Aim: Theory suggests that increasing productivity and climate stability towards the tropics favours specialization, thus contributing to the latitudinal richness gradient. A positive relationship between species richness and specialization should therefore emerge as a fundamental biogeographical pattern. However, land‐use and climate changes disproportionally increase the local extirpation risk for specialists, potentially weakening the relationship between richness and specialization. Here, we quantify empirically the richness–specialization prediction and test how 50 years of climate and land‐use change has affected the richness–specialization relationship. Location: USA. Time period: 1966–2015. Major taxa studied: Birds. Methods: We used the North American Breeding Bird Survey to quantify bird community richness and specialization to habitat and climate. We (a) quantify temporal change in the slope of the richness–specialization relationship, using a generalized mixed model; (b) assess how this change translates spatially, using generalized additive models; and (c) attribute spatio‐temporal change in the richness–specialization relationship to land use, climate and topographic drivers. Results: We found evidence for a positive but weak richness–specialization relationship in bird communities that greatly weakened over time. Given that specialization was not the main driver of richness, this relationship did not translate spatially into a linear spatial covariation between richness and specialization. Instead, the spatial covariation in richness and specialization followed a unimodal pattern, the peak of which shifted towards less specialized communities over time. These temporal changes were associated with precipitation change, decreasing temperature stability and land use. Main conclusions: Recent climate and land‐use changes have induced two contrasting types of community responses. In human‐dominated areas, the decoupling of richness and specialization drove a general trend for biotic homogenization. In areas of low human impact experiencing increasing climate harshness, specialization increased, whereas richness decreased. Our results offer new support for specialization as a key driver of macroecological diversity patterns and show that global changes are weakening this fundamental macroecological pattern.
Data used for the analyses of the paper
Data used in the GEB paper "Recent global changes have decoupled species richness from specialization patterns in North American birds"
The code using or producing the different data tables is available at https://github.com/AnneMimet/Ric-Spe-code
List of provided data tables (by alphabetic order):
- ci_coef_null.txt: Mean and confidence intervals of the slope of the Richness-Specialization relationship obtained from the null model (analysis (1))
- Community_data.txt: yearly data of community richness (“ric”) and specialization (“csi”) per route (“rte_id”), with associated longitude and latitude.
- Community_data_decades.txt: Data of community richness (“ric”) and specialization (“csi”) per route (“rte_id”), aggregated by decade (“year”: year 1 is the first decade – 1973-1983, year 2 is the second decade – 1985-1995; year 3 is the third decade – 2001-2011), with associated longitude and latitude
- correspondance.txt: Correspondence between the combination of kKoppen (“world_kopp”) and land use classes (“loc”) and the habitat classes obtained with the multivariate regression tree
- data_for_mrt.txt: Data used to compute species specialization, including species in columns (prefix “X”), the route id (“rte_id”), the Koppen climate class of the route (“world_kopp”) and the land use (“loc”)
- data_for_ssi.txt: Sub-sample of data_for_mrt.txt, used to compute species specialization
- data_for_tree.txt: Sub-sample of data_for_mrt.txt, used to create the habitat classes by multivariate regression tree (mrt)
- data_rtes.txt: Row data (averaged oer route from the North American Breeding Bird survey) of species abundance per route (“rte_id”) and year. All other table containing biodiversity data are derived from this table
- env.txt: Data of environmental variables for each route computed per decade (column “per”). Climate data are derived from the PRISM data base, land use data are derived from the HYDE database and topographic data are derived from the SRTM 30+ data base: spatial range of winter minimum temperature ("r_min_wint"), mean winter minimum temperature ("m_min_wint"), spatial range of winter mean temperature ("r_mean_wint"), mean winter mean temperature ("m_mean_wint"), spatial range of winter precipitation ("r_ppt_wint"), mean winter precipitation ("m_ppt_wint"), spatial range of minimum temperature during the breeding season ("r_min_breed"), mean minimum temperature during the breeding season ("m_min_breed"), spatial range of mean temperature during the breeding season ("r_mean_breed"), mean temperature during the breeding season ("m_mean_breed"), spatial range of precipitation during the breeding season ("r_ppt_breed"), precipitation during the breeding season ("m_ppt_breed"), temporal standard deviation of the minimum temperature during the breeding season ("tr_min_breed"), temporal standard deviation of the mean temperature during the breeding season ("tr_mean_breed"), temporal standard deviation of the precipitation during the breeding season ("tr_ppt_breed"), proportion of cropping area ("crop"), proportion of rangelands ("range"), proportion of pasture ("pasture"), proportion of irrigated croplands ("irrig"), proportion of urban areas ("urb"), population ("pop"), range in altitude ("alt_rge"), mean altitude ("alt_mean"), climate velocity ("velo_rte"), potential NPP ("npp0_rte"), human impact index ("impact")
- mat_ssi_random.txt: 500 randomizations of species specialization (only the first 200 were actually used in the analyses)
- rtes_XY: Latitude and longitude of the routes
- ssi.txt: Species specialization
data.zip