Ben Staton

Fisheries scientist with Bayesian tendencies. Enthusiastic R programmer. Here to learn about and share articles, tips, and tricks. Any opinions are my own.

[Ben Staton @bstat0n]

Pacific #salmon have been returning younger and smaller, but stock assessments have ignored the likely reductions in reproductive output (RO). In a new @cjfas paper, Catalano, Fleischman, @JanOhlberger, and I present a method that seeks to remedy this https://t.co/05PjCaZzdp 1/13

[Ben Staton @bstat0n]

Correcting the automatically generated transcripts of the narration on my recorded talks for #AFS152, and apparently our logit-linear model is actually a legit-linear model. I suspected it was legit, but always good to get the confirmation. New favorite link function discovered!

[Ben Staton @bstat0n]

@GrantAd17148426 This topic interests me too. Have you seen these papers & refs within? We've found it doesnt make much of a difference as long as the (co)variance in the 'observed' variables is carried through

https://t.co/gEBCIregpQ

https://t.co/vmFrdBJNpv

[Ben Staton @bstat0n]

@ColeMonnahan @TrevorABranch So does this mean we can just discard the draws that are divergent and make inference from the remaining samples? Or is it strictly essential to reparameterize/futz with settings until there are no/very few divergent transitions?

[Ben Staton @bstat0n]

Exciting news: GitHub *finally* supports TeX-like math expressions in markdown!

https://t.co/rFgTW87WU3

[Ben Staton @bstat0n]

All code and data to replicate our analyses are available and the code could be adapted for similar data sets or to simulation-test specific cases.

First post-grad lead-author pub in the books!✅ https://t.co/q6b1x4jGx1 5/5

[Ben Staton @bstat0n]

New paper on integrated modeling of covariate effects on detectability based on double-sampling designs with @OccupyRivers, @MattKaylor5, and others. We apply it to snorkel counts with paired mark-recap and challenge it with simulated data. https://t.co/Ms4YSagXM3 🧵1/5

[Ben Staton @bstat0n]

We found snorkel survey detection probability varies with habitat type (pool or not), depth, species, and visibility for salmonids in NE Oregon streams. The model often out-performed the non-integrated approach in simulation trials, especially when assumptions were violated. 4/5

[Ben Staton @bstat0n]

@spoonbill_hank @kbg_conserv @FisheriesRobert @SJC_fishy @DirtyDog650 @EvaThorstad @SpinyDag You might find something useful in the references we cite in the intro to this article where we talk about salmon run timing biology: https://t.co/TtTmZcS9Sg. Good example of staging behavior anyways

[Ben Staton @bstat0n]

Still, our work is a step in what we see as the right direction for salmon pop dyn models that inform harvest policies, and should be helpful for analysts wishing to account for demographic changes in other systems. All code and data are hosted at https://t.co/VmXDiG5DFd. 13/13

[Ben Staton @bstat0n]

Pacific #salmon have been returning younger and smaller, but stock assessments have ignored the likely reductions in reproductive output (RO). In a new @cjfas paper, Catalano, Fleischman, @JanOhlberger, and I present a method that seeks to remedy this https://t.co/05PjCaZzdp 1/13

[Ben Staton @bstat0n]

Our analysis was enabled by many assumptions that are difficult or impossible to evaluate with the available data, and violations in them would render our inferences less valid - we devoted a lengthy discussion paragraph to the many ways the model could give poor inference. 12/13