www.phidot.org

Hi,

Just hoping someone can confirm my understanding of the abundance estimate in MSORD is correct.

As in the example in Chapter 15 of the guide, I have transients and residents, with one observable nesting state and one unobservable skipping state.

It seems to me the estimate of abundance from this model in any season must be transient + resident adult female nesters, and not include the skipping females since most parameters for the skipping state are necessarily set to zero.

The one exception being survival in the skipping state which is set equal to survival in the nesting state for the sole reason of estimating probability of skipping, but not abundance of those skipping.

However, I've been asked to estimate total adult female population abundance (nesters + skippers).
I think this can't be done, but I'd be grateful for any advice.

You are correct that the abundance estimate from the MSORD model applies only to the number in observable states (nesters). This model is fully described in Kendall et al. (2020, Ecological Monographs). In that same paper we discuss the issue of estimating total abundance for this case you describe. If nesting probability were completely random (nesting probability this year is the same for those that did or did not nest last year), then total abundance would just be (nester abundance)/(nesting probability). The more common case of Markovian nesting probabilities requires a more complicated derivation based on the dynamics of the population, including an estimate of annual recruitment to the nesting state. We estimate this value for a population of hawksbill sea turtles in the paper, but we were comfortable doing so because we had reliable estimates of recruitment.

Hi,

As I understand it, in Kendall et al (2020) detection probabilities were close to 1.0 so observed new recruits were confidently substituted for estimated new recruits in the equation for total abundance.

In our MSORD (same as Chapter 15 in the guide, 2 states observed and unobserved, and separate survival for transients and residents), detection rates are often quite low (around 0.5, 0.6). As we need estimated recruits, I'm tempted to adjust observed recruits by detection rate for input to the total abundance equation. The assumption being that we missed new recruits at the same rate as we missed residents.

I'd be grateful to know if you think this approach is ok, or fundamentally flawed in some way?

What I'm thinking of doing is identifying transients in the sample, subtracting them from observed recruits and then dividing by the detection rate to give an estimate of new recruits in the population.

The added complication for us is that we have detection rates for each secondary occasion, so I'd be planning on estimating recruits by secondary occasion and summing them for the year.

If this is not ok, is there some way pent (probability of a new arrival) could be used instead?