Abstract:
Mark-recapture analysis of populations is becoming an important tool in population biology.
Mark-recapture methods can be used to estimate transition probabilities among life-stages
from capture histories of marked individuals for which stages can be determined at each
sampling occasion. This method is called a multi-stage mark-recapture (MSMR) method.
In this thesis, I describe advances I made in the MSMR method and present analyses that
apply this method to actual data.
The advances I made in the MSMR method are motivated by a need to provide a link
between mark-recapture data and demographic models such as matrix population models
and integrodifference models. I resolve some issues that are commonly encountered during
sampling, such as the fact that the sex or life-stage of some individuals is unknown during
some sampling occasions and that individuals become unobservable during some life-stages.
I introduce a stage-structure that permits simple conversion of estimated transition probabilities
into a matrix population model. I describe an algorithm to simplify programming
for parameter estimation. I also introduce a method to estimate the distribution of dispersal
displacements (a dispersal kernel) from mark-recapture data.
I apply some of the methods described above to data of the North Atlantic right whale
(Eubalaena glacialis) The right whales are considered one of the most endangered mammals.
The current population size is about 300 in the northwestern Atlantic, and the
number is declining. I applied the multi-stage mark-recapture statistics to the 17-year individual
sighting history data. Using the estimated transition probabilities, I constructed
a population projection matrix, which was used for further demographic analyses. I found
that the population was slowly increasing in 1980, but it started to decline slowly around
1992. I show that (1) this change was caused by increased mortality of females that have
just given birth, (2) protecting two females a year from the deaths is enough to prevent the
declining trend, and (3) demographic stochasticity is a more important factor influencing
their long-term viability than environmental stochasticity.
