Longevity, body dimension and reproductive mode drive differences in aquatic versus terrestrial life-history strategies

1. Aquatic and terrestrial environments display stark differences in key environmental factors and phylogenetic composition but their consequences for the evolution of species' life-history strategies remain poorly understood. 2. Here, we examine whether and how life-history strategies vary between terrestrial and aquatic species. We use demographic information for 685 terrestrial and 122 aquatic animal and plant species to estimate key life-history traits. We then use phylogenetically corrected least squares regression to explore potential differences in trade-offs between life-history traits between both environments. We contrast life-history strategies of aquatic versus terrestrial species in a principal component analysis while accounting for body dimensions and phylogenetic relationships. 3. Our results show that the same trade-offs structure terrestrial and aquatic life histories, resulting in two dominant axes of variation that describe species' pace of life and reproductive strategies. Terrestrial plants display a large diversity of strategies, including the longest-lived species in this study. Aquatic animals ex hibit higher reproductive frequency than terrestrial animals. When correcting for body size, mobile and sessile terrestrial organisms show slower paces of life than aquatic ones. 4. Aquatic and terrestrial species are ruled by the same life-history trade-offs, but have evolved different strategies, likely due to distinct environmental selective pressures. Such contrasting life-history strategies have important consequences for the conservation and management of aquatic and terrestrial species.

Life-history theory is rooted upon the concept of trade-offs as a unifying principle across the tree of life (Stearns, 1992). Given the limitations in available energy and physiological constraints, compromises among survival, development and reproduction are inescapable for any organism, whether aquatic or terrestrial (Stearns, 1992). Such constraints should result in a finite set of viable life-history strategies. The evolution of a life-history strategy in a given environment is then determined by two counteracting processes: environmental filtering and evolutionary history (Stearns, 1992). Environmental filtering stems from extrinsic factors favouring certain strategies over others. For example, aquatic environments enable the evolution of sessile animals due to the suspended nutrients and organic material in the water column. Such a strategy is not possible for terrestrial animals (Webb, 2012). On the other hand, evolutionary history represents the influence of phylogenetic relationships in determining the potential adaptations of a given species (Blomberg & Garland, 2002;Freckleton, 2000). Lifehistory strategies are then expected to be more similar, irrespective of environment, among closely related lineages.
According to life-history theory, the same trade-offs should be experienced by aquatic and terrestrial organisms. Comparative demographic studies have successfully identified and organized tradeoffs into a few major axes of trait co-variation (Gaillard et al., 1989;Salguero-Gómez, Jones, Jongejans, et al., 2016). A seminal concept in organizing such trait co-variation is the 'fast-slow continuum' (Stearns, 1992). In it, species are placed along a continuous axis bounded by two extremes: at the fast-living extreme, species develop quickly, are highly reproductive but have short life spans; while at the slow extreme, species have high survival rates, develop slowly and live long. However, an explicit comparison of the fast-slow continuum between aquatic and terrestrial species remains, to our knowledge, untested.
If trade-offs are universal, the strong environmental and phylogenetic dissimilarities between aquatic and terrestrial environments should result in different life-history strategies. For example, aquatic and terrestrial habitats impose differing selective pressures on body size. Indeed, aquatic endotherms have larger body sizes than terrestrial ones, due to the strict energetic demands of the aquatic environments (Gearty, McClain, & Payne, 2018). Such constrains must have consequences for aquatic life-history strategies, given that a large body size covaries positively with a slow pace of life (Gaillard et al., 1989;Healy et al., 2014). On the other hand, aquatic environments allow early life stages to feed and develop during the dispersal phase, promoting external reproduction (Burgess, Baskett, Grosberg, Morgan, & Strathmann, 2016;Bush, Hunt, & Bambach, 2016;Vermeij & Grosberg, 2017), while terrestrial species had to evolve reproductive systems independent of environmental water, such as internal fecundity or seeds (Bush et al., 2016;Grosberg, Vermeij, & Wainwright, 2012;Steele, Brink, & Scott, 2019). Therefore, aquatic species had to evolve strategies to counteract the uncertainty of recruitment success derived from external reproduction (Charnov & Schaffer, 1973;Tuljapurkar, Gaillard, & Coulson, 2009).
The colonization of land likely resulted in the evolution of life-history strategies to deal with higher temporal environmental variation (Dawson & Hamner, 2008;Ruokolainen, Lindén, Kaitala, & Fowler, 2009). On land, environmental variation is more stochastic and less temporally auto-correlated than in aquatic environments (Dawson & Hamner, 2008). Classical life-history theory predicts the evolution of longevity in constant environments (Lande et al., 2017). However, longevity can also be a strategy to deal with environmental variation (McDonald et al., 2017;Morris et al., 2008). For example, by spreading their reproductive output across several years, long-lived species are able to exploit favourable conditions in a stochastic environment, compen- Here, we test the hypotheses that (a) life-history trade-offs are universal across aquatic and terrestrial systems, and that (b) terrestrial species have evolved distinct life-history strategies compared to aquatic ones. We use high-resolution demographic data from 122 aquatic and 685 terrestrial species across the globe from the COMPADRE and COMADRE databases (Salguero-Gómez, Jones, Archer, et al., 2016;Salguero-Gómez et al., 2015). We estimate key life-history traits that reflect various moments of population turnover, as well as investments in survival, development and reproduction of each species. To test these hypotheses, we first determine whether correlations between life-history traits differ across environments as a way to examine whether trade-offs diverge between terrestrial versus aquatic species. Second, we explore the main axes of life-history variability shaping aquatic and terrestrial species. The presence of different life-history axes of variation and/or a distinct positioning of aquatic species compared to terrestrial ones within those axes would suggest dissimilar selection pressures occurring in terrestrial and aquatic environments. Given the scarcity of comparative studies and the lack of demographic information for many aquatic species, elucidating these questions is a key step towards understanding the evolution of life histories across environments.

| Demographic data and life-history traits
We calculated species' life-history strategies using demographic data describing the full life cycle of each species. This high-quality demographic information was obtained from the COMPADRE Plant Matrix Database (v. 5.0.1;Salguero-Gómez et al., 2015) and COMADRE Animal Matrix Database (v. 3.0.1;Salguero-Gómez, Jones, Archer, et al., 2016). These repositories archive demographic data as matrix population models (MPMs, hereafter) for over 700 plant and 400 animal species respectively. MPMs are summaries of organisms' demographic processes (i.e. vital rates such as survival, development and reproduction) that together determine their life-history strategies and resulting population dynamics (Caswell, 2001). For this reason, MPMs provide the ideal means to compare the vast array of life-history strategies (Franco & Silvertown, 2004;McDonald et al., 2017).
To compare life-history traits across aquatic and terrestrial species, we imposed a series of selection criteria to the available demographic data (see details in Appendix S2: Data selection in Supporting Information). These criteria resulted in 685 terrestrial species and 122 aquatic species used in this study (Appendix S1). To determine the marine, freshwater or terrestrial origin of species, we used the primary habitat reported in the World's Register of Marine Species (WORMS, www.marin espec ies.org) and the Catalogue of Life (CL, http://www.catal ogueo flife.org) databases. We estimated the total number of species in terrestrial, freshwater and marine environments based on the estimates provided in Grosberg et al. (2012), together with information provided in WORMS and CL. The number of species studied here represented a similar taxonomic coverage relative to the known biodiversity of the aquatic (~0.04%-0.03%) and terrestrial environment (~0.01%; Table S1 in Appendix S2).
Quantifying a species' life-history strategy requires detailed information regarding the timing, intensity, frequency and duration of key demographic processes across its life cycle (Capdevila & Salguero-Gómez, 2019;Stearns, 1992). To quantify species' life-history strategies, we calculated several life-history traits from each MPM using well-established methods (Salguero-Gómez, Jones, Jongejans, et al., 2016). We selected six life-history traits commonly used in comparative demography (Bielby et al., 2007;Gaillard et al., 2005;Salguero-Gómez, Jones, Jongejans, et al., 2016;Stearns, 1992). These traits include: generation time (T), age at sexual maturity (L α ), rate of senescence (H), mean vital rate of progressive development (γ), mean vital rate of sexual reproduction (φ) and degree of iteroparity (S; Table 1). Such traits provide insights into a species' population turnover, as well as of survival, developmental and reproductive strategies (detailed in Table 1).
For every species, we decomposed the MPM, A, into two subcomponents (Equation 1): the U matrix, which represents the survivaldependent vital rates (e.g. development, shrinkage, fission, etc) and the F matrix, containing the stage-specific per capita reproduction rates (Caswell, 2001;Morris & Doak, 2002). Those species showing clonality were removed from the analyses, in order to avoid potential over-estimation of survival rates. This decomposition facilitates the estimation of key life-history traits such as the time elapsed since, or to, a given demographic event (e.g. age at maturity, mean life expectancy; see Table 1).
The traits T, L α and R 0 were calculated using stage-from-age demographic decompositions (Caswell, 2001, pp. 124-127; see Table 1), where the beginning of life was a priori defined as the first nonpropagule stage in the life cycle of the organism (Burns et al., 2010).

TA B L E 1
Formulation of the life-history traits used to explore the variation in life-history strategies in the 685 terrestrial and 122 aquatic species studied. λ is the deterministic population growth rate, which corresponds to the dominant eigenvalue of the matrix A (Caswell, 2001); l x and m x are the age-specific survival and fertility schedules respectively; U and F are the submatrices of survival-and fertility-dependent processes respectively (Equation 1); U′ is the survival-independent matrix of transition probabilities (Caswell, 2001); w is the stable stage distribution of the matrix A, and i and j are the row and column entries of the matrix population model respectively

Life-history trait Definition Calculation
Turnover Generation time T Number of years required for an average individual in the population to replace itself This approach avoids uncertainties associated with the longevity of spores and seeds (Burns et al., 2010;Caswell, 2001;Salguero-Gómez, Jones, Jongejans, et al., 2016;Silvertown & Franco, 1993) and assures the comparability with species without such life cycle stages. To calculate S and H (Demetrius, 1974;Keyfitz, 1977), we first obtained the age-specific survivorship curve (l x ) and the agespecific fertility trajectory (m x ) following Caswell (2001, pp. 118-121), and implemented the formulae described in Table 1. The traits' progressive development (γ) and sexual reproduction (φ) summarize investments into development and reproduction annually for all stages across the life cycle weighted by the relative representation of stages under stationary conditions (Table 1).

| Phylogenetic analyses and trait comparisons
We accounted for and estimated the phylogenetic influence on the differences in life-history trait values within species and between aquatic and terrestrial environments. To do so, we constructed a species-level phylogenetic tree ( Figure S4 in Appendix S3) with data from Open Tree of Life (OTL, https://tree.opent reeof life.org, Hinchliff et al., 2015). OTL combines publicly available taxonomic and phylogenetic information across the tree of life (Hinchliff et al., 2015). Briefly, we built separate trees for our species of algae, plants and animals, using the rotl r package (Michonneau, Brown, & Winter, 2016). These trees were assembled in a supertree using the function bind.tree in the phytools package (Revell, 2012). To account for the phylogenetic relatedness of species, we computed the branch lengths and resolved polytomies (Revell, 2012). We also tested the sensitivity of our results to the choice of a particular set of branch lengths, by repeating our analyses setting all the branch lengths to one and using Pagel's branch length (Tables S5-S8 in Appendix S3). We did so using the software Mesquite 1.05 (Maddison & Maddison, 2001) and its PDAP module 1.06 (Midford, Garland Jr., & Maddison, 2005), for further details on the construction of the tree see Appendix S3.
To test whether life-history trait trade-offs are congruent between aquatic versus terrestrial species, we carried out a series of phylogenetic general least square (PGLS) analyses (Revell, 2010).
This approach allows us to accommodate residual errors according to a variance-covariance matrix that includes ancestral relationships between any pair of species from our phylogenetic tree (Revell, 2010(Revell, , 2012. The variance-covariance matrix represents the expected covariance between species' trait values, given a phylogenetic tree and under a specific model of evolution (see below; Revell, 2009). The expected covariance between species' trait values is directly proportional to the distance between the species and their most recent common ancestor, that is measured as the branch length of the phylogeny (Revell, 2009).
We implemented our set of PGLSs in R using the correlation structures provided by the package ape (Paradis, Claude, & Strimmer, 2004). We used a Brownian motion model of evolution, combined with the pgls function from the nlme package (Pinheiro, Bates, Debroy, Sarkar, & R Core Team, 2014). Separate PGLSs were fitted using Ornstein-Uhlenbeck model of evolution, which describes a Brownian model under the influence of friction (Uhlenbeck & Ornstein, 1930). Both models were compared using Akaike information criterion (Akaike, 1974); the Brownian motion model generally outperformed the Ornstein-Uhlenbeck model, but both showed similar results. Therefore, we only report the PGLS results from the Brownian motion model.

| Exploring dominant axes of life-history strategies
To explore the patterns of association among life-history traits for aquatic versus terrestrial species, we performed a series of principal components analysis (PCA). PCA is a multivariate analysis that reduces a set of correlated variables into linearly uncorrelated measurements, the so-called principal components (PCs). Life-history trait data were log-and z-transformed (M = 0, SD = 1) to fulfil normality assumptions of PCAs (Legendre & Legendre, 2012 (Blomberg & Garland, 2002;Freckleton, 2000). This metric varies between 0, when the observed patterns are not due to phylogenetic relationships, and 1 when the observed patterns can be explained by the employed phylogeny (Blomberg & Garland, 2002;Revell, 2010). The pPCA was estimated using the phyl.pca function from the r package phytools (Revell, 2012), assuming a Brownian motion model of evolution (Revell, 2010 Caswell, 2001;Jones et al., 2014), which was not the case for this species. In these cases, we imputed the missing data using the function amelia from the amelia package (Honaker, King, & Blackwell, 2011). This function uses a bootstrap expectationmaximization algorithm to impute missing data (Honaker et al., 2011). We then created 10 imputed datasets and ran analyses on each separately.

| Body dimension pPCA correction
Body weight and size are highly correlated with many life-history traits. Life-history studies typically correct for body dimension (e.g. size or weight) explicitly to unmask potential correlations in lifehistory traits once the effect of body dimension has been taken into account (e.g. Bielby et al., 2007;Gaillard et al., 1989). There are multiple ways to account for body dimension in life-history analyses (Jeschke & Kokko, 2009). Here, we used the residuals of the linear models between each life-history trait of interest and the body dimension of each species in the pPCA (Revell, 2009 Tables S3 and   S4, Appendix S2).

| Analysis validation
We examined the consistency of our results and explored the differ-  (Bush et al., 2016;Vermeij & Grosberg, 2017). We also performed a series of pPCA subsetting species into Animalia, and Plantae and Chromista kingdoms (brown algae). This distinction was made because animals and plants/algae differ in key physiological, trophic and developmental traits (Burgess et al., 2016;Grosberg et al., 2012). Such ecological differences between sessile/mobile and taxonomic kingdoms could have a potential impact on our hypothesis about how the evolution of life-history strategies differ in aquatic and terrestrial species.
We tested the sensitivity of our results to missing traits in the dataset using pPCAs and PGLS in two ways. First, we ran a pPCA and a PGLS only with the species with complete data (62 aquatic species, 477 terrestrial species, Tables S13 and S14, Appendix S4).
Then, we ran another pPCA and PGLS with all species, we were able to include species with missing data by imputing the necessary missing data (see details in Tables S9 and S10, Appendix S4). We also ran the pPCA and the PGLS with and without the outliers, the exclusion of outliers did not alter our main findings (see Tables S11 and S12 in Appendix S4 We evaluated the performance of the imputation method utilized in our study using the diagnostic tools available in the r package amelia

| Trade-offs are pervasive across environments
Life-history traits are shaped by the same trade-offs for terrestrial and aquatic organisms (Figure 1). Our PGLS analyses reveal a similar magnitude and the same direction of PGLS correlations between traits for aquatic and for terrestrial species (Figure 1 and Tables S10, S12 and S14, Appendix S4). Regardless of the environment, producing many recruits (high φ; Table 1) results in fast population turnover (low T). Species that postpone their first reproductive event (high L α ) have low senescence rates (high H; Figure 1). Species with fast development (high γ) achieve reproductive maturity early (low L α ) at the cost of high senescence rates (low H). Also, those species with high mean reproductive output (high φ) and frequent reproduction (high S) have low senescence rates (high H; Figure 1).

| Longevity is more prevalent in terrestrial environments
Together, the first two axes of our pPCA ( Table 1) Table S15 in Appendix S4), taxonomic kingdoms (Figure 3c,d and Table S16 in Appendix S4) and environments (Table S17, Appendix S4).
The sampled aquatic life-history strategies in our study are

| Mode-of-life, kingdom and body dimension drive key life-history differences across environments
The main axes of life-history variation remain unaltered across en- variation correspond to the fast-slow continuum and reproductive strategies in both sessile and mobile species (Figure 3a,b and Table S15 in Appendix S4), in Animalia and Plantae/Chromista kingdoms (Figure 3c,d and Table S16 in Appendix S4), and in terrestrial and aquatic species (Table S17, Appendix S4). These patterns remain the same after correcting for body weight in mobile species and body size in sessile species for both aquatic and terrestrial organisms ( Figure 4, Table S18 in Appendix S4).
Aquatic and terrestrial sessile species display significant dif-

| Ancestry does not shape cross-environmental life-history strategies
Overall, phylogenetic ancestry (i.e. phylogenetic inertia) plays a minor role in constraining life-history strategies between F I G U R E 3 The main axes of lifehistory variation remain constant, regardless of the degree of mobility/ sessility or taxonomic kingdom. Phylogenetically corrected principal component analysis of six life-history traits across 683 species. Trait definitions are shown in Table 1. Note that the fastslow continuum remains the dominant axis of variation across all partitions, explaining 49%-50% of the variation, followed by an axis of reproductive strategies, which explains ~21% of the variation in life-history traits. ns, nonsignificant; *p < 0.05; **p < 0.01.   Table 1. Note that the fast-slow continuum remains the dominant axis of variation, explaining ~51% of the variation, followed by an axis of reproductive strategies, which explains 18%-19% of the variation in life-history traits. ns: non-significant; *p < 0.05; **p < 0.01. environments. The estimates of Pagel's λ in our pPCA are indeed weak (0.26 ± 0.00). Such values of the phylogenetic signal remain weak across sessile species (λ = 0.18 ± 0.01; Table S15, Appendix S4), mobile species (λ = 0.36 ± 0.01; Table S15, Appendix S4), plants and algae (λ = 0.18 ± 0.01; Table S16, Appendix S4) and animals (λ = 0.31 ± 0.02; Table S16, Appendix S4). In addition, the phylogenetic signal is similar between terrestrial (λ = 0.24 ± 0.01; Table S17 in Appendix S4) and aquatic species (λ = 0.19 ± 0.02; Table S17 in Appendix S4).
The traits with the highest loading on the fast-slow continuum (T, H and L α ) are strongly phylogenetically linked to two leading traits of the reproductive-strategies axis (ɸ and S). Equally, the variation in age at maturity (L α ) is largely explained by its phylogenetic association with developmental rates (γ; Figure 1). For both aquatic and terrestrial species, reproductive traits (ɸ and S in Table 1) are systematically more labile (i.e. lower phylogenetic signal) than traits associated with survival (H, L α ), development (γ) or turnover (T). Generation time (T) and age at reproductive maturity (L α ) are strongly phylogenetically associated with the number of recruits produced (ɸ) and the degree of iteroparity (S; Figure 1).

| D ISCUSS I ON
Our results show that life-history strategies of terrestrial and aquatic organisms are organized along the same dominant axes of variation and are constrained by the same trade-offs, regardless of the environment. The aquatic species we have examined here have not evolved the high longevities attained by some of the terrestrial species. However, we have found that aquatic animals are more reproductive than the terrestrial ones. The relatively weak phylogenetic signal in our analyses suggests that these key life-history differences are not primarily explained by the differential taxonomic composition of both environments (Blomberg & Garland, 2002;Freckleton, 2000). Overall, we suggest that the contrasting environmental conditions between aquatic and terrestrial environments may play a major role in the observed lifehistory patterns and differences.

| Terrestrial plants and mobile animals show slower paces of life than aquatic ones
Our analyses reveal a greater diversity of life-history strategies in terrestrial compared to aquatic environments, for our studied species. This finding is congruent with the higher species richness (Costello & Chaudhary, 2017) (Steele et al., 2019;Wiens, 2015). These innovations allowed the exploitation of novel ecological niches, ultimately resulting in a sixfold increase in speciation rate (Costello & Chaudhary, 2017;Wiens, 2015). We argue that such adaptations are reflected in the vast diversity of life histories observed in the terrestrial environment relative to that in the aquatic environment in our study.
Plants and animals evolved different sets of adaptations to terrestrial and aquatic environments (Burgess et al., 2016;Steele et al., 2019) Correcting for body dimension allowed us to reveal a slower pace of life in terrestrial animals compared to aquatic ones. Terrestrial animals have been suggested to compensate for environmental uncertainties through the evolution of complex behaviours (e.g. sociality, nesting) and physiological adaptations (e.g. thermoregulation, internal fecundation; Grosberg et al., 2012;Steele et al., 2019). Such traits would explain the lack of significant differences in the fast-slow continuum between terrestrial and aquatic species. However, our results also show that, when correcting for body weight, aquatic animals have a faster pace of life than terrestrial ones. Water-dwelling endotherms have larger body mass than their terrestrial counterparts (Gearty et al., 2018). Hence, given the correlation of body dimension and the fast-slow continuum (Gaillard et al., 1989;Healy et al., 2014), for a given size, aquatic mobile animals are faster lived than terrestrial ones.

| Aquatic animals are more reproductive than terrestrial ones
Terrestrial and aquatic environments also differ in the repertoire of reproductive strategies. Aquatic colonizers of terrestrial environments had to evolve strategies to protect early developmental stages (e.g. to desiccation) and enable their development in non-aquatic environments (Burgess et al., 2016;Steele et al., 2019;Strathmann, 1990).
Plants, like many benthic aquatic species, have a sessile adulthood, so their dispersal relies on early developmental stages only. This mode-of-life promoted the evolution of flowers, pollination and seeds (Kenrick & Crane, 1997). Sessile mode-of-life resulted in the observed high reproductive outputs and frequencies in plants, despite the fact that they can also reach high longevities (e.g. McDonald et al., 2017;Salguero-Gómez, Jones, Jongejans, et al., 2016).
Aquatic animals show higher reproductive outputs and frequencies than terrestrial animals. This pattern is likely linked to the prevalence of external fertilization in aquatic environments, while internal fertilization is more common in terrestrial ones (Bush et al., 2016). Both viscosity and nutrient concentration are higher in seawater than in air (Dawson & Hamner, 2008), allowing propagules to remain suspended for long periods of time (Burgess et al., 2016;Strathmann, 1990). The release of progeny in the water column comes with a high early predation risk and mortality, and low establishment probability (Burgess et al., 2016;Strathmann, 1990). To compensate for such early mortality, aquatic species release high numbers of propagules frequently, resulting in highly reproductive life histories. Differently, most terrestrial animals retain female gametes on or in their bodies, with fertilization and early development being also internal (Bush et al., 2016;Steele et al., 2019), resulting in less reproductive strategies. Still, some aquatic species exist with internal fertilization, such as sharks or marine mammals (Steele et al., 2019), partly explaining the range of reproductive strategies observed in our study.

| Data limitations
Although the volume of data used in our study has a similar ratio to that of the biodiversity held in aquatic versus terrestrial environments (Table S1, Appendix S1), it still represents a limited fraction of the known diversity (Costello & Chaudhary, 2017;Grosberg et al., 2012). Importantly, here, we have focused mostly on macroscopic organisms, for which full demographic information is more readily available than for smaller species (Salguero-Gómez, Jones, Archer, et al., 2016;Salguero-Gómez et al., 2015). Organisms like insects, but also microscopic organisms, such as plankton or bacteria, are challenging subjects for demographic studies, so their data are scarce (Conde et al., 2019;Salguero-Gómez, Jones, Archer, et al., 2016). In addition, recently discovered extremely long-lived marine species (e.g. Somniosus microcephalus, Nielsen et al., 2016;Monorhaphis chuni, Jochum, Wang, Vennemann, Sinha, & Müller, 2012) are likely examples of slow strategies for which we do not have complete demographic data yet. Thus, the increase of studies quantifying the demographic processes of the full life cycle of species will likely shed more light on the differences between aquatic and terrestrial life histories.
In this study, we used demographic schedules as the common currency to quantify the life-history strategies of species.
Species life-history strategies are highly determined by the demographic processes of survival, development and reproduction (Caswell, 2001;Stearns, 1992). Researchers quantifying life-history strategies have used different approaches to compare species (e.g. fish in Winemiller & Rose, 1992; plants in Grime & Pierce, 2012;Westoby, 1998). These approaches have significantly contributed to improve our current understanding of life-history strategies both in terrestrial and aquatic environments (Grime & Pierce, 2012).
However, in some cases, these approaches use taxon-specific traits (such as the leaf-height-seed strategy scheme by Westoby, 1998), which would not allow us to compare across different taxonomic groups, such as animals and plants. For that reason, quantifying important moments of the life cycle of species with demographic data (Salguero-Gómez, Jones, Jongejans, et al., 2016) provides the ideal means to compare strategies across very different and distant taxonomic groups. We also demonstrate that considering incomplete demographic information (e.g. only investments in survival) can lead to the inaccurate characterization of the life-history strategy of a given species. Information on the fast-slow continuum explains 49.29% and 47.69% of the life-history variation in aquatic and terrestrial species respectively (Table S17 in Appendix S4). Demographic studies typically miss reproductive information because it is more challenging to collect and estimate. We show here that the current lack of data on reproductive rates prevents us from improving our understanding of life-history strategies by over 20% across environments (Table S17 in Appendix S4).

| CON CLUS IONS
Our study provides an entry point to comparative life-history studies between aquatic and terrestrial environments. Our findings evidence the existence of strong differences between the life-history strategies of aquatic and terrestrial species as a consequence of the colonization of terrestrial environments. Such contrasting lifehistory strategies are probably linked to the distinct responses to climate change (Pinsky, Eikeset, McCauley, Payne, & Sunday, 2019), exploitation (McCauley et al., 2015 or extinction rates (Webb & Mindel, 2015) observed in aquatic and terrestrial systems.
Understanding how patterns of life histories translate into differences in their response to disturbances will be crucial to improve management decisions and predict future biodiversity trends.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data and code supporting the results are deposited in the Dryad Digital Repository https://doi.org/10.5061/dryad.cvdnc jt1q . Matrix population models are available at www.compa dredb.org.