Urbanization is reshaping ecosystems worldwide, driving wildlife to navigate and adapt to novel and highly dynamic environments. The Eurasian tree sparrow (Passer montanus, ETS) serves as an exemplary human commensal, thriving in cities through exceptional behavioral and ecological flexibility. Here, we systematically investigated the nest-site selection strategies of ETSs across 645 buildings across 22 cities in northern China, integrating climatic, geographic, biotic, and anthropogenic variables at a macroecological scale. We found that both the availability and use of nest sites increased with building height, underscoring ETSs’ capacity to exploit vertical resources in dense urban landscapes. Notably, the preference for lower nest heights when nest sites were abundant suggests a strategy to reduce intraspecific competition and energy expenditure. Negative associations between nest-site use or preference and the normalized difference vegetation index indicate that ETSs favor anthropogenic over vegetated resources, likely to circumvent interspecific competition in urban green spaces. Additionally, altitudinal gradients modulated ETSs’ nesting responses: at lower elevations, higher building heights promoted nesting, whereas increased economic development (gross domestic product per cell) and noise suppressed it—signaling an avoidance of intense anthropogenic disturbance. Conversely, ETSs showed reduced competition at higher altitudes and increasingly relied on resources linked to urban prosperity. These findings reveal a nuanced behavioral adaptability in ETSs, allowing them to navigate trade-offs between anthropogenic benefits and environmental stressors across spatial gradients. Our study offers critical insights into the eco-behavioral mechanisms underlying urban adaptation and the evolution of commensalism, with important implications for biodiversity management and sustainable urban design.

Materials and Methods

2.1 Study Area and Species

We investigated the nest usage and preferences of the Eurasian tree sparrow (Passer montanus, ETS) across 22 cities in northern China (Fig. 1 in associated manuscript). These cities encompass a diverse range of natural environmental and socioeconomic conditions, including variations in elevation (-12 to 2328 m), temperature (12.2 to 20.5 °C during the breeding season), precipitation (24 to 110 mm during the breeding season), and human population density (37 to 1334 persons/km²). This diverse gradient provided a unique opportunity to examine the impact of both natural and anthropogenic factors on ETS nesting behavior.

2.2 Nest Site Surveys

Nest site surveys were conducted during the breeding seasons (April to July) of 2021 and 2022. Each city was systematically surveyed to identify ETS nests in urban areas. Potential nest sites, including building dimensions (height and width), air conditioner inlets, and other urban infrastructure, were inspected for ETS nest presence. Surveyors recorded GPS coordinates, height above ground, and the type of structure used for nesting.

The direction of the nests was observed and recorded on-site. Available nest density (NDa) was calculated as the number of potential nest sites (e.g., open air conditioner inlets) per meter along each building (nests/ building width). Similarly, occupied nest density (NDo) was determined as the number of nests actively used by adult ETSs or identified by the presence of nesting materials and eggs along each building (nests/ building width). Nest height preference (NHP) was calculated as the average height of occupied nests, weighted by density for each building.

2.3 Environmental Data Collection

Climatic and geographic variables were derived from a variety of reputable sources. Average air temperatures, precipitation, and wind speed during the warmest seasons that are critical for the breeding season, along with elevation data, were sourced from the WorldClim2 database (Fick & Hijmans, 2017), providing a comprehensive set of climatic parameters (Table S1). The normalized difference vegetation index (NDVI) during the warmest seasons was calculated from NASA's vegetation index and phenology (VIP) dataset, average over the period between 2000 and 2010 (0.05 Deg CMG).

Socioeconomic data were collected from authoritative databases. Population density was accessed through the Center for International Earth Science Information Network (CIESIN) at Columbia University, using the Gridded Population of the World, Version 4 (GPWv4). GDP per cell (0.083° resolution) data were sourced from a detailed gridded global dataset spanning 1990–2015 (Kummu et al., 2018) (Table S1). Noise levels were measured using the Sound Meter application (v 2.5.2) of the smartphone at the ground level directly below the nest sites, maintaining a vertical distance of 5 meters from the building facade. Each recording session lasted 1 min, with data collection consistently conducted between 8:00 h and 10:00 h daily. Artificial light at night data was accessed from the Harmonization of DMSP and VIIRS nighttime light data (Li & Zhou, 2017). The unit of measurement for DMSP NTL data is the digital number (DN), which represents the intensity of nighttime lights detected by the satellite sensor; values range from 0 to 63, with 0 indicating no detected light and 63 representing the maximum detectable light intensity. Urban land extent, representing the fraction of urban land, was acquired from SSP-consistent Global Spatial Urban Land Projections from 1/8-degree to 1-km resolution 2000-2100 (Gao & Pesaresi, 2021). All climatic, geographic, biotic, and socioeconomic data were extracted based on the geographic coordinates of each building.

2.4 Structural Equation Model Building

We employed a mixed-effects linear model to analyze the relationships between nest-related variables (NDa, NDo, and NHP) and environmental factors. Fixed effects included climatic and geographic factors (temperature, precipitation, wind speed, and altitude), biotic factors (NDVI, NPP, and nest direction), and anthropogenic factors (population density, GDP per cell, artificial light at night, urban land extent, building height, and noise level). Random effects were incorporated to account for city-specific variations. To accurately estimate the selection effect of nest use, we weighted 1/(NDa) for NDo and NDo for NHP. To address heterogeneity of variances across altitudinal gradients, given the data skew toward low and mid altitudes with fewer high-altitude observations – we stratified altitude into five quantile-based groups of equal sample size. We then implemented a variance-structured linear mixed model (LMM) via lme function in 'nlme' package (Pinheiro et al., 2012), incorporating group-specific residual variances using varIdent(form = ~1 | alt_group_quantile).​

Our strategy for SEM model modification involved systematically screening significant interconnections by performing piecewise tests of directed separation, using the 'piecewiseSEM' package (Lefcheck, 2016). Non-significant variables were iteratively removed, ensuring that all significant predictors were retained in the model. The model’s goodness of fit (GOF) was evaluated using Fisher's C statistic (with P > 0.05 indicating a good fit) and the Akaike Information Criterion (AIC), with lower scores indicating better model performance (Shipley, 2013). Standardized coefficients were used to assess the effects of predictors across varying scales. Subsequently, a full-factor structural equation model (SEM) was constructed by incorporating all significant predictors identified in the preliminary analysis (Table S2). To simplify the model, factors with partial R² < 0.3 were iteratively excluded until the simplified SEM met the following criteria: (1) all retained factors exhibited partial R² > 0.3, and (2) the model achieved the lower Fisher’s C statistic, indicating optimal parsimony (Table S3). The best-fitting SEM achieved the lowest Fisher's C statistic and AIC score (Table S4).

2.5 Interaction Analysis for Nest Use across Altitudes

To explore the effects of altitude on NDo and NHP, interaction terms for climatic, geographic, biotic, and anthropogenic factors were added to the base models. Interaction models were ranked by AIC, and the most informative model was selected. The Johnson-Neyman technique was used to estimate the range of significant slopes for interactions, implemented using the interactions package (Johnson & Fay, 1950; J. A. Long, 2019).

References

Fick, S. E., & Hijmans, R. J. (2017). WorldClim 2: new 1-km spatial resolution climate surfaces for global land areas. International Journal of Climatology, 37(12), 4302–4315. https://doi.org/10.1002/joc.5086

Gao, J., & Pesaresi, M. (2021). Downscaling SSP-consistent global spatial urban land projections from 1/8-degree to 1-km resolution 2000–2100. Scientific Data, 8, 281. https://doi.org/10.1038/s41597-021-01052-0

Johnson, P. O., & Fay, L. C. (1950). The Johnson-Neyman technique, its theory and application. Psychometrika, 15(4), 349–367. https://doi.org/10.1007/BF02288864

Kummu, M., Taka, M., & Guillaume, J. H. A. (2018). Gridded global datasets for Gross Domestic Product and Human Development Index over 1990–2015. Scientific Data, 5, 180004. https://doi.org/10.1038/sdata.2018.4

Lefcheck, J. S. (2016). piecewiseSEM: Piecewise structural equation modelling in R for ecology, evolution, and systematics. Methods in Ecology and Evolution, 7(5), 573–579. https://doi.org/10.1111/2041-210X.12512

Li, X., & Zhou, Y. (2017). A stepwise calibration of global DMSP/OLS stable nighttime light data (1992–2013). Remote Sensing, 9(6), 637. https://doi.org/10.3390/rs9060637

Long, J. A. (2019). interactions: Comprehensive, user-friendly toolkit for probing interactions. R package version 1.1.0. https://cran.r-project.org/package=interactions

Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., & R Core Team. (2012). nlme: Linear and nonlinear mixed effects models. R package version 3.1-103. https://CRAN.R-project.org/package=nlme

Shipley, B. (2013). The AIC model selection method applied to path analytic models compared using a d‑separation test. Ecology, 94(3), 560–564. https://doi.org/10.1890/12-0976.1