Sampling design considerations for demographic studies: a case of colonial seabirds

Abstract

For the purposes of making many informed conservation decisions, the main goal for data collection is to assess population status and allow prediction of the consequences of candidate management actions. Reducing the bias and variance of estimates of population parameters reduces uncertainty in population status and projections, thereby reducing the overall uncertainty under which a population manager must make a decision. In capture-recapture studies, imperfect detection of individuals, unobservable life-history states, local movement outside study areas, and tag loss can cause bias or precision problems with estimates of population parameters. Furthermore, excessive disturbance to individuals during capture-recapture sampling may be of concern because disturbance may have demographic consequences. We address these problems using as an example a monitoring program for Black-footed Albatross (Phoebastria nigripes) and Laysan Albatross (Phoebastria immutabilis) nesting populations in the northwestern Hawaiian Islands. To mitigate these estimation problems, we describe a synergistic combination of sampling design and modeling approaches. Solutions include multiple capture periods per season and multistate, robust design statistical models, dead recoveries and incidental observations, telemetry and data loggers, buffer areas around study plots to neutralize the effect of local movements outside study plots, and double banding and statistical models that account for band loss. We also present a variation on the robust capture-recapture design and a corresponding statistical model that minimizes disturbance to individuals. For the albatross case study, this less invasive robust design was more time efficient and, when used in combination with a traditional robust design, reduced the standard error of detection probability by 14% with only two hours of additional effort in the field. These field techniques and associated modeling approaches are applicable to studies of most taxa being marked and in some cases have individually been applied to studies of birds, fish, herpetofauna, and mammals.

A Synthesis of Tagging Studies Examining the Behaviour and Survival of Anadromous Salmonids in Marine Environments

“…However, tag loss/shed was measured in only 7.7% (n = 16) of studies and was not even acknowledged in 69.0% (n = 143) of studies where it could possibly have occurred. Double-tagging individuals could minimize the impacts of tag loss/shed on survival estimates [173], an approach that was employed in 28.5% (n = 59) of the studies. Assuming that loss/shed of the two tags are independent, information on the number of fish recaptured with one or both tags could be used to estimate the probability of tag loss/shed and then used to adjust survival estimates [173].…”

“…Double-tagging individuals could minimize the impacts of tag loss/shed on survival estimates [173], an approach that was employed in 28.5% (n = 59) of the studies. Assuming that loss/shed of the two tags are independent, information on the number of fish recaptured with one or both tags could be used to estimate the probability of tag loss/shed and then used to adjust survival estimates [173]. Alternatively, multistate capture-recapture models could be used to jointly estimate survival and tag/shed loss [174].…”

Dynamics of Breeding-Season Site Occupancy of the California Spotted Owl in Burned Forests

“…Local extinction from occupied sites and colonization of unoccupied sites over time are the dynamic processes that govern changes in occupancy, and the rates of these processes provide important information about the long-term sustainability of a population and effects of management (Mazarolle et al 2005, MacKenzie et al 2006. The development of statistical methods for estimating site occupancy and local rates of local extinction and colonization while accounting for imperfect detectability has increased the potential for understanding the effects of natural processes and human disturbance on wildlife populations (MacKenzie et al 2003, 2006, Kendall et al 2009). …”

Heterogeneity in detection probability along the breeding season in Black-legged Kittiwakes: implications for sampling design

“…Our results indicate that detailed knowledge about the population biology of the species under study can be extremely useful for optimizing sampling design (see also Kendall et al 2009;Converse et al 2009). …”