Forecasting species range dynamics with process‐explicit models: matching methods to applications

Knowing where species occur is fundamental to many ecological and environmental applications. Species distribution models (SDMs) are typically based on correlations between species occurrence data and environmental predictors, with ecological processes captured only implicitly. However, there is a growing interest in approaches that explicitly model processes such as physiology, dispersal, demography and biotic interactions. These models are believed to offer more robust predictions, particularly when extrapolating to novel conditions. Many process–explicit approaches are now available, but it is not clear how we can best draw on this expanded modelling toolbox to address ecological problems and inform management decisions. Here, we review a range of process–explicit models to determine their strengths and limitations, as well as their current use. Focusing on four common applications of SDMs – regulatory planning, extinction risk, climate refugia and invasive species – we then explore which models best meet management needs. We identify barriers to more widespread and effective use of process‐explicit models and outline how these might be overcome. As well as technical and data challenges, there is a pressing need for more thorough evaluation of model predictions to guide investment in method development and ensure the promise of these new approaches is fully realised.     

Unpacking the mechanisms captured by a correlative species distribution model to improve predictions of climate refugia

Climate refugia are regions that animals can retreat to, persist in and potentially then expand from under changing environmental conditions. Most forecasts of climate change refugia for species are based on correlative species distribution models (SDMs) using long‐term climate averages, projected to future climate scenarios. Limitations of such methods include the need to extrapolate into novel environments and uncertainty regarding the extent to which proximate variables included in the model capture processes driving distribution limits (and thus can be assumed to provide reliable predictions under new conditions). These limitations are well documented; however, their impact on the quality of climate refugia predictions is difficult to quantify. Here, we develop a detailed bioenergetics model for the koala. It indicates that range limits are driven by heat‐induced water stress, with the timing of rainfall and heat waves limiting the koala in the warmer parts of its range. We compare refugia predictions from the bioenergetics model with predictions from a suite of competing correlative SDMs under a range of future climate scenarios. SDMs were fitted using combinations of long‐term climate and weather extremes variables, to test how well each set of predictions captures the knowledge embedded in the bioenergetics model. Correlative models produced broadly similar predictions to the bioenergetics model across much of the species' current range – with SDMs that included weather extremes showing highest congruence. However, predictions in some regions diverged significantly when projecting to future climates due to the breakdown in correlation between climate variables. We provide unique insight into the mechanisms driving koala distribution and illustrate the importance of subtle relationships between the timing of weather events, particularly rain relative to hot‐spells, in driving species–climate relationships and distributions. By unpacking the mechanisms captured by correlative SDMs, we can increase our certainty in forecasts of climate change impacts on species.     

Relationship between heatwave‐induced forest die‐off and climatic suitability in multiple tree species

In recent decades, many forest die‐off events have been reported in relation to climate‐change‐induced episodes, such as droughts and heat waves. To understand how these extreme climatic events induce forest die‐off, it is important to find a tool to standardize the climatic conditions experienced by different populations during a specific climatic event, taking into account the historic climatic conditions of the site where these populations live (bioclimatic niche). In this study, we used estimates of climatic suitability calculated from species distribution models (SDMs) for such purpose. We studied forest die‐off across France during the 2003 heatwave that affected Western Europe, using 2,943 forest inventory plots dominated by 14 single tree species. Die‐off severity was estimated by Normalized Difference Vegetation Index (NDVI) loss using Moderate‐resolution Imaging Spectroradiometer remote sensor imagery. Climatic suitability at the local level during the historical 1979–2002 period (HCS), the episode time (2003; ECS) and suitability deviance during the historical period (HCS‐SD) were calculated for each species by means of boosted regression tree models using the CHELSA climate database and occurrences extracted from European forest inventories. Low HCS‐SD and high mean annual temperature explained the overall regional pattern of vulnerability to die‐off across different monospecific forests. The combination of high historical and low episode climatic suitability also contributed significantly to overall forest die‐off. Furthermore, we observed different species‐specific relationships between die‐off vulnerability and climatic suitability: Sub‐Mediterranean and Mediterranean species tended to be vulnerable in historically more suitable localities (high HCS), whereas Euro‐Siberian species presented greater vulnerability when the hot drought episode was more intense. We demonstrated that at regional scale, past climatic legacy plays an important role in explaining NDVI loss during the episode. Moreover, we demonstrated that SDMs‐derived indexes, such as HCS, ECS and HCS‐SD, could constitute a tool for standardizing the ways that populations and species experience climatic variability across time and space.     

A niche for biology in species distribution models

Why species are found where they are is a central question in biogeography. The most widely used tool for understanding the controls on distribution is species distribution modelling. Species distribution modelling is now a well‐established method in both the theoretical and applied ecological literature. In this special issue we examine the current state of the art in species distribution modelling and explore avenues for including more biological processes in such models. In particular we focus on physiological, demographic, dispersal, competitive and ecological‐modulation processes. This overview highlights opportunities for new species distribution model concepts and developments, as well as a statistical agenda for implementing such models.     

A problem with variable selection in a comparison of correlative and process‐based species distribution models: Comments on Higgins et al., 2020

Comments are presented on an article published in October 2020 in Ecology and Evolution (“Predictive ability of a process‐based versus a correlative species distribution model”) by Higgins et al. This analyzed natural distributions of Australian eucalypt and acacia species and assessed the adventive range of selected species outside Australia. Unfortunately, inappropriate variables were used with the MaxEnt species distribution model outside Australia, so that large climatically suitable areas in the Northern Hemisphere were not identified. Examples from a previous analysis and from the use of the freely available spatial portal of the Atlas of Living Australia are provided to illustrate how the problem can be overcome. The comparison of methods described in the Higgins et al. paper is worthwhile, and it is hoped that the authors will be able to repeat their analyses using appropriate variables with the correlative model.     

Multi‐model comparison highlights consistency in predicted effect of warming on a semi‐arid shrub

A number of modeling approaches have been developed to predict the impacts of climate change on species distributions, performance, and abundance. The stronger the agreement from models that represent different processes and are based on distinct and independent sources of information, the greater the confidence we can have in their predictions. Evaluating the level of confidence is particularly important when predictions are used to guide conservation or restoration decisions. We used a multi‐model approach to predict climate change impacts on big sagebrush (Artemisia tridentata), the dominant plant species on roughly 43 million hectares in the western United States and a key resource for many endemic wildlife species. To evaluate the climate sensitivity of A. tridentata, we developed four predictive models, two based on empirically derived spatial and temporal relationships, and two that applied mechanistic approaches to simulate sagebrush recruitment and growth. This approach enabled us to produce an aggregate index of climate change vulnerability and uncertainty based on the level of agreement between models. Despite large differences in model structure, predictions of sagebrush response to climate change were largely consistent. Performance, as measured by change in cover, growth, or recruitment, was predicted to decrease at the warmest sites, but increase throughout the cooler portions of sagebrush's range. A sensitivity analysis indicated that sagebrush performance responds more strongly to changes in temperature than precipitation. Most of the uncertainty in model predictions reflected variation among the ecological models, raising questions about the reliability of forecasts based on a single modeling approach. Our results highlight the value of a multi‐model approach in forecasting climate change impacts and uncertainties and should help land managers to maximize the value of conservation investments.     

Process,correlation and parameter fitting in species distribution models: a response to Kriticos et al

In a recent article (Dormann et al., 2012, Journal of Biogeography, 39, 2119–2131), we compared different approaches to species distribution modelling and depicted modelling approaches along an axis from purely ‘correlative’ to ‘forward process‐based’ models. In their correspondence, Kriticos et al. (2013, Journal of Biogeography, doi: 10.1111/j.1365‐2699.2012.02791.x ) challenge this view, claiming that our continuum representation neglects differences among models and does not consider the ability of fitted process‐based models to combine the advantages of both process‐based and correlative modelling approaches. Here we clarify that the continuum view resulted from recognition of the manifold differences between models. We also reinforce the point that the current trend towards combining different modelling approaches may lead not only to the desired combination of the advantages but also to the accumulation of the disadvantages of those approaches. This point has not been made sufficiently clear previously.     

Assessing forest vulnerability to climate warming using a process‐based model of tree growth: bad prospects for rear‐edges

Growth models can be used to assess forest vulnerability to climate warming. If global warming amplifies water deficit in drought‐prone areas, tree populations located at the driest and southernmost distribution limits (rear‐edges) should be particularly threatened. Here, we address these statements by analyzing and projecting growth responses to climate of three major tree species (silver fir, Abies alba; Scots pine, Pinus sylvestris; and mountain pine, Pinus uncinata) in mountainous areas of NE Spain. This region is subjected to Mediterranean continental conditions, it encompasses wide climatic, topographic and environmental gradients, and, more importantly, it includes rear‐edges of the continuous distributions of these tree species. We used tree‐ring width data from a network of 110 forests in combination with the process‐based Vaganov–Shashkin‐Lite growth model and climate–growth analyses to forecast changes in tree growth during the 21st century. Climatic projections were based on four ensembles CO₂ emission scenarios. Warm and dry conditions during the growing season constrain silver fir and Scots pine growth, particularly at the species rear‐edge. By contrast, growth of high‐elevation mountain pine forests is enhanced by climate warming. The emission scenario (RCP 8.5) corresponding to the most pronounced warming (+1.4 to 4.8 °C) forecasted mean growth reductions of ?10.7% and ?16.4% in silver fir and Scots pine, respectively, after 2050. This indicates that rising temperatures could amplify drought stress and thus constrain the growth of silver fir and Scots pine rear‐edge populations growing at xeric sites. Contrastingly, mountain pine growth is expected to increase by +12.5% due to a longer and warmer growing season. The projections of growth reduction in silver fir and Scots pine portend dieback and a contraction of their species distribution areas through potential local extinctions of the most vulnerable driest rear‐edge stands. Our modeling approach provides accessible tools to evaluate forest vulnerability to warmer conditions.     

Geographical patterns in prediction errors of species distribution models

Aim To describe and explain geographical patterns of false absence and false presence prediction errors that occur when describing current plant species ranges with species distribution models. Location Europe. Methods We calibrated species distribution models (generalized linear models) using a set of climatic variables and gridded distribution data for 1065 vascular plant species from the Atlas Florae Europaeae. We used randomly selected subsets for each species with a constant prevalence of 0.5, modelled the distribution 1000 times, calculated weighted averages of the model parameters and used these to predict the current distribution in Europe. Using a threshold of 0.5, we derived presence/absence maps. Comparing observed and modelled species distribution, we calculated the false absence rates, i.e. species wrongly modelled as absent, and the false presence rates, i.e. species wrongly modelled as present, on a 50 × 50 km grid. Subsequently, we related both error rates to species range properties, land use and topographic variability within grid cells by means of simultaneous autoregressive models to correct for spatial autocorrelation. Results Grid‐cell‐specific error rates were not evenly distributed across Europe. The mean false absence rate was 0.16 ± 0.12 (standard deviation) and the mean false presence rate was 0.22 ± 0.13. False absence rates were highest in central Spain, the Alps and parts of south‐eastern Europe, while false presence rates were highest in northern Spain, France, Italy and south‐eastern Europe. False absence rates were high when range edges of species accumulated within a grid cell and when the intensity of human land use was high. False presence rates were positively associated with relative occurrence area and accumulation of range edges. Main conclusions Predictions for various species are not only accompanied by species‐specific but also by grid‐cell‐specific errors. The latter are associated with characteristics of the grid cells but also with range characteristics of occurring species. Uncertainties of predictive species distribution models are not equally distributed in space, and we would recommend accompanying maps of predicted distributions with a graphical representation of predictive performance.     

Predicting the future of forests in the Mediterranean under climate change,with niche‐ and process‐based models: CO2 matters!

Assessing the potential future of current forest stands is a key to design conservation strategies and understanding potential future impacts to ecosystem service supplies. This is particularly true in the Mediterranean basin, where important future climatic changes are expected. Here, we assess and compare two commonly used modeling approaches (niche‐ and process‐based models) to project the future of current stands of three forest species with contrasting distributions, using regionalized climate for continental Spain. Results highlight variability in model ability to estimate current distributions, and the inherent large uncertainty involved in making projections into the future. CO₂ fertilization through projected increased atmospheric CO₂ concentrations is shown to increase forest productivity in the mechanistic process‐based model (despite increased drought stress) by up to three times that of the non‐CO₂ fertilization scenario by the period 2050–2080, which is in stark contrast to projections of reduced habitat suitability from the niche‐based models by the same period. This highlights the importance of introducing aspects of plant biogeochemistry into current niche‐based models for a realistic projection of future species distributions. We conclude that the future of current Mediterranean forest stands is highly uncertain and suggest that a new synergy between niche‐ and process‐based models is urgently needed in order to improve our predictive ability.     

Understanding species distribution in dynamic populations: a new approach using spatio‐temporal point process models

Understanding and predicting a species’ distribution across a landscape is of central importance in ecology, biogeography and conservation biology. However, it presents daunting challenges when populations are highly dynamic (i.e. increasing or decreasing their ranges), particularly for small populations where information about ecology and life history traits is lacking. Currently, many modelling approaches fail to distinguish whether a site is unoccupied because the available habitat is unsuitable or because a species expanding its range has not arrived at the site yet. As a result, habitat that is indeed suitable may appear unsuitable. To overcome some of these limitations, we use a statistical modelling approach based on spatio‐temporal log‐Gaussian Cox processes. These model the spatial distribution of the species across available habitat and how this distribution changes over time, relative to covariates. In addition, the model explicitly accounts for spatio‐temporal dynamics that are unaccounted for by covariates through a spatio‐temporal stochastic process. We illustrate the approach by predicting the distribution of a recently established population of Eurasian cranes Grus grus in England, UK, and estimate the effect of a reintroduction in the range expansion of the population. Our models show that wetland extent and perimeter‐to‐area ratio have a positive and negative effect, respectively, in crane colonisation probability. Moreover, we find that cranes are more likely to colonise areas near already occupied wetlands and that the colonisation process is progressing at a low rate. Finally, the reintroduction of cranes in SW England can be considered a human‐assisted long‐distance dispersal event that has increased the dispersal potential of the species along a longitudinal axis in S England. Spatio‐temporal log‐Gaussian Cox process models offer an excellent opportunity for the study of species where information on life history traits is lacking, since these are represented through the spatio‐temporal dynamics reflected in the model.     

Mutualism influences species distribution predictions for a bromeliad‐breeding anuran under climate change

Ecological niche models, or species distribution models, have been widely used to identify potentially suitable areas for species in future climate change scenarios. However, there are inherent errors to these models due to their inability to evaluate species occurrence influenced by non‐climatic factors. With the intuit to improve the modelling predictions for a bromeliad‐breeding treefrog (Phyllodytes melanomystax, Hylidae), we investigate how the climatic suitability of bromeliads influences the distribution model for the treefrog in the context of baseline and 2050 climate change scenarios. We used point occurrence data on the frog and the bromeliad (Vriesea procera, Bromeliaceae) to generate their predicted distributions based on baseline and 2050 climates. Using a consensus of five algorithms, we compared the accuracy of the models and the geographic predictions for the frog generated from two modelling procedures: (i) a climate‐only model for P. melanomystax and V. procera; and (ii) a climate‐biotic model for P. melanomystax, in which the climatic suitability of the bromeliad was jointly considered with the climatic variables. Both modelling approaches generated strong and similar predictive power for P. melanomystax, yet climate‐biotic modelling generated more concise predictions, particularly for the year 2050. Specifically, because the predicted area of the bromeliad overlaps with the predictions for the treefrog in the baseline climate, both modelling approaches produce reasonable similar predicted areas for the anuran. Alternatively, due to the predicted loss of northern climatically suitable areas for the bromeliad by 2050, only the climate‐biotic models provide evidence that northern populations of P. melanomystax will likely be negatively affected by 2050.     

Application of a process‐based model for predicting the productivity of Eucalyptus nitens bioenergy plantations in Spain

The feasibility of using plantation‐grown biomass to fuel bioenergy plants is in part dependent on the ability to predict the capacity of surrounding forests to maintain a sustainable supply. In this study, the potential productivity of Eucalyptus nitens (Deane and Maiden) Maiden plantations grown for bioenergy in a region of north‐west Spain was quantified using the 3‐PG process‐based model. The model was calibrated using detailed measurements from five permanent sample plots and validated using data from thirty‐five additional permanent sample plots; both sets represented the variability of climate and soils of the region. Plot scale analysis showed that the model was able to reasonably estimate above‐ground biomass and water use when compared with the observed data. Using a representative loam soil characteristic, a spatial analysis was then carried out to predict the potential productivity of E. nitens for bioenergy across a potential area for plantation establishment of 2550 km² and to evaluate different management scenarios related to rotation length and stocking. An increase of only 1.9% in mean annual increment (MAI) of above‐ground biomass (W_AGB) was found between stockings of 3000 and 5000 trees ha^?1; for the lower stocking, MAI of W_AGB increased 4% for rotation lengths between 6 and 8 years. Production was reduced by low summer rainfall and to a lesser extent by high summer and low winter temperatures, and vapour pressure deficit. Above‐ground biomass production was higher by around 12% when average rather than actual climate data were applied. The information from this study can be used to optimize forest management, determine regional relative potential productivity and contribute to decision‐making for bioenergy production from E. nitens plantations in north‐west Spain.     

Climate change and fire effects on a prairie–woodland ecotone: projecting species range shifts with a dynamic global vegetation model

Large shifts in species ranges have been predicted under future climate scenarios based primarily on niche‐based species distribution models. However, the mechanisms that would cause such shifts are uncertain. Natural and anthropogenic fires have shaped the distributions of many plant species, but their effects have seldom been included in future projections of species ranges. Here, we examine how the combination of climate and fire influence historical and future distributions of the ponderosa pine–prairie ecotone at the edge of the Black Hills in South Dakota, USA, as simulated by MC1, a dynamic global vegetation model that includes the effects of fire, climate, and atmospheric CO₂ concentration on vegetation dynamics. For this purpose, we parameterized MC1 for ponderosa pine in the Black Hills, designating the revised model as MC1‐WCNP. Results show that fire frequency, as affected by humidity and temperature, is central to the simulation of historical prairies in the warmer lowlands versus woodlands in the cooler, moister highlands. Based on three downscaled general circulation model climate projections for the 21st century, we simulate greater frequencies of natural fire throughout the area due to substantial warming and, for two of the climate projections, lower relative humidity. However, established ponderosa pine forests are relatively fire resistant, and areas that were initially wooded remained so over the 21st century for most of our future climate x fire management scenarios. This result contrasts with projections for ponderosa pine based on climatic niches, which suggest that its suitable habitat in the Black Hills will be greatly diminished by the middle of the 21st century. We hypothesize that the differences between the future predictions from these two approaches are due in part to the inclusion of fire effects in MC1, and we highlight the importance of accounting for fire as managed by humans in assessing both historical species distributions and future climate change effects.     

Integrating mechanistic and empirical model projections to assess climate impacts on tree species distributions in northwestern North America

Empirical and mechanistic models have both been used to assess the potential impacts of climate change on species distributions, and each modeling approach has its strengths and weaknesses. Here, we demonstrate an approach to projecting climate‐driven changes in species distributions that draws on both empirical and mechanistic models. We combined projections from a dynamic global vegetation model (DGVM) that simulates the distributions of biomes based on basic plant functional types with projections from empirical climatic niche models for six tree species in northwestern North America. These integrated model outputs incorporate important biological processes, such as competition, physiological responses of plants to changes in atmospheric CO₂ concentrations, and fire, as well as what are likely to be species‐specific climatic constraints. We compared the integrated projections to projections from the empirical climatic niche models alone. Overall, our integrated model outputs projected a greater climate‐driven loss of potentially suitable environmental space than did the empirical climatic niche model outputs alone for the majority of modeled species. Our results also show that refining species distributions with DGVM outputs had large effects on the geographic locations of suitable habitat. We demonstrate one approach to integrating the outputs of mechanistic and empirical niche models to produce bioclimatic projections. But perhaps more importantly, our study reveals the potential for empirical climatic niche models to over‐predict suitable environmental space under future climatic conditions.     

Estimating carbon fixation of plant organs for afforestation monitoring using a process‐based ecosystem model and ecophysiological parameter optimization

Afforestation projects for mitigating CO₂ emissions require to monitor the carbon fixation and plant growth as key indicators. We proposed a monitoring method for predicting carbon fixation in afforestation projects, combining a process‐based ecosystem model and field data and addressed the uncertainty of predicted carbon fixation and ecophysiological characteristics with plant growth. Carbon pools were simulated using the Biome‐BGC model tuned by parameter optimization using measured carbon density of biomass pools on an 11‐year‐old Eucommia ulmoides plantation on Loess Plateau, China. The allocation parameters fine root carbon to leaf carbon (FRC:LC) and stem carbon to leaf carbon (SC:LC), along with specific leaf area (SLA) and maximum stomatal conductance (g_smax) strongly affected aboveground woody (AC) and leaf carbon (LC) density in sensitivity analysis and were selected as adjusting parameters. We assessed the uncertainty of carbon fixation and plant growth predictions by modeling three growth phases with corresponding parameters: (i) before afforestation using default parameters, (ii) early monitoring using parameters optimized with data from years 1 to 5, and (iii) updated monitoring at year 11 using parameters optimized with 11‐year data. The predicted carbon fixation and optimized parameters differed in the three phases. Overall, 30‐year average carbon fixation rate in plantation (AC, LC, belowground woody parts and soil pools) was ranged 0.14–0.35 kg‐C m^?2 y^?1 in simulations using parameters of phases (i)–(iii). Updating parameters by periodic field surveys reduced the uncertainty and revealed changes in ecophysiological characteristics with plant growth. This monitoring method should support management of afforestation projects by carbon fixation estimation adapting to observation gap, noncommon species and variable growing conditions such as climate change, land use change.