Proximity shows as count in summary(fit)

[rebeca calanoce]

Hi Murray,

I’m new to SECR, so apologies if these are basic questions.

I’m working with 19 sites, ~30 camera traps per site/session (trapID format), with detector covariates Paired (Simple/Doble) and Effort (days active per camera), so I chose the proximity detector type. I treat detections as binary (one occasion per day). The captfile includes sex (F/M/NA), allowing me to use the hybrid mixture (hcov = "sex") and model sigma ~ h2. I don’t have a full usage matrix by occasion, but I do have the number of cameras per session.

1) Detector type changes to "count"
After reading and forcing proximity for all trap sets, summary(caps_all) shows Detector type = proximity and Occasions = 90. However, after fitting some models, summary(fit) reports Detector type = count (and sometimes Occasions = 1).
What could cause this change inside the fitted object?

2) HHN (a0) with unequal effort
Efford & Mowat (2014) note that using a0 as a surrogate for a is unreliable when sampling intensity varies. In my data, the number of active cameras differs between sessions (theft/failures). Without usage by occasion, is it reasonable to account for effort by adding a session covariate (e.g., numcams) to the a0 model? Or is it preferable to use HN (g0) in such cases and reserve HHN for when usage is available?

Any guidance on why “count” appears in the fitted object and on best practice for HHN vs HN with variable effort and no usage data would be greatly appreciated.

Thanks,
Rebeca.


Script:

caps_all <- read.capthist(
  captfile     = "all_caps.txt",
  trapfile     = list(AT="AT_trap.txt", CA="CA_trap.txt", ...),  # 19 sessions
  detector     = "proximity",
  fmt          = "trapID",
  covnames     = "sex",                    # F/M/NA
  trapcovnames = c("Paired","Effort"),     # detector covariates
  noccasions   = rep(90, 19),
  verify       = TRUE
)

# Two comparable models (robust HN vs HHN with a0)
fit_HN <- secr.fit(
  caps_all, mask = masklist,
  detectfn   = "HN",
  hcov       = "sex",
  model      = list(D ~ session, g0 ~ Paired, sigma ~ h2),
  verify     = FALSE, trace = FALSE
)

fit_HHN <- secr.fit(
  caps_all, mask = masklist,
  detectfn   = "HHN", details = list(param = 3),  
  hcov       = "sex",
  sessioncov = sessioncov,
  model      = list(D ~ session, a0 ~ Paired, sigma ~ h2),
  verify     = FALSE, trace = FALSE
)

# Summary(fit_HHN)$detector shows "count" here

[Murray Efford]

Hi Rebeca 

Well spotted.

To save time secr.fit automatically collapses proximity data to counts unless you set details=list(fast proximity=FALSE). Look around for advice on this e.g. SECR book.

Murray

--
You received this message because you are subscribed to the Google Groups "secr" group.
To unsubscribe from this group and stop receiving emails from it, send an email to secrgroup+...@googlegroups.com.
To view this discussion, visit https://groups.google.com/d/msgid/secrgroup/313099d4-f6f1-4fdb-a289-9f90dc887df1n%40googlegroups.com.