Biogeographic multi‐species occupancy models for large‐scale survey data

Ecologists often seek to infer patterns of species occurrence or community structure from survey data. Hierarchical models, including multi‐species occupancy models (MSOMs), can improve inference by pooling information across multiple species via random effects. Originally developed for local‐scale survey data, MSOMs are increasingly applied to larger spatial scales that transcend major abiotic gradients and dispersal barriers. At biogeographic scales, the benefits of partial pooling in MSOMs trade off against the difficulty of incorporating sufficiently complex spatial effects to account for biogeographic variation in occupancy across multiple species simultaneously. We show how this challenge can be overcome by incorporating preexisting range information into MSOMs, yielding a “biogeographic multi‐species occupancy model” (bMSOM). We illustrate the bMSOM using two published datasets: Parulid warblers in the United States Breeding Bird Survey and entire avian communities in forests and pastures of Colombia''s West Andes. Compared with traditional MSOMs, the bMSOM provides dramatically better predictive performance at lower computational cost. The bMSOM avoids severe spatial biases in predictions of the traditional MSOM and provides principled species‐specific inference even for never‐observed species. Incorporating preexisting range data enables principled partial pooling of information across species in large‐scale MSOMs. Our biogeographic framework for multi‐species modeling should be broadly applicable in hierarchical models that predict species occurrences, whether or not false absences are modeled in an occupancy framework.     

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.     

Predicting species distributions from checklist data using site‐occupancy models

Aim (1) To increase awareness of the challenges induced by imperfect detection, which is a fundamental issue in species distribution modelling; (2) to emphasize the value of replicate observations for species distribution modelling; and (3) to show how ‘cheap’ checklist data in faunal/floral databases may be used for the rigorous modelling of distributions by site‐occupancy models. Location Switzerland. Methods We used checklist data collected by volunteers during 1999 and 2000 to analyse the distribution of the blue hawker, Aeshna cyanea (Odonata, Aeshnidae), a common dragonfly in Switzerland. We used data from repeated visits to 1‐ha pixels to derive ‘detection histories’ and apply site‐occupancy models to estimate the ‘true’ species distribution, i.e. corrected for imperfect detection. We modelled blue hawker distribution as a function of elevation and year and its detection probability of elevation, year and season. Results The best model contained cubic polynomial elevation effects for distribution and quadratic effects of elevation and season for detectability. We compared the site‐occupancy model with a conventional distribution model based on a generalized linear model, which assumes perfect detectability (p = 1). The conventional distribution map looked very different from the distribution map obtained using site‐occupancy models that accounted for the imperfect detection. The conventional model underestimated the species distribution by 60%, and the slope parameters of the occurrence–elevation relationship were also underestimated when assuming p = 1. Elevation was not only an important predictor of blue hawker occurrence, but also of the detection probability, with a bell‐shaped relationship. Furthermore, detectability increased over the season. The average detection probability was estimated at only 0.19 per survey. Main conclusions Conventional species distribution models do not model species distributions per se but rather the apparent distribution, i.e. an unknown proportion of species distributions. That unknown proportion is equivalent to detectability. Imperfect detection in conventional species distribution models yields underestimates of the extent of distributions and covariate effects that are biased towards zero. In addition, patterns in detectability will erroneously be ascribed to species distributions. In contrast, site‐occupancy models applied to replicated detection/non‐detection data offer a powerful framework for making inferences about species distributions corrected for imperfect detection. The use of ‘cheap’ checklist data greatly enhances the scope of applications of this useful class of models.     

Predicting changes in the abundance of African wetland birds by incorporating abundance–occupancy relationships into habitat association models

Aim To map changes in the abundance of African wetland birds using remotely derived habitat data. We show that abundance–occupancy relationships can be coupled with habitat association models to map changes in abundance. As conservation resources are more easily allocated when spatial and temporal patterns of abundance are known, our method provides guidance for conservation planning. Location Papyrus, Cyperus papyrus, swamps in east central Africa. Methods Presence/absence surveys of six bird species in 93 wetlands were used to construct models predicting probability of occurrence from habitat characteristics. Densities were then determined from surveys in 23 additional wetlands and modelled as functions of occurrence probability. We then used satellite imagery to derive habitat characteristics remotely in two time periods (1984–87 and 2000–03) and used the modelled relationships between (1) habitat and occupancy and (2) occupancy and density, to infer changes in abundance in all c. 30,000 wetlands within the study area. Results Wetlands within the region declined by 8.6% between the two time periods, but by > 75% in regions of high human population density. Bird densities were also highest in these regions, which comprised wetlands subject to high levels of disturbance. The geographical coincidence of high densities and habitat loss and the existence of positive associations between bird density and occurrence meant that birds declined by much more than the average rate of their habitat. Main conclusions Targeting conservation efforts in areas with high drainage would protect a high proportion of the bird populations. Encouraging people to derive income from disturbance to which the birds are tolerant, rather than drainage, is likely to be an effective strategy. Because habitat characteristics are a key driver of abundance–occupancy relationships, we conclude that there is wide‐scale scope to couple abundance–occupancy relationships with remote habitat mapping to efficiently inform conservation planning.     

Ensemble modeling to predict habitat suitability for a large‐scale disturbance specialist

To conserve habitat for disturbance specialist species, ecologists must identify where individuals will likely settle in newly disturbed areas. Habitat suitability models can predict which sites at new disturbances will most likely attract specialists. Without validation data from newly disturbed areas, however, the best approach for maximizing predictive accuracy can be unclear (Northwestern U.S.A.). We predicted habitat suitability for nesting Black‐backed Woodpeckers (Picoides arcticus; a burned‐forest specialist) at 20 recently (≤6 years postwildfire) burned locations in Montana using models calibrated with data from three locations in Washington, Oregon, and Idaho. We developed 8 models using three techniques (weighted logistic regression, Maxent, and Mahalanobis D² models) and various combinations of four environmental variables describing burn severity, the north–south orientation of topographic slope, and prefire canopy cover. After translating model predictions into binary classifications (0 = low suitability to unsuitable, 1 = high to moderate suitability), we compiled “ensemble predictions,” consisting of the number of models (0–8) predicting any given site as highly suitable. The suitability status for 40% of the area burned by eastside Montana wildfires was consistent across models and therefore robust to uncertainty in the relative accuracy of particular models and in alternative ecological hypotheses they described. Ensemble predictions exhibited two desirable properties: (1) a positive relationship with apparent rates of nest occurrence at calibration locations and (2) declining model agreement outside surveyed environments consistent with our reduced confidence in novel (i.e., “no‐analogue”) environments. Areas of disagreement among models suggested where future surveys could help validate and refine models for an improved understanding of Black‐backed Woodpecker nesting habitat relationships. Ensemble predictions presented here can help guide managers attempting to balance salvage logging with habitat conservation in burned‐forest landscapes where black‐backed woodpecker nest location data are not immediately available. Ensemble modeling represents a promising tool for guiding conservation of large‐scale disturbance specialists.     

Humans alter habitat selection of birds on ocean‐exposed sandy beaches

Aim Resource‐selection functions (RSFs) can quantify and predict the density of animal populations across heterogeneous landscapes and are important conservation tools in areas subject to human disturbance. Sandy beach ecosystems have comparatively low habitat heterogeneity and structural relief in the intertidal zone, but intense human use. We aimed to develop predictive RSFs for birds on ocean‐exposed sandy beaches at two spatial scales, 25 ha (local scale) and 250 ha (landscape scale), and to test whether habitat selection of birds that commonly use the surf–beach–dune interface is influenced by the rates of human activities. Location Moreton and North Stradbroke Island, eastern Australia. Methods Avifauna and human activities were mapped on three sandy beaches covering 79 km of coastline for 15 months. Habitat characteristics of the surf–beach–dune interface were derived from remote sensing and ground surveys. RSFs were developed for 12 species of birds at two spatial scales: 25 ha (local scale) and 250 ha (landscape scale). Results  At local (25 ha) and landscape scales (250 ha), dune dimensions and the extent and type of vegetation structure were important predictors of bird density. Adding the frequency of human activities improved the predictive power of RSFs, suggesting that habitat selection of birds on beaches is modified by human use of these environments. Human activities occurred mostly in the mid‐ to lower intertidal zone of the beach, overlapping closely with the preferred habitats of Silver Gulls (Larus novaehollandiae), Pied Oystercatchers (Haematopus longirostris), Red‐capped Plovers (Charadrius ruficapillus) and endangered Little Terns (Sternula albifrons). Main conclusions In addition to demonstrating the appropriateness of RSFs to the surf–beach–dune interface, our results stress the need for systematic conservation planning for these ecosystems, where ecological values have traditionally been subsidiary to the maintenance of sand budgets and erosion control.     

Predicting effects of large‐scale reforestation on native and exotic birds

Aim

Ecological restoration is critical for recovering biodiversity and ecosystem services, yet designing interventions to achieve particular outcomes remains fraught with challenges. In the extensive regions where non‐native species are firmly established, it is unlikely that historical conditions can be fully reinstated. To what degree, and how rapidly, can human‐dominated areas be shifted via restoration into regimes that benefit target species, communities or processes?

Location

We explore this question in a >20‐year‐old reforestation effort underway at Hakalau Forest National Wildlife Refuge in montane Hawaii. This large‐scale planting of Acacia koa trees is designed to secure populations of globally threatened bird species by transitioning the site rapidly from pasture to native forest.

Methods

We surveyed all forest birds in multiple corridors of young planted trees, remnant corridors of mature trees along gulches and at sites within mature forest. Using a Bayesian hierarchical approach, we identified which factors (distance from forest, habitat type and surrounding tree cover) had the most important influence on native and exotic bird abundance in the reforestation area.

Results

We found that 90% of native and exotic bird species responded quickly, occupying corridors of native trees approximately a decade after planting. However, native and exotic forest birds responded to markedly different characteristics of the reforested area. Native bird abundance was strongly predicted by proximity to mature forest and remnant corridors; conversely, exotic bird abundance was best predicted by overall tree cover throughout the area reforested.

Main conclusions

Our results demonstrate that large‐scale tree planting in corridors adjacent to mature forest can catalyse rapid recovery (both increased abundance and expanded distribution) of forest birds and that it is possible to design reforestation to benefit native species in novel ecosystems.

     

Species traits and catchment‐scale habitat factors influence the occurrence of freshwater mussel populations and assemblages

相似文献   

Effects of point‐count duration on estimated detection probabilities and occupancy of breeding birds

Increasingly, point‐count data are used to estimate occupancy, the probability that a species is present at a given location; occupancy accounts for imperfect detection, the probability that a species is detected given that it is present. To our knowledge, effects of sampling duration on inferences from models of bird occupancy have not been evaluated. Our objective was to determine whether changing count duration from 5 to 8 min affected inferences about the occupancy of birds sampled in the Chesapeake Bay Lowlands (eastern United States) and the central and western Great Basin (western United States) in 2012 and 2013. We examined the proportion of species (two doves, one cuckoo, two swifts, five hummingbirds, 11 woodpeckers, and 122 passerines) for which estimates of detection probability were ≥ 0.3. For species with single‐season detection probabilities ≥ 0.3, we compared occupancy estimates derived from 5‐ and 8‐min counts. We also compared estimates for three species sampled annually for 5 yr in the central Great Basin. Detection probabilities based on both the 5‐ and 8‐min counts were ≥ 0.3 for 40% ± 3% of the species in an ecosystem. Extending the count duration from 5 to 8 min increased the detection probability to ≥ 0.3 for 5% ± 0.5% of the species. We found no difference in occupancy estimates that were based on 5‐ versus 8‐min counts for species sampled over two or five consecutive years. However, for 97% of species sampled over 2 yr, precision of occupancy estimates that were based on 8‐min counts averaged 12% ± 2% higher than those based on 5‐min counts. We suggest that it may be worthwhile to conduct a pilot season to determine the number of locations and surveys needed to achieve detection probabilities that are sufficiently high to estimate occupancy for species of interest.     

栖息地演变与人为干扰对升金湖越冬水鸟的影响

2002年10月—2003年4月,选择安徽升金湖6个不同干扰程度和栖息地质量的固定监测点,研究人为干扰对升金湖越冬水鸟分布格局的影响;同时结合前期调查数据,研究栖息地演变对升金湖越冬水鸟的影响.结果表明:同一越冬期不同监测点水鸟种类和数量与栖息地质量无显著相关关系,但水鸟数量与栖息地干扰程度显著负相关;不同年份升金湖越冬水鸟种类和数量与栖息地质量显著相关.升金湖越冬水鸟存在的主要威胁包括栖息地丧失、人为活动干扰以及生物杀灭剂的影响.提出了一些保护升金湖越冬水鸟的建议.     

Ecological principles of species distribution models: the habitat matching rule

Most species distribution models (SDMs) assume that habitats are closed, stable and without competition. In that environmental context, it is ecologically correct to assume that members of a species will be distributed in direct relation to the suitability of the habitat, that is, according to the so‐called habitat matching rule. This paper examines whether it is possible to maintain the assumption of the habitat matching rule in the following circumstances: (1) when habitats are connected and organisms can move between them, (2) when there are disturbances and seasonal cycles that generate instability, and (3) when there is inter‐specific and intra‐specific competition. Here I argue that it is possible as long as the following aspects are taken into account. In open habitats at equilibrium, in which habitat selection and competition operate, the habitat matching rule can be applied in some conditions, while competition tends to homogenize the species distribution in other environmental contexts. In the latter case, two methods can be used to incorporate these effects into SDMs: new parameters can be incorporated into the response functions, or the occurrence of proportions of categories of individuals (adult/young, male/female, or dominant/subordinate species in guilds) can be used instead of the occurrence of organisms. The habitat matching rule is not fulfilled in non‐equilibrium environments. The solution to this problem lies in the design of SDMs with two strategies that depend on scale. Locally, the disequilibrium can be encapsulated using average environmental conditions, with sufficiently large cells (in the case of metapopulations) and/or long enough sampling periods (in the case of seasonal cycles). At coarse scales, the use of presence‐only models can in some cases avoid the destabilizing effect of catastrophic historical processes. The matching law is a strong assumption of SDMs because it is based on population ecology theory and the principle of evolution by natural selection.     

Meta‐replication,sampling bias,and multi‐scale model selection: A case study on snow leopard (Panthera uncia) in western China

Replicated multiple scale species distribution models (SDMs) have become increasingly important to identify the correct variables determining species distribution and their influences on ecological responses. This study explores multi‐scale habitat relationships of the snow leopard (Panthera uncia) in two study areas on the Qinghai–Tibetan Plateau of western China. Our primary objectives were to evaluate the degree to which snow leopard habitat relationships, expressed by predictors, scales of response, and magnitude of effects, were consistent across study areas or locally landcape‐specific. We coupled univariate scale optimization and the maximum entropy algorithm to produce multivariate SDMs, inferring the relative suitability for the species by ensembling top performing models. We optimized the SDMs based on average omission rate across the top models and ensembles’ overlap with a simulated reference model. Comparison of SDMs in the two study areas highlighted landscape‐specific responses to limiting factors. These were dependent on the effects of the hydrological network, anthropogenic features, topographic complexity, and the heterogeneity of the landcover patch mosaic. Overall, even accounting for specific local differences, we found general landscape attributes associated with snow leopard ecological requirements, consisting of a positive association with uplands and ridges, aggregated low‐contrast landscapes, and large extents of grassy and herbaceous vegetation. As a means to evaluate the performance of two bias correction methods, we explored their effects on three datasets showing a range of bias intensities. The performance of corrections depends on the bias intensity; however, density kernels offered a reliable correction strategy under all circumstances. This study reveals the multi‐scale response of snow leopards to environmental attributes and confirms the role of meta‐replicated study designs for the identification of spatially varying limiting factors. Furthermore, this study makes important contributions to the ongoing discussion about the best approaches for sampling bias correction.     

North by north‐west: climate change and directions of density shifts in birds

There is increasing evidence that climate change shifts species distributions towards poles and mountain tops. However, most studies are based on presence–absence data, and either abundance or the observation effort has rarely been measured. In addition, hardly any studies have investigated the direction of shifts and factors affecting them. Here, we show using count data on a 1000 km south–north gradient in Finland, that between 1970–1989 and 2000–2012, 128 bird species shifted their densities, on average, 37 km towards the north north‐east. The species‐specific directions of the shifts in density were significantly explained by migration behaviour and habitat type. Although the temperatures have also moved on average towards the north north‐east (186 km), the species‐specific directions of the shifts in density and temperature did not correlate due to high variation in density shifts. Findings highlight that climate change is unlikely the only driver of the direction of species density shifts, but species‐specific characteristics and human land‐use practices are also influencing the direction. Furthermore, the alarming results show that former climatic conditions in the north‐west corner of Finland have already moved out of the country. This highlights the need for an international approach in research and conservation actions to mitigate the impacts of climate change.     

Using species distribution models to define nesting habitat of the eastern metapopulation of double‐crested cormorants

When organisms with similar phenotypes have conflicting management and conservation initiatives, approaches are needed to differentiate among subpopulations or discrete groups. For example, the eastern metapopulation of the double‐crested cormorant (Phalacrocorax auritus) has a migratory phenotype that is culled because they are viewed as a threat to commercial and natural resources, whereas resident birds are targeted for conservation. Understanding the distinct breeding habitats of resident versus migratory cormorants would aid in identification and management decisions. Here, we use species distribution models (SDM: Maxent) of cormorant nesting habitat to examine the eastern P. auritus metapopulation and the predicted breeding sites of its phenotypes. We then estimate the phenotypic identity of breeding colonies of cormorants where management plans are being developed. We transferred SDMs trained on data from resident bird colonies in Florida and migratory bird colonies in Minnesota to South Carolina in an effort to identify the phenotype of breeding cormorants there based on the local landscape characteristics. Nesting habitat characteristics of cormorant colonies in South Carolina more closely resembled those of the Florida phenotype than those of birds of the Minnesota phenotype. The presence of the resident phenotype in summer suggests that migratory and resident cormorants will co‐occur in South Carolina in winter. Thus, there is an opportunity for separate management strategies for the two phenotypes in that state. We found differences in nesting habitat characteristics that could be used to refine management strategies and reduce human conflicts with abundant winter migrants and, at the same time, conserve less common colonies of resident cormorants. The models we use here show potential for advancing the study of geographically overlapping phenotypes with differing conservation and management priorities.     

Quantifying the importance of multi‐scale management and environmental variables on moorland bird abundance

UK moorlands are semi‐natural habitats managed for a mix of livestock, game shooting and forestry, among other activities. An assessment of the importance of characteristics that correlate with moorland bird populations of high conservation importance can inform appropriate management at appropriate locations. We use hierarchical partitioning to assess the absolute and relative importance of climate, topography, soil, landscape management (wider scale habitat configuration of forestry and agriculture) and site‐based management (indices of predator control, and vegetation characteristics) in determining the abundance of a suite of upland birds in four regions of the UK. Unmeasured region‐specific effects often made the largest contribution to models. Physical characteristics (climate, topography, soil) were important and on average explained a similar amount of variation to site‐based management. However, there was considerable interspecific variation in the importance of both. Landscape‐scale variables were generally of low importance. An index of predator control was positively correlated with the abundance of Red Grouse Lagopus lagopus scotica and two waders. Vegetation characteristics (composition and structure) were, together, strong correlates of the abundance of passerine species. Vegetation characteristics were as important as indices of predator control for waders and grouse. The importance of regional effects, physical characteristics and variables relating to management suggest targeting site‐based management (such as predator control or vegetation management) to the areas where physical characteristics are most favourable. The most beneficial management will vary between species, necessitating a mosaic of management options across upland areas to benefit all species.     

Wrong,but useful: regional species distribution models may not be improved by range‐wide data under biased sampling

Species distribution modeling (SDM) is an essential method in ecology and conservation. SDMs are often calibrated within one country's borders, typically along a limited environmental gradient with biased and incomplete data, making the quality of these models questionable. In this study, we evaluated how adequate are national presence‐only data for calibrating regional SDMs. We trained SDMs for Egyptian bat species at two different scales: only within Egypt and at a species‐specific global extent. We used two modeling algorithms: Maxent and elastic net, both under the point‐process modeling framework. For each modeling algorithm, we measured the congruence of the predictions of global and regional models for Egypt, assuming that the lower the congruence, the lower the appropriateness of the Egyptian dataset to describe the species' niche. We inspected the effect of incorporating predictions from global models as additional predictor (“prior”) to regional models, and quantified the improvement in terms of AUC and the congruence between regional models run with and without priors. Moreover, we analyzed predictive performance improvements after correction for sampling bias at both scales. On average, predictions from global and regional models in Egypt only weakly concur. Collectively, the use of priors did not lead to much improvement: similar AUC and high congruence between regional models calibrated with and without priors. Correction for sampling bias led to higher model performance, whatever prior used, making the use of priors less pronounced. Under biased and incomplete sampling, the use of global bats data did not improve regional model performance. Without enough bias‐free regional data, we cannot objectively identify the actual improvement of regional models after incorporating information from the global niche. However, we still believe in great potential for global model predictions to guide future surveys and improve regional sampling in data‐poor regions.