Table 1. Evaluation of our species distribution modeling practices against the best practices that have been proposed for this field (Araújo et al., 2019).

Guideline	Standard	Justification
Response variables	(A) Sampling: bronze	Best data available; municipalities, local governments, and states choose who to test. Positive tests only reported
	(B) Identification: gold	Assuming best practices in testing and reporting
	(C) Spatial accuracy: bronze	County assignments provide a rough georeference for each record, but do not precisely describe where transmission of the virus occurred. Spatial accuracy unknown. Occurrences limited to identifiable county level localities
	(D) Environmental extent: deficient	Limiting the study area to the continental U.S. is unlikely to adequately test environmental boundaries
	(E) Geographic extent: bronze	Study area to include current range in the U.S.
Predictor variables	(A) Selection of candidates: bronze/deficient	Unclear and not well documented correlations between SARS-CoV-2 transmission and climate. At best, distal variables with weak, indirect control on the distribution
	(B) Spatial and temporal resolution: deficient	Variables sampled from a 2.5 arcminute grid for all cells within 5km of each occurrence point. Mean value used for modeling. Monthly climate averages as predictors for end of March occurrence data
	(C) Uncertainty: bronze	Temporal and spatial uncertainty in occurrence data has unquantified potential effects on the model output.
	(A) Model Complexity: silver	ENMeval for model testing and selection (maximize testing AUC and minimize AICc in the case of ties) using internal cross validation through the block resampling method
	(B) Treatment of response bias: silver	Internal cross validation to evaluate bias effects in different models
	(C) Treatment of collinearity: bronze	“Approximate methods are applied” — Predictor variables hand selected from monthly climate data available to avoid collinearity (i.e., used only Tmax and not Tavg or Tmin)
	(D) Uncertainty: bronze	Multiple Maxent model parameters tested, but only the optimal model presented
Model evaluation	(A) Evaluation of model assumptions: gold/silver	Select robust models from all tested models with ENMeval
	(B) Evaluation of model outputs: silver	Evaluated against multiple, non-independent, geographically structured sub-samples
	(C) Measures of model performance: silver	Suite of model performance metrics performed via ENMeval
Summary	Mode of the scores: bronze	Model building and testing is generally robust, but data and geographic scope are incomplete at this time