Covariate processing
We reclassified the NLCD 2016 (U.S. Geological Survey 2019) raster layer
from 15 original landcovers to four landcover types representing
important habitat types for deer or for human exposure to the LD hazard:
water/wetland/herbaceous (wetland-herb), open and low-intensity
development (low-dev), medium-high intensity development (med-hi-dev),
and forested landcover. To explicitly investigate how deer habitat
selection can influence human encounters with the LD hazard, we
classified residential blocks based on their attractiveness to deer and
suitability for tick survival. We selected the block level to match the
spatial scale of deer relocations (average step length, for males:
146m/2hours; for females: 84m/1hour). To construct this landcover layer,
we estimated total yard area, canopy height, landcover type proportions
within each block, and landscape connectivity for deer within 100m
surrounding each block. We standardized the block-level variables and
used K-means clustering to define the block types (Supporting
Information, Figure S2). Block-1 characterized blocks with high
vegetation, low impervious cover, high connectivity for deer, and large
yard areas. Block-2 described blocks with low vegetation, high
impervious cover, low connectivity for deer, and small yards. We merged
block assignments with the reclassified landcover layer to use in the
second and third order selection models (Figure 1). We validated the
association between block types and tick habitat suitability using a
dataset from our prior work assessing tick presence in residential yards
bordering urban green spaces from May-June 2018 and 2019 on SI (Gregoryet al. 2022, Supporting Information). Of 245 household samples
around urban forests, 92.41% of yards where ticks were collected were
in block-1 (n = 223 households) and 7.59% (n = 22
households) in block-2. Further, 90% of yards where I.
scapularis were collected were in block-1. This confirmed that block
assignment successfully identified residences that were accessible to
deer and support I. scapularis survival. Lastly, due to low
representation of deer use across all six landcover types, we created a
second landcover layer of coarser thematic resolution where the six
landcover types were re-classified into either ‘natural’ (forest and
wetland-herb) or ‘urban’ (low-dev, med-hi-dev, block-1, and block-2).