Temporal resolutions in species distribution models of highly mobile marine animals: Recommendations for ecologists and managers

While ecologists have long recognized the influence of spatial resolution on species distribution models (SDMs), they have given relatively little attention to the influence of temporal resolution. Considering temporal resolutions is critical in distribution modelling of highly mobile marine animals, as they interact with dynamic oceanographic processes that vary at time‐scales from seconds to decades. We guide ecologists in selecting temporal resolutions that best match ecological questions and ecosystems, and managers in applying these models. We group the temporal resolutions of environmental variables used in SDMs into three classes: instantaneous, contemporaneous and climatological. We posit that animal associations with fine‐scale and ephemeral features are best modelled with instantaneous covariates. Associations with large scale and persistent oceanographic features are best modelled with climatological covariates. Associations with mesoscale features are best modelled with instantaneous or contemporaneous covariates if ephemeral processes are present or interannual variability occurs, and climatological covariates if seasonal processes dominate and interannual variability is weak.     

Detection of plot-level changes in ectomycorrhizal communities across years in an old-growth mixed-conifer forest

Understanding spatial and temporal patterns present in ectomycorrhizal fungal community structure is critical to understanding both the scale and duration of the potential impact these fungi have on the plant community. While recent studies consider the spatial structure of ectomycorrhizal communities, few studies consider how this changes over time. Ectomycorrhizal root biomass and the similarity of community composition were measured at scales up to 20 cm replicated in nine plots and over 3 yr. Soil cores were additionally stratified into three depths. Annual occurrence of the dominant ectomycorrhizal species was constant at larger spatial scales but varied more across years at a fine spatial scale. Turnover of ectomycorrhizal species between years was observed frequently at scales < 20 cm. The ectomycorrhizal community within a plot was more similar across years than it was to other plots sampled in the same year. Our results demonstrate the dynamic nature of the ectomycorrhizal community even in the absence of large-scale disturbances. The potential role of root turnover and drought stress is discussed.     

生态学中的尺度问题——尺度上推

尺度推绎是生态学理论和应用的核心。如何在一个异质景观中进行尺度推绎仍然是一个悬而未决的科学难题,是对当今生态学家在全球变化背景下研究环境问题的重大挑战。就目前的研究,一般可分为四大类尺度推绎途径:空间分析法(如分维分析法和小波分析法)、基于相似性的尺度上推方法、基于局域动态模型的尺度上推方法、随机(模型)法。基于相似性的尺度上推方法来源于生物学上的异量关联,可将其思想延伸至空间上,研究物种丰富度、自然河网、地形特征、生态学格局或过程变量和景观指数等。基于局域动态模型的尺度上推方法需要首先确定是否进行跨尺度推绎,以及是否考虑空间单元之间的水平相互作用和反馈,然后再应用具体的方法或途径,如简单聚合法、有效值外推法、直接外推法、期望值外推、显式积分法和空间相互作用模拟法等。随机(模型)法以其它尺度上推方法为基础,根据研究的是单个景观,还是多个景观,采用不同的途径。理解、定量和降低尺度推绎结果的不确定性已经变得越来越重要,但相关研究仍然极少。以上所有有关尺度推绎的方法、途径和结果分析共同构成了尺度推绎的概念框架。     

Spatial patterns of weeds along a gradient of landscape complexity

Processes that drive spatial patterning among plant species are of ongoing interest, mostly because these patterns have implications for the structure and function of plant communities. We investigated the spatial strategies of weeds focusing on how spatial patterns of weeds are mediated by agricultural landscape complexity and species life-history attributes. We quantified the spatial distribution of 110 weed species using data collected in ten landscapes in central western France along a gradient of landscape complexity, from structurally complex (numerous small fields) to structurally simple (few large fields). We then related differences observed in species’ distribution patterns to ecological attributes of species for resource exploitation and dispersion. Our study reveals that weeds were spatially aggregated at the landscape scale. Their spatial patterns are related to the frequency of occurrence of weeds but surprisingly not directly to the seed dispersal type, nor to the degree of habitat specialization. We show that landscape complexity had no direct effect on the spatial patterning of weeds but through interactions with species attributes. Our results point to the importance of interactions between landscape complexity and species attributes in the spatial patterning of weed species even in intensively managed fields. These patterns appear to be a consequence of the spatial arrangement of landscape elements as well as the result of landscape filtering on species attributes.     

Predicted effects of climate factors on mountain species are not uniform over different spatial scales

The selection of relevant factors and appropriate spatial scale(s) is fundamental when modelling species response to climate change. We evaluated whether the effects of climate factors on species distribution/occurrence are consistently modelled over different spatial scales in birds, and used a two‐scale approach to identify species–climate correlations unlikely to represent causal effects. We used passerine birds inhabiting mountain grassland in the Apennines (Italy) as a model. We surveyed four grassland species at 400 sampling points, and built habitat selection models (territory scale) and distribution models (seven algorithms, landscape scale). We compared the effect of climatic predictors on occurrence/distribution highlighted by models over the two spatial scales, and with the effects supposed a priori based on the climatic niche of each species. Models at the territory level included at least one climatic predictor for three species; the observed effect of climatic predictors was seldom consistent with supposed effects. At the broadest scale, distribution models for all species included climatic predictors, with varying consistence with supposed effects and findings at the finer scale. Despite the importance of climate for species distribution, occurrence could be more directly related to other factors, with important implications for understanding/predicting the impacts of climate/environmental changes. Our approach revealed key variables for grassland birds, and highlighted the scale‐dependent perceived importance of climate. At the local scale, climate effects were weak or hard to interpret. We found a general lack of consistence between supposed and observed effects at the territory level, and between landscape and territory models. Our results show the importance of predicting the potential effect of climatic factors prior to the analyses, carefully selecting ecologically meaningful variables and scales, and evaluating the nature and scale of climate–species links. We call for caution when predicting under future climates, especially when mechanistic effects and consistency across scales are lacking.     

A Multi-Scale Distribution Model for Non-Equilibrium Populations Suggests Resource Limitation in an Endangered Rodent

Species distributions are known to be limited by biotic and abiotic factors at multiple temporal and spatial scales. Species distribution models, however, frequently assume a population at equilibrium in both time and space. Studies of habitat selection have repeatedly shown the difficulty of estimating resource selection if the scale or extent of analysis is incorrect. Here, we present a multi-step approach to estimate the realized and potential distribution of the endangered giant kangaroo rat. First, we estimate the potential distribution by modeling suitability at a range-wide scale using static bioclimatic variables. We then examine annual changes in extent at a population-level. We define “available” habitat based on the total suitable potential distribution at the range-wide scale. Then, within the available habitat, model changes in population extent driven by multiple measures of resource availability. By modeling distributions for a population with robust estimates of population extent through time, and ecologically relevant predictor variables, we improved the predictive ability of SDMs, as well as revealed an unanticipated relationship between population extent and precipitation at multiple scales. At a range-wide scale, the best model indicated the giant kangaroo rat was limited to areas that received little to no precipitation in the summer months. In contrast, the best model for shorter time scales showed a positive relation with resource abundance, driven by precipitation, in the current and previous year. These results suggest that the distribution of the giant kangaroo rat was limited to the wettest parts of the drier areas within the study region. This multi-step approach reinforces the differing relationship species may have with environmental variables at different scales, provides a novel method for defining “available” habitat in habitat selection studies, and suggests a way to create distribution models at spatial and temporal scales relevant to theoretical and applied ecologists.     

Effects of spatial scale and sample size in GPS-based species distribution models: are the best models trivial for red deer management?

Species distribution models (SDMs) are popular in conservation and management of a wide array of taxa. Often parameterized with coarse GIS-based environmental maps, they perform well in macro-ecological settings but it is debated if the models can predict distribution within broadly suitable “known” habitats of interest to local managers. We parameterized SDMs with GIS-derived environmental variables and location data from 82 GPS-collared female red deer (Cervus elaphus) from two study areas in Norway. Candidate GLM models were fitted to address the effect of spatial scale (landscape vs. home range), sample size, and transferability between study areas, with respect to predictability (AUC) and explained variance (Generalized R ² and deviance). The landscape level SDM captured variation in deer distribution well and performed best on all diagnostic measures of model quality, caused mainly by a trivial effect of avoidance of non-habitat (barren mountains). The home range level SDMs were far less predictable and explained comparatively little variation in space use. Landscape scale models stabilized at the low sample size of 5–10 individuals and were highly transferrable between study areas implying a low degree of individual variation in habitat selection at this scale. It is important to have realistic expectations of SDMs derived from digital elevation models and coarse habitat maps. They do perform well in highlighting potential habitat on a landscape scale, but often miss nuances necessary to predict more fine-scaled distribution of wildlife populations. Currently, there seems to be a trade-off between model quality and usefulness in local management.     

Habitat selection by a large herbivore at multiple spatial and temporal scales is primarily governed by food resources

Habitat selection can be considered as a hierarchical process in which animals satisfy their habitat requirements at different ecological scales. Theory predicts that spatial and temporal scales should co‐vary in most ecological processes and that the most limiting factors should drive habitat selection at coarse ecological scales, but be less influential at finer scales. Using detailed location data on roe deer Capreolus capreolus inhabiting the Bavarian Forest National Park, Germany, we investigated habitat selection at several spatial and temporal scales. We tested 1) whether time‐varying patterns were governed by factors reported as having the largest effects on fitness, 2) whether the trade‐off between forage and predation risks differed among spatial and temporal scales and 3) if spatial and temporal scales are positively associated. We analysed the variation in habitat selection within the landscape and within home ranges at monthly intervals, with respect to land‐cover type and proxys of food and cover over seasonal and diurnal temporal scales. The fine‐scale temporal variation follows a nycthemeral cycle linked to diurnal variation in human disturbance. The large‐scale variation matches seasonal plant phenology, suggesting food resources being a greater limiting factor than lynx predation risk. The trade‐off between selection for food and cover was similar on seasonal and diurnal scale. Habitat selection at the different scales may be the consequence of the temporal variation and predictability of the limiting factors as much as its association with fitness. The landscape of fear might have less importance at the studied scale of habitat selection than generally accepted because of the predator hunting strategy. Finally, seasonal variation in habitat selection was similar at the large and small spatial scales, which may arise because of the marked philopatry of roe deer. The difference is supposed to be greater for wider ranging herbivores.     

The relative effects of local management and landscape context on weed seed predation and carabid functional groups

Farmland biodiversity and its associated ecosystem services are affected by agricultural activities at multiple spatial scales. Among these services, the regulation of weeds by invertebrate seed predators has received much attention recently but little is known about the relative effect of local management and landscape context of fields on this process. We monitored seed predation on four common weed species and carabid communities in 28 winter-cereals fields during five consecutive weeks in spring 2010. These fields were situated in contrasted landscape contexts and varied in terms of intensity of pesticide treatments and soil tillage regimes. Weed seed predation was strongly and positively related to the Shannon diversity of (strictly) granivorous carabids and to the activity–density of omnivorous carabids but negatively to the richness of omnivorous carabids. Weed seed predation and granivore diversity were positively related to landscape diversity and the proportion cover of temporary grassland within a 1000 m radius around focal fields and were negatively affected by the intensity of local pesticide treatments. No-till systems sheltered higher diversity of granivorous carabids but did not show higher seed predation rates. We showed that landscape composition factors had a higher relative influence than local practices factors on weed seed predation service. Consequently, weed management strategies should not only consider the management of single fields but also the surrounding landscape to preserve carabid biodiversity and enhance weed seed predation service.     

Patterns of resource tracking by avian frugivores at multiple spatial scales: two case studies on discordance among scales

Scaling is relevant for the analysis of plant‐frugivore interaction, since the ecological and evolutionary outcomes of seed dispersal depend on the spatial and temporal scale at which frugivory patterns emerge. We analyse the relationship between fruit abundance and frugivore activity at local and landscape spatial scales in two different systems composed, respectively, by the bird‐dispersed woody plants Juniperus communis and Bursera fagaroides, and their frugivore assemblages. We use a hierarchical approach of nested patchiness of fruit‐resource, where patches are defined by individual plants within site, at the local scale, and by sites within region, at the landscape scale. The structure of patches is also described in terms of contrast (differences in fruit availability among patches) and aggregation (spatial distribution of patches). For J. communis, frugivore activity was positively related to fruit availability at the landscape scale, this pattern seldom emerging at the local scale; conversely, B. fagaroides showed a general trend of positive local pattern that disappeared at the landscape scale. These particular trends might be partially explained by differences in contrast and aggregation. The strong contrast among plants within site together with a high aggregation among sites would promote the B. fagaroides pattern to be only local, whereas in J. communis, low aggregation among sites within region would favour a sharp landscape‐scale pattern. Both systems showed discordant patterns of fruit‐resource tracking among consecutive spatial scales, but the sense of discordance differed among systems. These results, and the available multi‐scale frugivory data, suggest that discordance among successive scales allows to link directly frugivory patterns to resource‐tracking mechanisms acting at particular scales, resulting, thus, more informative than concordance observational data, in which landscape patterns might result from accumulated effect of local mechanisms. In this context, we propose new methodological approaches for a better understanding of the hierarchical behavioural mechanisms underpinning the multi‐scale resource tracking by frugivores.     

A dynamic multi‐scale occupancy model to estimate temporal dynamics and hierarchical habitat use for nomadic species

Distribution models are increasingly being used to understand how landscape and climatic changes are affecting the processes driving spatial and temporal distributions of plants and animals. However, many modeling efforts ignore the dynamic processes that drive distributional patterns at different scales, which may result in misleading inference about the factors influencing species distributions. Current occupancy models allow estimation of occupancy at different scales and, separately, estimation of immigration and emigration. However, joint estimation of local extinction, colonization, and occupancy within a multi‐scale model is currently unpublished. We extended multi‐scale models to account for the dynamic processes governing species distributions, while concurrently modeling local‐scale availability. We fit the model to data for lark buntings and chestnut‐collared longspurs in the Great Plains, USA, collected under the Integrated Monitoring in Bird Conservation Regions program. We investigate how the amount of grassland and shrubland and annual vegetation conditions affect bird occupancy dynamics and local vegetation structure affects fine‐scale occupancy. Buntings were prevalent and longspurs rare in our study area, but both species were locally prevalent when present. Buntings colonized sites with preferred habitat configurations, longspurs colonized a wider range of landscape conditions, and site persistence of both was higher at sites with greener vegetation. Turnover rates were high for both species, quantifying the nomadic behavior of the species. Our model allows researchers to jointly investigate temporal dynamics of species distributions and hierarchical habitat use. Our results indicate that grassland birds respond to different covariates at landscape and local scales suggesting different conservation goals at each scale. High turnover rates of these species highlight the need to account for the dynamics of nomadic species, and our model can help inform how to coordinate management efforts to provide appropriate habitat configurations at the landscape scale and provide habitat targets for local managers.     

Not seeing the forest for the trees: Generalised linear model out-performs random forest in species distribution modelling for Southeast Asian felids

Species Distribution Models (SDMs) are a powerful tool to derive habitat suitability predictions relating species occurrence data with habitat features. Two of the most frequently applied algorithms to model species-habitat relationships are Generalised Linear Models (GLM) and Random Forest (RF). The former is a parametric regression model providing functional models with direct interpretability. The latter is a machine learning non-parametric algorithm, more tolerant than other approaches in its assumptions, which has often been shown to outperform parametric algorithms. Other approaches have been developed to produce robust SDMs, like training data bootstrapping and spatial scale optimisation. Using felid presence-absence data from three study regions in Southeast Asia (mainland, Borneo and Sumatra), we tested the performances of SDMs by implementing four modelling frameworks: GLM and RF with bootstrapped and non-bootstrapped training data. With Mantel and ANOVA tests we explored how the four combinations of algorithms and bootstrapping influenced SDMs and their predictive performances. Additionally, we tested how scale-optimisation responded to species' size, taxonomic associations (species and genus), study area and algorithm. We found that choice of algorithm had strong effect in determining the differences between SDMs' spatial predictions, while bootstrapping had no effect. Additionally, algorithm followed by study area and species, were the main factors driving differences in the spatial scales identified. SDMs trained with GLM showed higher predictive performance, however, ANOVA tests revealed that algorithm had significant effect only in explaining the variance observed in sensitivity and specificity and, when interacting with bootstrapping, in Percent Correctly Classified (PCC). Bootstrapping significantly explained the variance in specificity, PCC and True Skills Statistics (TSS). Our results suggest that there are systematic differences in the scales identified and in the predictions produced by GLM vs. RF, but that neither approach was consistently better than the other. The divergent predictions and inconsistent predictive abilities suggest that analysts should not assume machine learning is inherently superior and should test multiple methods. Our results have strong implications for SDM development, revealing the inconsistencies introduced by the choice of algorithm on scale optimisation, with GLM selecting broader scales than RF.     

Integrating anthropogenic factors into regional‐scale species distribution models—A novel application in the imperiled sagebrush biome

Species distribution models (SDMs) that rely on regional‐scale environmental variables will play a key role in forecasting species occurrence in the face of climate change. However, in the Anthropocene, a number of local‐scale anthropogenic variables, including wildfire history, land‐use change, invasive species, and ecological restoration practices can override regional‐scale variables to drive patterns of species distribution. Incorporating these human‐induced factors into SDMs remains a major research challenge, in part because spatial variability in these factors occurs at fine scales, rendering prediction over regional extents problematic. Here, we used big sagebrush (Artemisia tridentata Nutt.) as a model species to explore whether including human‐induced factors improves the fit of the SDM. We applied a Bayesian hurdle spatial approach using 21,753 data points of field‐sampled vegetation obtained from the LANDFIRE program to model sagebrush occurrence and cover by incorporating fire history metrics and restoration treatments from 1980 to 2015 throughout the Great Basin of North America. Models including fire attributes and restoration treatments performed better than those including only climate and topographic variables. Number of fires and fire occurrence had the strongest relative effects on big sagebrush occurrence and cover, respectively. The models predicted that the probability of big sagebrush occurrence decreases by 1.2% (95% CI: ?6.9%, 0.6%) when one fire occurs and cover decreases by 44.7% (95% CI: ?47.9%, ?41.3%) if at least one fire occurred over the 36 year period of record. Restoration practices increased the probability of big sagebrush occurrence but had minimal effect on cover. Our results demonstrate the potential value of including disturbance and land management along with climate in models to predict species distributions. As an increasing number of datasets representing land‐use history become available, we anticipate that our modeling framework will have broad relevance across a range of biomes and species.     

Regional distribution of genetically modified organisms (GMOs)—Up-scaling the dispersal and persistence potential of herbicide resistant oilseed rape (Brassisca napus)

Most genetically modified (GM) crop plants are designed to be grown on large areas. However, empirical investigations for risk assessment are limited in their temporal and spatial extent. In the case of GM crop plants it is difficult to test the relevance of anticipated risks on the same spatial scale as the intended use. Processes which are difficult to assess experimentally include combinatory effects, interactions between different integration levels, persistence, long distance dispersal and occurrence of rare events. To a limited extent, it is possible to combine results of investigations on small spatial scales in a way that large-scale and long-term implications on the regional scale can be analysed by using modelling and extrapolation approaches. It is thus possible to indicate some of the involved risks which are not accessible otherwise.In this paper we present the results of an extrapolation methodology comprising several scales from the field size up to the landscape level. This methodology aimed at analysing the implications of a large-scale release of genetically modified oilseed rape (GM OSR). The approach consisted of an extrapolation scheme beginning with a landscape analysis which generated representative scenarios considering climate and OSR cultivation characteristics. For the spatial extent of several fields this information was applied in an individual-based model representing ontogeny, dispersal and persistence of cultivated, volunteers and feral oilseed rape. In a final step, simulation results were extrapolated to the region of Northern Germany.Here we focus on the model results which were extrapolated to the regional level by applying a set of ecological indicators which allowed to assess potential implications on this level. These indicators included the number and distribution of flowering GM plants and the dynamics of GM OSR seeds in the soil seedbank. Specific results related to the long-term dynamics in the seedbank and volunteer development. Model results emphasise the long-term consequences of GM OSR cultivation and the explicit necessity to regard high variability in potential GMO admixture. This has to be considered when developing landscape management schemes for co-existence.The extrapolation approach presented here, integrates different traits to assess effects of GMOs on large spatial scales with respect to persistence and dispersal. The developed methodology is equally applicable for other crops, regions and different agricultural conditions.     

Using Species Distribution Models For Fungi

Species distribution models (SDMs) are an emerging tool in the study of fungi, and their use is expanding across species and research topics. To summarise progress to date and to highlight important considerations for future users, we review 283 studies that apply SDMs to fungi. We found that macrofungi, lichens, and pathogenic microfungi are most often studied. While many studies only aim to model species response to environmental covariates, the use of SDMs for explicitly predicting fungal occurrence in space and time is growing. Many studies collect fungal occurrence data, but the use of pre-collected records from reference collections and citizen science programs is increasing. Challenges of applying SDMs to fungi include detection and sampling biases, and uncertainties in identification and taxonomy. Further, finding environmental covariates at appropriate spatial and temporal scales is important, as fungi can respond to fine-scale environmental patterns. Fine-scale covariate data can be difficult to gather across space, but we show remote-sensing measurements are viable for fungi SDMs. For those fungi interacting with host species, host information is also important, and can be used as covariates in SDMs. We also highlight that competition among fungi, and dispersal, can affect observed distributions, with the latter particularly prominent for invasive fungi. We show how one can account for these processes in models, when suitable data are available. Finally, we note that environmental DNA records create new opportunities and challenges for future modelling efforts, and discuss the difficulties in predicting invasions and climate change impacts. The application of SDMs to fungi has already provided interesting lessons on how to adapt modelling tools for specific questions, and fungi will continue to be relevant test subjects for further technical development of SDMs.     

A multi-scale comparison of trait linkages to environmental and spatial variables in fish communities across a large freshwater lake

Species present in communities are affected by the prevailing environmental conditions, and the traits that these species display may be sensitive indicators of community responses to environmental change. However, interpretation of community responses may be confounded by environmental variation at different spatial scales. Using a hierarchical approach, we assessed the spatial and temporal variation of traits in coastal fish communities in Lake Huron over a 5-year time period (2001–2005) in response to biotic and abiotic environmental factors. The association of environmental and spatial variables with trophic, life-history, and thermal traits at two spatial scales (regional basin-scale, local site-scale) was quantified using multivariate statistics and variation partitioning. We defined these two scales (regional, local) on which to measure variation and then applied this measurement framework identically in all 5 study years. With this framework, we found that there was no change in the spatial scales of fish community traits over the course of the study, although there were small inter-annual shifts in the importance of regional basin- and local site-scale variables in determining community trait composition (e.g., life-history, trophic, and thermal). The overriding effects of regional-scale variables may be related to inter-annual variation in average summer temperature. Additionally, drivers of fish community traits were highly variable among study years, with some years dominated by environmental variation and others dominated by spatially structured variation. The influence of spatial factors on trait composition was dynamic, which suggests that spatial patterns in fish communities over large landscapes are transient. Air temperature and vegetation were significant variables in most years, underscoring the importance of future climate change and shoreline development as drivers of fish community structure. Overall, a trait-based hierarchical framework may be a useful conservation tool, as it highlights the multi-scaled interactive effect of variables over a large landscape.     

The spatial scaling of beta diversity

Beta diversity is an important concept used to describe turnover in species composition across a wide range of spatial and temporal scales, and it underpins much of conservation theory and practice. Although substantial progress has been made in the mathematical and terminological treatment of different measures of beta diversity, there has been little conceptual synthesis of potential scale dependence of beta diversity with increasing spatial grain and geographic extent of sampling. Here, we evaluate different conceptual approaches to the spatial scaling of beta diversity, interpreted from ‘fixed’ and ‘varying’ perspectives of spatial grain and extent. We argue that a ‘sliding window’ perspective, in which spatial grain and extent covary, is an informative way to conceptualize community differentiation across scales. This concept more realistically reflects the varying empirical approaches that researchers adopt in field sampling and the varying scales of landscape perception by different organisms. Scale dependence in beta diversity has broad implications for emerging fields in ecology and biogeography, such as the integration of fine‐resolution ecogenomic data with large‐scale macroecological studies, as well as for guiding appropriate management responses to threats to biodiversity operating at different spatial scales.     

A Bayesian approach for multiscale modeling of the influence of seasonal and annual habitat variation on relative abundance of ring-necked pheasant roosters

Finding ecologically relevant relationships between environmental covariates and response variables requires determining appropriate scales of effect. While considering multiple spatial scales of effect in hierarchical models has been the focus of recent studies, the effect of spatiotemporal scales, and temporal resolution of data on habitat suitability and species abundance has received less attention. We investigated relationships between ring-necked pheasant rooster abundance and environmental covariates with the goal of identifying important variables and their scales of effect in South Dakota, U.S.A. Using a suite of remote sensing data, we examined whether seasonal environmental conditions influence pheasant relative abundance and how survey conditions might affect detectability of roosters. To select optimal scales of effect and the best subset of covariates simultaneously, we employed a Reversible-Jump Monte Carlo Markov Chain (RJMCMC) approach in a Bayesian framework. We explored sources of uncertainty in data and controlled them through consideration of random effects. The use of seasonal covariates in addition to annual covariates revealed differential effects on species abundance. The proportion of grasslands on the landscape was an important covariate in models in all years, with rooster abundance generally being highest at intermediate levels of grassland density at local scales of effect. Pheasant abundance was also positively related to the proportion of small grain crop cover on the landscape at >2 km scales. Spring gross primary productivity and percentage of herbaceous wetlands on the landscape, both at a large scale (8 km), were the most important covariates in the wet years of 2018 and 2019 and were positively related to pheasant abundance. Grasslands at intermediate levels of density explained variability of pheasant abundance. However, other variables important to pheasant relative abundance varied among years depending on prevailing weather and climate conditions. Our workflow to model relationships between relative abundance and habitat components for pheasants can also be employed to model count data for other species to inform management decisions.     

Testing projected wild bee distributions in agricultural habitats: predictive power depends on species traits and habitat type

Species distribution models (SDM) are increasingly used to understand the factors that regulate variation in biodiversity patterns and to help plan conservation strategies. However, these models are rarely validated with independently collected data and it is unclear whether SDM performance is maintained across distinct habitats and for species with different functional traits. Highly mobile species, such as bees, can be particularly challenging to model. Here, we use independent sets of occurrence data collected systematically in several agricultural habitats to test how the predictive performance of SDMs for wild bee species depends on species traits, habitat type, and sampling technique. We used a species distribution modeling approach parametrized for the Netherlands, with presence records from 1990 to 2010 for 193 Dutch wild bees. For each species, we built a Maxent model based on 13 climate and landscape variables. We tested the predictive performance of the SDMs with independent datasets collected from orchards and arable fields across the Netherlands from 2010 to 2013, using transect surveys or pan traps. Model predictive performance depended on species traits and habitat type. Occurrence of bee species specialized in habitat and diet was better predicted than generalist bees. Predictions of habitat suitability were also more precise for habitats that are temporally more stable (orchards) than for habitats that suffer regular alterations (arable), particularly for small, solitary bees. As a conservation tool, SDMs are best suited to modeling rarer, specialist species than more generalist and will work best in long‐term stable habitats. The variability of complex, short‐term habitats is difficult to capture in such models and historical land use generally has low thematic resolution. To improve SDMs’ usefulness, models require explanatory variables and collection data that include detailed landscape characteristics, for example, variability of crops and flower availability. Additionally, testing SDMs with field surveys should involve multiple collection techniques.