Conservation management and sustainable harvest quotas are sensitive to choice of climate modelling approach for two marine gastropods

Aim

To establish the robustness of two alternative methods for predicting the future ranges and abundances for two wild‐harvested abalone species (Haliotis rubra Donovan 1808 and H. laevigata Leach 1814): single atmosphere–ocean general circulation model (GCM) or ensemble‐averaged GCM forecasts.

Location

South Australia.

Methods

We assessed the ability of 20 GCMs to simulate observed seasonal sea surface temperature (SST) between 1980–1999, globally, and regionally for the Indian and Pacific Oceans south of the Equator. We used model rankings to characterize a set of representative climate futures, using three different‐sized GCM ensembles and two individual GCMs (the Parallel Climate Model and the Community Climate System Model, version 3.0). Ecological niche models were then coupled to physiological information to compare forecast changes in area of occupancy, population size and harvest area based on forecasts using the various GCM selection methods, as well as different greenhouse gas emission scenarios and climate sensitivities.

Results

We show that: (1) the skill with which climate models reproduce recent SST records varies considerably amongst GCMs, with multimodel ensemble averages showing closer agreement to observations than single models; (2) choice of GCM, and the decision on whether or not to use ensemble‐averaged climate forecasts, can strongly influence spatiotemporal predictions of range, abundance and fishing potential; and (3) comparable hindcasting skill does not necessarily guarantee that GCM forecasts and ecological and evolutionary responses to these forecast changes, will be similar amongst closely ranked models.

Conclusion

By averaging across an ensemble of seven highly ranked skilful GCMs, inherent uncertainties stemming from GCM differences are incorporated into forecasts of change in species range, abundance and sustainable fishing area. Our results highlight the need to make informed and explicit decisions on GCM choice, model sensitivity and emission scenarios when exploring conservation options for marine species and the sustainability of future harvests using ecological niche models.

     

Back to the future: using historical climate variation to project near‐term shifts in habitat suitable for coast redwood

Studies that model the effect of climate change on terrestrial ecosystems often use climate projections from downscaled global climate models (GCMs). These simulations are generally too coarse to capture patterns of fine‐scale climate variation, such as the sharp coastal energy and moisture gradients associated with wind‐driven upwelling of cold water. Coastal upwelling may limit future increases in coastal temperatures, compromising GCMs’ ability to provide realistic scenarios of future climate in these coastal ecosystems. Taking advantage of naturally occurring variability in the high‐resolution historic climatic record, we developed multiple fine‐scale scenarios of California climate that maintain coherent relationships between regional climate and coastal upwelling. We compared these scenarios against coarse resolution GCM projections at a regional scale to evaluate their temporal equivalency. We used these historically based scenarios to estimate potential suitable habitat for coast redwood (Sequoia sempervirens D. Don) under ‘normal’ combinations of temperature and precipitation, and under anomalous combinations representative of potential future climates. We found that a scenario of warmer temperature with historically normal precipitation is equivalent to climate projected by GCMs for California by 2020–2030 and that under these conditions, climatically suitable habitat for coast redwood significantly contracts at the southern end of its current range. Our results suggest that historical climate data provide a high‐resolution alternative to downscaled GCM outputs for near‐term ecological forecasts. This method may be particularly useful in other regions where local climate is strongly influenced by ocean–atmosphere dynamics that are not represented by coarse‐scale GCMs.     

Strengthening forecasts of climate change impacts with multi‐model ensemble averaged projections using MAGICC/SCENGEN 5.3

Climate output from general circulation models (GCMs) is being used with increasing frequency to explore potential climate change impacts on species’ distributional range shifts and extinction probability. However, different GCMs do not perform equally well in their ability to hindcast the key climatic factors that potentially influence species distributions. Previous research has demonstrated that multi‐model ensemble forecasts perform better than any single GCM in simulating observed conditions at a global scale. MAGICC/SCENGEN 5.3 is a freeware climate model ‘emulator’ that generates multi‐model ensemble forecasts, conditional on regional and/or global performance, for up to twenty GCMs. In combination with a new application ‘M/SGridder’, this software can be used to produce down‐scaled ensemble forecasts, which minimize climate‐model‐related uncertainty, for a range of ecological problems.     

Uncertainties in the projection of species distributions related to general circulation models

Ecological Niche Models (ENMs) are increasingly used by ecologists to project species potential future distribution. However, the application of such models may be challenging, and some caveats have already been identified. While studies have generally shown that projections may be sensitive to the ENM applied or the emission scenario, to name just a few, the sensitivity of ENM‐based scenarios to General Circulation Models (GCMs) has been often underappreciated. Here, using a multi‐GCM and multi‐emission scenario approach, we evaluated the variability in projected distributions under future climate conditions. We modeled the ecological realized niche (sensu Hutchinson) and predicted the baseline distribution of species with contrasting spatial patterns and representative of two major functional groups of European trees: the dwarf birch and the sweet chestnut. Their future distributions were then projected onto future climatic conditions derived from seven GCMs and four emissions scenarios using the new Representative Concentration Pathways (RCPs) developed for the Intergovernmental Panel on Climate Change (IPCC) AR5 report. Uncertainties arising from GCMs and those resulting from emissions scenarios were quantified and compared. Our study reveals that scenarios of future species distribution exhibit broad differences, depending not only on emissions scenarios but also on GCMs. We found that the between‐GCM variability was greater than the between‐RCP variability for the next decades and both types of variability reached a similar level at the end of this century. Our result highlights that a combined multi‐GCM and multi‐RCP approach is needed to better consider potential trajectories and uncertainties in future species distributions. In all cases, between‐GCM variability increases with the level of warming, and if nothing is done to alleviate global warming, future species spatial distribution may become more and more difficult to anticipate. When future species spatial distributions are examined, we propose to use a large number of GCMs and RCPs to better anticipate potential trajectories and quantify uncertainties.     

sars: an R package for fitting,evaluating and comparing species–area relationship models

The species–area relationship (SAR) constitutes one of the most general ecological patterns globally. A number of different SAR models have been proposed. Recent work has shown that no single model universally provides the best fit to empirical SAR datasets: multiple models may be of practical and theoretical interest. However, there are no software packages available that a) allow users to fit the full range of published SAR models, or b) provide functions to undertake a range of additional SAR‐related analyses. To address these needs, we have developed the R package ‘sars’ that provides a wide variety of SAR‐related functionality. The package provides functions to: a) fit 20 SAR models using non‐linear and linear regression, b) calculate multi‐model averaged curves using various information criteria, and c) generate confidence intervals using bootstrapping. Plotting functions allow users to depict and scrutinize the fits of individual models and multi‐model averaged curves. The package also provides additional SAR functionality, including functions to fit, plot and evaluate the random placement model using a species–sites abundance matrix, and to fit the general dynamic model of oceanic island biogeography. The ‘sars’ R package will aid future SAR research by providing a comprehensive set of simple to use tools that enable in‐depth exploration of SARs and SAR‐related patterns. The package has been designed to allow other researchers to add new functions and models in the future and thus the package represents a resource for future SAR work that can be built on and expanded by workers in the field.     

Climate and host plant availability impact the future distribution of the bean leaf beetle (Cerotoma trifurcata)

The bean leaf beetle, Cerotoma trifurcata, has become a major pest of soybean throughout its North American range. With a changing climate, there is the potential for this pest to further expand its distribution and become an increasingly severe pest in certain regions. To examine this possibility, we developed bioclimatic envelope models for both the bean leaf beetle, and its most important agronomic host plant, soybean (Glycine max). These two models were combined to examine the potential future pest status of the beetle using climate change projections from multiple general circulation models (GCMs) and climate change scenarios. Despite the broad tolerances of soybean, incorporation of host plant availability substantially decreased the suitable and favourable areas for the bean leaf beetle as compared to an evaluation based solely on the climate envelope of the beetle, demonstrating the importance of incorporating biotic interactions in these predictions. The use of multiple GCM–scenario combinations also revealed differences in predictions depending on the choice of GCM, with scenario choice having less of an impact. While the Norwegian model predicted little northward expansion of the beetle from its current northern range limit of southern Ontario and overall decreases in suitable and favourable areas over time, the Canadian and Russian models predict that much of Ontario and Quebec will become suitable for the beetle in the future, as well as Manitoba under the Russian model. The Russian model also predicts expansion of the suitable and favourable areas for the beetle over time. Two predictions that do not depend on our choice of GCM include a decrease in suitability of the Mississippi Delta region and continued favourability of the southeastern United States.     

On the Applicability of Temperature and Precipitation Data from CMIP3 for China

Global Circulation Models (GCMs) contributed to the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (AR4) and are widely used in global change research. This paper assesses the performance of the AR4 GCMs in simulating precipitation and temperature in China from 1960 to 1999 by comparison with observed data, using system bias (B), root-mean-square error (RMSE), Pearson correlation coefficient (R) and Nash-Sutcliffe model efficiency (E) metrics. Probability density functions (PDFs) are also fitted to the outputs of each model. It is shown that the performance of each GCM varies to different degrees across China. Based on the skill score derived from the four metrics, it is suggested that GCM 15 (ipsl_cm4) and GCM 3 (cccma_cgcm_t63) provide the best representations of temperature and precipitation, respectively, in terms of spatial distribution and trend over 10 years. The results also indicate that users should apply carefully the results of annual precipitation and annual temperature generated by AR4 GCMs in China due to poor performance. At a finer scale, the four metrics are also used to obtain best fit scores for ten river basins covering mainland China. Further research is proposed to improve the simulation accuracy of the AR4 GCMs regarding China.     

Spatial processes and evolutionary models: a critical review

Evolution is a fundamentally population level process in which variation, drift and selection produce both temporal and spatial patterns of change. Statistical model fitting is now commonly used to estimate which kind of evolutionary process best explains patterns of change through time using models like Brownian motion, stabilizing selection (Ornstein–Uhlenbeck) and directional selection on traits measured from stratigraphic sequences or on phylogenetic trees. But these models assume that the traits possessed by a species are homogeneous. Spatial processes such as dispersal, gene flow and geographical range changes can produce patterns of trait evolution that do not fit the expectations of standard models, even when evolution at the local‐population level is governed by drift or a typical OU model of selection. The basic properties of population level processes (variation, drift, selection and population size) are reviewed and the relationship between their spatial and temporal dynamics is discussed. Typical evolutionary models used in palaeontology incorporate the temporal component of these dynamics, but not the spatial. Range expansions and contractions introduce rate variability into drift processes, range expansion under a drift model can drive directional change in trait evolution, and spatial selection gradients can create spatial variation in traits that can produce long‐term directional trends and punctuation events depending on the balance between selection strength, gene flow, extirpation probability and model of speciation. Using computational modelling that spatial processes can create evolutionary outcomes that depart from basic population‐level notions from these standard macroevolutionary models.     

Metacommunity‐scale biodiversity regulation and the self‐organised emergence of macroecological patterns

There exist a number of key macroecological patterns whose ubiquity suggests that the spatio‐temporal structure of ecological communities is governed by some universal mechanisms. The nature of these mechanisms, however, remains poorly understood. Here, we probe spatio‐temporal patterns in species richness and community composition using a simple metacommunity assembly model. Despite making no a priori assumptions regarding biotic spatial structure or the distribution of biomass across species, model metacommunities self‐organise to reproduce well‐documented patterns including characteristic species abundance distributions, range size distributions and species area relations. Also in agreement with observations, species richness in our model attains an equilibrium despite continuous species turnover. Crucially, it is in the neighbourhood of the equilibrium that we observe the emergence of these key macroecological patterns. Biodiversity equilibria in models occur due to the onset of ecological structural instability, a population‐dynamical mechanism. This strongly suggests a causal link between local community processes and macroecological phenomena.     

Relationships between body size and geographical range size among Australian mammals: has human impact distorted macroecological patterns?

Extinction and artificial reduction in the size of geographical ranges of many species have occurred extensively across the globe because of human activities. In particular, Australian mammals have suffered heavily in the last two hundred years, with the highest number of reported cases of mammal extinctions anywhere. In the present study, we investigated the extent to which human impact has affected contemporary macroecological patterns in Australian terrestrial mammals. After examining patterns relating to body size and range size among the contemporary mammal fauna, we removed the effects of the last two hundred years of human impact by exploring patterns in the pre‐European assemblage. This permitted us to determine whether contemporary macroecological patterns are distortions of pre‐European patterns. In contrast to the expected pattern of a significant positive relationship between body size and range size, our results showed no significant association for the complete fauna in both cross‐species and phylogenetic analyses, even when data were corrected for species extinctions and range reductions. Analyses within families and among species with the same dietary strategy revealed three significant positive relationships (Macropodidae, Peramelidae, and herbivores) and one significant negative relationship (insectivores) within the contemporary assemblage that disappeared when the pre‐European assemblage was analysed. A positive relationship also emerged in the pre‐European assemblage for the Vombatidae that was not apparent in the contemporary fauna. Thus, correcting for human impact revealed important distortions among contemporary macroecological relationships that have been brought about by human‐induced range reduction and extinction. These findings not only provide further evidence that the Australian continent presents a unique and valuable opportunity with which to test the generality of macroecological patterns, but they also have important ramifications for the analysis and interpretation of contemporary macroecological datasets.     

Spatial variability in species' potential distributions during the Last Glacial Maximum under different Global Circulation Models: Relevance in evolutionary biology

Ecological niche models (ENM) have been used to reconstruct potential distributions during the Last Glacial Maximum (LGM)—or other time periods—and this use is increasingly common in zoological studies. For this reason, we urgently need understanding factors affecting these predictions. Here, we examine how the use of different Global Circulation Models (GCMs) affects the variability in species' potential distributions during the LGM and how the degree of model extrapolation and its associated uncertainty depends on the GCM used. We develop these issues using two North American shrews, Notiosorex crawfordi and Cryptotis alticola, inhabiting two environmentally different regions. First, we compared paleoclimates in these two regions simulated by three GCMs: Community Climate System Model (CCSM), Model for Interdisciplinary Research on Climate (MIROC), and the Max‐Planck‐Institute für Meteorologie model (MPI). Then, we used maxent to estimate the LGM potential distribution of these two mammals under the three GCMs to assess the spatial variability and extrapolation uncertainty associated with idiosyncrasies of GCM. MIROC estimated noticeably more different climatic conditions than CCSM and MPI in the study areas during the LGM, and its pattern of environmental conditions was distributed differently. The MIROC scenario suggested a remarkable different prediction of potential distribution for both species, being more dramatic for the high mountain shrew, C. alticola. In particular, climatic differences among GCMs resulted in differences in the factors that limit and drive the potential distribution of the species during the LGM. Equally dramatic was the disagreement of extrapolation areas among GCMs. MIROC showed a greater number of pixels where extrapolation is required in both regions. Our findings should be taken into consideration when identifying areas of endemism, dynamic geographic barriers, and glacial refugia. When projecting into alternative scenarios of LGM climate, the idiosyncrasies of each GCM should be explicitly taken into account.     

A reply to the ‘critical evaluation of the Oscillayers methods and dataset’

Brown, Hill & Haywood (2020) offered a critical evaluation of the theory and methods of the Oscillayers approach and attempted to test its utility by reproducing global circulation model (GCM)‐based palaeoclimatic reconstructions of two variables (Bio1 and Bio12) for four different time points (130 ka, 787 ka, 3.2 Ma and 3.3 Ma). They concluded that Oscillayers show poor agreement with independent GCMs and thus do not provide a robust approximation of palaeoclimate throughout the Plio‐Pleistocene. Here, I demonstrate that the authors underestimated the ability of Oscillayers to reproduce independent GCMs by not taking into account inter‐framework differences between the models used to generate the Oscillayers and PaleoClim datasets. However, upon correcting this systematic error, differences in Bio1 and Bio12 between Oscillayers and PaleoClim GCMs are less than ± 1°C or ± 50 mm, on average, in 35.9% (range: 11.8–59.0%) or 46.7% (20.8–66.0%) of values, respectively. Thus, the agreement between Oscillayers and PaleoClim is, on average, c. 1.5–2 times higher than estimated by Brown et al. (2020) and mostly well above the inter‐model agreement between two commonly used GCMs (CCSM and MIROC) for the Last Glacial Maximum. Consequently, I conclude that the Oscillayers approach does provide reasonably robust approximations of palaeoclimate throughout the Plio‐Pleistocene. Some clarifications are given.     

Model selection and information theory in geographical ecology

Aim   Although parameter estimates are not as affected by spatial autocorrelation as Type I errors, the change from classical null hypothesis significance testing to model selection under an information theoretic approach does not completely avoid problems caused by spatial autocorrelation. Here we briefly review the model selection approach based on the Akaike information criterion (AIC) and present a new routine for Spatial Analysis in Macroecology (SAM) software that helps establishing minimum adequate models in the presence of spatial autocorrelation.
Innovation    We illustrate how a model selection approach based on the AIC can be used in geographical data by modelling patterns of mammal species in South America represented in a grid system ( n  = 383) with 2° of resolution, as a function of five environmental explanatory variables, performing an exhaustive search of minimum adequate models considering three regression methods: non-spatial ordinary least squares (OLS), spatial eigenvector mapping and the autoregressive (lagged-response) model. The models selected by spatial methods included a smaller number of explanatory variables than the one selected by OLS, and minimum adequate models contain different explanatory variables, although model averaging revealed a similar rank of explanatory variables.
Main conclusions    We stress that the AIC is sensitive to the presence of spatial autocorrelation, generating unstable and overfitted minimum adequate models to describe macroecological data based on non-spatial OLS regression. Alternative regression techniques provided different minimum adequate models and have different uncertainty levels. Despite this, the averaged model based on Akaike weights generates consistent and robust results across different methods and may be the best approach for understanding of macroecological patterns.     

Challenges in the application of geometric constraint models

Discerning the processes influencing geographical patterns of species richness remains one of the central goals of modern ecology. Traditional approaches to exploring these patterns have focused on environmental and ecological correlates of observed species richness. Recently, some have suggested these approaches suffer from the lack of an appropriate null model that accounts for species ranges being constrained to occur within a bounded domain. Proponents of these null geometric constraint models (GCMs), and the mid-domain effect these models produce, argue their utility in identifying meaningful gradients in species richness. This idea has generated substantial debate. Here we discuss what we believe are the three major challenges in the application of GCMs. First, we argue that there are actually two equally valid null models for the random placement of species ranges within a domain, one of which actually predicts a uniform distribution of species richness. Second, we highlight the numerous decisions that must be made to implement a GCM that lead to marked differences in the predictions of the null model. Finally, we discuss challenges in evaluating the importance of GCMs once they have been implemented.     

On the selection of phylogenetic eigenvectors for ecological analyses

Among the statistical methods available to control for phylogenetic autocorrelation in ecological data, those based on eigenfunction analysis of the phylogenetic distance matrix among the species are becoming increasingly important tools. Here, we evaluate a range of criteria to select eigenvectors extracted from a phylogenetic distance matrix (using phylogenetic eigenvector regression, PVR) that can be used to measure the level of phylogenetic signal in ecological data and to study correlated evolution. We used a principal coordinate analysis to represent the phylogenetic relationships among 209 species of Carnivora by a series of eigenvectors, which were then used to model log‐transformed body size. We first conducted a series of PVRs in which we increased the number of eigenvectors from 1 to 70, following the sequence of their associated eigenvalues. Second, we also investigated three non‐sequential approaches based on the selection of 1) eigenvectors significantly correlated with body size, 2) eigenvectors selected by a standard stepwise algorithm, and 3) the combination of eigenvectors that minimizes the residual phylogenetic autocorrelation. We mapped the mean specific component of body size to evaluate how these selection criteria affect the interpretation of non‐phylogenetic signal in Bergmann's rule. For comparison, the same patterns were analyzed using autoregressive model (ARM) and phylogenetic generalized least‐squares (PGLS). Despite the robustness of PVR to the specific approaches used to select eigenvectors, using a relatively small number of eigenvectors may be insufficient to control phylogenetic autocorrelation, leading to flawed conclusions about patterns and processes. The method that minimizes residual autocorrelation seems to be the best choice according to different criteria. Thus, our analyses show that, when the best criterion is used to control phylogenetic structure, PVR can be a valuable tool for testing hypotheses related to heritability at the species level, phylogenetic niche conservatism and correlated evolution between ecological traits.     

Interspecific geographic range size–body size relationship and the diversification dynamics of Neotropical furnariid birds

Among the earliest macroecological patterns documented, is the range and body size relationship, characterized by a minimum geographic range size imposed by the species’ body size. This boundary for the geographic range size increases linearly with body size and has been proposed to have implications in lineages evolution and conservation. Nevertheless, the macroevolutionary processes involved in the origin of this boundary and its consequences on lineage diversification have been poorly explored. We evaluate the macroevolutionary consequences of the difference (hereafter the distance) between the observed and the minimum range sizes required by the species’ body size, to untangle its role on the diversification of a Neotropical species‐rich bird clade using trait‐dependent diversification models. We show that speciation rate is a positive hump‐shaped function of the distance to the lower boundary. The species with highest and lowest distances to minimum range size had lower speciation rates, while species close to medium distances values had the highest speciation rates. Further, our results suggest that the distance to the minimum range size is a macroevolutionary constraint that affects the diversification process responsible for the origin of this macroecological pattern in a more complex way than previously envisioned.     

Are stacked species distribution models accurate at predicting multiple levels of diversity along a rainfall gradient?

We use observed patterns of species richness and composition of ant communities along a 1000 mm rainfall gradient in northern Australian savanna to assess the accuracy of species richness and turnover predictions derived from stacked species distribution models (S‐SDMs) and constrained by macroecological models (MEMs). We systematically sampled ants at 15 sites at 50 km intervals along the rainfall gradient in 2012 and 2013. Using the observed data, we created MEMs of species richness, composition and turnover. We built distribution models for 135 of the observed species using data from museum collections and online databases. We compared two approaches of stacking SDMs and three modelling algorithms to identify the most accurate way of predicting richness and composition. We then applied the same beta diversity metrics to compare the observed versus predicted patterns. Stacked SDMs consistently over‐predicted local species richness, and there was a mismatch between the observed pattern of richness estimated from the MEM, and the pattern predicted by S‐SDMs. The most accurate richness and turnover predictions occurred when the stacked models were rank‐ordered by their habitat suitability and constrained by the observed MEM richness predictions. In contrast with species richness, the predictions obtained by the MEM of community similarity, composition and turnover matched those predicted by the S‐SDMs. S‐SDMs regulated by MEMs may therefore be a useful tool in predicting compositional patterns despite being unreliable estimators of species richness. Our results highlight that the choice of species distribution model, the stacking method used, and underlying macroecological patterns all influence the accuracy of community assembly predictions derived from S‐SDMS.     

Designing ecological climate change impact assessments to reflect key climatic drivers

Identifying the climatic drivers of an ecological system is a key step in assessing its vulnerability to climate change. The climatic dimensions to which a species or system is most sensitive – such as means or extremes – can guide methodological decisions for projections of ecological impacts and vulnerabilities. However, scientific workflows for combining climate projections with ecological models have received little explicit attention. We review Global Climate Model (GCM) performance along different dimensions of change and compare frameworks for integrating GCM output into ecological models. In systems sensitive to climatological means, it is straightforward to base ecological impact assessments on mean projected changes from several GCMs. Ecological systems sensitive to climatic extremes may benefit from what we term the ‘model space’ approach: a comparison of ecological projections based on simulated climate from historical and future time periods. This approach leverages the experimental framework used in climate modeling, in which historical climate simulations serve as controls for future projections. Moreover, it can capture projected changes in the intensity and frequency of climatic extremes, rather than assuming that future means will determine future extremes. Given the recent emphasis on the ecological impacts of climatic extremes, the strategies we describe will be applicable across species and systems. We also highlight practical considerations for the selection of climate models and data products, emphasizing that the spatial resolution of the climate change signal is generally coarser than the grid cell size of downscaled climate model output. Our review illustrates how an understanding of how climate model outputs are derived and downscaled can improve the selection and application of climatic data used in ecological modeling.     

Climate change and voltinism in Californian insect pest species: sensitivity to location,scenario and climate model choice

Experimental studies of the impact of climatic change are hampered by their inability to consider multiple climate change scenarios and indeed often consider no more than simple climate sensitivity such as a uniform increase in temperature. Modelling efforts offer the ability to consider a much wider range of realistic climate projections and are therefore useful, in particular, for estimating the sensitivity of impact predictions to differences in geographical location, and choice of climate change scenario and climate model projections. In this study, we used well‐established degree‐day models to predict the voltinism of 13 agronomically important pests in California, USA. We ran these models using the projections from three Atmosphere–Ocean Coupled Global Circulation Models (AOCGCMs or GCMs), in conjunction with the SRES scenarios. We ran these for two locations representing northern and southern California. We did this for both the 2050s and 2090s. We used anova to partition the variation in the resulting voltinism among time period, climate change scenario, GCM and geographical location. For these 13 pest species, the choice of climate model explained an average of 42% of the total variation in voltinism, far more than did geographical location (33%), time period (17%) or scenario (1%). The remaining 7% of the variation was explained by various interactions, of which the location by GCM interaction was the strongest (5%). Regardless of these sources of uncertainty, a robust conclusion from our work is that all 13 pest species are likely to experience increases in the number of generations that they complete each year. Such increased voltinism is likely to have significant consequences for crop protection and production.