Detecting and projecting species-level responses to variation in weather and changes in climate

Background/Question/Methods

An ecosystem’s representative biological structure, composition, and function reflect long-term climatic drivers. However, most measurements of structure, composition, and function also reflect short-term weather. For example, butterflies and birds often are touted as highly responsive to climate change. Although this may be true, the responsiveness of butterflies and birds to weather at resolutions of hours, days, and years creates considerable noise in field data. Moreover, species-climate relations are not static. Instead, phenotypic plasticity and adaptive evolution increase the probability of persistence in variable environments. Thus, the biology of most taxonomic groups complicates efforts to identify species-climate relations. Additionally, studies often are not designed explicitly to examine species-climate relations as opposed to relations with other environmental variables or stressors.

We conducted 13 years of point counts of birds in four mountain ranges in the central Great Basin and two years of point counts in three ranges in the western Great Basin. We surveyed butterflies across almost 20 years in the central Great Basin and two years in the western Great Basin. We modeled probabilities of detection and occupancy of these taxonomic groups to characterize species-environment relations and to explore the effects of sampling methods on biological inferences, including responses to weather and climate

Results/Conclusions

The distributions, abundances, and phenology of butterflies and birds varied considerably over time. In a given location, regardless of spatial extent, species richness typically was more consistent among years than species identity. Sampling design, such as survey duration or number of surveys, had different effects on inferences about individual species (e.g., occupancy) and assemblages (e.g., species richness). For example, point counts of 8 min duration did not result in statistically different estimates of occupancy of individual species of birds than point counts of 5 min duration. However, estimates of species richness were greater when based on 8-min than 5-min counts. The extent to which annual changes in occupancy were correlated with changes in weather varied substantially among species.

The probability of detecting and projecting species-level responses to weather and climate will be increased by matching sampling methods and data to objectives. For example, decisions about land-use allocation may require different information than efforts to quantify habitat quality. Because no model can rescue weak data, response variables, covariates, and statistical methods should be selected in part on the basis of the biology of the species and the resolution and accuracy of data for the relevant climatic variables.