Estimating reproductive and juvenile survival rates when offspring ages are uncertain: a novel multievent mark-resight model with beluga whale case study

Himes Boor, G.K., T.L. McGuire, A.J. Warlick, R.L. Taylor, S.J. Converse, J.R. McClung, A.D. Stephens

For questions regarding the code, contact Gina Himes Boor

Citation:

Himes Boor, G.K., T.L. McGuire, A.J. Warlick, R.L. Taylor, S.J. Converse, J.R. McClung, A.D. Stephens. 2023. Estimating reproductive and juvenile survival rates when offspring ages are uncertain: a novel multievent mark-resight model with beluga whale case study. Methods in Ecology and Evolution 14(2):631-642.

---

Abstract

1. Understanding the survival and reproductive rates of a population is critical to determining its long-term dynamics and viability. Mark-resight models are often used to estimate these demographic rates, but estimation of survival and reproductive rates is challenging, especially for wide-ranging, patchily distributed, or cryptic species. In particular, existing mark-resight models cannot accommodate data from populations in which offspring remain with parents for multiple years, are not always detected, and cannot be aged with certainty.
2. Here we describe a Bayesian multievent mark-resight modelling framework that uses all available adult and adult-offspring sightings (including sightings with older offspring of uncertain age) to estimate reproductive rates and survival rates of adults and juveniles. We extend existing multievent mark-resight models that typically only incorporate adult breeding state uncertainty by additionally accounting for age uncertainty in unmarked offspring and uncertainty in the duration of the mother-offspring association. We describe our model in general terms and with a simple illustrative example, then apply it in a more complex empirical setting using thirteen years of photo-ID data from a critically endangered population of beluga whales (Delphinapterus leucas). We evaluated model performance using simulated data under a range of sample sizes, and adult and offspring detection rates.
3. Applying our model to the beluga data yielded precise estimates for all demographic rates of interest despite substantial uncertainty in calf ages, including non-breeder survival and reproductive rates lower than that estimated for other beluga populations. Simulations suggested our model yields asymptotically unbiased parameter estimates with good precision and low bias even with moderate sample sizes and detection rates.
4. This work represents an important new development in multievent mark-resight modeling, allowing estimation of reproductive and juvenile survival rates for populations with extended adult – offspring associations and uncertain offspring ages (e.g., some marine mammals, elephants, bears, great apes, bats, and birds). Our model facilitated estimation of robust demographic rates for an endangered beluga population that were previously inestimable (e.g., non-breeder and juvenile survival, reproductive rate) and that will yield new insights into this population’s continued decline.

---

CODE:

Model:
CIBW_ME_Model-ms_code.R: R code for the multievent mark-recapture model developed to estimate survival and reproductive rates from sighting history data of marked adults that are sometimes observed with their unmarked offspring of unknown age. The model was originally developed for photo-ID data from Cook Inlet beluga whales, but it can be adapted for other populations with similar characteristics (i.e., extended parental care of offspring that can not always be aged with certainty).

Simulations:
CIBW_ME_sim.R: R code for the accompanying simulation analysis for the multievent mark-recapture model examining model performance across varying detection rates and sample population sizes.

DATA:

ms_SH_data.csv: formatted Cook Inlet beluga whale photo-ID mark-recapture data for running the multievent model described above. The data were collected by Dr. Tamara McGuire and colleagues at The Cook Inlet Beluga Whale Photo-ID Project and should not be used outside of this analysis without express permission from Dr. McGuire.

Additional Required Files:

start_mat-ms_SH_data.csv: formatted starting latent matrix required by JAGS to run the multievent model (see Inputs README file for more information about this file)

---

Funding

Funding for development of this model came from the North Pacific Research Board (project # 1718). Cook Inlet beluga photo-ID data were collected using a variety of funding sources.