counts, abundance, and trap-nights about neon4cast-beetles

[cboettig]

relatedly -- is it odd not to also be adjusting richness for sampling effort (i.e. trap nights)? e.g. some sites will have a lot less trap-nights in a given bout, month, or year, and surely this contributes to ability to detect rare species at that site?

[taddallas]

That makes sense as a way to calculate trap nights, from what I understand after a brief read. I really am curious how much trap-nights would actually influence richness. In my mind, richness should be more robust to sampling error, as if we assume that each species has the same probability of being sampled across traps, trap-nights become a bit less important in estimating overall richness. We could explore this by removing certain sampling periods from the raw data and estimating richness. If I were to bet on it, I would argue that we'd have to exclude around 20% or so of sampled traps to have an appreciable difference on species richness. But I might just be overestimating our ability to get at those rare species.

[cboettig]

@taddallas great point, we should look at the data! Since there's already some natural variation in trapnights across the data, for a first pass we can see how this correlates with effort. Using the quick tables in my Rmds,

df <- left_join(richness, effort) %>%
  mutate(month = lubridate::month(month, label=TRUE)) 
df %>%
  ggplot(aes(trapnights, n)) + geom_point() + facet_wrap(~month) +
  geom_smooth(method='lm', formula= y~x)

err.. I dunno what that says. Certainly bouts at sites with really low trapnights will produce low estimates of richness. No idea why there's some outliers with huge trapnights, but a wild guess that those are driven by the "whoops I thought you picked up last months traps!, this trap has been out here for ages!". In any event outliers seem to get only average levels of richness.

Maybe this needs to be plotted by site to get a better sense as to whether we are saturating species richness or not (there's probably some well-developed theory to answer that question). I really don't have any viable biological intuition to go on, but given how many traps seem to show up empty or at least get less than one beetle a night, catching beetles seems to be harder than I thought, and so I'd guess you'd need a hell of a lot of trap effort to catch all the species present.

I suppose one could spitball the expected Carabid species richness from some occurrence maps for the "area" (iNaturalist? map of life?) According to wikipedia, there's roughly 2000 species of north american carabids, while it looks like there's 744 carabids in the NEON data. And 129 of those unique species have been seen only once by all the NEON traps.