Skip to main content

    Power analyses of CEMP indices for penguins at Admiralty Bay and fur seals at Cape Shirreff and Seal Island

    Request Meeting Document
    Document Number:
    WG-EMM-03/52
    Author(s):
    G.M. Watters, R.P. Hewitt, W.Z. Trivelpiece and M.E. Goebel (USA)
    Agenda Item(s)
    Abstract

    We used the program MONITOR to conduct a set of power analyses for CEMP indices developed by the US AMLR Program. We consider CEMP indices A2, A3, A5, A6, A7, and A8 for penguins at Admiralty Bay, and indices C1 and C2 for fur seals at Cape Shirreff and Seal Island. Our power analyses describe the likelihood of rejecting null hypotheses stating that the slopes of linear models relating the values of transformed CEMP indices to breeding year are equal to zero when, in fact, these trends actually exist. We estimated the power of rejecting these null hypotheses at three levels of ? (the probability of rejecting the null hypotheses when they were, in fact, true), three periods of monitoring, and 21 levels of annual percent change in the CEMP indices. We conducted a number of “combined analyses” in which observations for different species, sexes, colonies/rookeries, and periods during the breeding chronology were treated as separate “plots” that might be monitored in an attempt to detect some type of overall trend. We displayed the results of our analyses in a tabular format designed to identify general patterns rather than specific outcomes. In general, power to detect change was increased when ? levels, periods of monitoring, and levels of change were increased. We found that many of the CEMP indices considered here are contaminated by so much observation error that there would often be less than a 50:50 chance of detecting an actual trend over a period of 5 years. Nevertheless, detecting trends over a period of 20 years may be feasible for most indices. Detecting trends in indices A8 and C2 may be extremely difficult under any combination of ?, observation period, and level of change, but detecting trends in log-abundance of breeding penguins (one possible transformation of Index A3) will, apparently, be relatively easy . Treating observations for different species, sexes, colonies/rookeries, and periods of the breeding chronology as separate plots both degraded (e.g., combining species for Index A2 and periods of the breeding chronology for Index A5) and enhanced (e.g., combining species for Index A6 and chinstrap penguin colonies for Index A3) the ability to detect real trends. It is difficult, however, to know whether the “combined” approach is warranted because we are uncertain of the degrees to which different species, sexes, colonies, etc. are jointly affected by some overall trend. We caution that our analyses should, at this point, still be considered exploratory since there are underlying questions regarding the appropriate use of MONITOR that we have not been able to resolve.