Fig 1.
Study sites and moth diversity patterns across New York City.
(A) On-the-ground sampling locations for 12 locations in Brooklyn and Queens, New York City, USA, where we sampled Lepidoptera, and National Land Cover (NLCD) data for New York City. Insets represent magnified views of study sites. (B) Species diversity in 1 km hexagonal bins derived from iNaturalist data.
Fig 2.
Land cover composition, principal components analysis, and site imagery of sampling locations.
(A) Stacked bar plot of NLCD land cover composition at on-the-ground sampling sites. Colors represent the proportion of each land cover type in 150 m buffers around each sampling location. (B) Violin plots represent the sum of urban NLCD land cover proportions (developed land of high, medium, and low intensity) for our on-the-ground sampling locations and all hex-bins used with iNaturalist data. (C) Satellite images of sampling locations in Brooklyn and Queens, New York. Satellite images courtesy of the USDA, USGS The National Map: Orthoimagery.
Fig 3.
Lepidoptera family diversity from field sampling and iNaturalist data.
(A) Stacked bar plot of Lepidoptera Family diversity per site for on-the-ground sampling. The left side of the plot has number of moths sampled per site (N). To the right of the chart is the date each site was surveyed and the corresponding moon phase. Bolded names represent families that were observed in both our field surveys and on iNaturalist. (B) Number of observations of each Lepidoptera Family across all iNaturalist hex bins. Bars are colored by Families observed. Bolded family names represent observations found in both field data and iNaturalist datasets. (C) Images of some moths taken during on-the-ground sampling.
Fig 4.
Relationships between Simpson’s Diversity Index and NLCD land cover.
(A) Simpson’s diversity (1-D) as it relates to land cover summarized by principal components analysis (PC1, PC2, PC3), with 95% confidence intervals (shaded area). Significance levels are denoted by asterisks: p < 0.001 (***). (B) Land cover loadings for each principal component in the corresponding plot in part A immediately above each. PC1 represents more highly urbanized land and less open space and water. PC2 represents more low intensity developed land and open space with less herbaceous land cover. PC3 represents more deciduous forest and shrub with less highly developed land and water.
Fig 5.
Pathways linking urban land cover to Lepidoptera diversity revealed by structural equation modeling.
Relationships among predictor variables and Simpson’s diversity generated by the structural equation model using reflective syntax. The top row of land cover classifications makes up the observed variable group, while the middle row represents the 3 latent variables. The thin single-headed arrows (pointing from latent to observed) denote the loadings of each land cover. Dashed lines represent the marker variables used to scale the latent construct. The double-headed arrows represent covariance between variables while the bolded lower arrows demonstrate the structural regression paths. Significance levels are denoted by asterisks: p < 0.05 (*), p < 0.01 (**), p < 0.001 (***). Nodes represent variables included in the model, while edges indicate associations between them. Thicker lines represent stronger relationships, with positive effects shown in green and negative effects in red. Edge weights are proportional to the strength of standardized regression coefficients.