravainen, N., Weeks, T. L., Barber, R. A., Walkden, P. A., 970 MacGregor, H. E. A., Jones, S. E. I., Vi ncent, C., Phillips, A. G., Marples, N. M., 971 Montaño-Centellas, F. A., Leandro-Silva, V., Claramunt, S., … Schleuning, M. 972 (2022). AVONET: morphological, ecological and geographical data for all birds. 973 Ecology Letters, 25(3), 581–597. https://doi.org/10.1111/ele.13898 974 Tuljapurkar, S. (1982). Population-dynamics in variable environments. 2. Correlated 975 environments, sensitivity analsyis and dynamics. Theoretical Population Biology , 976 21(1), 114–140. https://doi.org/10.1016/0040-5809(82)90009-0 977 Tuljapurkar, S., Horvitz, C. C., & Pascarella, J. B. (2003). The many growth rates and 978 elasticities of populations in random environments. The American Naturalist , 979 162(4), 489–502. https://doi.org/10.1086/378648 980 Wong, M., & Balshine, S. (2011). The evolution of cooperative breeding in the African 981 cichlid fish, Neolamprologus pulcher. Biological Reviews of the Cambridge 982 Philosophical Society , 86(2), 511–530. https://doi.org/10.1111/j.1469-983 185X.2010.00158.x 984 Young, A. J., & Bennett, N. C. (2013). Intra-sexual selection in cooperative mammals 985 and birds: why are females not bigger and better armed? Philosophical 986 Transactions of the Royal Society of London. Series B, Biological Sciences , 987 368(1631), 20130075. https://doi.org/10.1098/rstb.2013.0075 988 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 1 Table 1. Populations located in climates with low predictability for precipitation regimes show a disproportionate 989 increase in their demographic buffering abilities. Estimates and associated P-values (P < 0.05 in bold) for a battery of 990 phylogenetic generalised least square (pgls) models evaluating the effects of the degree of sociality and climate on the 991 elasticity of the stochastic population growth rate (λ s) to (A) the total change in mean and variance of all vital rates in each 992 population, (B) changes in the mean of each vital rate, and ( C) change in the variance of each vital rate, separately. PC1 993 and PC2 are defined in Figure 2, and correspond to a continuum of predictability in precipitation and of constancy of 994 temperature, respectively. A significant interaction between sociality and either principal component axis indicates that the 995 relationship between demographic buffering and sociality changes along climatic regimes. 996 997 Stochastic elasticity to Predictor Estimate P Predictor Estimate P A. Total Mean - Tμ Sociality 0.221 0.541 Sociality 0.052 0.878 PC1 0.894 0.165 PC2 -0.268 0.770 Soc × PC1 -0.231 0.286 Soc × PC2 0.22 0.472 Variance - Tσ Soc 1.778 0.016 Soc 0.634 0.497 PC1 4.233 0.026 PC2 -1.362 0.585 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 2 Soc × PC1 -1.431 0.027 Soc × PC2 1.290 0.131 B. Changes in mean Juvenile survival - |Eσ J μ | Soc 0.423 0.446 Soc 0.052 0.911 PC1 0.768 0.424 PC2 -0.816 0.516 Soc × PC1 -0.411 0.216 Soc × PC2 0.631 0.141 Maturation - |Eγ μ | Soc -0.503 0.538 Soc -0.724 0.337 PC1 1.027 0.468 PC2 -1.278 0.524 Soc × PC1 -0.345 0.475 Soc × PC2 0.551 0.408 Adult survival - |Eσ A μ | Soc 0.183 0.751 Soc 0.028 0.960 PC1 1.150 0.257 PC2 0.032 0.983 Soc × PC1 -0.235 0.491 Soc × PC2 0.069 0.889 Reproductio n - |Eφ μ | Soc 0.382 0.656 Soc -0.216 0.778 PC1 2.262 0.141 PC2 0.216 0.916 Soc × PC1 -0.677 0.193 Soc × PC2 0.482 0.481 C. Juvenile Soc 0.518 0.631 Soc 0.112 0.911 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 3 Changes in variance survival - |Eσ J σ | PC1 1.791 0.342 PC2 -1.321 0.625 Soc × PC1 -0.607 0.346 Soc × PC2 0.595 0.507 Maturation - |Eγ σ | Soc -0.156 0.888 Soc -0.413 0.691 PC1 1.790 0.356 PC2 -1.711 0.540 Soc × PC1 -0.497 0.451 Soc × PC2 0.569 0.538 Adult survival - |Eσ A σ | Soc 0.283 0.384 Soc 0.150 0.613 PC1 0.525 0.352 PC2 0.151 0.849 Soc × PC1 -0.152 0.425 Soc × PC2 0.056 0.832 Reproductio n - |Eφ σ | Soc 1.223 0.334 Soc 0.430 0.708 PC1 3.401 0.129 PC2 -0.574 0.852 Soc × PC1 -0.976 0.197 Soc × PC2 0.879 0.393 998 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 1 Figure legends 999 Figure 1. Representation of the 87 natura l populations of the 66 animal species 1000 examined in this study to test the social buffering hypothesis. A. Geographic 1001 location of the populations. B. Phylogenetic relationships of the species. The phylogeny 1002 was modified to accommodate intra-specific spatial replication, indicated by the name of 1003 the species followed by the number of the population. In total, we examined patterns of 1004 demographic buffering across 12 animal taxonomic classes. Each species was 1005 classified into one of the five levels of sociality shown in the insert in panel A, following 1006 Salguero-Gómez (Salguero-Gómez, 2024). 1007 1008 Figure 2. The examined 87 animal populations are organised along two axes of 1009 climatic variation: (PC1) predictability in precipitation and (PC2) constancy of 1010 temperature. Principal component analyses of the first two principal components, 1011 displaying the loadings of climatic products derived from the NASA POWER database. 1012 Climatic variables are: (1) Mean annual total precipitation ( /g1842/g3364), (2) Variance annual total 1013 precipitation ( Δ T), (3) Predictability of annual precipitation ( Ppred), (4) Mean annual 1014 temperature ( T/i1), (5) Variance annual temperature ( Δ T); and (6) Constancy of 1015 temperature ( Tcons). The first principal component (PC1), which explains 50% of the 1016 variance, positions populations along a continuum of climates with more variable 1017 precipitation (right) vs. more predictable precipitation (left). The second principal 1018 component (PC2), explaining 18% of variance, places populations along a continuum of 1019 higher mean annual precipitation (bottom) vs. climates with higher constancy in 1020 temperature (top). Constancy quantifies the extent to which a climatic variable remains 1021 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 2 stable over time. Predictability is the sum of constancy and contingency, where 1022 contingency reflects the extent to which fluctuations are structured and recurrent. The 1023 colours of each dot represent the classification of sociality (See Figure 1). Silhouettes 1024 represent a subset of the animal species shown on the PCA, obtained from phylopic 1025 (Keesey, 2020). 1026 1027 Figure 3. Increases in sociality are associated with canalisation in the selective 1028 pressures of demographic performance and increases in demographic buffering. 1029 Sum of stochastic elasticities of population growth rate ( λ s) to changes in the ( A) 1030 variance (Tσ ) and (B ) mean (T μ ) of all vital rates in a given population, classified by 1031 degree of sociality. Response variable is the absolute value of the residuals of Tσ and Tμ 1032 against adult body mass of each animal species. Tσ is log10-transformed. P values (top-1033 right) correspond to a phylogenetic ANOVA, and post-hoc Tukey test letters positioned 1034 on top of each group indicate whether groups are significantly different, after 1035 phylogenetic corrections. 1036 1037 Figure 4. As species become more social, their population performance becomes 1038 less sensitive to changes in juven ile survival and reproduction. Stochastic 1039 elasticity of population growth rate (λ s) to changes in the variance of four vital rates (vr ) 1040 (|Evr σ |): (| Evr σ |): juvenile survival (| Eσ J σ |), maturation (| Eγ σ |), adult survival (| Eσ A σ |), and 1041 reproduction (| Eφ σ |), grouped by degree of sociality. Response variables are the 1042 residuals of E vr σ against adult body mass of each animal species, and log 10-1043 transformed. P values correspond to a phylogenetic ANOVA, and post-hoc Tukey test 1044 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 3 letters positioned on top of each group indicate whether groups are significantly 1045 different, after phylogenetic corrections. 1046 1047 Figure 5. In more social species, vital rates are more tightly regulated. Stochastic 1048 elasticity of population growth rate ( λ s) to changes in the mean of four vital rates ( vr) 1049 (|Evr μ |): juvenile survival (| Eσ J μ |), maturation (| Eγ μ |), adult survival (| Eσ A μ |), and 1050 reproduction (| Eφ μ |), grouped by degree of sociality. Response variables are the 1051 absolute values of the residuals of Evr σ against adult body mass of each animal species. 1052 P values correspond to a phylogenetic ANOVA, and post-hoc Tukey test letters 1053 positioned on top of each group indicate whether groups are significantly different, after 1054 phylogenetic corrections. 1055 1056 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 4 Figure 1 1057 1058 1059 4 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 5 Figure 2 1060 1061 1062 1063 1064 5 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 6 Figure 3 1065 1066 6 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 7 Figure 4 1067 1068 1069 1070 7 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 8 Figure 5 1071 1072 1073 1074 8 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 9 Supplementary Online Materials 1075 1076 Support for the social buffering hypothesis, especially under 1077 unpredictable precipitation regimes 1078 1079 Table of contents 1080 1081 Table S1. Sensitivity of results to temporal replication in studies 1082 Table S2. Metadata and source of matrix population models 1083 Table S3. Degree of sociality across the climatic space 1084 1085 Figure S1. Sampling frequency of the matrix population models 1086 Figure S2. Matrix population model availability per study 1087 Figure S3. Matrix population model dimensionality 1088 Figure S4. Climate driver collinearities 1089 Figure S5. PCA screeplot 1090 Figure S6. Climatic predictability of demographic buffering 1091 1092 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 10 Table S1. Sensitivity of results to temporal replication in studies 1093 Our overall results are mostly insensitive to the duration of the study. Here, study 1094 duration is defined as the number of matrix population models (MPMs) available in each 1095 examined population. Battery of pgls models examining the relationships between the 1096 different stochastic elasticities of stochastic population growth rate ( λ s) and sociality, 1097 with study duration as a covariate. Note that the latter is only borderline significant in 1098 three occasions: Tσ , |Eσ A μ | and |Eσ A σ |. 1099 Stochastic elasticity to Predictor Estimate P A. Total Mean - Tμ Sociality 0.079 0.768 Duration -0.200 0.040 Variance - Tσ Sociality 0.807 0.399 Duration -0.451 0.062 B. Changes in mean Juvenile survival - |Eσ J μ | Sociality 0.117 0.818 Duration -0.086 0.487 Maturation - |Eγ μ | Sociality -0.738 0.318 Duration -0.023 0.894 Adult survival - |E σ A μ | Sociality 0.052 0.911 Duration -0.241 0.046 Reproduction - |E φ μ | Sociality -0.062 0.938 Duration -0.144 0.458 C. Changes in Juvenile survival - Sociality 0.108 0.906 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 11 variance |Eσ J σ | Duration -0.334 0.141 Maturation - |Eγ σ | Sociality -0.474 0.630 Duration -0.206 0.389 Adult survival - |Eσ A σ | Sociality 0.185 0.470 Duration -0.134 0.042 Reproduction - |E φ σ | Sociality 0.592 0.593 Duration -0.417 0.131 1100 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 1 Table S2. Metadata and source of matrix population models 1101 Sources of the demographic data obtained from COMADR E for the 66 examined species, together with other 1102 pertinent details. Population details the number of populations available per species in this study. Dimension details the 1103 number of stages the original matrix population models (MPMs) had before being collapsed to a set of 2×2 MPMs (see 1104 Methods). Sociality details the assigned level of our sociality continuum, with the explanation for each of the five levels 1105 provided in the Methods too. 1106 1107 1108 Species Common name Phylum Class Order Family Population Authors Journal DOI_ISBN Year Dimension Sociality Acyrth osipho n pisum Pea aphid Arthrop oda Insecta Aphido morph a Aphidid ae 1 Hamda ; Jevtic; Lasko wski Ecotoxi cology 10.100 7/s106 46- 012- 0904-5 2012 6 Gregari ous .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 2 Agarici a agaricit es Tan lettuce- leaf coral Cnidari a Anthoz oa Sclerac tinia Agaricii dae 1 Hughe s; Tanner Ecolog y 10.189 0/0012 - 9658(2 000)08 1[2250: RFLHA L]2.0.C O;2 2000 3 Coloni al Alces alces Moose Chorda ta Mamm alia Artioda ctyla Cervid ae 1 Ballard ; Whitm an; Reed Wildlife Monog r https:// www.js tor.org/ stable/ 383071 3 1991 3 Solitary Amblo Rock Chorda Actinop Percifo Centrar 1 People Master NA 2010 3 Gregari .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 3 plites rupestri s bass ta terygii rmes chidae s Thesis ous Ampeli sca abdita Amphi pod Arthrop oda Malaco straca Amphi poda Ampeli scidae 1 Kuhn; Munns; Serbst; Edwar ds; Cantw ell; Gleaso n; Pelletie r; Berry Enviro n Toxicol Chem 10.100 2/etc.5 620210 425 2002 8 Gregari ous Anser caerule Snow goose Chorda ta Aves Anserif ormes Anatid ae 1 Cooch; Rockw Ecol Monog 10.189 0/0012 2001 5 Coloni al .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 4 scens ell; Brault r - 9615(2 001)07 1[0377: RAOD RT]2.0. CO;2 Anthro poides paradis eus Blue crane Chorda ta Aves Gruifor mes Gruida e 1 Altweg g; Anders on Funct Ecol 10.111 1/j.136 5- 2435.2 009.01 563.x 2009 5 Gregari ous Astrobl epus ubidiai Andea n catfish Chorda ta Actinop terygii Silurifo rmes Astrobl epidae 1 Vélez- Espino Ecol Freshw Fish 10.111 1/j.160 0- 2005 7 Gregari ous .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 5 0633.2 005.00 084.x Boloria eunomi a Bog fritillary Arthrop oda Insecta Lepido ptera Nymph alidae 1 Radch uk; Turlure ; Schtick zelle J Anim Ecol 10.111 1/j.136 5- 2656.2 012.02 029.x 2012 5 Solitary Brachy teles hypoxa nthus Norther n muriqui Chorda ta Mamm alia Primat es Atelida e 2 Morris; Altman n; Brock man; Cords; Fediga Am Nat 10.108 6/6574 43 2011 10 Social .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 6 n; Pusey; Stoinsk i; Bronik owski; Alberts ; Strier Caenor habditi s elegan s NA Nemat oda Secern entea Rhabdi tida Rhabdi tidae 1 Li; Ju; Liao; Liao Ecotoxi cology 10.100 7/s106 46- 014- 1267- x) 2014 6 Gregari ous Callosp ermop Golden - Chorda ta Mamm alia Rodent ia Sciurid ae 1 Hostetl er; PLOS ONE 10.137 1/jourN 2012 6 Solitary .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 7 hilus laterali s mantle d ground squirrel Kneip; Van Vuren; Oli Al.pon e.0034 379 Capitell a capitat a Polych aete Annelid a Polych aeta NA Capitell idae 1 Hanse n; Forbes ; Forbes Funct Ecol 10.104 6/j.136 5- 2435.1 999.00 299.x 1999 2 Solitary Cebus capuci nus White- faced capuch in monke y Chorda ta Mamm alia Primat es Cebida e 2 Morris; Altman n; Brock man; Cords; Am Nat 10.108 6/6574 43 2011 8 Social .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 8 Fediga n; Pusey; Stoinsk i; Bronik owski; Alberts ; Strier Centro cercus minimu s Gunnis on sage- grouse Chorda ta Aves Charad riiforme s Stercor ariidae 1 Davis; Hooten ; Phillips ; Dohert y Ecol Evol 10.100 2/ece3. 1290 2014 4 Solitary .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 9 Cercop ithecus mitis Blue monke y Chorda ta Mamm alia Primat es Cercop ithecid ae 1 Morris; Altman n; Brock man; Cords; Fediga n; Pusey; Stoinsk i; Bronik owski; Alberts ; Strier Am Nat 10.108 6/6574 43 2011 9 Social Chloro Vervet Chorda Mamm Primat Cercop 1 Isbell; Int J 10.100 2009 2 Social .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 10 cebus aethiop s ta alia es ithecid ae Young; Jaffe; Carlso n; Chanc ellor Primat ol 7/s107 64- 009- 9332-7 Chroso mus oreas Mounta in redbell y dace Chorda ta Actinop terygii Cyprini formes Cyprini dae 1 People s Master Thesis NA 2010 2 Comm unal Ciconia ciconia White stork Chorda ta Aves Ciconiif ormes Ciconii dae 1 Schau b; Pradel; Lebret on Biol Conser v 10.101 6/j.bioc on.200 3.11.00 2 2004 2 Gregari ous Clinoco Woolly Chorda Actinop Scorpa Cottida 2 Davis; Mar 10.335 2002 3 Solitary .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 11 ttus analis sculpin ta terygii eniform es e Levin Ecol Prog Ser 4/meps 234229 Clinost omus fundulo ides Rosysi de dace Chorda ta Actinop terygii Cyprini formes Cyprini dae 1 People s Master Thesis NA 2010 2 Comm unal Colias alexan dra Queen Alexan dra's sulphur Arthrop oda Insecta Lepido ptera Pierida e 1 Hayes Oecolo gia 10.100 7/BF00 349187 1981 7 Solitary Cottus bairdii Mottled sculpin Chorda ta Actinop terygii Scorpa eniform es Cottida e 1 People s Master Thesis NA 2010 3 Solitary Erythro cebus Patas monke Chorda ta Mamm alia Primat es Cercop ithecid 1 Isbell; Young; Int J Primat 10.100 7/s107 2009 2 Social .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 12 patas y ae Jaffe; Carlso n; Chanc ellor ol 64- 009- 9332-7 Etheos toma flabella re Fantail darter Chorda ta Actinop terygii Percifo rmes Percid ae 1 People s Master Thesis NA 2010 3 Solitary Eulam prus tympan um Water skink Chorda ta Reptilia Squam ata Scincid ae 1 Blomb erg; Shine Austral Ecol 10.104 6/j.144 2- 9993.2 001.01 120.x 2001 5 Solitary Falco Lesser Chorda Aves Falconi Falconi 1 Hiraldo J Appl 10.230 1996 2 Comm .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 13 nauma nni kestrel ta formes dae ; Negro; Donaz ar; Gaona Ecol 7/2404 688 unal Falco peregri nus Peregri ne falcon Chorda ta Aves Falconi formes Falconi dae 1 Altweg g; Jenkin s; Abadi Ibis 10.111 1/ibi.12 125 2013 5 Solitary Forpus passeri nus Green- rumpe d parrotl ets Chorda ta Aves Psittaci formes Psittaci dae 1 Sander cock; Beissin ger J Appl Stat 10.108 0/0266 476012 010881 8 2002 2 Social Gopher Desert Chorda Reptilia Testudi Testudi 1 Perez- Oikos 10.111 2011 4 Comm .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 14 us agassi zii tortoise ta nes nidae Heydri ch; Oli; Brown 1/j.160 0- 0706.2 011.19 735.x unal Gorilla beringe i beringe i Mounta in gorilla Chorda ta Mamm alia Primat es Homini dae 1 Morris; Altman n; Brock man; Cords; Fediga n; Pusey; Stoinsk i; Am Nat 10.108 6/6574 43 2011 11 Social .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 15 Bronik owski; Alberts ; Strier Heliose ris cuculla ta Sunray lettuce coral Cnidari a Anthoz oa Sclerac tinia Agaricii dae 1 Hughe s; Tanner Ecolog y 10.189 0/0012 - 9658(2 000)08 1[2250: RFLHA L]2.0.C O;2 2000 3 Coloni al Homo sapien s Human Chorda ta Mamm alia Primat es Homini dae 1 Nicol- Harper; Dooley Popul Ecol 10.100 7/s101 44- 2018 18 Social .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 16 sapien s ; Packm an; Mueller ; Bijak; Hodgs on; Townle y; Ezard 018- 0620-y Lagopu s leucura White- tailed ptarmig an Chorda ta Aves Gallifor mes Phasia nidae 1 Wilson; Martin BMC Ecol 10.118 6/1472 -6785- 12-9 2012 2 Solitary Lagopu s muta Japane se rock Chorda ta Aves Gallifor mes Phasia nidae 1 Wilson; Martin BMC Ecol 10.118 6/1472 2012 2 Solitary .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 17 ptarmig an -6785- 12-9 Lepus americ anus Snows hoe hare Chorda ta Mamm alia Lagom orpha Leporid ae 1 Meslo w; Keith J Wildlife Manag e 10.230 7/3799 557 1968 4 Solitary Macac a mulatta Rhesu s macaq ue Chorda ta Mamm alia Primat es Cercop ithecid ae 1 Kessler ; Pachec o; Rawlin gs; Ruiz- Lambri des; Delgad Am J Primat ol 10.100 2/ajp.2 2323 2014 5 Social .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 18 o; Sabat Membr anipora membr anacea Sea mat Bryozo a Gymno laemat a Cheilos tomida Membr anipori dae 2 Harvell ; Caswel l; Simpso n Oecolo gia 10.100 7/BF00 323539 1990 5 Coloni al Microtu s oecono mus Root vole Chorda ta Mamm alia Rodent ia Murida e 1 Johann esen; Aars; Andrea ssen; Ims Popul Ecol 10.100 7/s101 44- 003- 0139-7 2003 3 Solitary Nocom is Bluehe ad Chorda ta Actinop terygii Cyprini formes Cyprini dae 1 People s Master Thesis NA 2010 3 Comm unal .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 19 leptoce phalus chub Notam acropu s eugenii Tamm ar wallaby Chorda ta Mamm alia Diproto dontia Macrop odidae 3 Chamb ers; Bencini Wildlife Res 10.107 1/WR1 0080 2010 2 Gregari ous Orbicel la annular is Caribb ean star coral Cnidari a Anthoz oa Sclerac tinia Faviida e 2 Hughe s; Tanner Ecolog y 10.189 0/0012 - 9658(2 000)08 1[2250: RFLHA L]2.0.C O;2 2000 3 Coloni al .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 20 Orcinu s orca Killer whale Chorda ta Mamm alia Cetace a Delphi nidae 2 Vélez- Espino; Ford; Araújo; Ellis; Parken ; Balcom b Can Tech Report Fish & Aq Sci 978-1- 100- 23563- 9 2014 7 Social Ovis aries Soay sheep Chorda ta Mamm alia Artioda ctyla Bovida e 1 Clutton -Brock; Price; Albon; Jewell J Anim Ecol 10.230 7/5330 1992 6 Gregari ous Paguru s Long- clawed Arthrop oda Malaco straca Decap oda Paguri dae 1 Damia ni Ecolog y 10.189 0/04- 2005 3 Solitary .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 21 longica rpus hermit crab 0956 Pan troglod ytes schwei nfurthii Easter n chimpa nzee Chorda ta Mamm alia Primat es Homini dae 1 Morris; Altman n; Brock man; Cords; Fediga n; Pusey; Stoinsk i; Bronik owski; Alberts Am Nat 10.108 6/6574 43 2011 17 Social .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 22 ; Strier Papio cynoce phalus Olive baboon Chorda ta Mamm alia Primat es Cercop ithecid ae 1 Morris; Altman n; Brock man; Cords; Fediga n; Pusey; Stoinsk i; Bronik Am Nat 10.108 6/6574 43 2011 8 Social .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 23 owski; Alberts ; Strier Param uricea clavata Violesc ent sea- whip; Red gorgoni an Cnidari a Anthoz oa Alcyon acea Plexau ridae 2 Linares ; Doak Mar Ecol Prog Ser 10.335 4/meps 08437 2010 7 Coloni al Pimep hales promel as Fathea d minno w Chorda ta Actinop terygii Cyprini formes Cyprini dae 1 Schwin dt NA NA 2013 4 Comm unal Plexau Gorgon Cnidari Anthoz Alcyon Plexau 1 Lasker Oecolo 10.100 1991 3 Coloni .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 24 ra homom alla ian coral a oa acea ridae gia 7/BF00 318316 al Porites astreoi des Caribb ean coral reef Cnidari a Anthoz oa Sclerac tinia Poritid ae 1 Edmun ds Mar Ecol Prog Ser 10.335 4/meps 08595 2010 3 Coloni al Propith ecus verrea uxi Verrea ux's sifaka Chorda ta Mamm alia Primat es Indriida e 1 Morris; Altman n; Brock man; Cords; Fediga n; Pusey; Am Nat 10.108 6/6574 43 2011 8 Social .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 25 Stoinsk i; Bronik owski; Alberts ; Strier Pseud odiplori a strigos a Symm etrical brain coral Cnidari a Anthoz oa Sclerac tinia Faviida e 1 Edmun ds Mar Ecol Prog Ser 10.335 4/meps 08595 2010 3 Coloni al Pygosc elis adeliae Adelie pengui n Chorda ta Aves Spheni sciform es Spheni scidae 1 Hinke; Trivelpi ece; Trivelpi ece Ecosph ere 10.100 2/ecs2. 1666 2017 2 Coloni al .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 26 Rana tempor aria Europe an commo n frog Chorda ta Amphi bia Anura Ranida e 1 Campb ell; Garner ; Tessa; Scheel e; Griffith s; Wilfert; Harriso n PeerJ NA 2018 11 Gregari ous Scelop orus grammi cus Mesqui te lizard Chorda ta Reptilia Squam ata Phryno somati dae 2 Ménde z–de la Cruz; Zúñiga Can J Zool 10.113 9/Z08- 124 2008 3 Solitary .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 27 -Vega; Cuellar Scolytu s ventrali s Fir engrav er beetle Arthrop oda Insecta Coleop tera Curculi onidae 1 Berrym an Can Entom ol 10.403 9/Ent1 051465 -11 1973 6 Solitary Strix occide ntalis Norther n spotted owl Chorda ta Aves Strigifo rmes Strigid ae 1 LaHay e; Zimme rman; Gutiérr ez Auk 10.164 2/0004 - 8038(2 004)12 1[1056: TVITV R]2.0. CO;2 2004 3 Solitary .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 28 Suricat a suricatt a Meerka t Meerka t Mamm alia Carniv ora Herpes tidae 1 Conqu et,Ozg ul, Blumst ein, Armita ge, Oli, Martin, Clutton -Brock, Paniw Ecosph ere 10.100 2/ecy.3 894 2023 4 Social Turdus torquat us Ring ouzel Chorda ta Aves Passeri formes Turdid ae 1 Sim; Rebec ca; Ludwig ; Grant; J Anim Ecol 10.111 1/j.136 5- 2656.2 010.01 2011 2 Solitary .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 29 Reid 750.x Umbon ium costatu m NA Mollus ca Gastro poda Vetigas tropod a Trochid ae 1 Noda; Nakao J Anim Ecol 10.230 7/5722 1996 6 Comm unal Ursus maritim us Polar bear Chorda ta Mamm alia Carniv ora Ursida e 1 Hunter; Caswel l; Runge; Regehr ; Amstru p; Stirling Ecolog y 10.189 0/09- 1641 2010 6 Solitary Vireo atricapi Black- capped Chorda ta Aves Passeri formes Vireoni dae 11 Walker ; Biol Conser 10.101 6/j.bioc 2016 2 Solitary .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 30 lla vireo Marzluf f; Cimpri ch v on.201 6.09.01 6. Xenos aurus agreno n Knob- scaled lizard Chorda ta Reptilia Squam ata Xenos auridae 1 Zamor a- Abrego ; Chang; Zúñiga -Vega; Nieto- Montes de Oca; Johnso Herpet ologica 10.165 5/09- 005.1 2010 4 Solitary .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 31 n Xenos aurus grandis Crevic e- dwellin g lizard Chorda ta Reptilia Squam ata Xenos auridae 1 Zúñiga -Vega; Valver de; Rojas- Gonzal ez; Lemos- Espinal Copeia 10.164 3/0045 - 8511(2 007)7[ 324:A OTPD O]2.0. CO;2 2007 4 Solitary Xenos aurus platyce ps Flathea d knob- scaled lizard Chorda ta Reptilia Squam ata Xenos auridae 2 Rojas- Gonzal ez; Jones; Amphi bia- Reptilia 10.116 3/1568 538087 841249 2008 4 Solitary .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 32 Zúñiga -Vega; Lemos- Espinal 92 1109 1110 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 1 Table S3. Degree of sociality across the climatic space 1111 The degree of sociality is not predicted by the climatic principal component axis. 1112

of a pgls model predicting the degree of sociality as a function of the positioning 1113 of each population along PC1 (Precipita tion predictability; Figure 2) and PC2 1114 (temperature constancy). 1115 Estimate S. E. t P Intercept 2.616 1.1876 2.2032 0.0436 PC1 0.1150 0.2581 0.4453 0.6624 PC2 0.2682 0.3579 0.7493 0.4652 PC1 × PC2 0.0165 0.1845 0.0894 0.9299 1116 1117 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 2 Figure S1. Sampling frequency of the matrix population models 1118 Most of the matrix population models (MPMs) in our selected set of matrix1119 population models from the COMADRE Animal Matrix Database were sampled1120 once a year. Species in our study with an original sampling frequency that is not1121 annual: Caenorhabditis elegans (0.000114), Capitella capitata (0.0192), Xenosaurus1122 agrenon (0.08), Membranopira mebranaceae (0.12), Acyrthosiphon pisum (0.02),1123 Clinocottus analis (0.25), Ampelisca abdita (0.03), Eulamprus tympanum (0.08),1124 Sceloporus grammicus (0.5), Montastrae annularis (5), and Homo sapiens sapiens (5). 1125 1126 1127 2 ix ed ot us 2), 8), . .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 3 Figure S2. Matrix population model availability per study 1128 Number of matrix population models (MPMs) per studied population in our1129 selected subset of 66 species from the COMADRE Animal Matrix. Total MPM =1130 955. Total number of separate populations = 87. 1131 1132 3 ur = .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 4 Figure S3. Matrix population model dimensionality 1133 The dimensionality of our examined matr ix population models (MPMs) varied1134 considerably between 2 and 18 stages. As such, we collapsed all models to a two -1135 stage MPM to allow for stage comparability in vital rates. 1136 1137 4 ed - .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 5 Figure S4. Climate driver collinearities 1138 Some of the examined climatic drivers are st rongly collinear. Pairwise correlations1139 (top-right) and scatterplots (bottom-left) of the statis tics of monthly precipitation and1140 temperature derived from the NASA POWER database to quantify the degree of1141 environmental stochasticity to which our 87 natural populations have been exposed.1142 Temperature statistics include: : mean; Δ T: variance: T cons: constancy; T cont:1143 contingency; and Tpred: predictability. Ditto for precipitation (P). Constancy quantifies the1144 extent to which a climatic variable remains stable over time (1 - variance/mean);1145 contingency quantifies the extent to which fluctuations are str uctured and recurrent;1146 predictability = constancy + contingency, as per Colwell (Colwell, 1974) . Font size of1147 spearman correlation coefficient is proportional to its values. 1148 1149 5 ns nd of d. : he n); nt; of .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 6 Figure S5. PCA screeplot 1150 Screeplot of the six principal components of the climatic PCA. The blue dashed1151 line separates the principal components axis with associated eigenvalue > 1, which we1152 retained for the next steps in our analyses, as per the Kaiser criterion. 1153 1154 1155 1156 6 ed e .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint 7 Figure S6. Climatic predictability of demographic buffering 1157 The predictability of annual precipitation patterns predicts the ability of animal1158 populations to buffer against extreme climatic events. Scaled effect sizes (negative:1159 red; positive: blue) and P values for the correlations between the climatic PCAs (Figure1160 2 - PC1: precipitation predictability, PC2: temperature constancy, and their interaction)1161 and: (A) the sum of total stochastic elasticities to the mean (|Tμ |) and the variance (|Tσ |),1162 (B) the stochastic elasticities to changes in vital rate means (|Eμ |), and (C) to changes in1163 vital rate variances (| Eσ |). Vital rates are: juvenile survival ( σ J), maturation ( γ ), adult1164 survival (σ A), and reproduction (φ ) (Eq. 1). 1165 1166 7 al e: re n) |), in ult .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 4, 2025. ; https://doi.org/10.1101/2025.04.30.651380doi: bioRxiv preprint