Monitoring cetacean abundance is often challenged by insufficient funding and/or effort. Large sample sizes are required to obtain reliable estimates via Capture Mark Recapture (CMR) models. Dataset simulations and power analysis can be used to assess 1) minimum sample size to achieve reliable estimates within one sampling season, and 2) the number of repeated estimates required to detect abundance trends. We used photo-identification data from June-September 2012 to estimate abundance of striped dolphins Stenella coeruleoalba in the Gulf of Corinth, Greece, using CMR models. To evaluate minimum sample size needed within one sampling season, we used dataset simulations in program MARK and compared accuracy and precision obtained with increasing sample sizes. We then used Gerrodette’s inequality model to perform a power analysis and compare the smallest annual percent declines detectable by decennial monitoring plans based on 1) one estimate per year, 2) one estimate every two years, and 3) one estimate every five years. The effect of different levels of precision of the estimates (CV=0.01-0.04) was also investigated. Population abundance was estimated at 1,309 (SE=62.45; CV=0.05; 95% CI=1,192–1,437) for open models and at 1,293 (SE=53.66; CV=0.04; 95% CI=1,192–1,403) for closed models. These estimates are substantially larger that those obtained in a previous CMR study on 2009 data, and need to be validated. To obtain reasonably accurate estimates within one sampling season, simulations indicated a minimum sample size of 8 capture occasions, with minimum capture probabilities of 0.5 for closed models and 0.7 for open models. Furthermore, a power analysis indicated that the smallest annual declines (5%–12.5%) could be detected by monitoring plans providing one estimate every two years. High levels of precision are required (CV=0.01 to detect a 5% annual decline). This information is being used to set monitoring goals and optimize the available resources.
Optimizing abundance estimates of striped dolphins in the Gulf of Corinth, Greece / Santostasi, NINA LUISA; Bonizzoni, Silvia; Bearzi, Giovanni. - ELETTRONICO. - (2015). (Intervento presentato al convegno 29th Annual Conference of the European Cetacean Society tenutosi a St. Julians, Malta).
Optimizing abundance estimates of striped dolphins in the Gulf of Corinth, Greece
Nina Luisa Santostasi
;
2015
Abstract
Monitoring cetacean abundance is often challenged by insufficient funding and/or effort. Large sample sizes are required to obtain reliable estimates via Capture Mark Recapture (CMR) models. Dataset simulations and power analysis can be used to assess 1) minimum sample size to achieve reliable estimates within one sampling season, and 2) the number of repeated estimates required to detect abundance trends. We used photo-identification data from June-September 2012 to estimate abundance of striped dolphins Stenella coeruleoalba in the Gulf of Corinth, Greece, using CMR models. To evaluate minimum sample size needed within one sampling season, we used dataset simulations in program MARK and compared accuracy and precision obtained with increasing sample sizes. We then used Gerrodette’s inequality model to perform a power analysis and compare the smallest annual percent declines detectable by decennial monitoring plans based on 1) one estimate per year, 2) one estimate every two years, and 3) one estimate every five years. The effect of different levels of precision of the estimates (CV=0.01-0.04) was also investigated. Population abundance was estimated at 1,309 (SE=62.45; CV=0.05; 95% CI=1,192–1,437) for open models and at 1,293 (SE=53.66; CV=0.04; 95% CI=1,192–1,403) for closed models. These estimates are substantially larger that those obtained in a previous CMR study on 2009 data, and need to be validated. To obtain reasonably accurate estimates within one sampling season, simulations indicated a minimum sample size of 8 capture occasions, with minimum capture probabilities of 0.5 for closed models and 0.7 for open models. Furthermore, a power analysis indicated that the smallest annual declines (5%–12.5%) could be detected by monitoring plans providing one estimate every two years. High levels of precision are required (CV=0.01 to detect a 5% annual decline). This information is being used to set monitoring goals and optimize the available resources.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
