Premium
Estimating population size for Hawaiian monk seals using haulout data
Author(s) -
Harting Albert L.,
Baker Jason D.,
Johanos Thea C.
Publication year - 2017
Publication title -
the journal of wildlife management
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.94
H-Index - 111
eISSN - 1937-2817
pISSN - 0022-541X
DOI - 10.1002/jwmg.21303
Subject(s) - abundance (ecology) , population , range (aeronautics) , percentile , statistics , population size , biology , demography , ecology , mathematics , materials science , sociology , composite material
Obtaining a range‐wide abundance estimate for the Hawaiian monk seal ( Neomonachus schauinslandi ; monk seal) has formerly not been possible because of data limitation at sites where the only available data are infrequent counts. We describe a new method for obtaining abundance estimates for these sites that relies upon the proportion of the non‐pup population observed in standardized counts at other sites with known abundance. We converted these proportions ( p ), compiled over all sites and years having full population enumeration, to haulout correction factors (CF), where CF = 1/ p . We then applied these CF values to counts at sites with unknown population size to provide a bounded distribution of population estimates. We used 2,179 CFs from 44 site‐years at intensively studied sites. We used these CFs for population estimation at 2 sites, Necker and Nihoa Islands, Hawaii, USA, where infrequent counts occurred and total abundance was unknown. The resulting population estimates ( x ¯and 5 and 95 percentiles) indicated that abundance had increased at Nihoa from 2001 ( x ¯ = 31 non‐pups, 5–95% = 21–47) to 2015 ( x ¯ = 116, 5–95% = 79–177), but there was no increase evident at Necker. We conducted method validation by randomly selecting 1–5 of the observed counts at site‐years with full enumeration, applying the CFs to the observed count, and testing whether the true (known) value lay within the 5–95% range of the resulting distribution of estimates. When we used only 1 count, the method succeeded in capturing the true value in >85% of 1,000 randomizations for single islands and 65–87% of the randomizations in atolls. Performance markedly improved when we used >1 randomly selected count to estimate population size. For the monk seal, this methodology represents a significant advancement because it enables a more complete estimate of the species’ abundance and a better assessment of the relative importance of different regions of its range to the overall recovery initiative. © 2017 The Wildlife Society.