Data for: Recent changes in thermal niche position and breadth of bird assemblages in Spain in relation to increasing temperatures

Aim: Animal communities around the world are responding to climate change by altering their taxonomic composition, mainly through an increase in the colonisation rate of warm-dwelling species and the local extinction of cold-dwelling ones. We assessed whether the taxonomic composition of bird assemblages in peninsular Spain has changed in accordance with the recent increase in temperature. We also evaluated the role of species' thermal affinities and population dynamics in these changes.

Location: Peninsular Spain.

Taxon: Birds.

Methods: We compared assemblages reported in the last Spanish breeding bird atlases (1998–2002 vs 2014–2019) in 10x10 km squares. We described species’ thermal niches by overlaying global species breeding distributions and world temperature metrics (based on mean, minimum, maximum and range), and then aggregated them to obtain a set of community thermal indices for each assemblage (CTIs, and CTR for ranges). Long-term average temperatures and local current temperatures were related to changes in CTIs using spatial GLMMs, which considered habitat change. We identified the species most responsible for variation in assemblages and regressed species’ influence on thermal affinities and population dynamics.

Results: CTIs increased with temperature and warm-dwelling species became more prevalent to the detriment of cold-dwelling ones. However, we found a counteracting effect of temperature and habitat. Cold-dwelling forest species were among the most influential species, mainly through colonisation, while warm-dwelling farmland species contributed through local extinctions (both attenuated local increases in CTI). The mean thermal breadth of assemblages (CTR) decreased with temperatures.

Main conclusions: The taxonomic composition of bird assemblages shifted in line with the main expectations due to global change (thermophilisation), mainly due to local colonisation of warm-dwelling species, although it did not show the pattern of thermal homogenization suggested elsewhere. Our results add further evidence of the interplay between climate warming and land-use change in the ongoing adjustment of animal communities.

Methods
The dataset is a dataframe that comprises the Community Thermal Indices (response variable) and the environmental and geographic variables employed as predictors of the spatial GLMM. This model related the temperatures to the changes of CTI, considering the habitat (forest) change.

The Community Thermal Indices were computed from the Species Thermal Indices. We obtained four thermal indices for each species (Species Thermal Index – STI) by combining the global species’ distribution and the climate information. The STI1 (i) shows the mean temperature of the breeding season (April-July) throughout the species’ distribution range. Similarly, the STI2 (ii) is the average of the maximum temperatures above the percentile 95 in July, and the STI3 (iii) is the average minimum temperature below the percentile 05 in April in the species’ breeding distribution range. These three indices represent a species’ thermal affinity. On the other hand, the fourth index (iv) (Species Thermal Range - STR) represents the average thermal range (April-July) throughout the distribution area and can be understood as species thermal breadth. It is computed as STI3-STI2.

We calculated a set of community thermal indices (CTI) for the assemblage of bird species in each of the 10x10km UTM grid squares of each of the breeding bird atlases. We obtained four different CTIs: CTI1, CTI2, CTI3, and CTR. The first three were calculated as the average of the STI, STI2, and STI3 of the species present in the assemblage, respectively. The CTR (Community Thermal Range) is based on the average temperature range of the species (STR) that make up the assemblage and thus informs on the average niche breadth (Gaget et al., 2020). We calculated CTIs for each of the four-year periods covered by the atlases.

The dataset also includes the standardized and unstandardized local temperature (ºC) and forest cover (ha) for each grid square and for each breeding bird atlas. It also includes the standardized and unstandardized coordinates of each grid square in decimal degrees (WGS84). Local temperatures were obtained from Chelsa (v.2.1., Karger et al., 2017), averaging data for each five-year sampling period in each square. We used the CORINE Land Cover Accounting Layers built for the years 2000 and 2018, to link forest cover with the community indices for the first and second sampling periods, respectively