Callum R. Lawson

Prevalence, thresholds and the performance of presence–absence models

1.The use of species distribution models to understand and predict species’ distributions necessitates tests of fit to empirical data. Numerous performance metrics have been proposed, many of which require continuous occurrence probabilities to be converted to binary “present or absent” predictions using threshold transformations. It is widely accepted that both continuous and binary performance metrics should be independent of prevalence (the proportion of locations that are occupied). However, because these metrics have been mostly assessed on a case-specific basis, there are few general guidelines for measuring performance. 2.Here, we develop a conceptual framework for classifying performance metrics, based on whether they are sensitive to prevalence, and whether they require binary predictions. We use this framework to investigate how these performance metric properties influence the predictions made by the models they select. 3.A literature survey reveals that binary metrics are widely employed and that prevalence-independent metrics are used more frequently than prevalence-dependent metrics. However, we show that prevalence-dependent metrics are essential to assess the numerical accuracy of model predictions, and are more useful in applications that require occupancy estimates. Furthermore, we demonstrate that in comparison with continuous metrics, binary metrics often select models that have reduced ability to separate presences from absences, make predictions which over- or under-estimate occupancy, and give misleading estimates of uncertainty. Importantly, models selected using binary metrics will often be of reduced practical use even when applied to ecological problems that require binary decision-making. 4.We suggest that SDM performance should be assessed using prevalence-dependent performance metrics whenever the absolute values of occurrence predictions are important, and that continuous metrics should be used instead of binary metrics whenever possible. We thus recommend the wider application of prevalence-dependent continuous metrics, particularly likelihood-based metrics such as Akaikes information criterion (AIC), to assess the performance of presence-absence models.

Identifying the best climatic predictors in ecology and evolution

1.Ecologists and many evolutionary biologists relate variation in physiological, behavioral, life-history, demographic, population and community traits to variation in weather, a key environmental driver. However, identifying which weather variables (e.g. rain, temperature, El Nino index), over which time period (e.g. recent weather, spring or year-round weather) and in what ways (e.g. mean, threshold of temperature) they affect biological responses is by no means trivial, particularly when traits are expressed at different times among individuals. 2.A literature review shows that a systematic approach for identifying weather signals is lacking and that the majority of studies select weather variables from a small number of competing hypotheses that are founded on unverified a priori assumptions. This is worrying because studies that investigate the nature of weather signals in detail suggest that signals can be complex. Using suboptimal or wrongly-identified weather signals may lead to unreliable projections and management decisions. 3.We propose a four-step approach which allows for more rigorous identification and quantification of weather signals (or any other predictor variable for which data is available at high temporal resolution), easily implementable with our new R package ‘climwin’. We compare our approach with conventional approaches and provide worked examples. 4.Although our more exploratory approach also has some drawbacks–such as the risk of overfitting and bias that our simulations show can occur at low sample and effect sizes—these issues can be addressed with the right knowledge and tools. 5.By developing both the methods to fit critical weather windows to a wide range of biological responses and the tools to validate them and determine sample size requirements, our approach facilitates the exploration and quantification of the biological effects of weather in a rigorous, replicable, and comparable way, while also providing a benchmark performance to compare other approaches to.

Population turnover, habitat use and microclimate at the contracting range margin of a butterfly

Climate change is expected to drive patterns of extinction and colonisation that are correlated with geographic gradients in the climate, such as latitude and elevation. However, local population dynamics also depend on the fine-scale effects of vegetation and topography on resource availability and microclimate. Understanding how this fine-scale variation influences population survival in the face of changing climatic favourability could provide clues for adapting conservation to climate change. Here, we document a long-term decline of the butterfly Parnassius apollo in the Sierra de Guadarrama mountain range in central Spain, and examine recent population turnover and habitat use by the species to make inferences about its ecology and conservation. A decline since the 1960s throughout the elevation range suggests a regional deterioration in favourability for the species. Since 2006, local habitat quality has been the main correlate of population persistence, with populations that persisted from 2006 to 2012 associated with high availability of larval host plants. At a finer resolution, the larval distribution in a network of suitable habitat in 2011 and 2012 was most closely related to bare ground cover. Thus, although slope, aspect and elevation lead to considerable variation in microhabitat temperatures during the period of P. apollo larval development, vegetation structure appears to have been the most critical factor for local habitat use and population persistence. The results show that site selection and management retain key roles in conservation despite the broad-scale effects of environmental change.