Disentangling drivers of spatial autocorrelation in species distribution models

Mielke, Konrad P.; Claassen, Tom; Busana, Michela; Heskes, Tom; Huijbregts, Mark A. J.; Koffijberg, Kees; Schipper, Aafke M.

doi:10.1111/ecog.05134

Cited by 20 publications

(15 citation statements)

References 51 publications

(54 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This large distance implies that the influence of endogenous processes is likely difficult to detect. By considering a measure of topographic distance, we attempted to highlight the influence of such processes but obtained similar results to Euclidean distancebased analyses, confirming that these processes are complex to detect empirically, unless additional data are considered (Mielke et al 2020). Our results, however, suggest that SAC is likely of exogenous origin in this study area.…”

Section: Discussionsupporting

confidence: 65%

Low spatial autocorrelation in mountain biodiversity data and model residuals

et al. 2021

View full text Add to dashboard Cite

Spatial autocorrelation (SAC) is a common feature of ecological data where observations tend to be more similar at some geographic distance(s) than expected by chance. Despite the implications of SAC for data dependencies, its impact on the performance of species distribution models (SDMs) remains controversial, with reports of both strong and negligible impacts on inference. Yet, no study has comprehensively assessed the prevalence and the strength of SAC in the residuals of SDMs over entire geographic areas. Here, we used a large-scale spatial inventory in the western Swiss Alps to provide a thorough assessment of the importance of SAC for (1) 850 species belonging to nine taxonomic groups, (2) six predictors commonly used for modeling species distributions, and (3) residuals obtained from SDMs fitted with two algorithms with the six predictors included as covariates. We used various statistical tools to evaluate (1) the global level of SAC, (2) the spatial pattern and spatial extent of SAC, and (3) whether local clusters of SAC can be detected. We further investigated the effect of the sampling design on SAC levels. Overall, while environmental predictors expectedly displayed high SAC levels, SAC in biodiversity data was rather low overall and vanished rapidly at a distance of~5-10 km. We found low evidence for the existence of local clusters of SAC. Most importantly, model residuals were not spatially autocorrelated, suggesting that inferences derived from SDMs are unlikely to be affected by SAC. Further, our results suggest that the influence of SAC can be reduced by a careful sampling design. Overall, our results suggest that SAC is not a major concern for rugged mountain landscapes.

show abstract

Section: Discussionsupporting

confidence: 65%

Low spatial autocorrelation in mountain biodiversity data and model residuals

et al. 2021

View full text Add to dashboard Cite

show abstract

“…A major issue is that spatial clustering of conspecific, or separation of heterospecific, occurrence records can be affected by multiple factors, which are often difficult to disentangle. These include: i) habitat suitability (Phillips, Anderson, & Schapire, 2006), ii) dispersal limitation (Glor & Warren, 2010), iii) interactions between individuals of the same or different species such as conspecific attraction, competitive exclusion or mutualism (Mielke et al, 2020), or iv) sampling bias, where occurrence records are more likely to be collected from more easily accessible or intensively studied areas (Phillips et al, 2009). One approach to overcome these issues has been to use null species occurrences to define the expectations if only the inherent spatial structure within species has shaped their distributions (Beale et al, 2008;Algar, Mahler, Glor, & Losos, 2013).…”

Section: Introductionmentioning

confidence: 99%

Fauxcurrence: simulating multi-species occurrences for null models in species distribution modelling and biogeography

Osborne

Fell

Atkins

et al. 2021

Preprint

View full text Add to dashboard Cite

Defining appropriate null expectations for species distribution hypotheses is important because sampling bias and spatial autocorrelation can produce realistic, but ecologically meaningless, geographic patterns. Generating null species occurrences with similar spatial structure to observed data can help overcome these problems, but existing methods focus on single or pairs of species and do not incorporate between-species spatial structure that may occlude comparative biogeographic analyses. Here, we describe an algorithm for generating randomised species occurrence points that mimic the within- and between-species spatial structure of real datasets and implement it in a new R package - fauxcurrence. The algorithm can be implemented on any geographic domain for any number of species, limited only by computing power. To demonstrate its utility, we apply the algorithm to two common analysis-types: testing the fit of species distribution models (SDMs) and evaluating niche-overlap. The method works well on all tested datasets within reasonable timescales. We found that many SDMs, despite a good fit to the data, were not significantly better than null expectations and identified only two cases (out of a possible 32) of significantly higher niche divergence than expected by chance. The package is user-friendly, flexible and has many potential applications beyond those tested here, such as joint SDM evaluation and species co-occurrence analysis, spanning the areas of ecology, evolutionary biology and biogeography.

show abstract

“…2b ). We note that the shrub expansion pattern can also be influenced by other factors unaccounted for, leading to a spatial correlation pattern unexplained by the considered covariates 56 . Nonetheless, additionally accounting for spatial correlation of shrub expansion patterns using a spatial regression (Eq.…”

Section: Resultsmentioning

confidence: 96%

Dispersal and fire limit Arctic shrub expansion

et al. 2022

View full text Add to dashboard Cite

Arctic shrub expansion alters carbon budgets, albedo, and warming rates in high latitudes but remains challenging to predict due to unclear underlying controls. Observational studies and models typically use relationships between observed shrub presence and current environmental suitability (bioclimate and topography) to predict shrub expansion, while omitting shrub demographic processes and non-stationary response to changing climate. Here, we use high-resolution satellite imagery across Alaska and western Canada to show that observed shrub expansion has not been controlled by environmental suitability during 1984–2014, but can only be explained by considering seed dispersal and fire. These findings provide the impetus for better observations of recruitment and for incorporating currently underrepresented processes of seed dispersal and fire in land models to project shrub expansion and climate feedbacks. Integrating these dynamic processes with projected fire extent and climate, we estimate shrubs will expand into 25% of the non-shrub tundra by 2100, in contrast to 39% predicted based on increasing environmental suitability alone. Thus, using environmental suitability alone likely overestimates and misrepresents shrub expansion pattern and its associated carbon sink.

show abstract

Disentangling drivers of spatial autocorrelation in species distribution models

Cited by 20 publications

References 51 publications

Low spatial autocorrelation in mountain biodiversity data and model residuals

Low spatial autocorrelation in mountain biodiversity data and model residuals

Fauxcurrence: simulating multi-species occurrences for null models in species distribution modelling and biogeography

Dispersal and fire limit Arctic shrub expansion

Contact Info

Product

Resources

About