Are comparisons of species distribution models biased? Are they biologically meaningful?

A major problem in ecology is to understand how environmental requirements change over space and time. To this end, numerous authors have attempted to use comparisons of species’ distributions as a surrogate for comparisons of environmental requirements. Unfortunately, it is currently unclear when comparisons of species’ distributions produce reliable inferences about changes in environmental requirements. To address this problem, I develop an analytic model that identifies the conditions under which a comparison of species’ distribution models can serve as surrogate for a comparison of environmental requirements. This work demonstrates that 1) comparisons of species’ distributions typically produce biased comparisons of environmental requirements, 2) assuming distribution models are fit appropriately, it is possible to compare environmental requirements of distinct taxa, 3) there are multiple biologically relevant questions we can address using comparisons of distribution models, with each question corresponding to a distinct measure of the difference between distribution models. By developing an analytic model for comparisons of species’ distributions this work helps to clarify and remedy poorly understood sources of error associated with existing methods.     

Do differences in understory light contribute to species distributions along a tropical rainfall gradient?

In tropical forests, regional differences in annual rainfall correlate with differences in plant species composition. Although water availability is clearly one factor determining species distribution, other environmental variables that covary with rainfall may contribute to distributions. One such variable is light availability in the understory, which decreases towards wetter forests due to differences in canopy density and phenology. We established common garden experiments in three sites along a rainfall gradient across the Isthmus of Panama in order to measure the differences in understory light availability, and to evaluate their influence on the performance of 24 shade-tolerant species with contrasting distributions. Within sites, the effect of understory light availability on species performance depended strongly on water availability. When water was not limiting, either naturally in the wetter site or through water supplementation in drier sites, seedling performance improved at higher light. In contrast, when water was limiting at the drier sites, seedling performance was reduced at higher light, presumably due to an increase in water stress that affected mostly wet-distribution species. Although wetter forest understories were on average darker, wet-distribution species were not more shade-tolerant than dry-distribution species. Instead, wet-distribution species had higher absolute growth rates and, when water was not limiting, were better able to take advantage of small increases in light than dry-distribution species. Our results suggest that in wet forests the ability to grow fast during temporary increases in light may be a key trait for successful recruitment. The slower growth rates of the dry-distribution species, possibly due to trade-offs associated with greater drought tolerance, may exclude these species from wetter forests.     

Do foliar fungal communities of Norway spruce shift along a tree species diversity gradient in mature European forests?

Foliar fungal species are diverse and colonize all plants, though whether forest tree species composition influences the distribution of these fungal communities remains unclear. Fungal communities include quiescent taxa and the functionally important and metabolically active taxa that respond to changes in the environment. To determine fungal community shifts along a tree species diversity gradient, needles of Norway spruce were sampled from trees from four mature European forests. We hypothesized that the fungal communities and specific fungal taxa would correlate with tree species diversity. Furthermore, the active fungal community, and not the total community, would shift along the tree diversity gradient. High-throughput sequencing showed significant differences in the fungal communities in the different forests, and in one forest, tree diversity effects were observed, though this was not a general phenomenon. Our study also suggests that studying the metabolically active community may not provide additional information about community composition or diversity.     

Does the interpolation accuracy of species distribution models come at the expense of transferability?

Model transferability (extrapolative accuracy) is one important feature in species distribution models, required in several ecological and conservation biological applications. This study uses 10 modelling techniques and nationwide data on both (1) species distribution of birds, butterflies, and plants and (2) climate and land cover in Finland to investigate whether good interpolative prediction accuracy for models comes at the expense of transferability – i.e. markedly worse performance in new areas. Models’ interpolation and extrapolation performance was primarily assessed using AUC (the area under the curve of a receiver characteristic plot) and Kappa statistics, with supplementary comparisons examining model sensitivity and specificity values. Our AUC and Kappa results show that extrapolation to new areas is a greater challenge for all included modelling techniques than simple filling of gaps in a well‐sampled area, but there are also differences among the techniques in the degree of transferability. Among the machine‐learning modelling techniques, MAXENT, generalized boosting methods (GBM), and artificial neural networks (ANN) showed good transferability while the performance of GARP and random forest (RF) decreased notably in extrapolation. Among the regression‐based methods, generalized additive models (GAM) and generalized linear models (GLM) showed good transferability. A desirable combination of good prediction accuracy and good transferability was evident for three modelling techniques: MAXENT, GBM, and GAM. However, examination of model sensitivity and specificity revealed that model types may differ in their tendencies to either increased over‐prediction of presences or absences in extrapolation, and some of the methods show contrasting changes in sensitivity vs specificity (e.g. ANN and GARP). Among the three species groups, the best transferability was seen with birds, followed closely by butterflies, whereas reliable extrapolation for plant species distribution models appears to be a major challenge at least at this scale. Overall, detailed knowledge of the behaviour of different techniques in various study settings and with different species groups is of utmost importance in predictive modelling.     

Are niche‐based species distribution models transferable in space?

Aim To assess the geographical transferability of niche‐based species distribution models fitted with two modelling techniques. Location Two distinct geographical study areas in Switzerland and Austria, in the subalpine and alpine belts. Methods Generalized linear and generalized additive models (GLM and GAM) with a binomial probability distribution and a logit link were fitted for 54 plant species, based on topoclimatic predictor variables. These models were then evaluated quantitatively and used for spatially explicit predictions within (internal evaluation and prediction) and between (external evaluation and prediction) the two regions. Comparisons of evaluations and spatial predictions between regions and models were conducted in order to test if species and methods meet the criteria of full transferability. By full transferability, we mean that: (1) the internal evaluation of models fitted in region A and B must be similar; (2) a model fitted in region A must at least retain a comparable external evaluation when projected into region B, and vice‐versa; and (3) internal and external spatial predictions have to match within both regions. Results The measures of model fit are, on average, 24% higher for GAMs than for GLMs in both regions. However, the differences between internal and external evaluations (AUC coefficient) are also higher for GAMs than for GLMs (a difference of 30% for models fitted in Switzerland and 54% for models fitted in Austria). Transferability, as measured with the AUC evaluation, fails for 68% of the species in Switzerland and 55% in Austria for GLMs (respectively for 67% and 53% of the species for GAMs). For both GAMs and GLMs, the agreement between internal and external predictions is rather weak on average (Kulczynski's coefficient in the range 0.3–0.4), but varies widely among individual species. The dominant pattern is an asymmetrical transferability between the two study regions (a mean decrease of 20% for the AUC coefficient when the models are transferred from Switzerland and 13% when they are transferred from Austria). Main conclusions The large inter‐specific variability observed among the 54 study species underlines the need to consider more than a few species to test properly the transferability of species distribution models. The pronounced asymmetry in transferability between the two study regions may be due to peculiarities of these regions, such as differences in the ranges of environmental predictors or the varied impact of land‐use history, or to species‐specific reasons like differential phenotypic plasticity, existence of ecotypes or varied dependence on biotic interactions that are not properly incorporated into niche‐based models. The lower variation between internal and external evaluation of GLMs compared to GAMs further suggests that overfitting may reduce transferability. Overall, a limited geographical transferability calls for caution when projecting niche‐based models for assessing the fate of species in future environments.     

Are local patterns of anthropoid primate diversity related to patterns of diversity at a larger scale?

Aims (1) To determine the relationship between local and regional anthropoid primate species richness. (2) To establish the spatial and temporal scale at which the ultimate processes influencing patterns of primate species coexistence operate. Location Continental landmasses of Africa, South America and Asia (India to China, and all islands as far south as New Guinea). Methods The local–regional species richness relationship for anthropoid primates is estimated by regressing local richness against regional richness (independent variable). Local richness is estimated in small, replicate local assemblages sampled in regions that vary in total species richness. A strong linear relationship is taken as evidence that local assemblages are unsaturated and local richness results from proportional sampling of the regional pool. An asymptotic curvilinear relationship is interpreted to reflect saturated communities, where strong biotic interactions limit local richness and local processes structure the species assemblage. As a further test of the assumption of local assemblage saturation, we looked for density compensation in high‐density local primate assemblages. Results The local–regional species richness relationship was linear for Africa and South America, and the slope of the relationship did not differ between the two continents. For Asia, curvilinearity best described the relationship between local and regional richness. Asian primate assemblages appear to be saturated and this is confirmed by density compensation among Asian primates. However, density compensation was also observed among African primates. The apparent assemblage saturation in Asia is not a species–area phenomenon related to the small size of the isolated islands and their forest blocks, since similar low local species richness occurs in large forests on mainland and/or peninsular Asia. Main conclusions In Africa and South America local primate assemblage composition appears to reflect the influence of biogeographic processes operating on regional spatial scales and historical time scales. In Asia the composition of primate assemblages are by‐and‐large subject to ecological constraint operating over a relatively small spatial and temporal scale. The possible local influence of the El Niño Southern Oscillations on the evolution and selection of life‐history characteristics among Asian primates, and in determining local patterns of primate species coexistence, warrants closer inspection.     

Does soil moisture availability explain liana seedling distribution across a tropical rainfall gradient?

Liana density tends to increase with decreasing rainfall and increasing seasonality. However, the pattern of liana distribution may be due to differences in soil water retention capacity, not rainfall and seasonality per se. We tested the effect of rainfall and soil substrate with respect to the distribution of liana seedlings in six sites across a rainfall gradient from the wet Atlantic to the dry Pacific in central Panama. Soils were either limestone, with low water‐holding capacity, or laterite, with higher water‐holding capacity. We sampled liana seedlings at each site using three 1 × 100 m transect. We found that relative liana seedling density was higher on limestone soils compared to laterite soils regardless of the amount of rainfall. Furthermore, liana community composition on limestone soils was more similar to dry forest sites than to adjacent wet and moist forest sites. Liana seedling species diversity relative to trees was significantly higher in a low‐fertility dry forest site compared to a high‐fertility forest, but did not differ from the other sites. Thus, liana seedling density and community structure may be driven more by soil type and thus by soil moisture availability than strictly by mean annual rainfall and the seasonality of rainfall.     

Is vascular plant species diversity a predictor of bryophyte species diversity in Mediterranean forests?

This study aimed to (i) investigate the congruence among the species composition and diversity of bryophytes and vascular plants in forests; (ii) test if site prioritization for conservation aims by the maximization of the pooled number of vascular plant species is effective to maximize the pooled number of bryophyte species. The study was performed in six forests in Tuscany, Italy. Four-hundred and twenty vascular plant species (61 of which were woody) and 128 bryophyte species were recorded in 109 plots. Despite the good predictive value of the compositional patterns of both woody plants and total vascular with respect to the compositional pattern of bryophytes, the species richness of the latter was only marginally related to the species richness of the former two. Bryophyte rare species were not spatially related to rare plant species and neither coincided with the sites of highest plant species richness. The species accumulation curves of bryophytes behaved differently with respect to those of woody plants or total vascular plants. Reserve selection analysis based on the maximization of the pooled species richness of either woody plants or total vascular plants were not effective in maximizing the pooled species richness of bryophytes. This study indicates that species diversity of vascular plants is not likely to be a good indicator of the bryophyte species diversity in Mediterranean forests.     

Are non-sexual models appropriate for predicting the impact of virus-vectored immunocontraception?

In response to the need to efficiently control mammal pest populations while avoiding unnecessary suffering, applied and theoretical ecologists have recently focused on virus-vectored immunocontraception (VVIC). So far, modellers have only considered a non-sexual approach (models of sexually reproducing populations without explicitly discerning between the sexes), which appears dubious in view of the sex-specificity of VVIC agents. In this paper, we derive and compare predictions of non-sexual and two-sex models of the spread of a VVIC agent in a host population in order to assess the adequacy of non-sexual models in this context. Our results show that predictions of non-sexual and two-sex models generally diverge and that non-sexual models often fail to predict the control impact of VVIC. We thus recommend using two-sex models, especially if the mating system and life history of the target species are known. Our analysis also shows that female-specific viruses generally give better results than male-specific ones, and suggests that virus choice should focus more on its sterilizing power rather than transmission efficiency.     

Surrogate species versus landscape metric: does presence of a raptor species explains diversity of multiple taxa more than patch area?

Prioritizing conservation areas is a central theme in conservation biology. The use of surrogate species and landscape metrics to identify areas with high conservation value is common. However, few studies have examined the relative efficacy of these two surrogates. In this study, we compared the efficacy of the presence/absence (PA) of a top predator (Eastern Marsh Harrier, Circus spilonotus) and wetland patch area on species richness, total abundance, and community composition of birds, plants, and small mammals, but species richness and community composition only for plants, in a fragmented wetland landscape. Although harrier PA was an effective indicator of the distribution of birds, it had no significant efficacy on the distributions of plants and small mammals. Patch area was more effective indicator of the distributions of plants and small mammals. These results suggest that surrogate species can be more effective indicators than landscape surrogates when there are ecological linkages between surrogate species and the focal taxa (e.g., similarity of habitat requirements between surrogate species and focal taxa or hetero-specific attraction). On the other hand, landscape surrogates would be useful when ecological knowledge about the relationships is limited.     

Do traits of plant species predict the efficacy of species distribution models for finding new occurrences?

Species distribution models (SDMs) are used to test ecological theory and to direct targeted surveys for species of conservation concern. Several studies have tested for an influence of species traits on the predictive accuracy of SDMs. However, most used the same set of environmental predictors for all species and/or did not use truly independent data to test SDM accuracy. We built eight SDMs for each of 24 plant species of conservation concern, varying the environmental predictors included in each SDM version. We then measured the accuracy of each SDM using independent presence and absence data to calculate area under the receiver operating characteristic curve (AUC) and true positive rate (TPR). We used generalized linear mixed models to test for a relationship between species traits and SDM accuracy, while accounting for variation in SDM performance that might be introduced by different predictor sets. All traits affected one or both SDM accuracy measures. Species with lighter seeds, animal‐dispersed seeds, and a higher density of occurrences had higher AUC and TPR than other species, all else being equal. Long‐lived woody species had higher AUC than herbaceous species, but lower TPR. These results support the hypothesis that the strength of species–environment correlations is affected by characteristics of species or their geographic distributions. However, because each species has multiple traits, and because AUC and TPR can be affected differently, there is no straightforward way to determine a priori which species will yield useful SDMs based on their traits. Most species yielded at least one useful SDM. Therefore, it is worthwhile to build and test SDMs for the purpose of finding new populations of plant species of conservation concern, regardless of these species’ traits.     

Does freshwater macroinvertebrate diversity along a pH‐gradient reflect adaptation to low pH?

1. The impacts of anthropogenic surface water acidification are much better known than those of natural acidity. Recent studies have indicated biodiversity is not degraded and species composition unaltered in naturally acidic compared to circumneutral watercourses.
2. Here, we use a geographically extensive dataset comprising sites in more than 200 Swedish streams to test whether the lack of effects on macroinvertebrate species diversity is due to exaptation and adaptation to natural acidity.
3. To this end, we modelled pH associated with spring flood episodes, which inflict the most challenging hydrochemical conditions to the biota. We compared taxonomic richness and species composition along the modelled pH gradient in northern Sweden, where acidity is largely natural, with southern Sweden, a region influenced by significant anthropogenic acidification.
4. We found Plecoptera richness did not respond to varying pH either in northern or southern Sweden. Ephemeroptera richness was sensitive to pH in both regions, while that of Trichoptera increased with increasing pH in southern Sweden, but decreased in the north. The taxonomic composition of Plecoptera changed along the pH gradient in both regions, whereas that of Ephemeroptera and Trichoptera changed more strongly with pH in southern Sweden.
5. Our results support the hypothesis that stream invertebrates are able to tolerate low pH through exaptation or adaptation, but that this capability varies among taxonomic groups.     

Does disturbance drive the collapse of biotic interactions at the severe end of a diversity–biomass gradient?

It has been recently proposed that the decrease in diversity towards the severe end of the humped-back diversity–biomass model of Grime was driven by a collapse of facilitation due to extreme conditions of either stress or physical disturbance. In order to test the hypothesis that disturbance is the primary direct factor driving the collapse of interactions occurring along environmental severity gradients, we conducted a removal experiment in the highly stressed French coastal dunes along a gradient of disturbance due to sand burial. Four dune species were used as targets and transplanted with and without neighbours in four communities along the gradient. The experiment was conducted twice, a dry and an average year. Results of the experiment showed that during the dry year the effect of the environment was prominent and only one species was facilitated for survival in the least disturbed community. During the average year, interactions for growth were important only in the same community, with positive or negative responses depending on the natural position of the target species within the coastal dune gradient. In accordance with our hypothesis, most interactions for both survival and growth were observed in the least disturbed community exhibiting the highest diversity. There were no interactions in the most disturbed community with the lowest diversity.     

Are the speeds of species invasions regulated? The importance of null models

Understanding the mechanisms and patterns that govern the invasion of species is essential for coping with global change of the biological world. A recent study highlights the possibility, based on data from a wide range of different taxa, that the invasion speed of species could be governed by a regulatory process. In principle, it is possible that mechanisms such as Allee effects could cause the invasion fronts to be regulated, such that the change in the rate of spread is negatively related to the current rate. This is very similar to how some populations are regulated around an equilibrium size, and finding the regulation structure if true, is of both pure and applied interest. However, here we will argue that the methods used so far are incomplete, thus even though there is a theoretical possibility that the speed of species invasions are regulated, more scrutiny is needed for its detection. Analysing changes of the ratio of current and past rate of spread against current ratios may give the impression of regulation in null models that are in fact unregulated. In addition we show that the apparent pattern is highly influenced by the spatial scale of investigation. Our results show that detecting regulatory patterns in species invasions is similarly non-trivial as is detecting density-dependence per se, but necessary, given the importance of this problem.     

Are invasive species a major cause of extinctions?

The link between species invasions and the extinction of natives is widely accepted by scientists as well as conservationists, but available data supporting invasion as a cause of extinctions are, in many cases, anecdotal, speculative and based upon limited observation. We pose the question, are aliens generally responsible for widespread extinctions? Our goal is to prompt a more critical synthesis and evaluation of the available data, and to suggest ways to take a more scientific, evidence-based approach to understanding the impact of invasive species on extinctions. Greater clarity in our understanding of these patterns will help us to focus on the most effective ways to reduce or mitigate extinction threats from invasive species.     

What drives the species richness patterns of non‐volant small mammals along a subtropical elevational gradient?

The biodiversity of non‐volant small mammals along an extensive subtropical elevational gradient was studied for the first time on Gongga Mountain, the highest mountain in Hengduan Mountain ranges in China, located in one of the 25 global biodiversity hotspots. Non‐volant small mammals were replicate sampled in two seasons at eight sampling sites between 1000 and 4200 m elevation on the eastern slope of Gongga Mountain. In all, 726 individual small mammals representing 25 species were documented in 28 800 trap nights. The species richness pattern for non‐volant small mammals along the elevational gradients was hump‐shaped with highest richness at mid‐elevations. However, different richness patterns emerged between endemic and non‐endemic species, between larger‐ranged and smaller‐ranged species and between rodents and insectivores. Temperature, precipitation, plant species richness and geometric constraints (mid‐ domain effect) were most significant in explaining species richness patterns. Based on the analysis of simple ordinary least squares (OLS) and stepwise multiple regressions, the overall richness pattern, as well as the pattern of insectivores, endemic species and larger‐ranged species showed strong correlation with geometric constraint predictions. However, non‐endemic species richness was more strongly correlated with temperature, while rodent richness was correlated with plant species richness. Our study shows that no single key factor can explain all richness patterns of non‐volant small mammals. We need to be cautious in summarizing a general richness pattern of large species groups (e.g. small mammals or mammals) from species in smaller groups having different ecological distributions and life histories. Elevational richness patterns and their driving factors for small mammals are more likely dependent on what kind of species we study.     

Biodiversity and ecosystem services−A case study for the assessment of multiple species and functional diversity levels in a cultural landscape

The expansion of large-scale plantations has a major impact on landscapes in the Tropics and Subtropics. Crops like soy bean, oil palm and rubber have led to drastic changes in land cover over the past decades, thereby altering ecosystem functions and services (ESS). Associated shifts in ESS such as climate regulation, erosion and water cycles, biodiversity as well as soil fertility or the provisioning of raw materials have been assessed through several models and software solutions (InVEST, ARIES, MIMES). However, suitable methods for the integration of a range of biodiversity assessments in agricultural landscapes are scarce.With this study, we introduce a methodology for incorporating multiple levels of species diversity into models to allow an integrated evaluation of ESS. We collected data sets from both published and unpublished sources on the distribution of vascular plants, selected pollinator groups, ground beetles, ungulates as well as amphibians, mammals, reptiles and birds in rubber-dominated landscapes, with a focus on our study sites in Southwest China and Thailand. Based on this information, we developed a common classification scheme that enables the integration of different facets of biodiversity (species diversity and functional diversity) to complement an interdisciplinary ESS assessment.Species diversity data were normalized against the most divers habitats reported (using habitat scores) to assess the impact of rubber cultivation on multiple levels of biodiversity. This resulted in a comparable matrix of different land use types and their suitability as habitat for the respective species groups allowing the aggregation of very diverse indicators. The findings were applied to two alternative land use scenarios in southern China to highlight the potential effects of land use and management decisions on species and functional diversity. Our results highlighted that the conservation oriented scenario did score higher for habitat suitability in both total species (+5%) as well as IUCN Red List species (+6%) assessments compared to the current state or business as usual scenarios (-2% and −3% compared to current state).The process presented here allows for an application within established ESS software programs, in our case InVEST, using aggregated indices while additionally providing enhanced opportunities for comparable, spatially explicit assessments of the expected impact of the analyzed scenarios on specific species groups.     

Pattern of local plant species richness along a gradient of landscape topographical heterogeneity: result of spatial mass effect or environmental shift?

Several processes are hypothesised to mediate the relationship between local (microsite) plant species richness and the topographical heterogeneity of the surrounding landscape. In a topographically heterogeneous landscape with various habitats occurring close to each other, local species richness may be enriched by species from surrounding habitats due to the spatial mass effect (sink‐source dynamics). In contrast, increased habitat fragmentation due to spatial heterogeneity may have a negative effect on local species richness. The spatial mass effect is thought to be more pronounced in communities with a higher ratio of generalists, as generalists are more likely to establish viable populations in sink habitats. To reveal the pattern of local species richness along a gradient of landscape topographical heterogeneity at middle altitudes of the Bohemian Massif, we used 2551 forest vegetation plots stored in the Czech National Phytosociological Database. We developed an analytical approach relating the pattern of local species richness of vegetation types to the gradient of landscape topographical heterogeneity. An increase or decrease in species richness with increasing landscape heterogeneity was related to changes in the generalist/specialist ratio, and also to changes in soil reaction and productivity estimated through Ellenberg indicator values. Local species richness along a gradient of increasing landscape heterogeneity increased in nutrient‐poor vegetation and decreased in nutrient‐rich vegetation. Nutrient‐poor vegetation types, such as thermophilous and acidophilous oak forests, also had a high proportion of habitat generalists, supporting the hypothesis that increased richness in heterogeneous landscapes may result from the spatial mass effect. However, the same pattern may be explained by a shift in environmental conditions along the landscape heterogeneity gradient, such as increasing productivity of nutrient‐rich vegetation types or increasing soil reaction of most vegetation types in more heterogeneous landscapes. We discuss available evidence and conclude that these two explanations need not be mutually exclusive.