Skip to main content
Dryad

Invasion success and tolerance to urbanization in birds

Cite this dataset

González-Lagos, César; Cardador, Laura; Sol Rueda, Daniel (2021). Invasion success and tolerance to urbanization in birds [Dataset]. Dryad. https://doi.org/10.5061/dryad.dr7sqv9zr

Abstract

Cities are considered hotspot of biological invasions, yet it remains unclear why non-indigenous species are so successful in environments that most local native species do not tolerate. Here, we explore the intriguing possibility that humans may be unintentionally introducing species preadapted to persist in such environments. Combining data on historical introductions with information of avian assemblages along urban-wildland gradients, we found that avian species that in their native range proliferate in human-altered environments have been more likely to be transported and introduced to new locations than species confined to the wildland. We also found that such urban dwellers had higher chances to become established because they already had adaptations to cope with novel environments. These findings suggest that the pathway of introduction selects for species preadapted to persist in novel environments, providing an explanation for why non-indigenous birds are so successful in cities. Because the tendency to introduce species associated with human-altered environments continues, there is an urgent need to develop new regulations to prevent future introductions.

Methods

We drew on openly available data sources. For Dataset S1, we assess for "occurrence of birds in human-altered environments" by using the habitat classification scheme from the International Union for Conservation of Nature (IUCN). This classification scheme includes data for habitat occurrence of 9,903 bird species (see IUCN habitat classification scheme 3.1; http://www.iucnredlist.org/technical-documents/classification-schemes). The IUCN data allowed us to classify bird species as occurring or not across any type of artificial terrestrial habitat such as arable land, pastureland, plantations, rural, urban and subtropical/tropical heavily degraded former forest (n = 4,515 species occurring in these human-altered environments, "humanDisturbed" column in Dataset S1). We also classified species as occurring or not in urban areas - the most modified environments (n = 905 species, "urban" column in Dataset S1). We also assess for "bird tolerance to urban areas in the native range". Because occurrences of avian species in urban environments depends in part from their abundance in nearby rural and natural habitats, we used a dataset of avian assemblages available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.2rbnzs7jf to compute two metrics of urban tolerance. The first metric was estimated as the log-log difference in the mean number of individuals of a species recorded between highly urbanized habitat (HUR) and nearby natural vegetation (NVG) of a given study location (‘hereafter’ urban tolerance index or UTI, Sol et al. 2017). High positive values of the index indicate high tolerance to the environmental alterations associated with urbanization, while low negative values indicate species less tolerant to urbanization (Evans et al. 2011)("UTI" column in Dataset S1). Values close to zero are more uncertain as they may indicate either no effect of urbanization or that the species is too scarce in the region to assess its tolerance (Sol et al., 2014). To tackle this issue, we used a second, more restrictive, metric to estimate urban tolerance based on community simulations. This allowed us to effectively separate species that tolerate urbanization (exploiters) from those that do not (avoiders). A species was considered an urban ‘exploiter’ if its observed abundance in the highly urbanized habitat was equal to or greater than the 95th percentile of the abundance expected by chance, and an urban ‘avoider’ if it was equal to or lower than the 5th percentile; the species for which information was insufficient to be assigned to one or another category (i.e. their abundances in urban areas were distributed as expected by random) were not considered. Because we were interested in the intrinsic ability of a species to tolerate urbanization, we considered a species as an exploiter when at least one of its native populations was classified in this category ("tolerance" column in Dataset S1). Additionally, we estimated bird “invasion potential” (i.e. tendency of a species to succeed wherever it is introduced) as the species-level random coefficients of a generalized linear mixed model that includes establishment success as the dependent variable, introduction effort as fixed effect, and taxonomy (Orders, Families within orders and Species within families) and country of introduction as random effects (Sol et al. 2005, 2012b). For this, we used a global data source of historical avian introductions from a previous study (Sol et al. 2012b). This dataset provides information on establishment success (i.e. whether a given introduction event was successful or failed to produce a self-sustaining population) and introduction effort for 832 events for 202 bird species (Data available at https://doi.org/10.1126/science.1221523 as Supporting Information). A high score in the invasion potential means that the species tends to succeed wherever it is introduced and a low score that it tends to fail when differences in introduction effort are controlled statistically ("invasion.potential" column in the Dataset S1). Finnaly, we gathered information from previously assembled data sets for species traits that preadapt birds to cope with novel environments (Sol et al. 2012b, 2017a, Sayol et al. 2016) and could also affect urban tolerance (Sol et al. 2014, 2017a). The traits considered were: i) the brood value ("broodValue" column in Dataset S1), an index that represents the value of current reproduction with respect to the value of future reproduction (Bókony et al. 2009), estimated as log10{1/(broods per year x lifespan)}; ii) relative brain size, which is a surrogate of behavioral flexibility (Lefebvre et al. 1997, Reader and Laland 2002) and was estimated as the residuals of a phylogenetically corrected least squares log-log regression between brain ("Brain" column in Dataset S1) and body size ("Body" column in Dataset S1) using the phyl.resid function of the phytools r-package (Revell 2009); and iii) niche breadth, characterized by the frequency of use (not used = 0, occasionally used = 0.1, frequently used = 0.5 and almost exclusively used = 1) of different categories of habitat and diet, expressed as Rao's quadratic entropy (De Cáceres et al. 2011). The categories used to define habitat breadth were forest, woodland, open vegetation (including tundra and grassland) and aquatic habitats (including marshes and wetlands; see Wilman et al. (2014))("HA" column in Dataset S1) . For diet breadth ("DA" column in Dataset S1), we considered the percentage of use of invertebrates, endotherms, ectotherms, fishes, scavengers, fruits, nectar, seeds, plants and unknow items.

For Dataset S2, we used CITES information on bird trade. Because patterns of introduction change over time, pathways that were rare in the past, like the trade in pet birds, are now the most common (Abellán et al. 2016, Dyer et al. 2017). To test whether there is an ongoing trend to introduce species that are associated with humans in their native ranges, we used information on the total number of live individuals per species that were legally traded from 1975 to 2013 according to the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES) data (n = 1074 species; "Imports" column in Dataset S2)(www.cites.org). While CITES information is not exempt of potential bias such as underrepresentation of some taxonomic groups or occasional failures in the estimation of the number of individuals involved in trade, it still represents the only global formal convention tracking data on international trade in wildlife in a structured and verifiable way (Reino et al. 2017). Indeed previous analyses have shown a direct link between trade and introduction numbers (Reino et al. 2017, Cardador et al. 2019). For each CITES trade event, information on the number of traded individuals came from the importer country (28.7%), the exporter country (19.6%) or both (51.7%). This is not likely to affect our results, since when available, information provided by the importer and the exporter countries was highly correlated (Pearson correlation coefficient: r = 0.91, P > 0.001, n = 7,606). In these cases, the maximum number of individuals reported was used for analyses (Dataset S2).

Usage notes

In addition to Dataset S1 and S2, we provide Dataset S3 containing the code and country for each study location, and aslo their latitude and longitude. We provide an R script to replicate the main analyses of the manuscript "Invasion success and tolerance to urbanization in birds" using these dataset and other available dataset (doi are provided in the r script).

Funding

Agencia Nacional de Investigación y Desarrollo, Award: FONDECYT 11160271

Agencia Nacional de Investigación y Desarrollo, Award: ANID PIA/BASAL FB0002

Catalan Government and EU COFUND, Award: 801370

Ministerio de Asuntos Económicos y Transformación Digital, Award: CGL2017-90033-P

Catalan Government and EU COFUND, Award: 801370