Fig. 3

Major axes of plumage pattern variation, summarized with principal components analysis (PCA). From granularity analysis, these values are used for species-level approaches. a Pattern energy spectra for exemplar species characterized by maximally divergent PC scores (i = Dryobates villosus, ii = Mulleripicus funebris, iii = Colaptes fernandinae, iv = Melanerpes erythrocephalus). These show information on relative contributions of different granularities to overall appearance. b Pattern maps of exemplar species enable visualization of energy at subsets of isotropic band-pass filter sizes (2, 32, 64, 128, 256, and 512 pixels). c Principal component loadings of pattern energy spectra across filter size for all species reveals how pattern elements of different sizes influence PC scores. d Pattern PC1 and PC2, collectively, account for 82.1% of variation across woodpeckers. Species (i–iv) illustrate the extremes of variation along PC1 and PC2: (i) exhibits high energy scores across most pattern element sizes, with small, medium, and large pattern elements; (ii) has low energy scores across the spectrum, with few pattern elements of any size; (iii) has many small pattern elements and few of any other sizes; (iv) has only medium and large size pattern elements. Illustrations © HBW Alive/Lynx Edicions