Life history predicts global population responses to the weather in terrestrial mammals
Curation statements for this article:-
Curated by eLife
Evaluation Summary:
In this manuscript, the authors use long-term population records for 157 mammal species to investigate how populations respond to annual weather anomalies, whether the responses are explained by species' life-history traits, and whether responses vary among species and biomes. They find that populations of shorter-lived species that have larger litter sizes respond more to weather anomalies than longer-lived species with smaller litter sizes. Their results can help understand and predict how different species may respond to climate change, and ultimately, what makes species more sensitive to climate change.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
- Ecology (eLife)
Abstract
With the looming threat of abrupt ecological disruption due to a changing climate, predicting which species are most vulnerable to environmental change is critical. The life-history of a species is an evolved response to its environmental context, and therefore a promising candidate for explaining differences in climate-change responses. However, we need broad empirical assessments from across the world's ecosystems to explore the link between life history and climate-change responses. Here, we use long-term abundance records from 157 species of terrestrial mammals and a two-step Bayesian meta-regression framework to investigate the link between annual weather anomalies, population growth rates, and species-level life history. Overall, we found no directional effect of temperature or precipitation anomalies or variance on annual population growth rates. Furthermore, population responses to weather anomalies were not predicted by phylogenetic covariance, and instead there was more variability in weather responses for populations within a species. Crucially, however, long-lived mammals with smaller litter sizes had smaller absolute population responses to weather anomalies compared with their shorter living counterparts with larger litters. These results highlight the role of species-level life history in driving responses to the environment.
Article activity feed
-
-
Author Response
Reviewer #1 (Public Review):
As we lack empirical data of the response of most species to environmental changes, developing predictive tools based on traits that are easier to access or infer may help us developing better management tools. This is the case even for terrestrial mammals, a rather well studied group but with a large study bias towards temperate Europe and North America. This study uses maximum longevity, litter size and body mass to predict the sign and size of the relationships between annual temperature and precipitation anomalies and population growth rates, using the Living Planet database for times series of abundance and Chelsa for weather anomalies. The authors use a Bayesian framework to relate the size and absolute magnitude of the relationships between detrended population growth rates and …
Author Response
Reviewer #1 (Public Review):
As we lack empirical data of the response of most species to environmental changes, developing predictive tools based on traits that are easier to access or infer may help us developing better management tools. This is the case even for terrestrial mammals, a rather well studied group but with a large study bias towards temperate Europe and North America. This study uses maximum longevity, litter size and body mass to predict the sign and size of the relationships between annual temperature and precipitation anomalies and population growth rates, using the Living Planet database for times series of abundance and Chelsa for weather anomalies. The authors use a Bayesian framework to relate the size and absolute magnitude of the relationships between detrended population growth rates and weather anomalies, the framework accounting for the uncertainty in estimates as well as phylogenetic dependencies. They did not find any systematic effects -- on average the slopes of the relationships were close to 0 -- but the magnitude of the coefficients decreases for species with high maximum longevity and low litter size. Therefore, this study points to possible predictions of the magnitude of the response to weather variability using simple demographic indices such as longevity and litter size. The study has clear limitations that are common to similar "meta-regressions" using publicly available databases, but they are not ignored when discussing the results. One would hope that such limitations would lead to improving the quality of such databases, both in terms of taxonomic and geographic coverage as well as quality of data.
We would like to thank Reviewer 1 for their overall positive feedback and constructive comments on the method and our predictions. We have now included complementary analyses based on high-quality subsets (≥ 20-year records; using life history traits estimated from structured population models), have clarified our set of hypotheses and discussed our results accordingly. Detailed responses are given below.
I would like to challenge the authors in terms of why one would expect relationships of a given sign or magnitude. First with respect to sign of relationships, even for the same species and the same weather parameters, one could expect different signs depending on where the study is done with regards to the climatic niche. If one is close to the warm (or wet) edge, any positive temperature (or precipitation) anomalies would probably have a negative effect, but the reverse would happen when close to the cold or dry edge. There are studies showing such demographic and growth rate variability differences. I find therefore hard to interpret the sign of such weather anomalies and what it tells us about the "effect" of weather variability.
We think that this is an important point to discuss with respect to the importance of within-species variability in population dynamics. Certainly, from the results L203-206 it is clear that populations of the same species can have responses of differing signs. It is also interesting to note that this may be the result of a population’s position in the climatic niche. However, aside from exploring this for species with long-term demographic monitoring across the range, we do not feel that exploring this was in the scope of the current study across species. We agree fully however that adding this perspective to studies of how populations are responding to changing climates is critical. As well as the paper mentioned below by Gaillard et al. (2013), recent work in Plantago lancelota with extensive spatial replication has also begun to reveal these within-range dynamics as a function of latitudinal or climatic gradients (Römer et al. 2021). We have added further discussion of this to the manuscript L330-340. We believe that this point adds to the context of our results highlighting variability within-species. In addition, we have clarified in the introduction that no clear directional responses of populations to weather anomalies was expected among and within species L133-135.
Römer, G., Christiansen, D. M., de Buhr, H., Hylander, K., Jones, O. R., Merinero, S., ... & Dahlgren, J. P. (2021). Drivers of large‐scale spatial demographic variation in a perennial plant. Ecosphere, 12(1), e03356.
Second with regards to the magnitude, it is clear that the maximum growth rate is strongly linked to maximum longevity and litter size -- slow species have a much lower maximum rate of growth than fast species. So, one would expect that variability of population growth rates is larger in fast species than slow species, and therefore the magnitude of their response to environmental variability. Now the question might also be whether weather variability explains a smaller or larger proportion of the variability in population growth rates -- that is, does weather have a relatively larger influence in fast species than slow species? You might have the answer but with the multiple standardizations of the response and predictor variables it is not obvious (that is, when you standardize the response and predictor variables, coefficients are correlations, but this is across species, not for a given population).
The reviewer raises a very interesting and important point on whether the patterns we observe are simply a result of larger variability in growth rates in short-lived species. We have two responses to this point: 1) while there is indeed larger variation in the population growth rates of short-lived species, we believe that this variability is likely an evolved life-history strategy in response to the environment, and thus a key component of patterns we observe, 2) we also feel that our use of models that included annual effects, and state-space models with explicit process-noise terms, account for any confounding effect of this variation.
To address the first point in more detail, we expect that life-histories (and thus population dynamics) are evolved responses to the environment (Stearns, 1992). For ‘fast’ organisms therefore, their intrinsic life-history strategy results in boom-bust population dynamics relative to ‘slow’ species. This is clearly observable in transient or non-asymptotic dynamics, where short-lived species more often have short-term population dynamics with a greater magnitude (Stott et al. 2011). On this point, we therefore argue that this variation in population growth is part of what we are trying to capture. Anomalies in the weather are therefore expected to act more strongly in ‘fast’ species. Following this point and the comments of Reviewer #3, we have now included more explicit hypotheses in terms of life-history L133-144.
For the second point, while we may expect this variability to be the result of dynamics we are trying to capture, this does not preclude other sources of variation in population size confounding the patterns we could observe. For example, hunting pressure may influence both short-term population variability and long-term trends. As a result, we aimed to capture this residual variation using auto-regressive terms for year in our GAMs. While these terms do not explicitly model variability in population growth, they do account for a component of the trend, with variation (error around the trend, which is expected to be larger for fast species), and auto-regressive components of population change. Moreover, we did additional analyses using a state-space modelling approach. In the state-space approach, process noise, which in our case would equate to variability in population growth, is explicitly modelled and accounted for. We therefore believe that our analyses account for residual variability in population growth rates. State space models were also highly correlated with our auto-regressive GAMs, and we can therefore conclude that we do not expect that this variability influences our findings. We have now asserted this in the Methods section L531-535.
Stearns, S.C., 1992. The evolution of life histories (No. 575 S81).
Stott, I., Townley, S. and Hodgson, D.J., 2011. A framework for studying transient dynamics of population projection matrix models. Ecology Letters, 14(9), pp.959-970.
Your analyses remove trends -- that is, climate or other systematic change as opposed to weather anomalies (yearly differences) -- and trends might be the main concerns in terms of conservation. This is made clear in the discussion but perhaps not as much in the introduction where you seem to focus on climate change (the title reflects this well, however, as you mention weather, not climate). This confusion between weather and climate is often made in the literature, when reference is made to climate effects rather than weather effects.
We agree with the reviewer that climate and weather are often conflated in ecological studies. We apologise for this oversight in the introduction, and agree that the narrative and link to weather was not made explicit in the previous version. Following this point and the suggestions of Reviewer #3, we have now restructured large sections of the introduction to improve the clarity of our hypotheses. To address this point, we have now included specific introduction of different components of climate that species populations may respond to, including short-term extreme weather patterns as we explore in this study. Please find this revised section L80-97.
Finally, I would like to see a measure of how good is the prediction you can make using traits. You may have "significant effects" but not helping much in terms of prediction (see PB Adler et al. 2011 in Science, for an example with species richness and productivity).
On this point we disagree with the reviewer. The core of our analysis framework was to examine the predictive performance of models. We do not report any significant effects, and instead use Bayesian inference. Throughout the analysis framework, we used explicit tests of out-of-sample predictive performance with leave-one-out cross validation (Vehtari et al. 2017). This is asserted in the manuscript title and results section when introducing our spatial analysis L188-191. Cross validation was combined with model selection to test the predictive performance of a set of candidate models with respect to base models excluding predictors of interest. This predictive performance framework was not applied to examine the directional effects (question 1), as these models did not contain key predictors. However, model selections using predictive performance were done throughout questions 2 and 3, to explore spatial and life-history effects. We highlight this point in both the results L188-191 and methods sections L608-615. In the case of life-history, we found that relative to the base model, out-of-sample predictions were improved when including univariate life-history traits relative to the base model, and thus life-history traits aid in predicting weather responses.
We did not explore the relative predictive performance of life-history traits with respect to other traits such as dietary specialisation, which have been shown to be important in climate responses (Pacifici et al. 2017). We believe that this would have been out of scope for the purpose of the current study, where we aimed to test specific hypotheses established in life-history theory.
Pacifici, M., Visconti, P., Butchart, S.H., Watson, J.E., Cassola, F.M. and Rondinini, C., 2017. Species’ traits influenced their response to recent climate change. Nature Climate Change, 7(3), pp.205-208.
Vehtari, A., Gelman, A. and Gabry, J., 2017. Practical Bayesian model evaluation using leave-one-out cross-validation and WAIC. Statistics and computing, 27(5), pp.1413-1432.
Reviewer #2 (Public Review):
Jackson et al. present a global analysis of the effects of life history on the response of terrestrial mammal populations to weather, showing that litter size and longevity significantly alter how populations respond to anomalies in temperature and rainfall. The topic is highly interesting, as it has implications for what data we should monitor to make more reliable predictions about species' responses to climatic change, and how we should prioritise which species to conserve by identifying those which might be at greatest risk.
The authors comprehensively validate their results with substantial secondary analyses, and I believe that their assertions are supported by the results presented here. Whilst global scale analyses such as this provide useful generalities, they should be taken as that: an investigation of the general trends observed across large spatial scales, and caution should be taken extrapolating too far away from the species which have been analysed for this study.
We thank the reviewer for their positive feedback, and agree with not drawing too many generalities from our findings. In the first paragraph of the discussion L253-262, we now explicitly refer to the results in the context of mammal population-dynamics/conservation.
Reviewer #3 (Public Review):
In this study, the authors aim to investigate how mammalian species are likely to respond to climate change. To this end, they investigate the effects of weather anomalies on the growth rates of mammalian populations. They use long-term population records for 157 terrestrial mammals from the Living Planet database. They explore three different questions using a two-step modelling approach: (1) whether temperature and precipitation anomalies have significant effects on population growth rates across species; (2) whether responses differ among species and biomes; and (3) whether life-history traits explain species responses to weather anomalies.
The work undertaken in this manuscript is of broad appeal in the field and has the potential to inform conservation. Overall, the methodology is sound and the modelling framework robust; the authors took care to test the robustness of their models by fitting alternative sets of models. The two-step design of this study is interesting and the choice of the study system is relevant for the questions the authors aim to tackle. The authors also paid attention to some important points that are at times overlooked such as resolving taxonomy before running their analyses. I also appreciated the fact that the authors made their code available.
We thank the reviewer for their positive feedback on the manuscript, which highlights many of our key goals with the paper.
I nevertheless think that, in its present form, the main weakness of this manuscript is the clarity of the writing, the framing of the study and the overall flow. I found the manuscript at times a bit difficult to follow. That said, I think there is much scope for the authors to improve it. First, I think the work would benefit from better explanation of the underlying hypotheses. Second, in some places I think the authors go into a lot of details at the expense of clarity. As such, I think the authors should strive to better balance clarity with detailed information (notably in the results and methods; adding summary sentences, for example, could help clarify these sections). Third, I think there is room for improvement in the narrative and the flow of the introduction and the discussion. Finally, I think stronger justifications are sometimes required regarding specific points of the analysis.
I believe that the conclusions of this work are supported by the data and the analyses, and think they are of interest and relevant to the field. However, I think the discussion should highlight the main limitations of the study. In particular, I think the biases in the data should be discussed, and notably whether these biases are expected to affect the results (and if so, in what way).
To conclude, I think that beyond the aforementioned weaknesses of this study, the results and the methods are of interest for the field. I think the modelling framework is applicable to other study systems and relevant to the field as well.
We warmly thank the reviewer for their positive words and thorough constructive feedback. We have extensively re-worked large sections of the manuscript (particularly the discussion and introduction) based on these points, and done our best to address all of them. Generally, we have strived to improve the clarity and succinctness of the manuscript.
-
Evaluation Summary:
In this manuscript, the authors use long-term population records for 157 mammal species to investigate how populations respond to annual weather anomalies, whether the responses are explained by species' life-history traits, and whether responses vary among species and biomes. They find that populations of shorter-lived species that have larger litter sizes respond more to weather anomalies than longer-lived species with smaller litter sizes. Their results can help understand and predict how different species may respond to climate change, and ultimately, what makes species more sensitive to climate change.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share …
Evaluation Summary:
In this manuscript, the authors use long-term population records for 157 mammal species to investigate how populations respond to annual weather anomalies, whether the responses are explained by species' life-history traits, and whether responses vary among species and biomes. They find that populations of shorter-lived species that have larger litter sizes respond more to weather anomalies than longer-lived species with smaller litter sizes. Their results can help understand and predict how different species may respond to climate change, and ultimately, what makes species more sensitive to climate change.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)
-
Reviewer #1 (Public Review):
As we lack empirical data of the response of most species to environmental changes, developing predictive tools based on traits that are easier to access or infer may help us developing better management tools. This is the case even for terrestrial mammals, a rather well studied group but with a large study bias towards temperate Europe and North America. This study uses maximum longevity, litter size and body mass to predict the sign and size of the relationships between annual temperature and precipitation anomalies and population growth rates, using the Living Planet database for times series of abundance and Chelsa for weather anomalies. The authors use a Bayesian framework to relate the size and absolute magnitude of the relationships between detrended population growth rates and weather anomalies, the …
Reviewer #1 (Public Review):
As we lack empirical data of the response of most species to environmental changes, developing predictive tools based on traits that are easier to access or infer may help us developing better management tools. This is the case even for terrestrial mammals, a rather well studied group but with a large study bias towards temperate Europe and North America. This study uses maximum longevity, litter size and body mass to predict the sign and size of the relationships between annual temperature and precipitation anomalies and population growth rates, using the Living Planet database for times series of abundance and Chelsa for weather anomalies. The authors use a Bayesian framework to relate the size and absolute magnitude of the relationships between detrended population growth rates and weather anomalies, the framework accounting for the uncertainty in estimates as well as phylogenetic dependencies. They did not find any systematic effects -- on average the slopes of the relationships were close to 0 -- but the magnitude of the coefficients decreases for species with high maximum longevity and low litter size. Therefore, this study points to possible predictions of the magnitude of the response to weather variability using simple demographic indices such as longevity and litter size. The study has clear limitations that are common to similar "meta-regressions" using publicly available databases, but they are not ignored when discussing the results. One would hope that such limitations would lead to improving the quality of such databases, both in terms of taxonomic and geographic coverage as well as quality of data.
I would like to challenge the authors in terms of why one would expect relationships of a given sign or magnitude. First with respect to sign of relationships, even for the same species and the same weather parameters, one could expect different signs depending on where the study is done with regards to the climatic niche. If one is close to the warm (or wet) edge, any positive temperature (or precipitation) anomalies would probably have a negative effect, but the reverse would happen when close to the cold or dry edge. There are studies showing such demographic and growth rate variability differences. I find therefore hard to interpret the sign of such weather anomalies and what it tells us about the "effect" of weather variability. Second with regards to the magnitude, it is clear that the maximum growth rate is strongly linked to maximum longevity and litter size -- slow species have a much lower maximum rate of growth than fast species. So, one would expect that variability of population growth rates is larger in fast species than slow species, and therefore the magnitude of their response to environmental variability. Now the question might also be whether weather variability explains a smaller or larger proportion of the variability in population growth rates -- that is, does weather have a relatively larger influence in fast species than slow species? You might have the answer but with the multiple standardizations of the response and predictor variables it is not obvious (that is, when you standardize the response and predictor variables, coefficients are correlations, but this is across species, not for a given population).
Your analyses remove trends -- that is, climate or other systematic change as opposed to weather anomalies (yearly differences) -- and trends might be the main concerns in terms of conservation. This is made clear in the discussion but perhaps not as much in the introduction where you seem to focus on climate change (the title reflects this well, however, as you mention weather, not climate). This confusion between weather and climate is often made in the literature, when reference is made to climate effects rather than weather effects.
Finally, I would like to see a measure of how good is the prediction you can make using traits. You may have "significant effects" but not helping much in terms of prediction (see PB Adler et al. 2011 in Science, for an example with species richness and productivity).
-
Reviewer #2 (Public Review):
Jackson et al. present a global analysis of the effects of life history on the response of terrestrial mammal populations to weather, showing that litter size and longevity significantly alter how populations respond to anomalies in temperature and rainfall. The topic is highly interesting, as it has implications for what data we should monitor to make more reliable predictions about species' responses to climatic change, and how we should prioritise which species to conserve by identifying those which might be at greatest risk.
The authors comprehensively validate their results with substantial secondary analyses, and I believe that their assertions are supported by the results presented here. Whilst global scale analyses such as this provide useful generalities, they should be taken as that: an …
Reviewer #2 (Public Review):
Jackson et al. present a global analysis of the effects of life history on the response of terrestrial mammal populations to weather, showing that litter size and longevity significantly alter how populations respond to anomalies in temperature and rainfall. The topic is highly interesting, as it has implications for what data we should monitor to make more reliable predictions about species' responses to climatic change, and how we should prioritise which species to conserve by identifying those which might be at greatest risk.
The authors comprehensively validate their results with substantial secondary analyses, and I believe that their assertions are supported by the results presented here. Whilst global scale analyses such as this provide useful generalities, they should be taken as that: an investigation of the general trends observed across large spatial scales, and caution should be taken extrapolating too far away from the species which have been analysed for this study.
-
Reviewer #3 (Public Review):
In this study, the authors aim to investigate how mammalian species are likely to respond to climate change. To this end, they investigate the effects of weather anomalies on the growth rates of mammalian populations. They use long-term population records for 157 terrestrial mammals from the Living Planet database. They explore three different questions using a two-step modelling approach: (1) whether temperature and precipitation anomalies have significant effects on population growth rates across species; (2) whether responses differ among species and biomes; and (3) whether life-history traits explain species responses to weather anomalies.
The work undertaken in this manuscript is of broad appeal in the field and has the potential to inform conservation. Overall, the methodology is sound and the …
Reviewer #3 (Public Review):
In this study, the authors aim to investigate how mammalian species are likely to respond to climate change. To this end, they investigate the effects of weather anomalies on the growth rates of mammalian populations. They use long-term population records for 157 terrestrial mammals from the Living Planet database. They explore three different questions using a two-step modelling approach: (1) whether temperature and precipitation anomalies have significant effects on population growth rates across species; (2) whether responses differ among species and biomes; and (3) whether life-history traits explain species responses to weather anomalies.
The work undertaken in this manuscript is of broad appeal in the field and has the potential to inform conservation. Overall, the methodology is sound and the modelling framework robust; the authors took care to test the robustness of their models by fitting alternative sets of models. The two-step design of this study is interesting and the choice of the study system is relevant for the questions the authors aim to tackle. The authors also paid attention to some important points that are at times overlooked such as resolving taxonomy before running their analyses. I also appreciated the fact that the authors made their code available.
I nevertheless think that, in its present form, the main weakness of this manuscript is the clarity of the writing, the framing of the study and the overall flow. I found the manuscript at times a bit difficult to follow. That said, I think there is much scope for the authors to improve it. First, I think the work would benefit from better explanation of the underlying hypotheses. Second, in some places I think the authors go into a lot of details at the expense of clarity. As such, I think the authors should strive to better balance clarity with detailed information (notably in the results and methods; adding summary sentences, for example, could help clarify these sections). Third, I think there is room for improvement in the narrative and the flow of the introduction and the discussion. Finally, I think stronger justifications are sometimes required regarding specific points of the analysis.
I believe that the conclusions of this work are supported by the data and the analyses, and think they are of interest and relevant to the field. However, I think the discussion should highlight the main limitations of the study. In particular, I think the biases in the data should be discussed, and notably whether these biases are expected to affect the results (and if so, in what way).
To conclude, I think that beyond the aforementioned weaknesses of this study, the results and the methods are of interest for the field. I think the modelling framework is applicable to other study systems and relevant to the field as well.
-
-