Phenology estimates
We gridded our study area into 25 x 25-km equal area cells using the
North America Albers Equal Area Conic projection. Community science
observations can be biased by organized, public observation events
(e.g., iNaturalist’s City Nature Challenge) generating an unusually high
number of observations that do not reflect the actual seasonal abundance
of a species. We therefore filtered our dataset to include only one
observation per day of each species in a year, given each cell. Next,
the number of observations for each cell-by-year-by-species combination
was counted and deemed usable if at least 10 observations were
documented. For each unique cell-year-species combination, we estimated
the 0.05 and 0.95 sample quantiles using the quantile() function within
the stats R package (R Core Team, 2020) to represent the
emergence (first appearance) and termination (last evidence) of adult
insect activity. These quantiles are demonstrated to be more robust
estimates of phenology than estimating the absolute bounds of a
phenophase (Belitz et al. 2020). We calculated the duration of adult
insect activity as the difference between the termination and emergence.
In total, we used 228,423 records to generate 5,469 emergence,
termination, and duration estimates across 626 unique grid cells for 284
species. Over 97% had a basis of record listed as human observation,
indicating the vast majority of the data were generated by community
scientists.