Report

Life history responses of meerkats to seasonal changes in extreme environments

See allHide authors and affiliations

Science  08 Feb 2019:
Vol. 363, Issue 6427, pp. 631-635
DOI: 10.1126/science.aau5905

Timing matters

How a species responds to rapid climate change is complicated. Paniw et al. used long-term data on the Kalahari meerkat, an arid specialist, to explore how predicted changes might affect population persistence over time. Warming and rainfall changes in one part of the year had a negative impact on survival and persistence, whereas similar changes during another part of the year had the opposite effect. Understanding such variability will be essential as we attempt to understand the broader influence of climate change.

Science, this issue p. 631

Abstract

Species in extreme habitats increasingly face changes in seasonal climate, but the demographic mechanisms through which these changes affect population persistence remain unknown. We investigated how changes in seasonal rainfall and temperature influence vital rates and viability of an arid environment specialist, the Kalahari meerkat, through effects on body mass. We show that climate change–induced reduction in adult mass in the prebreeding season would decrease fecundity during the breeding season and increase extinction risk, particularly at low population densities. In contrast, a warmer nonbreeding season resulting in increased mass and survival would buffer negative effects of reduced rainfall during the breeding season, ensuring persistence. Because most ecosystems undergo seasonal climate variations, a full understanding of species vulnerability to global change relies on linking seasonal trait and population dynamics.

Ecosystems subject to climatic extremes, such as arid regions, which cover >40% of the terrestrial landmass, are among the most vulnerable to climate change (13). Changes in rainfall-drought cycles (2, 4) and increasing temperatures (5) are likely to severely affect population dynamics of arid environment species (6). This is because key vital rates—such as survival, growth, and reproduction—of these species respond unusually strongly to seasonal and interannual climatic patterning (2, 5, 7). Despite this increased vulnerability, population viability analyses of arid environment specialists are scarce (6). Studies on seasonal changes in vital rates and phenotypic traits, which strongly mediate climatic effects on viability (810), are thus far absent.

In this study, we used long-term demographic and trait (body mass) data of meerkats (Suricata suricatta) from the Kalahari Desert in southern Africa to investigate how future changes in seasonal rainfall and temperature may affect vital rates and population persistence. Meerkats are an ideal study species because their vital rates can be measured precisely (11) and respond strongly to climatic factors, generating large variation in population size (8, 12). In addition, meerkats are cooperative breeders: Young born to a dominant female are co-reared by nonbreeding helpers (13), and the number of helpers in the population increases reproductive success (8, 14). This relationship allows us to assess how interactions between population density, which is influenced predominantly by the number of helpers, and climate affect vital rates (12, 15). To establish a basis for detailed projections of population change over time, we first used 20 years of individual data from female meerkats to fit generalized additive models (GAMs) (8). In these models, vital rates (survival, growth, reproduction, stage transitions, and emigration) of nine life history stages (pups, juveniles, and subadults, as well as nonpregnant, pregnant, and litter-weaning helpers and dominants; fig. S1) were fitted as functions of body mass, population density, season (month of year), interannual rainfall and temperature deviations (from seasonal means), and interactions among these drivers (8) (table S1).

Our results from the most parsimonious GAMs agreed with previous findings, showing strong seasonal effects, both positive and negative, of all considered variables on meerkat vital rates (8, 12). For instance, monthly growth in adult helpers was generally highest (P < 0.01) in the rainy season when food resources were most abundant (Fig. 1A). Interactions between population density and rainfall and temperature deviation mediated these seasonal effects. High density and rainfall amount increased helpers’ growth (Fig. 1A). High density under low temperatures, however, decreased the probability of dominant reproduction at the onset of the dry season (April to May) (Fig. 1B). Overall, a higher body mass had a consistently positive effect on vital rates, strongly mediating environmental impacts (8). Higher rainfall amounts also positively affected vital rates (12, 16), whereas the effect of higher temperatures was positive only in dry and cool seasons (April to August) (table S2 and fig. S18). Under most environmental conditions, population density showed a negative effect on survival and on emigration of adult helpers (12). In dominants, however, survival increased with density at the onset of the breeding season but decreased at high densities in the nonbreeding season, when resources were scarce. The highest reproductive output was achieved at intermediate densities (table S2). Population density therefore both amplified and compensated for negative responses to the environment, depending on the vital rate affected (15) (Fig. 1).

Fig. 1 Interactive effects of seasonality, population density, and rainfall and temperature variation on meerkat vital rates and population dynamics.

(A) Line colors and different plot panels depict predictions of log(body mass) (grams) in the next month reproduction rates given average log(mass) in the current month using maximum (+) and minimum (–) observed temperature and rainfall deviation from seasonal averages, respectively. Shaded areas show 95% prediction intervals. Plot backgrounds highlight the rainy (October to April; darker color) and dry (May to September; lighter color) seasons. (B) Average (lines) ± 95% bootstrap confidence intervals (shaded areas) projected population densities (individuals per square kilometer) obtained from modeling the relationships described in (A).

We next assessed how changes in climate variation across seasons might affect population dynamics through direct effects on vital rates and through effects mediated by body mass and density. We first used the most parsimonious GAMs of vital rates to assemble a density-dependent, environment-specific, mass-stage–classified integral projection model for each study month and year (17) (supplementary materials). This method enabled us to project trait and population dynamics simultaneously in discrete 1-month intervals, which could then be integrated over the entire year (fig. S2). Our population model assumed that past conditions affecting meerkats were captured by the current mass distribution and were propagated through time, allowing us to assess trait-mediated population processes (10). These assumptions were justified, as we could not detect life history trade-offs (i.e., fitness decreases due to high growth or reproductive effort) (supplementary text) (16).

The population model replicated observed seasonal population and mass fluctuations (1997–2016) with high accuracy (for total population density: Pearson’s correlation coefficient r = 0.74, P < 0.001) (Fig. 1B) and did not extrapolate beyond biologically realistic values of masses (figs. S3 and S4). Model projections were also robust, showing low uncertainty due to parameter estimates (tables S3 and S4). We then used this model to project population dynamics for 50 years on the basis of 12 scenarios of changes in rainfall (drier) and temperature (hotter) extremes. These scenarios were derived from four projections of greenhouse gas emissions, showing a plausible range of season-specific climate change in the Kalahari (figs. S5 and S6).

Projected changes in climatic patterning—particularly increases in extreme events—escalated the risk of population quasi-extinction (<20 individuals or <5 dominants) up to 55% by 2066 (Fig. 2, A and B). Simultaneous changes in rainfall and temperature led to a higher extinction risk (figs. S7 and S8), highlighting potentially detrimental compound climate change effects on the viability of arid environment species (18). Density was important in regulating persistence (Fig. 2C); projections that kept populations at either low or high densities resulted in significantly higher probability of quasi-extinction (to 100%; P < 0.01) and decreased time to extinction, on average, by 20 years (SE = 10; P < 0.01). At low population densities, when the benefits of cooperation decrease and the environment favors the settlement of new groups (12, 19), emigration rates were relatively high. Consequently, both the overall number of helpers and the degree of reproductive success were reduced (20) (fig. S14). At high densities, when resource competition and intergroup conflicts increase, particularly in times of scarce rainfall (table S2), reproduction and survival rates were reduced, increasing the risk of extinction (21). Maintaining densities at intermediate levels, when the benefits of cooperation are largest, eliminated extinction risk (Fig. 2C), similar to patterns seen in other social species (20).

Fig. 2 Projected density and viability of meerkats under changes in rainfall and temperature variation.

(A) Average (lines) population density (individuals per square kilometer) ± 95% projection interval (shaded areas) based on 280 and 120 simulations in which the population persisted or went extinct, respectively. The plot background highlights the rainy (October to April) and dry (May to September) seasons. (B) Cumulative probabilities of quasi-extinction under four scenarios of greenhouse gas Representative Concentration Pathways (RCPs). Shaded areas show 95% projection intervals among sequential versus stochastic projections of climate. (C) Effects on extinction probability of imposing constant low, intermediate, and high densities during projections.

As well as influencing viability, projected changes in climate substantially altered population structure due to changes in the distribution of mass. In projections where climatic extremes became more likely but the population persisted, the proportion of nonpregnant dominant females increased (compared with baseline simulations) at the end of the breeding season (April to July), as did the proportion of pregnant dominant females and that of females with dependent litters at the onset of the breeding season (August to November) (Fig. 3A). These changes occurred as a result of mass increases of helpers (from 600 up to 670 g) and dominants in the dry season (May to September) (Fig. 3B), which led to higher reproductive output in subsequent months and compensated for the loss of adult helpers from the population (Fig. 3A). These changes in the proportion of helpers and dominants indicate a reduction in average group size under climate change (21). In contrast, in projections where the population collapsed, the proportion of nonpregnant helpers dropped to 20% (from 40% in baseline simulations), as helpers emigrated more readily under low population densities (Fig. 3). Density was further reduced by a lower reproductive output that resulted from decreased masses of reproductive females at the onset of the breeding season (Fig. 3B), providing a potential early warning signal for an impending population crash (22). Therefore, for social species, and cooperative breeders in particular, density feedbacks (23) may exacerbate a breakdown of social groups under climate change (21).

Fig. 3 Projected changes in population structure and trait dynamics for meerkats under climate change.

Seasonal distribution of proportion of different life history stages (A) and average log(body mass) (grams) within each stage (B). Box plots show the distribution of values across years and simulations grouped on the basis of different simulations of future rainfall and temperature variation. The stages are pups, juveniles (Juv), and subadults (SubA), as well as nonpregnant (NP), pregnant (P), and litter-weaning (L) helpers (H) and dominants (D). Plot backgrounds highlight the rainy (October to April) and dry (May to September) seasons.

To explore the seasonal demographic mechanisms behind projected quasi-extinctions, we assessed how perturbations of our population model affected population growth. We replicated simulations that had resulted in extinction but maintained the effects of rainfall and temperature deviations at observed past values (1997–2016) for either specific vital rates or simultaneously for all vital rates, accounting for covariation (fig. S11). We maintained the effects either for the entire year or across four seasons: rainy (October to April), dry (May to September), hot (November to February), and cool (June to August). Our results demonstrate highly season-specific contributions of demography to extinction. The time to extinction can be slowed by 4 years, on average (SE = 2.1), if the reproduction of dominants is not affected by climate change in the rainy season (Fig. 4 and fig. S10). On the other hand, increasing warming can potentially decrease emigration, particularly in the dry and cool seasons, despite potential mass gains of prospective emigrants. Maintaining emigration rates under the observed rainfall and temperature regimes therefore leads to faster extinction (Fig. 4). These results agreed with analytical perturbations of population growth, which showed high but seasonally variable relative contributions of helper and dominant vital rates to population growth (figs. S12 and S13).

Fig. 4 Seasonal differences in probability of quasi-extinction under climate-change simulations.

Averages (points) ± 1 SE (error bars) changes in the time (years) to extinction across 120 simulations when a given vital rate is affected by observed (1997–2016) rather than projected (2017–2066) rainfall and temperature variation. The observed variation was maintained over an entire year or specifically for the rainy (October to April), dry (May to September), hot (November to February), or cool (June to August) seasons.

This work emphasizes that assessing the dynamics of seasonal influences on phenotypic traits may be key to understanding how changes in demography and population structure can ensure population persistence when rainfall and temperature patterns change (5, 10). In particular, we show that climate-driven changes in body mass affect vital rates differently in different seasons. These trait-mediated effects can either buffer populations from extinction (15) or exacerbate extinction risk under climate-density interactions for species for which density positively affects vital rates (15, 24). Seasonal, demographic analyses that include phenotypic trait changes are therefore required to gain much-needed information about population responses to global change, such as interactions of climatic components that are increasingly extreme (18, 24, 25).

Supplementary Materials

www.sciencemag.org/content/363/6427/631/suppl/DC1

Materials and Methods

Supplementary Text

Figs. S1 to S18

Tables S1 to S4

References (2657)

R Scripts S1 to S3

Data S1 to S18

References and Notes

Acknowledgments: We are grateful to the many volunteers and field managers, in particular T. Vink, of the Kalahari Meerkat Project (KMP) for their contribution to data collection; and to M. Manser for her contribution to the organization of the KMP. Data collection was supported logistically by the Mammal Research Institute of the University of Pretoria. We also thank the Trustees of the Kalahari Research Centre and the Directors of the Kalahari Meerkat Project for access to the data used in this paper, D. Gaynor for access to prior analysis of climate effects on meerkat dynamics and for discussion, and S. Albon for comments on the analysis. Funding: Data used in this paper were collected under ERC Advanced Grants (294494 and 742808) to T.C.B. Analysis of data was funded by an ERC Starting Grant (33785) and a Swiss National Science Foundation Grant (31003A_182286) to A.O. and an ERC Advanced Grant (742808) to T.C.-B. Author contributions: T.C.-B. led the long-term study and data collection; M.P. and A.O. conceived the ideas for the paper and its structure; M.P., A.O., N.M., and G.C. designed the analyses; M.P. conducted the analyses and wrote the manuscript; and all authors discussed the results and commented on the manuscript. Competing interests: The authors declare no competing interests. Data and materials availability: The parameters and datasets generated and analyzed during our study, which are required to build and project meerkat population dynamics, are freely available in the GitHub repository: https://github.com/MariaPaniw/meerkats. All analyses in this study were performed using the freely available statistical software environment R. All R scripts necessary to run the analyses are available at the GitHub site noted above.
View Abstract

Stay Connected to Science

Navigate This Article