Land use change can elevate disease risk by creating conditions beneficial to species that carry zoonotic pathogens. Observations of concordant global trends in pathogen prevalence and disease incidence have engendered concerns that urbanization could increase transmission risk of some pathogens. Yet host-pathogen relationships underlying transmission risk have not been well characterized within cities, even where contact between humans and species capable of transmitting pathogens of concern occur. We addressed this deficit by testing the hypothesis that areas in cities experiencing greater population loss and infrastructure decline (i.e., counter-urbanization) can support a greater diversity of host species and a larger and more diverse pool of pathogens. We did so by characterizing pathogenic Leptospira infection relative to rodent host richness and abundance across a mosaic of abandonment in post-Katrina New Orleans (Louisiana, USA). We found that Leptospira infection loads were highest in areas that harbored higher rodent species richness. Areas with greater host co-occurrence also harbored a greater number of hosts, including the most competent hosts, indicating that Leptospira infection is amplified by increases in overall and relative host abundance. Evidence of shared infection among rodent hosts indicates that cross-species transmission of Leptospira likely increases infection at sites with greater host syntopy. Additionally, evidence that rodent co-occurrence and abundance and Leptospira infection load parallel abandonment suggests that counter-urbanization can elevate zoonotic disease risk within cities, particularly in underserved communities that are burdened with disproportionate concentrations of derelict properties.

All animals were collected in the field using live trapping as also outlined in https://doi.org/10.1016/j.landurbplan.2019.103710. Tissue samples were collected immediately at necropsy and frozen at -80C. To identify pathogenic Leptospira to species, we sequenced a partial region of the glmU gene using genomic kidney DNA from all individuals that tested positive for Leptospira infection in the qPCR screen. After PCR amplification following a standard protocol, we cleaned all PCR products with ExoSAP-IT (Affymetrix, Santa Clara, CA, USA) and completed final paired sequencing reactions consisting of 3.75 µL PCR grade H₂0, 3.75 µL 5uM MgCl₂, 1.0 µL each of 10 mM forward and reverse glmU primers (respectively) and 0.5 µL BigDye terminator (Applied Biosystems, Foster, CA, USA). We cleaned reactions using Sephadex columns prior to electrophoresis on an ABI 3730xl (Applied Biosystems). We aligned trimmed sequences with GenBank archived sequences of the glmU gene for all pathogenic Leptospira. We then constructed phylogenetic trees with all new and representative archived sequences using Bayesian Inference in MrBayes 3.2.6. Trees were built using the GTR+G model, and a Markov chain Monte Carlo analysis with four chains running for 8x10⁶ generations. Trees were sampled every 1000 generations with the first 1000 trees discarded as burn-in. Convergence was determined when the final deviation of split frequencies fell below 0.02. We assigned sequences to Leptospira species based on relationships recovered in the final tree, with confirmation from BLAST comparisons.

Information explaining each column is included in the metadata tab. Environmental data is only included for the subset of sites at which we targeted rats and mice (as explained in the manuscript). Data for every animal tested for Leptospira is included in the Leptospira tab. Not all animals tested were included in the final analysis.