Integrating species distribution models (SDM) in spatially explicit GIS-tools to support nature-sensitive urban planning

Modelling species distributions with remote sensing data: bridging disciplinary perspectives

AimImproving species distribution models (SDMs) and species abundance models (SAMs) of woody shrubs is critical for predicting biodiversity changes in the Arctic, which is experiencing especially high warming rates. Yet, it remains relatively unexplored if SDMs and SAMs can explain local scale patterns. We aim to identify predictor differences for the distribution versus abundance of two widespread Arctic shrub species with high resolution models and to compare validation approaches to assess the models’ predictive abilities.LocationNuup Kangerlua (NK) and Kangerluarsunnguaq (K), two adjacent fjords in Southwest Greenland.MethodsWe conducted two separate field surveys in either fjord to construct high resolution (~90 m) SDMs and SAMs forBetula nanaandSalix glauca, analysing the predictive influences of local scale climate, topography and soil moisture indicators. We then alternatively trained and validated models in either NK or K fjord and compared these results with the common split‐sample validation approach. Finally, we assessed if including local scale biotic predictors improves SAM performance.ResultsTemperature extremes and precipitation best predicted the distributions of both species, whereas insolation and soil moisture indicators best predicted abundances. Compared to split‐sample validation, both SDM and SAM performance was substantially reduced with separate survey validation. Regardless of validation approach, models performed poor to moderately well, and including local scale biotic parameters improved SAM performance.Main conclusionsSubstantial differences in model performance between validation approaches highlight the usefulness of using a separate survey for validating model predictive performance. We discuss various factors that might have caused poor model performance, such as not capturing all relevant predictors or enough local scale heterogeneity in predictor or response variables. We emphasise the need to include predictors relevant at the spatial scale of study, such as local scale biotic interactions, for improved predictions at high spatial resolution.

Climate change’s impact on biodiversity is expected to be significant in the twenty-first century. Climate change will influence ecologically sensitive areas, and managing these changes will be critical. This chapter focuses on the utilization of species distribution models (SDMs) in assessing climate change impacts and its associated variables on species distribution, leading to population shift, migration, and species vulnerability. The review concentrates on several species distribution models (SDMs), its application in various ecosystems and their management, the gaps in the models and modelling techniques, and the challenges in their applicability. To investigate the variables utilized for modelling the future projections of the species distribution, several SDMs were explored. Additionally, the most commonly used SDM parameters are assessed in relation to their data inputs. However, the applicability of this metric is also evaluated for various ecosystems. Further, different SDMs were contrasted regarding how their algorithms utilized the input variables. A conventional review was conducted to examine the applicability of various SDMs in relation to climate change. The assessment concentrates on (1) climate change impacts on biodiversity and related ecologically sensitive hotspots, (2) various SDMs employed for biodiversity management, (4) SDM variables used to account for climate change, (5) the parameters and factors that influence the outcomes of SDMs, (6) how SDMs are applied in different ecosystems, and (7) a comparative of different SDMs currently used with the algorithms and variables they employ. Our research includes the discussion of gaps and challenges with the use of different SDM models, such as the lack of appropriate data and the noninclusion of biotic factors. But it also discusses the future perspectives and direction of research that needs to be conducted. Given our analysis, the use of SDMs will be critical in comprehending the future effect of climate change on species dispersal and distribution in the future; however there is a need to improve the robustness of these models so accurate assessments and predictions can be made.

news and update ISSN 1948‐6596 update Recent advances in using species distributional models to un‐ derstand past distributions The first wave of climate‐driven, paleoecological, species distribution models (SDMs) focused pri‐ marily on testing global climate models (GCMs; e.g., Bartlein et al. 1986). Recently, a second wave of research, using increasingly sophisticated SDMs, expanded the use of paleo‐SDMs into bio‐ geographic questions. This new focus is contribut‐ ing substantially to discussions about outstanding questions in biogeography, such as stability of niches through time and fundamental spatiotem‐ poral controls on geographic patterns of biodiver‐ sity. The value of paleo‐data in biogeographically oriented distributional models gained prominence when Nogues‐Bravo (2009) highlighted the use of SDMs for investigating past distributions of spe‐ cies. Nogues‐Bravo (2009) outlined areas where paleo‐data can help understand and test some of the assumptions underlying SDMs, highlighted methodological considerations when creating and projecting SDMs through time, and recommended a set of best practices when extending SDMs to include paleo‐data. Since then, many new studies have utilized paleo‐SDMs, alone or in combination with other approaches, and the number of papers relying on paleo SDMs has grown rapidly (e.g. 27 studies reviewed by Nogues‐Bravo 2009; 82 stud‐ ies reviewed by Svenning et al. 2011). Two recent review papers (Svenning et al. 2011; Varela et al. 2011) deepen the discussion about paleo‐SDMs by synthesizing this new research, discussing ad‐ vances made in paleo‐biogeography, and outlining the challenges going forward. While there are many points of agreement between the two re‐ views, reading them in parallel makes obvious some areas of disagreement that point to unre‐ solved issues with paleo‐SDMs. SDMs based on modern distribution data are influenced by a number of different uncertain‐ ties, including incomplete distribution informa‐ tion, simplified or uncertain mechanistic relation‐ ships between distribution and climate, and ex‐ trapolation into no‐analog climates, to name a few (e.g. Guisan & Zimmermann 2000; Guisan & Thuiller 2005; Elith et al. 2006; Jimenez‐Valverde et al. 2008). Paleo‐SDMs are subject to the same limitations, but often to a greater extent. Many of these technical issues are discussed in detail by Varela et al. (2011), with the goal of establishing a “robust and scientifically‐based theoretical and methodological framework” for the use of SDMs in paleobiology. They focus largely on the particu‐ lars of paleontological data, and highlight issues of spatial, temporal, taphonomic, and collection bias that should be considered when modeling and interpreting paleo‐SDMs. Varela et al. (2011) ad‐ vocate cautious use of paleo‐SDMs, arguing that paleo‐SDMs are promising but that key gaps in knowledge currently limit their widespread appli‐ cation. The use of SDMs in paleobiology has grown rapidly despite these limitations. Svenning et al. (2011) synthesize the many ways that SDMs have been and could be applied to outstanding ques‐ tions in paleoecology. Their review is broader (82 papers vs. 42 reviewed by Varela et al. 2011) and focuses in particular on the integration of SDMs with genetic data. They outline four primary appli‐ cations of SDMs, including the use of paleo‐SDMs to test hypotheses about glacial refugia, the end‐ Pleistocene megafaunal extinctions, Holocene pa‐ leoecology, and deep‐time biogeography. Sven‐ ning et al. (2011) provide a more optimistic view of the use of SDMs in paleobiology, perhaps be‐ cause they focus more on applications of paleo‐ SDMs and less on potential issues with the under‐ lying data. However, some recommendations by Varela et al. (2011) imply that work highlighted by Svenning et al. (2011) should not be attempted at large spatial or temporal scales at this time. As one example, both groups correctly argue for cau‐ tion when using and interpreting statistically‐ downscaled climate simulations. However, Varela et al. (2011) take a highly conservative approach and argue that statistical downscaling should not frontiers of biogeography 3.4, 2012 — © 2012 the authors; journal compilation © 2012 The International Biogeography Society

Planet Earth hosts an incredible biological diversity. Estimated numbers of species occurring on Earth range from 5 to 11 million eukaryotic species including 400,000-450,000 species of plants. Much of this biodiversity remains poorly known and many species have not yet been named or even been discovered. This is not surprising, as the majority of species is known to be rare and ecosystems are generally dominated by a limited number of common species.
\n
\nTropical rainforests are the most species-rich terrestrial ecosystems on Earth. The general higher level of species richness is often explained by higher levels of energy near the Equator (latitudinal diversity gradient). However, when comparing tropical rainforest biomes, African rainforests host fewer plant species than either South American or Asian ones. The Central African country of Gabon is situated in the Lower Guinean phytochorical region. It is largely covered by what is considered to be the most species-rich lowland rainforest in Africa while the government supports an active conservation program. As such, Gabon is a perfect study area to address that enigmatic question that has triggered many researchers before: “What determines botanical species richness?”.
\n
\nIn the past 2.5 million years, tropical rainforests have experienced 21 cycles of global glaciations. They responded to this by contracting during drier and cooler glacials into larger montane and smaller riverine forest refugia and expanding again during warmer and wetter interglacials. The current rapid global climate change coupled with change of land use poses new threats to the survival of many rainforest species. The limited availability of resources for conservation forces governments and NGOs to set priorities. Unfortunately, for many plant species, lack of data on their distribution hampers well-informed decision making in conservation.
\n
\nSpecies distribution models (SDMs) offer opportunities to bridge at least partly this knowledge gap. SDMs are correlative models that infer the spatial distribution of species using only a limited set of known species occurrence records coupled with high resolution environmental data. SDMs are widely applied to study the past, present and future distribution of species, assess the risk of invasive species, infer patterns of species richness and identify hotspots, as well as to assess the impact of climate change. The currently available methods form a pipeline, with which data are selected and cleaned, models selected, parameterized, evaluated and projected to other areas and climatic scenarios, and biodiversity patterns are computed from these SDMs. In this thesis, SDMs of all Gabonese plant species were generated and patterns of species richness and of weighted endemism were computed (chapter 4 & 5).
\n
\nAlthough this pipeline enables the rapid generation of SDMs and inferring of biodiversity patterns, its effective use is limited by several matters of which three are specifically addressed in this thesis. Not knowing the true distribution limits the opportunities to assess the accuracy of models and assess the impact of assumptions and limitations of SDMs. The use of simulated species has been advocated as a method to systematically assess the impact of specific matters of SDMs (virtual ecologist). Following this approach, in chapter 2, I present a novel method to simulate large numbers of species that each have their own unique niche.
\n
\nOne matter of SDMs that is usually ignored but has been shown to be of great impact on model accuracy is the number of species occurrence records used to train a model. In chapter 2, I quantify the effect of sample size on model accuracy for species of different range size classes. The results show that the minimum number of records required to generate accurate SDMs is not uniform for species of every range size class and that larger sample sizes are required for more widespread species. By applying a uniform minimum number of records, SDMs of narrow-ranged species are incorrectly rejected and SDMs of widespread species are incorrectly accepted. Instead, I recommend to identify and apply the unique minimum numbers of required records for each individual species. The method presented here to identify the minimum number of records for species of particular range size classes is applicable to any species group and study area.
\n
\nThe range size or prevalence is an important plant feature that is used in IUCN Red List classifications. It is commonly computed as the Extent Of Occurrence (EOO) and Area Of Occupancy (AOO). Currently, these metrics are computed using methods based on the spatial distribution of the known species occurrences. In chapter 3, using simulated species again, I show that methods based on the distribution of species occurrences in environmental parameter space clearly outperform those based on spatial data. In this chapter, I present a novel method that estimates the range size of a species as the fraction of raster cells within the minimum convex hull of the species occurrences, when all cells from the study area are plotted in environmental parameter space. This novel method outperforms all ten other assessed methods. Therefore, the current use of EOO and AOO based on spatial data alone for the purpose of IUCN Red List classification should be reconsidered. I recommend to use the novel method presented here to estimate the AOO and to estimate the EOO from the predicted distribution based on a thresholded SDM.
\n
\nIn chapter 4, I apply the currently best possible methods to generate accurate SDMs and estimate the range size of species to the large dataset of Gabonese plant species records. All significant SDMs are used here to assess the unique contribution of narrow-ranged, widespread, and randomly selected species to patterns of species richness and weighted endemism. When range sizes of species are defined based on their full range in tropical Africa, random subsets of species best represent the pattern of species richness, followed by narrow-ranged species. Narrow-ranged species best represent the weighted endemism pattern. Moreover, the results show that the applied criterion of widespread and narrow-ranged is crucial. Too often, range sizes of species are computed on their distribution within a study area defined by political borders. I recommend to use the full range size of species instead. Secondly, the use of widespread species, of which often more data are available, as an indicator of diversity patterns should be reconsidered.
\n
\nThe effect of global climate change on the distribution patterns of Gabonese plant species is assed in chapter 5 using SDMs projected to the year 2085 for two climate change scenarios assuming either full or no dispersal. In Gabon, predicted loss of plant species ranges from 5% assuming full dispersal to 10% assuming no dispersal. However, these numbers are likely to be substantially higher, as for many rare, narrow-ranged species no significant SDMs could be generated. Predicted species turnover is as high as 75% and species-rich areas are predicted to loose many species. The explanatory power of individual future climate anomalies to predicted future species richness patterns is quantified. Species loss is best explained by increased precipitation in the dry season. Species gain and species turnover are correlated with a shift from extreme to average values of annual temperature range.
\n
\nIn the final chapter, the results are placed in a wider scientific context. First, the results on the methodological aspects of SDMs and their implications of the SDM pipeline are discussed. The method presented in this thesis to simulate large numbers of species offers opportunities to systematically investigate other matters of the pipeline, some of which are discussed here. Secondly, the factors that shape the current and predicted future patterns of plant species richness in Gabon are discussed including the location of centres of species richness and of weighted endemism in relation to the hypothesized location of glacial forest refugia. Factors that may contribute to the lower species richness of African rainforests compared with South American and Asian forests are discussed. I conclude by reflecting on the conservation of the Gabonese rainforest and its plant species as well as on the opportunities SDMs offer for this in the wider socio-economic context of a changing world with growing demand for food and other ecosystem services.

A species distribution model for Paracentrotus lividus: predicted projections of habitat suitability

Climate change significantly alters species distributions. Numerous studies project the future distribution of species using Species Distribution models (SDMs), most often using coarse resolutions. Working at coarse resolutions in forest ecosystems fails to capture landscape-level dynamics, spatially explicit processes, and temporally defined events that act at finer resolutions and that can disproportionately affect future outcomes. Dynamic Forest Landscape Models (FLMs) can simulate the survival, growth, and mortality of (stands of) trees over long time periods at small resolutions. However, as they are able to simulate at fine resolutions, study landscapes remain relatively small due to computational constraints. The large amount of feedbacks between biodiversity, forest, and ecosystem processes cannot completely be captured by FLMs or SDMs alone. Integrating SDMs with FLMs enables a more detailed understanding of the impact of perturbations on forest landscapes and their biodiversity. Several studies have used this approach at landscape scales, using fine resolutions. Yet, many scientific questions in the fields of biogeography, macroecology, conservation management, among others, require a focus on both large scales and fine resolutions. Here, drawn from literature and experience, we provide our perspective on the most important challenges that need to be overcome to use integrated frameworks at spatial scales larger than the landscape and at fine resolutions. Future research should prioritize these challenges to better understand drivers of species distributions in forest ecosystems and effectively design conservation strategies under the influence of changing climates on spatially and temporally explicit processes. We further discuss possibilities to address these challenges.

There is an acknowledged need to combine species distribution and macro-ecological models with phylogenetic information, particularly when biogeographic research incorporates multiple species, explores phenotypic traits, or is spatially dynamic. Our aim is to present a new approach to multi-species joint modeling that applies spatially explicit phylogenetic regression to simultaneously predict species occurrence probability and the geographic distribution of interspecific continuous morphological traits. We developed a multi-tiered Bayesian geostatistical model that incorporates a species phylogeny, morphometric traits, and environmental variables to jointly estimate traits and geographic distributions for six species of South American leaf-eared mice (genus: Phyllotis). Covariates are included with the model to control for genetic relatedness, specimen age, specimen sex, and repeated measures errors. To help gauge model performance, we compared our approach to predictions made using several other species distribution modeling applications. Our integrated modeling framework demonstrated improved accuracy over alternative species distribution modeling techniques as judged by model sensitivity, specificity, and the true skill statistic. The inclusion of trait-based covariates and model terms to account for genetic relatedness, repeated measures, and spatial error were determined important as judged by credible intervals and parsimony metrics. Species distribution models and trait-based approaches that do not account for spatial dependencies, phylogenetic relationships, or repeated measures sampling errors may produce parameter estimates with smaller uncertainty than is warranted and produce predictions with significant error. Our study offers tools to address spatially and phylogenetically structured species data and presents an approach to integrating biological comparative methods in biogeographic research.

Species distribution models (SDMs) are widely used to hindcast or forecast suitable habitat conditions during climate change. Although distant populations of a given species may show local adaptations to diverging environmental conditions, traditional SDMs disregard intraspecific variation. Yet, incorporating genetic information into SDMs could improve predictions. Here we aimed at investigating whether genetically informed SDMs would outperform traditional SDMs. Using published information on the spatial genetic structure of the European Beech Fagus sylvatica L. (1753), we built lineage-specific SDMs for each phylogenetic group of the species. We then combined all lineage-specific SDMs into a single genetically informed SDM that we compared against a traditional SDM approach. We finally compared SDMs’ predictions against independent datasets of present-day distribution as well as fossil distribution data from the Mid-Holocene, using six metrics of model performance. We found that aggregating lineage-specific SDMs into a single genetically informed SDM increased model performances to identify suitable areas currently occupied by F. sylvatica. In comparison to a traditional SDM, the genetically informed SDM we built for F. sylvatica assigned higher probabilities of occurrence during the Mid-Holocene at locations where fossil records were found. Aggregating lineage-specific SDMs into a single genetically informed SDM seems to outperform the traditional SDM approach, especially so when the aim is to identify potentially suitable areas of occupancy. This could be particularly useful for the identification of cryptic refugia that remain undetected by traditional SDMs. Genetically informed SDMs have the potential to improve our understanding of species redistribution under climate change.

Smallholder poultry production dominated by indigenous chickens is an important source of livelihoods for most rural households in Ethiopia. The long history of domestication and the presence of diverse agroecologies in Ethiopia create unique opportunities to study the effect of environmental selective pressures. Species distribution models (SDMs) and Phenotypic distribution models (PDMs) can be applied to investigate the relationship between environmental variation and phenotypic differentiation in wild animals and domestic populations. In the present study we used SDMs and PDMs to detect environmental variables related with habitat suitability and phenotypic differentiation among nondescript Ethiopian indigenous chicken populations. 34 environmental variables (climatic, soil, and vegetation) and 19 quantitative traits were analyzed for 513 adult chickens from 26 populations. To have high variation in the dataset for phenotypic and ecological parameters, animals were sampled from four spatial gradients (each represented by six to seven populations), located in different climatic zones and geographies. Three different ecotypes are proposed based on correlation test between habitat suitability maps and phenotypic clustering of sample populations. These specific ecotypes show phenotypic differentiation, likely in response to environmental selective pressures. Nine environmental variables with the highest contribution to habitat suitability are identified. The relationship between quantitative traits and a few of the environmental variables associated with habitat suitability is non-linear. Our results highlight the benefits of integrating species and phenotypic distribution modeling approaches in characterization of livestock populations, delineation of suitable habitats for specific breeds, and understanding of the relationship between ecological variables and quantitative traits, and underlying evolutionary processes.

ISSN 1948-6596 news and update workshop summary The application of species distribution models in the megadi- verse Neotropics poses a renewed set of research questions Species distribution models: applications, challenges and perspectives – Belo Horizonte, Brazil, 29 th –30 th August 2011 The community of researchers and technicians interested in biogeography is large and growing in Brazil, with members coming from fields as di- verse as ecology, evolution and conservation. The increment in the number of postgraduate pro- grams in ecology and evolutionary biology is linked to many research questions about the causes and dynamics of species’ ranges, as well as about their consequences for short-term evolu- tionary processes. As a result, the use (and abuse) of species distribution models (SDMs) as a tool in research and technical studies has grown rapidly in recent years. Systems that integrate biodiver- sity databases (e.g. SpeciesLink 1 ) allow one to ob- tain distributional information about many spe- cies; and easy-to-use SDM software (e.g. MaxEnt 2 , Phillips et al. 2004, or openModeller 3 , Munoz et al. 2011) is also available online. The easy avail- ability of data and SDM tools provides a powerful means for answering questions about the geo- graphic distribution of species. Potential users include non-specialist researchers, untrained post- graduate students and government technicians. But is this helping to develop a sound and solidly grounded knowledge of the distribution of Neotropical diversity? In order to unite the Brazilian community of SDM users and provide them with a better under- standing of the technique, the Postgraduate Course on Plant Biology of the Universidade Fed- eral de Minas Gerais organized a workshop in Belo Horizonte last August. This workshop allowed more than 100 students and technicians to meet and discuss with researchers working with SDMs. The main conclusions of the meeting were that there is a growing interest in using the technique to study species’ distributions and find unknown populations of rare species, and that there is a need for a code of good practice in both field sur- veys and SDM applications. These conclusions have been already discussed elsewhere (Kamino et al. 2011). Here we develop further one of the problems identified during the workshop: the lack of clear questions in many studies using SDMs; in other words, the mere application of species dis- tribution modelling as a fashionable technique often ‘justifies’ a study. Reviews summarizing the most important challenges for SDMs (e.g. Araujo and Guisan 2006, Zimmermann et al. 2010, Peterson et al. 2011) typically present how users view the field and what problems they perceive in each step in the modelling process. However, things move fast in this emerging field of research because, while the number of studies using SDMs increases steadily (see Fig. 1 in Lobo et al. 2010), the technique is also used to address completely new questions. Thus, the challenges to their application are them- selves changing. We believe that the most impor- tant of these challenges are theoretically rather than methodologically grounded, although we recognize that both kinds of problems overlap to some extent. Here we outline the basic questions that we believe SDM users must take into account while studying current species’ distributions. Soberon (2007) provides perhaps the best starting point to understand the theoretical prob- lems of SDMs (see also Colwell and Rangel 2009; Soberon 2010). The first problem is the definition of a clear research question. Here it is important to discriminate between theoretical questions such as “Why is this species here?” – which are more interesting in the long run – from those that are eminently practical such as “Where is this spe‐ cies?” – which unfortunately seem to be the most common. But even practical questions stimulated 1 http://www.splink.org.br; last accessed 10/11/2011 2 http://www.cs.princeton.edu/~schapire/maxent/; last accessed 10/11/2011 3 http://openmodeller.sourceforge.net/; last accessed 10/11/2011 frontiers of biogeography 4.1, 2012 — © 2012 the authors; journal compilation © 2012 The International Biogeography Society

Моделирование распространения, численности и выживаемости видов: новые возможности и методы

The species distribution models (SDMs) simulate and predict the potential distribution of species in geographical space by quantifying the relationships between species distribution and environmental variables, and extrapolating these relationships to unknown landscape units, which makes them important tools in ecology, biogeo-graphy, and conservation biology. Current SDMs mainly take abiotic factors as prediction variables, whereas biotic factors, especially species interactions, are often ignored due to the difficulties in data quantification and modeling. Incorporating species interactions into SDMs is considered as the main challenge of SDMs. We reviewed the influence of species interactions on species distribution simulations, clarified the necessity of incorporating species interactions into SDMs, summarized four main ways to incorporate species interactions into SDMs, analyzed their strengths and limitations, and discussed the future development direction of incorporating species interactions into SDMs. The study showed that incorporating species interaction into SDMs was based on the premise that the spatial scale of species distribution simulation was consistent with that of species interactions, and that the training data should be collected from large environmental heterogeneous space to ensure the diversity of species interactions in heterogeneous habitats. In order to eliminate the influence of multicollinearity on the prediction of SDMs, all abiotic and biotic factors should be fully considered and accurately quantified. Modeling the complex population/community dynamics would be an important development direction of incorporating species interactions into SDMs.

Species distribution models (SDMs) relate species observations to mapped environmental variables to estimate the realized niche of species and predict their distribution. SDMs are key tools for projecting the impact of climate change on species and have been used in many biodiversity assessments. However, when fitted within spatial extents that do not encompass the whole species range (i.e. subrange), the estimated realized environmental niche can be truncated, which can lead to wrong or inaccurate predictions. A simple solution to this niche truncation consists in fitting SDMs at a spatial extent that encompasses the whole species range, but this often implies using a spatial resolution too coarse for local conservation assessments. To keep a fine resolution, a solution is to fit spatially nested SDMs (N‐SDMs), where a whole range, coarse‐grain SDM is combined with a subrange, fine‐grain SDM. N‐SDMs have demonstrated superior performance to subrange (truncated) SDMs in projecting species distributions under climate change and have accordingly regained considerable interest. Here, we review developments, applications and effectiveness of N‐SDMs. We present and discuss existing methods and tools to fit N‐SDMs, and assess when N‐SDMs are not needed. We highlight strengths and weaknesses of N‐SDMs, underline their importance in reducing niche truncation, and identify remaining challenges and future perspectives. Our review highlights that subrange SDMs most often lead to niche truncation and thus to incorrect spatial projections, a problem that can be overcome by using N‐SDMs. We show that the various N‐SDM methods come with their strengths and weaknesses and should be selected depending on the intended goal of the study. Synthesis. N‐SDMs are key tools to develop untruncated regional climate change forecasts of species distributions at fine resolution over restricted extent. While several N‐SDM approaches were proposed, there is currently no universal solution suggesting that further developments and testing are crucial if we are to derive robust future projections of species distributions, at least until SDMs can be applied for most species at high resolution over large geographic extents.

news and update ISSN 1948‐6596 update Choosing the right path for species distribution modeling Species distribution models (SDM) have been widely used to address the lack of knowledge about species’ distributions (i.e. the so‐called Wal‐ lacean shortfall; Lomolino 2004). However, SDM show great variability in their predictions due to the uncertainties that accumulate during the modeling process (Barry & Elith 2006). These un‐ certainties are generally separated in three major classes: data, model and predictions. Uncertainty in the data refers to variation arising from incom‐ plete knowledge about the patterns and proc‐ esses studied, bias in the sampling procedure, and quality and choice of covariates; it applies to data used to build and fit the models, as well as to as‐ sess their success. Model uncertainty is due to discrepancies in the assumptions and algorithms used to fit the data. Finally, given that the true distribution is unknown and the relationship be‐ tween species’ presence and the environment may differ in studied and unstudied geographic regions, we are uncertain whether a prediction is a perfect fit and even if it were, it may not be so at a different time or place. Currently, SDM applica‐ tions lack approaches to manage such uncertainty, which could result in using them without consider‐ ing the reliability of their outputs with the subse‐ quent cost in the quality of the research based in these analyses. If, on the contrary, such uncer‐ tainty is acknowledged but is thought to be un‐ manageable, SDM would be flagged as unreliable for conservation planning or basic research de‐ spite their potential to provide good quality bio‐ geographical data. Beale and Lennon (2012) review the main sources of uncertainty associated with SDM and highlight research directions to improve SDM pre‐ dictions. These authors categorize SDM along an axis ranging from those that are purely statistical and try to identify process from pattern (i.e. niche ‐based models) to those that identify directly the processes and mechanisms to then generate the distributional pattern (i.e. process‐based models). For practical reasons the authors classified SDM into three types: niche‐based distribution models (which estimate the niche from the species’ geo‐ graphic distribution and re‐project it on a geo‐ graphic space), demographic models (which corre‐ late demographic parameters of the species with climate or weather to characterize its distribution) and process‐based models (which identify the physiological responses of the species and use them to determine the geographical distribution). The authors argue that, for all model types, the most critical sources of uncertainty are model uncertainty and prediction uncertainty. While the type of model is clearly relevant to niche‐based models for its influence in niche identification, the type of model also is important for demographic models to identify the actual links between popu‐ lation growth and weather, and for process‐based models to correctly estimate parameters. The un‐ certainties associated with predictions also affect all model types because of current deficiencies in the measures of the model fit. Beale and Lennon (2012) state that the performance of models can be improved by the incorporation of uncertainty in environmental covariates and by development of measures of model fit that take into account model complexity but are not affected by preva‐ lence and spatial autocorrelation. Specifically for niche‐based SDMs, Beale and Lennon (2012) identify that the quality of dis‐ tribution data presents particular challenges, since these data are not only used for model fit but also for model building. They argue, however, that suitable tools exist to assess the uncertainty at all steps of the modeling process of niche‐based models (data quality, choice of covariates, model‐ ing technique and evaluation). Once these uncer‐ tainties are estimated, it is possible to incorporate spatial error terms into the model building. This allows evaluation of the effects of uncertainties in the predictions, or even obtaining SDM results corrected by the underlying uncertainty. Process‐based models, however, require detailed species‐specific information which often is unavailable, and present the additional difficulty of identifying the interaction between species and frontiers of biogeography 4.3, 2012 — © 2012 the authors; journal compilation © 2012 The International Biogeography Society