Read the draft final research report estimating the demographic parameters of New Zealand sea lions in the Auckland Islands from 1997/98 to 2010/11.
New Zealand sea lionTag-resight data from female New Zealand sea lions on Enderby Island, collected between 1997/98-2010/11, was analysed to estimate survival and breeding
- Multi-state mark-recapture models were used for the estimation within a Bayesian statistical framework.
- Demographic parameters were presumed to vary annually, and also due to age and breeding status. Females with differing numbers of tags or other markings were presumed to have different daily resighting rates. Daily resighting rates were also presumed to vary annually and by age.
- Two definitions of breeder were used based upon the recorded status of a female; a strict and more liberal definition.
- For each data set, four models were compared to determine which best represented the data. The four models represented combinations of two factors being considered; 1)
whether age relationships were best described using 3 age-classes (0-3, 4-14 and 5+) or a linear-logistic relationship; and 2) whether the number of resights per season per
females was best described by a binomial or zero-inflated binomial distribution.
- For both data sets, the models with the zero-inflated binomial distribution performed better, though there was little separation between the age-classes and linear-logistic
- Average survival probabilities of prime-age females (around 9-years old) was approximately 0.90 for non-breeding females and 0.95 for breeding females, using
both data sets. Survival was lower for females that were much older or younger than this. There was no discernible temporal pattern in survival probabilities.
- Average breeding probabilities of prime-age females was approximately 0.30 and 0.65 for females that did not breed and did breed in the previous year respectively,
using the stricter definition of breeder. Using the more liberal definition of breeder, average breeding probabilities were, approximately 0.35 and 0.68 for females that did
not breed and did breed in the previous year respectively. While there were some apparent temporal patterns in estimated breeding probabilities, these are possibly
confounded with a changing age-structure or tagged females.
- An exploratory analysis was conducted using an alternative approach to representing the nature of the relationship between age and demographic parameters that does not
require the general parametric form to be defined a priori, allowing the data to have greater influence in determining the general form of the relationship. The exploratory
models fit the data better than those used in the primary analysis although a number of issues remain to be fully considered before a full analysis is recommended using this
approach. The general method does hold some promise for future analyses.
Author: Darryl I. MacKenzie
Research Report by Proteus Wildlife Research Consultants