search

ebuhle / LeatherbackNesting

Analyses of trends in Florida leatherback sea turtle nesting phenology and success
1 stars 0 forks source link

Questions about Zone, Distance to High Water Line, and Distance to Toe of Dune #6

Open ebuhle opened 3 years ago

ebuhle commented 3 years ago

Hey @hoffmannsarahlouise, I don't know if I'll catch you before you head out to do actual ecology, but I have some hopefully easy questions about ecology in silico.

  1. What does Zone refer to? The data dictionary just says "location". I was thinking it might be standard vertical zonation, but it's not associated with either distance to HWL or distance to dune in any obvious way at any beach. Is it just a (longitudinal) spatial survey unit within the beach? And if so, should we be including it as a random effect?

  2. I thought I understood distance to HWL and distance to dune (in m, it looks like), but I'm perplexed by the negative values. Negative distance to HWL I can imagine would mean a nest in the intertidal zone (which surprises me, but there aren't very many of them). But what does a negative distance to toe of dune mean? It's above the toe, higher in the dune?

Thanks for helping me get up to speed. FWIW, I ask because it looks like there may be some interesting patterns in emergence rate as a function of position on the shore. They've taken a little while to tease out, and they're going to require a somewhat fancier model -- one that treats the probability of nest failure (zero emergence) separately from the emergence rate given non-failure. Before I commit, I want to make sure I know what I'm doing.

hoffmannsarahlouise commented 3 years ago

I did not know that those emails don't post back to here..... sorry! My working knowledge of github is pretty limited!

  1. The zones refer to locations along the beach. I know, for example, that zones 1-7 are public beach and the others are private. We looked at the effect of zone on nest density and success for another analysis on a different species, but did not find an effect. The whole beach itself is pretty small compared to the actual nesting footprint in Florida and I would be surprised if there was a signfiicant effect. Things that may differ between zones to some degree are lighting, densitiy of people on the beach, and possibly dune composition.
  2. You have this definition correct. Turtles will lay nests in the intertidal zone sometimes. Often those nests will be relocated above the HWL. I'm not sure we know exactly why that happens, could be that a female as attempted nesting and false crawled a couple nights in a row and just needs to get the eggs out? Could also be that there was some lighting disorientation or something like that. Similarly, turtles will also nest above the dune. It's less common with leatherbacks but does happen. Let me pass that one on to Sarah and Justin just to confirm.

Thanks for continuing to work on it! It's really exciting to see some trends emerging! Let me know if you have any other questions!

ebuhle commented 3 years ago

Excellent, thanks. I eventually figured out the answer to (1) by plotting latitude against zone (duh). I haven't tried including zone as a random effect, but it doesn't look like it would do much to "explain" the patterns in emergence success. That's another Issue (literally) and one that I'm still puzzling over. The sneak preview: contrary to what I wrote above, it's not just that emergence rate is zero-inflated; there's actually a minor mode of low values (< 10% or so) in addition to the main mode around 75%, but so far I have not found any predictors that distinguish the blob of low values from more successful nests.

While I'm asking remedial questions about the data, though, here's one more. You mentioned in #4 that you were surprised not to see evidence of a trend in nesting phenology. I subsequently took a closer look at the Date of First Nest sheet in Historical Leatherback Data_updated1.18.2020.xlsx and discovered one potential reason for that surprise. That sheet plots an annual time series of DOY of the first nest. It's hard to compare directly to our models predicting the mean DOY, but more importantly I realized that the plot includes data back to 2001. And if you restrict the time series to 2007-2020, like our full data set, the R^2 of the linear trend drops from ~40% to ~5%. So clearly the contrast in the full 20-yr time series is driving the trend in nest timing.

Likewise, the Remigration sheet contains detection histories (perfectly formatted for CJS, by the way!) back to 2000.

So my question is, do (the royal) we have access to the full data set back to 2000 or 2001?

I did not know that those emails don't post back to here..... sorry! My working knowledge of github is pretty limited!

No worries -- email replies to the auto-notifications do post here, but they come with all sorts of email formatting (which I've been editing out of your posts) and of course they don't let you use Markdown or link to code and results in the repo -- not to mention the most useful feature of all! :turtle: I'm also not sure how much of that formatting and linked content actually shows up in the notification emails themselves, since I rarely look at them. By all means, let me know if you have GitHub questions. Although designed for software development, it's a really useful platform for collaborative data analysis projects too.