Concrete habitat: Impervious surface in nest vicinity is associated with avian fitness decline in two urban adapters

Description

Notes

Methods

Study area

Data were collected from 500 Schwegler woodcrete-nestboxes (type 1b, with a 32 mm-diameter entrance hole, suitable for great tits and blue tits) set in a 50m grid within eight study sites in and outside the city of Warsaw, Poland (Fig.1). Our set up comprehends study sites from highly urbanized locations close to the city center (e.g., an office area and two residential sites – blocks of flats typical of post-war Eastern European architecture, built in the late 40's and 50's of the 20th century) and from green urban spaces (an urban park and two urban woodlands). The study system is also composed of two study areas located outside the city: a suburban village and a natural forest. More details relative to each study site location are reported in earlier publications (Corsini et al. 2019, 2020).

Breeding data

Great tit and blue tit breeding data were collected for five years (2017-2021). Each year, nestboxes were inspected weekly, starting from March until mid-July with the only exception of one urban woodland (the cemetery), which could not be visited regularly in 2020 due to pandemic-related access restrictions (Corsini et al. 2022). Overall, such visiting frequency allowed for estimating the date of the first egg laid and clutch size, as these species lay one egg daily. Hatching date was determined by checking the nestboxes one day before expected hatching or every other day after that, until hatching occurred. Parents were caught, aged (as first year breeders or older), measured, and ringed when nestlings were 10-13 days old. Fifteen days after hatching, alive nestlings were counted, individually weighed and ringed. Twenty-five days after hatching, nestboxes were checked to record fledging success. Only first broods initiated within 30 days after the first brood recorded for each species in a given year and site were included in the study (Van Balen 1973).

Quantifying urbanization

Urbanization was measured as the percentage of Impervious Surface Area (ISA) in a 100m-radius from each nestbox. Namely, a 20-m pixels resolution ISA-raster file was downloaded from Copernicus Land Monitoring Services (https://land.copernicus.eu/sitemap), and averaged for each nestbox via buffer-analyses in qGIS v.3.10 (Team 2016) (see Szulkin et al. (2020)). The ISA layer contained all built-up areas which replaced original natural cover with impervious or any other artificial surface. The 100m-radius was selected based on the food foraging distance covered by blue tit parents while feeding the nestlings (Tremblay et al. 2004).

Statistical analyses

Statistical analyses were performed in R (v. 4.1.2). Because of species-specific differences in life-history, great tits and blue tits were analyzed separately. To test the association between number of fledged offspring and ISA, we fitted Generalized Linear Mixed Effects Models (GLMMs). Namely, we included all nests where at least one egg hatched. Because data were over dispersed, we used a quasi-Poisson error-structure. We included the continuous predictors ISA, Lay date (1^st of April recorded as 1), and Female Body Mass (as heavier females produce larger clutches and start breeding earlier in the season (Perrins & McCleery 1989)). We also fitted Year (5 levels) as a fixed effect. When including Site (8 levels) as a random effect to control for inter-site variation in avian breeding, we found that the between sites variance was close to zero, and therefore removed it from the models. Interactions between Year and ISA, and between Lay date and Female body mass were tested and removed from the models if not significant. Further, as the influence of ISA on Clutch size may influence the variability in fledging success, we conducted supplementary analyses using the same structure as previously indicated, incorporating Clutch size as a continuous predictor in the models. Potential multicollinearity issues were verified using the vif-function (car package (Fox et al. 2012)), which always yielded values below 2. Plots were built in ggplot2 (Wickham 2011) after calculating predictions in ggeffects (Lüdecke et al. 2020). Unit decreases in fledged offspring were calculated using the function avg_comparisons in the package marginaleffects (Arel-Bundock, 2023).