The stated sample sizes struck me as far smaller than what one would get from such extensive surveys and what one would want for the kind of analyses done here. Quick exploration of the data from the original sources produced much larger numbers of samples, and brings me to wonder if data processing errors might have affected results?

The paper says "A total of 2,237 individuals ... were captured from the Louisianan biogeographic province..." I downloaded the data and summed T_ABS (defined in the original metadata as "# of individuals collected of the taxon") to get 43,131 individuals. The paper also says "a total of 4,120 individuals" (for combined LA and VA data), but summing T_ABN in my copy tells me there were 79,087 individuals.

However, my quick summary agreed with other details provided in the paper. Maps from my copy matched Figure 1 very well. Min and max in the M_LEN field in my copy of the LA data are 2.2 and 91.18, and the paper says, "fork length: min. = 2.2 cm; max. = 91.18 cm". Though methods for dealing with non-fish records, undefined species codes, and taxonomy issues were not provided in the paper, my independent processing produced a species list differing by only one from that of Appendix S1.

But then the M_LEN field, defined in original metadata as "Mean length (cm) of fish in taxon” hinted at another possible explanation of differences. The values in that field are clearly means of lengths of all measured individuals in each sample, not min and max lengths of individuals as indicated in the paper, making me think perhaps numbers of individuals in the paper actually were numbers of samples. However, my version of the data has 3,187 samples from LA and 1,850 from VA, vs the paper’s 4,120.