Exploring timescales of predictability in species distributions

Abstract

Accurate forecasts of how animals respond to climate-driven environmental change are needed to prepare for future redistributions, however, it is unclear which temporal scales of environmental variability give rise to predictability of species distributions. We examined the temporal scales of environmental variability that best predicted spatial abundance of a marine predator, swordfish Xiphias gladius, in the California Current. To understand which temporal scales of environmental variability provide biological predictability, we decomposed physical variables into three components: a monthly climatology (long-term average), a low frequency component representing interannual variability, and a high frequency (sub-annual) component that captures ephemeral features. We then assessed each component's contribution to predictive skill for spatially-explicit swordfish catch. The monthly climatology was the primary source of predictability in swordfish spatial catch, reflecting the spatial distribution associated with seasonal movements in this region. Importantly, we found that the low frequency component (capturing interannual variability) provided significant skill in predicting anomalous swordfish distribution and catch, which the monthly climatology cannot. The addition of the high frequency component added only minor improvement in predictability. By examining models' ability to predict species distribution anomalies, we assess the models in a way that is consistent with the goal of distribution forecasts -to predict deviations of species distributions from their average historical locations. The critical importance of low frequency climate variability in describing anomalous swordfish distributions and catch matches the target timescales of physical climate forecasts, suggesting potential for skillful ecological forecasts of swordfish distributions across short (seasonal) and long (climate) timescales. Understanding sources of prediction skill for species environmental responses gives confidence in our ability to accurately predict species distributions and abundance, and to know which responses are likely less predictable, under future climate change. This is important as climate change continues to cause an unprecedented redistribution of life on Earth.

Decomposing the spatial and temporal effects of climate on bird populations in northern European mountains

“…Species with negative associations with the temporal climate components may be among the first to be negatively impacted by climate warming in our study regions, as colder breeding seasons are expected to become rarer in the future (Bärring et al, 2017). (Brodie et al, 2021). For some bird species in the US, SDMs with dynamic covariates described the breeding distributions and their dynamic changes better than SDMs with static covariates (Bateman et al, 2016) (Elston et al, 2017;Pearce-Higgins et al, 2015).…”

“…It is also vital to understand the pattern of species distributions in the intervening time, the variation in distribution patterns in response to climate variation at different temporal scales, and the mechanisms causing the variation in distribution pattern. For example, long‐term average conditions were adequate to explain the average distribution and catch of a fish species, but deviations in average catch were best explained by interannual variability of the marine environment (Brodie et al, 2021 ). For some bird species in the US, SDMs with dynamic covariates described the breeding distributions and their dynamic changes better than SDMs with static covariates (Bateman et al, 2016 ).…”

Applications of telemetry to fish habitat science and management

“…In coastal marine systems (including estuaries), tides cause dramatic variations in the amount of wetted area, water depth, and salinity, driving fish movements, species compositions, and habitat characteristics. Fishes and other animals may also be more vulnerable to human stressors during certain periods, such as spawning, necessitating dynamic spatial-temporal management actions (Pecl et al 2006;Hobday et al 2010;Brodie et al 2021). Spatial-temporal habitat models would therefore be highly useful for decision making surrounding fish habitat, including the timing and location of anthropogenic disturbances (e.g., in-water works or activities), and could also guide alternative scientific sampling efforts (Larocque et al 2020).…”

Toward Regional Marine Ecological Forecasting Using Global Climate Model Predictions From Subseasonal to Decadal Timescales: Bottlenecks and Recommendations

“…Of course, such statistical analysis can only identify correlations, which do not necessarily mean causality, and the relationships may be nonstationary. Furthermore, a causal relationship is not enough for prediction, because if physical conditions that influence marine species are unpredictable, then biological targets are also not predictable (Brodie et al, 2021).…”