Spatial variation of climatic and non-climatic controls on species distribution: the range limit of Tsuga heterophylla

Aim To assess which climatic variables control the distribution of western hemlock (Tsuga heterophylla), how climatic controls vary over latitude and between disjunct coastal and interior sub‐distributions, and whether non‐climatic factors, such as dispersal limitation and interspecific competition, affect range limits in areas of low climatic control. Location North‐western North America. Methods We compared four bioclimatic variables [actual evapotranspiration (AET), water deficit (DEF), mean temperature of the coldest month (MTCO), and growing degree‐days (GDD5)] with the distribution of T. heterophylla at a 2‐km grid cell resolution. The distribution is based on a zonal ecosystem classification where T. heterophylla is the dominant late‐successional species. For each bioclimatic variable and at each degree of latitude, we calculated the threshold that best defines the T. heterophylla distribution and assessed the extent to which T. heterophylla was segregated to one end of the bioclimatic gradient. We also fitted two forms of multivariate bioclimatic models to predict the T. heterophylla distribution: a simple threshold model and a complex Gaussian mixture model. Each model was trained separately on the coastal and interior distributions, and predicted areas outside of the T. heterophylla distribution (overprediction) were evaluated with respect to known outlier populations. Results Actual evapotranspiration was the most accurate predictor across the T. heterophylla distribution; other variables were important only in certain areas. There was strong latitudinal variation in the thresholds of all variables except AET, and the interior distribution had wider bioclimatic thresholds than the coastal distribution. The coastal distribution was predicted accurately by both bioclimatic models; areas of overprediction rarely occurred > 10 km from the observed distribution and generally matched small outlier populations. In contrast, the interior distribution was poorly predicted by both models; areas of overprediction occurred up to 140 km from the observed distribution and did not match outlier populations. The greatest overprediction occurred in Idaho and Montana in areas supporting species that typically co‐exist with T. heterophylla. Main conclusions The high predictive capacity of AET is consistent with this species’ physiological requirements for a mild and humid climate. Spatial variation of MTCO, GDD5 and DEF thresholds probably reflects both the correlation of these variables with AET and ecotypic variation. The level of overprediction in portions of the interior suggests that T. heterophylla has not completely expanded into its potential habitat. Tsuga heterophylla became common in the interior 2000–3500 years ago, compared with > 9000 years ago in the coastal region. The limited time for dispersal, coupled with frequent fires at the margins of the distribution and competition with disturbance‐adapted species, may have retarded range expansion in the interior. This study demonstrates that bioclimatic modelling can help identify various climatic and non‐climatic controls on species distributions.     

Do we need land‐cover data to model species distributions in Europe?

Aim To assess the influence of land cover and climate on species distributions across Europe. To quantify the importance of land cover to describe and predict species distributions after using climate as the main driver. Location The study area is Europe. Methods (1) A multivariate analysis was applied to describe land‐cover distribution across Europe and assess if the land cover is determined by climate at large spatial scales. (2) To evaluate the importance of land cover to predict species distributions, we implemented a spatially explicit iterative procedure to predict species distributions of plants (2603 species), mammals (186 species), breeding birds (440 species), amphibian and reptiles (143 species). First, we ran bioclimatic models using stepwise generalized additive models using bioclimatic variables. Secondly, we carried out a regression of land cover (LC) variables against residuals from the bioclimatic models to select the most relevant LC variables. Finally, we produced mixed models including climatic variables and those LC variables selected as decreasing the residual of bioclimatic models. Then we compared the explanatory and predictive power of the pure bioclimatic against the mixed model. Results (1) At the European coarse resolution, land cover is mainly driven by climate. Two bioclimatic axes representing a gradient of temperature and a gradient of precipitation explained most variation of land‐cover distribution. (2) The inclusion of land cover improved significantly the explanatory power of bioclimatic models and the most relevant variables across groups were those not explained or poorly explained by climate. However, the predictive power of bioclimatic model was not improved by the inclusion of LC variables in the iterative model selection process. Main conclusion Climate is the major driver of both species and land‐cover distributions over Europe. Yet, LC variables that are not explained or weakly associated with climate (inland water, sea or arable land) are interesting to describe particular mammal, bird and tree distributions. However, the addition of LC variables to pure bioclimatic models does not improve their predictive accuracy.     

Land use improves spatial predictions of mountain plant abundance but not presence‐absence

Question: Does a land‐use variable improve spatial predictions of plant species presence‐absence and abundance models at the regional scale in a mountain landscape? Location: Western Swiss Alps. Methods: Presence‐absence generalized linear models (GLM) and abundance ordinal logistic regression models (LRM) were fitted to data on 78 mountain plant species, with topo‐climatic and/or land‐use variables available at a 25‐m resolution. The additional contribution of land use when added to topo‐climatic models was evaluated by: (1) assessing the changes in model fit and (2) predictive power, (3) partitioning the deviance respectively explained by the topo‐climatic variables and the land‐use variable through variation partitioning, and (5) comparing spatial projections. Results: Land use significantly improved the fit of presence‐absence models but not their predictive power. In contrast, land use significantly improved both the fit and predictive power of abundance models. Variation partitioning also showed that the individual contribution of land use to the deviance explained by presence‐absence models was, on average, weak for both GLM and LRM (3.7% and 4.5%, respectively), but changes in spatial projections could nevertheless be important for some species. Conclusions: In this mountain area and at our regional scale, land use is important for predicting abundance, but not presence‐absence. The importance of adding land‐use information depends on the species considered. Even without a marked effect on model fit and predictive performance, adding land use can affect spatial projections of both presence‐absence and abundance models.     

Exploring consensus in 21st century projections of climatically suitable areas for African vertebrates

Africa is predicted to be highly vulnerable to 21st century climatic changes. Assessing the impacts of these changes on Africa's biodiversity is, however, plagued by uncertainties, and markedly different results can be obtained from alternative bioclimatic envelope models or future climate projections. Using an ensemble forecasting framework, we examine projections of future shifts in climatic suitability, and their methodological uncertainties, for over 2500 species of mammals, birds, amphibians and snakes in sub‐Saharan Africa. To summarize a priori the variability in the ensemble of 17 general circulation models, we introduce a consensus methodology that combines co‐varying models. Thus, we quantify and map the relative contribution to uncertainty of seven bioclimatic envelope models, three multi‐model climate projections and three emissions scenarios, and explore the resulting variability in species turnover estimates. We show that bioclimatic envelope models contribute most to variability, particularly in projected novel climatic conditions over Sahelian and southern Saharan Africa. To summarize agreements among projections from the bioclimatic envelope models we compare five consensus methodologies, which generally increase or retain projection accuracy and provide consistent estimates of species turnover. Variability from emissions scenarios increases towards late‐century and affects southern regions of high species turnover centred in arid Namibia. Twofold differences in median species turnover across the study area emerge among alternative climate projections and emissions scenarios. Our ensemble of projections underscores the potential bias when using a single algorithm or climate projection for Africa, and provides a cautious first approximation of the potential exposure of sub‐Saharan African vertebrates to climatic changes. The future use and further development of bioclimatic envelope modelling will hinge on the interpretation of results in the light of methodological as well as biological uncertainties. Here, we provide a framework to address methodological uncertainties and contextualize results.     

Climatic niche of the Saker Falcon Falco cherrug: predicted new areas to direct population surveys in Central Asia

Accurate species distribution data across remote and extensive geographical areas are difficult to obtain. Here, we use bioclimatic envelope models to determine climatic constraints on the distribution of the migratory Saker Falcon Falco cherrug to identify areas in data-deficient regions that may contain unidentified populations. Sakers live at low densities across large ranges in remote regions, making distribution status difficult to assess. Using presence-background data and eight bioclimatic variables within a species distribution modelling framework, we applied MaxEnt to construct models for both breeding and wintering ranges. Occurrence data were spatially filtered and climatic variables tested for multicollinearity before selecting best fit models using the Akaike information criterion by tuning MaxEnt parameters. Model predictive performance tested using the continuous Boyce index (B) was high for both breeding (B_TEST = 0.921) and wintering models (B_TEST = 0.735), with low omission rates and minimal overfitting. The Saker climatic niche was defined by precipitation in the warmest quarter in the breeding range model, and mean temperature in the wettest quarter in the wintering range model. Our models accurately predicted areas of highest climate suitability and defined the climatic constraints on a wide-ranging rare species, suggesting that climate is a key determinant of Saker distribution across macro-scales. We recommend targeted population surveys for the Saker based on model predictions to areas of highest climatic suitability in key regions with distribution knowledge gaps, in particular the Qinghai-Tibet plateau in western China. Further applications of our models could identify protected areas and reintroduction sites, inform development conflicts, and assess the impact of climate change on distributions.     

Land-cover data improve bioclimatic models for anurans and turtles at a regional scale

Aim We investigated whether accounting for land cover could improve bioclimatic models for eight species of anurans and three species of turtles at a regional scale. We then tested whether accounting for spatial autocorrelation could significantly improve bioclimatic models after statistically controlling for the effects of land cover. Location Nova Scotia, eastern Canada. Methods Species distribution data were taken from a recent (1999–2003) herpetofaunal atlas. Generalized linear models were used to relate the presence or absence of each species to climate and land‐cover variables at a 10‐km resolution. We then accounted for spatial autocorrelation using an autocovariate or third‐order trend surface of the geographical coordinates of each grid square. Finally, variance partitioning was used to explore the independent and joint contributions of climate, land cover and spatial autocorrelation. Results The inclusion of land cover significantly increased the explanatory power of bioclimatic models for 10 of the 11 species. Furthermore, including land cover significantly increased predictive performance for eight of the 11 species. Accounting for spatial autocorrelation improved model fit for rare species but generally did not improve prediction success. Variance partitioning demonstrated that this lack of improvement was a result of the high correlation between climate and trend‐surface variables. Main conclusions The results of this study suggest that accounting for the effects of land cover can significantly improve the explanatory and predictive power of bioclimatic models for anurans and turtles at a regional scale. We argue that the integration of climate and land‐cover data is likely to produce more accurate spatial predictions of contemporary herpetofaunal diversity. However, the use of land‐cover simulations in climate‐induced range‐shift projections introduces additional uncertainty into the predictions of bioclimatic models. Further research is therefore needed to determine whether accounting for the effects of land cover in range‐shift projections is merited.     

The role of land cover in bioclimatic models depends on spatial resolution

Aim We explored the importance of climate and land cover in bird species distribution models on multiple spatial scales. In particular, we tested whether the integration of land cover data improves the performance of pure bioclimatic models. Location Finland, northern Europe. Methods The data of the bird atlas survey carried out in 1986–89 using a 10 × 10 km uniform grid system in Finland were employed in the analyses. Land cover and climatic variables were compiled using the same grid system. The dependent and explanatory variables were resampled to 20‐km, 40‐km and 80‐km resolutions. Generalized additive models (GAM) were constructed for each of the 88 land bird species studied in order to estimate the probability of occurrence as a function of (1) climate and (2) climate and land cover variables. Model accuracy was measured by a cross‐validation approach using the area under the curve (AUC) of a receiver operating characteristic (ROC) plot. Results In general, the accuracies of the 88 bird–climate models were good at all studied resolutions. However, the inclusion of land cover increased the performance of 79 and 78 of the 88 bioclimatic models at 10‐km and 20‐km resolutions, respectively. There was no significant improvement at the 40‐km resolution. In contrast to the finer resolutions, the inclusion of land cover variables decreased the modelling accuracy at 80km resolution. Main conclusions Our results suggest that the determinants of bird species distributions are hierarchically structured: climatic variables are large‐scale determinants, followed by land cover at finer resolutions. The majority of the land bird species in Finland are rather clearly correlated with climate, and bioclimate envelope models can provide useful tools for identifying the relationships between these species and the environment at resolutions ranging from 10 km to 80 km. However, the notable contribution of land cover to the accuracy of bioclimatic models at 10–20‐km resolutions indicates that the integration of climate and land cover information can improve our understanding and model predictions of biogeographical patterns under global change.     

The influence of climate on the distribution of lichens: a case study in a borderline area (Liguria,NW Italy)

The purposes of this article are to quantify the relationship between epiphytic lichen distribution and macroclimatic variables in the study area and to provide a case study for evaluating the predictive role of epiphytic lichens as bioclimatic indicators. The study was carried out in the Liguria region (NW-Italy), a small (5432 km²) borderline area, where phytoclimatic features range from the dry Mediterranean to the Alpine in a few kilometers. Epiphytic lichen diversity was sampled using a standardized protocol [Asta et al (2002) In: Nimis et al (eds) Monitoring with lichens: monitoring lichens. Kluwer, Dordrecht]. Abundance of the species in the sampling sites was related to macroclimatic parameters (yearly average temperature and rainfall) and non-parametric multivariate models were calculated to find significative relationships among predictive and response variables. A total of 59 species showed highly significant relation with macroclimatic parameters. Four groups were selected, by means of a cluster analysis, related to four climatic niches (warm-humid, cold-humid, mesothermic-humid, warm-dry). Distributional pattern of the groups in the survey area showed a good correspondence with the bioclimatic units of Liguria region described by Nimis [(2003) Checklist of the Lichens of Italy 3.0. University of Trieste, Dept of Biology. http://www.dbiodbs.univ.trieste.it. Cited 1 Jun 2006]. A significant subset of epiphytic lichen species in the study area have been proved to be efficient bioclimatic indicator and it is supposed to give good results to monitor climatic changes, in a long-term perspective.     

Bioclimatic model of tree radial growth: application to the French Mediterranean Aleppo pine forests

The potential effects of global changes on forests are of increasing concern. Dendrochronology, which deals with long-term records of tree growth under natural environmental conditions, can be used to evaluate the impact of climatic change on forest productivity. However, assessment of climatic change impacts must be supported by accurate and reliable models of the relationships between climate and tree growth. In this study, a bioclimatic model is used to explore the relationships between tree radial growth and bioclimatic variables closely related to the biological functioning of a tree. This model is at an intermediate level of complexity between purely empirical and process-based models. The method is illustrated with data for 21 Aleppo pine (Pinus halepensis Mill.) stands grown under a Mediterranean climate in south-east France. The results show that Aleppo pine growth is mainly controlled by soil water availability during the growing season. The bioclimatic variable which best expresses the observed inter-annual tree growth variations is the actual evapotranspiration (AET). Four parameters were adjusted to simulate dendrochronological data: the soil water capacity, the wilting point, the minimum temperature for photosynthesis, and the end of the growing season. The bioclimatic model gives better results than the standard response function and provides better insight into the functional processes involved in tree growth. The convincing results obtained by the bioclimatic model as well as the limited numbers of parameters it requires demonstrate the feasibility of using it to explore future climatic change impacts on Aleppo pine forests.     

Importance of soil nutrients in the distribution of forest communities on a large geographical scale

Aim Soil nutrient content plays a key role in plant growth through mineral nutrition and toxicity. Its impact on plant species and community distribution is studied on a large geographical scale through surrogates like topography or geology. We investigated the importance of soil pH and C:N ratio, as direct nutritional gradients, to determine, with climatic factors, the spatial distribution of plant communities over large territories. Location We studied the distribution of six beech habitats of the NATURA 2000 network throughout France (550,000 km²). Methods Models were calibrated with 2108 floristic plots classified in the NATURA 2000 system and including climatic and topographic variables and soil nutritional measurements carried out in a laboratory. Logistic regression was used to model habitat distribution according to environmental variables. Climatic layers, a digital elevation model and maps of soil pH and nitrogen content, created using plant indicator values and large floristic databases, were used to map the sites suitable for beech communities. Distribution models were evaluated with an independent set of 2091 phytosociological plots. Results pH and nitrogen supply were the key distribution drivers for four of the six beech communities on a national scale. Their use in the distribution models distinguished within homogeneous climatic territories a gradient of nutritional conditions from acidic areas, suitable for nutrient‐poor beech communities, to calcareous areas suitable for nutrient‐rich ones. Predicted maps of beech habitats fit the spatial distribution of validation plots. Main conclusions Soil pH and nitrogen supply strongly improve predictions of forest community distribution carried out with climatic variables on a broad geographical scale. They allow delineation of areas with nutritional conditions suitable for each community, as well as the realization of predictive high‐resolution maps over large areas useful for sustainable and conservation management. Nomenclature Tutin & Heywood (2001 ) Flora Europaea. Cambridge University Press, Cambridge.     

Does seasonal fine‐tuning of climatic variables improve the performance of bioclimatic envelope models for migratory birds?

We examined the influence of 'seasonal fine-tuning' of climatic variables on the performance of bioclimatic envelope models of migrating birds. Using climate data and national bird atlas data from a 10 × 10 km uniform grid system in Finland, we tested whether the replacement of one 'baseline' set of variables including summer (June–August) temperature and precipitation variables with climate variables tailored ('fine-tuned') for each species individually improved the bird-climate models. The fine-tuning was conducted on the basis of time of arrival and early breeding of the species. Two generalized additive models (GAMs) were constructed for each of the 63 bird species studied, employing (1) the baseline climate variables and (2) the fine-tuned climate variables. Model performance was measured as explanatory power (deviance change) and predictive power (area under the curve; AUC) statistics derived from cross-validation. Fine-tuned climate variables provided, in many cases, statistically significantly improved model performance compared to using the same baseline set of variables for all the species. Model improvements mainly concerned bird species arriving and starting their breeding in May–June. We conclude that the use of the fine-tuned climate variables tailored for each species individually on the basis of their arrival and critical breeding periods can provide important benefits for bioclimatic modelling.     

Evaluating the use of macroscale variables as proxies for local aquatic variables and to model stream fish distributions

相似文献   

Modelling relationships between species spatial abundance patterns and climate

Aim To move towards modelling spatial abundance patterns and to evaluate the relative impacts of climatic change upon species abundances as opposed to range extents. Location Southern Africa, including Lesotho, Namibia, South Africa, Swaziland and Zimbabwe. Methods Quantitative response surface models were fitted for 78 bird species, mostly endemic (68) or near‐endemic to the region, to model relationships between species reporting rates (i.e. the proportion of checklists reporting a species for a particular grid cell), as recorded by the Southern African Bird Atlas Project, and four bioclimatic variables derived from climatic data for the period 1961–90. With caution, reporting rates can be used as a proxy for abundance. Models were used to project potential impacts of a series of projected climatic change scenarios upon species abundance patterns and range extents. Results Most models obtained were robust with good predictive power. Projections of potential future abundance patterns indicate that the magnitude of impacts upon a proxy for abundance are greater than those upon range extent for the majority of species (82% by 2071–2100). For most species (74%) both abundance and range extent are projected to decrease by 2100. Impacts are especially severe if species are unable to realize projected range changes; when only the area of a species' simulated present range is considered, overall abundance decreases of more than 80% are projected for 19 (24%) of species examined. Main conclusions Our results indicate that projected climatic changes are likely to elicit greater relative changes in species abundances than range extents. For most species examined changes were decreases, suggesting the impacts upon biodiversity are likely generally to be negative. These results also suggest that previous estimates of the proportion of species at increased risk of extinction as a result of climatic change may, in some cases, be under‐estimates.     

Regional climate oscillations and local topography shape genetic polymorphisms and distribution of the giant columnar cactus Echinopsis terscheckii in drylands of the tropical Andes

Aim

We sought to determine if the present fragmentary distribution of the giant columnar cactus Echinopsis terscheckii in tropical drylands is a relict of a previously more widespread range during cold and dry phases of the Last Glacial Maximum (LGM).

Location

Tropical and subtropical dry ecotonal areas of northern and central Andes of Argentina.

Methods

We combined ecological niche models (ENM) with molecular polymorphisms of isozymes and DNA sequences. We collected samples from 30 individuals at 24 locations for genetic analysis covering a wide range of environmental conditions. We sequenced the nuclear ITS and three non‐coding regions of the chloroplast DNA and we resolved 15 isozyme loci. Potential distribution was modelled using 88 E. terscheckii presence training records and a reduced set of 10 modern bioclimatic variables. LGM and the Mid‐Holocene distributions were derived by projecting bioclimatic data under present to past environmental conditions according to CCSM4 and MIROC‐ESM Global Climate Models.

Results

We detected high isozyme diversity towards the south. The multivariate cluster analysis yielded two groups of populations that were geographically concordant with the DNA haplotypes located north and south of a divide at 27°S. Distribution models show range expansion during the LGM in two north and south areas separated by a gap of low suitability at 27°S. Suitable areas in the south were close to current populations, while in the north, populations survived in more disjunct locations that probably suffered from founder effects. In contrast, Mid‐Holocene bioclimatic conditions were relatively unsuitable in the south.

Main conclusions

Our results suggest that the divergence of north and south groups of E. terscheckii populations reflect long‐lasting persistence through climatic cycles that were reinforced by the presence of an orogenic divide at mid‐latitudes. Latitudinally divergent groups of populations should be treated as distinct evolutionary significant units that deserve independent conservation actions. Increased genetic diversity and inbreeding towards the south may guide setting up priorities for the long‐term protection of a dominant element of drylands as E. terscheckii.     

Modelling the distribution of a threatened habitat: the California sage scrub

Aim Using predictive species distribution and ecological niche modelling our objectives are: (1) to identify important climatic drivers of distribution at regional scales of a locally complex and dynamic system – California sage scrub; (2) to map suitable sage scrub habitat in California; and (3) to distinguish between bioclimatic niches of floristic groups within sage scrub to assess the conservation significance of analysing such species groups. Location Coastal mediterranean‐type shrublands of southern and central California. Methods Using point localities from georeferenced herbarium records, we modelled the potential distribution and bioclimatic envelopes of 14 characteristic sage scrub species and three floristic groups (south‐coastal, coastal–interior disjunct and broadly distributed species) based upon current climate conditions. Maxent was used to map climatically suitable habitat, while principal components analysis followed by canonical discriminant analysis were used to distinguish between floristic groups and visualize species and group distributions in multivariate ecological space. Results Geographical distribution patterns of individual species were mirrored in the habitat suitability maps of floristic groups, notably the disjunct distribution of the coastal–interior species. Overlap in the distributions of floristic groups was evident in both geographical and multivariate niche space; however, discriminant analysis confirmed the separability of floristic groups based on bioclimatic variables. Higher performance of floristic group models compared with sage scrub as a whole suggests that groups have differing climate requirements for habitat suitability at regional scales and that breaking sage scrub into floristic groups improves the discrimination between climatically suitable and unsuitable habitat. Main conclusions The finding that presence‐only data and climatic variables can produce useful information on habitat suitability of California sage scrub species and floristic groups at a regional scale has important implications for ongoing efforts of habitat restoration for sage scrub. In addition, modelling at a group level provides important information about the differences in climatic niches within California sage scrub. Finally, the high performance of our floristic group models highlights the potential a community‐level modelling approach holds for investigating plant distribution patterns.     

Precipitation of the warmest quarter and temperature of the warmest month are key to understanding the effect of climate change on plant species diversity in Southern African savannah

In this study, we test for the key bioclimatic variables that significantly explain the current distribution of plant species richness in a southern African ecosystem as a preamble to predicting plant species richness under a changed climate. We used 54,000 records of georeferenced plant species data to calculate species richness and spatially interpolated climate data to derive nineteen bioclimatic variables. Next, we determined the key bioclimatic variables explaining variation in species richness across Zimbabwe using regression analysis. Our results show that two bioclimatic variables, that is, precipitation of the warmest quarter (R² = 0.92, P < 0.001) and temperature of the warmest month (R² = 0.67, P < 0.001) significantly explain variation in plant species richness. In addition, results of bioclimatic modelling using future climate change projections show a reduction in the current bio‐climatically suitable area that supports high plant species richness. However, in high‐altitude areas, plant richness is less sensitive to climate change while low‐altitude areas show high sensitivity. Our results have important implications to biodiversity conservation in areas sensitive to climate change; for example, high‐altitude areas are likely to continue being biodiversity hotspots, as such future conservation efforts should be concentrated in these areas.     

Biotic interactions improve prediction of boreal bird distributions at macro-scales

Aim The role of biotic interactions in influencing species distributions at macro‐scales remains poorly understood. Here we test whether predictions of distributions for four boreal owl species at two macro‐scales (10 × 10 km and 40 × 40 km grid resolutions) are improved by incorporating interactions with woodpeckers into climate envelope models. Location Finland, northern Europe. Methods Distribution data for four owl and six woodpecker species, along with data for six land cover and three climatic variables, were collated from 2861 10 × 10 km grid cells. Generalized additive models were calibrated using a 50% random sample of the species data from western Finland, and by repeating this procedure 20 times for each of the four owl species. Models were fitted using three sets of explanatory variables: (1) climate only; (2) climate and land cover; and (3) climate, land cover and two woodpecker interaction variables. Models were evaluated using three approaches: (1) examination of explained deviance; (2) four‐fold cross‐validation using the model calibration data; and (3) comparison of predicted and observed values for independent grid cells in eastern Finland. The model accuracy for approaches (2) and (3) was measured using the area under the curve of a receiver operating characteristic plot. Results At 10‐km resolution, inclusion of the distribution of woodpeckers as a predictor variable significantly improved the explanatory power, cross‐validation statistics and the predictive accuracy of the models. Inclusion of land cover led to similar improvements at 10‐km resolution, although these improvements were less apparent at 40‐km resolution for both land cover and biotic interactions. Main conclusions Predictions of species distributions at macro‐scales may be significantly improved by incorporating biotic interactions and land cover variables into models. Our results are important for models used to predict the impacts of climate change, and emphasize the need for comprehensive evaluation of the reliability of species–climate impact models.     

Inclusion of soil data improves the performance of bioclimatic envelope models for insect species distributions in temperate Europe

Aim To analyse the effect of the inclusion of soil and land‐cover data on the performance of bioclimatic envelope models for the regional‐scale prediction of butterfly (Rhopalocera) and grasshopper (Orthoptera) distributions. Location Temperate Europe (Belgium). Methods Distributional data were extracted from butterfly and grasshopper atlases at a resolution of 5 km for the period 1991–2006 in Belgium. For each group separately, the well‐surveyed squares (n = 366 for butterflies and n = 322 for grasshoppers) were identified using an environmental stratification design and were randomly divided into calibration (70%) and evaluation (30%) datasets. Generalized additive models were applied to the calibration dataset to estimate occurrence probabilities for 63 butterfly and 33 grasshopper species, as a function of: (1) climate, (2) climate and land‐cover, (3) climate and soil, and (4) climate, land‐cover and soil variables. Models were evaluated as: (1) the amount of explained deviance in the calibration dataset, (2) Akaike’s information criterion, and (3) the number of omission and commission errors in the evaluation dataset. Results Information on broad land‐cover classes or predominant soil types led to similar improvements in the performance relative to the climate‐only models for both taxonomic groups. In addition, the joint inclusion of land‐cover and soil variables in the models provided predictions that fitted more closely to the species distributions than the predictions obtained from bioclimatic models incorporating only land‐cover or only soil variables. The combined models exhibited higher discrimination ability between the presence and absence of species in the evaluation dataset. Main conclusions These results draw attention to the importance of soil data for species distribution models at regional scales of analysis. The combined inclusion of land‐cover and soil data in the models makes it possible to identify areas with suitable climatic conditions but unsuitable combinations of vegetation and soil types. While contingent on the species, the results indicate the need to consider soil information in regional‐scale species–climate impact models, particularly when predicting future range shifts of species under climate change.     

Species traits are associated with the quality of bioclimatic models

Aim To analyse the effects of nine species trait variables on the accuracy of bioclimatic envelope models built for 98 butterfly species. Location Finland, northern Europe. Methods Data from a national butterfly atlas monitoring scheme (NAFI) collected from 1991–2003 with a resolution of 10 × 10 km were used in the analyses. Generalized additive models (GAMs) were constructed for 98 butterfly species to predict their occurrence as a function of climatic variables. Modelling accuracy was measured as the cross‐validation area under the curve (AUC) of the receiver–operating characteristic plot. Observed variation in modelling accuracy was related to species traits using multiple GAMs. The effects of phylogenetic relatedness among butterflies were accounted for by using generalized estimation equations. Results The values of the cross‐validation AUC for the 98 species varied between 0.56 and 1.00 with a mean of 0.79. Five species trait variables were included in the GAM that explained 71.4% of the observed variation in modelling accuracy. Four variables remained significant after accounting for phylogenetic relatedness. Species with high mobility and a long flight period were modelled less accurately than species with low mobility and a short flight period. Large species (>50 mm in wing span) were modelled more accurately than small ones. Species inhabiting mires had especially poor models, whereas the models for species inhabiting rocky outcrops, field verges and open fells were more accurate compared with other habitats. Main conclusions These results draw attention to the importance of species traits variables for species–climate impact models. Most importantly, species traits may have a strong impact on the performance of bioclimatic envelope models, and certain trait groups can be inherently difficult to model reliably. These uncertainties should be taken into account by downweighting or excluding species with such traits in studies applying bioclimatic modelling and making assessments of the impacts of climate change.     

Using functional traits to predict species growth trajectories,and cross‐validation to evaluate these models for ecological prediction

Modeling plant growth using functional traits is important for understanding the mechanisms that underpin growth and for predicting new situations. We use three data sets on plant height over time and two validation methods—in‐sample model fit and leave‐one‐species‐out cross‐validation—to evaluate non‐linear growth model predictive performance based on functional traits. In‐sample measures of model fit differed substantially from out‐of‐sample model predictive performance; the best fitting models were rarely the best predictive models. Careful selection of predictor variables reduced the bias in parameter estimates, and there was no single best model across our three data sets. Testing and comparing multiple model forms is important. We developed an R package with a formula interface for straightforward fitting and validation of hierarchical, non‐linear growth models. Our intent is to encourage thorough testing of multiple growth model forms and an increased emphasis on assessing model fit relative to a model's purpose.