An analysis of historical bottlenose dolphin (Tursiops truncatus) strandings in the Mississippi Sound, USA using classification and regression trees (CART)

Contenido principal del artículo

J. L. Pitchford
B. J. S. Serafin
D. Shannon
A. T. Coleman
M. Solangi

Resumen

Trends in bottlenose dolphin (Tursiops truncatus) strandings can be used to examine several factors associated with mortality and life history and are essential for detecting unusual mortality events (UMEs). This study characterised stranding trends in the Mississippi Sound (MS) region of the northern Gulf of Mexico (GoM) from 1996–2009 using kernel density estimation (KDE) and classification and regression tree (CART) analysis. An annual mean of 26.1 strandings (n = 14), SD = 13.7, 95% CI [18.2, 34.0] and a peak in strandings during spring (March–May) were evident from our analyses. Neonates stranded almost exclusively in spring indicating that this is the dominant breeding and calving season in this area. Spatial distributions revealed that the majority of dolphins stranded along central and western portions of the MS Sound near Gulfport, MS and on Ship Island during the spring and summer months, but were more often found in the eastern MS Sound during winter and autumn. Our CART analyses indicated that 1996, which contained a declared UME, was anomalous from other years as the number of adult, sub-adult and juvenile strandings was relatively high during the autumn and winter. Further, our analyses showed that the location of those strandings on Ship Island in autumn and winter was unique from all other years in the historical record. These results represent historical conditions that can be used as a baseline for future studies of the effects of environmental disturbances, including UMEs, in MS. This research also demonstrates the versatility and usefulness of CART for describing historical trends, detecting departures from the norm and explaining UMEs within the framework of a single analysis. This approach represents an objective assessment tool that could be used to assist governmental agencies with determining the onset of a UME and could help support or refute the cause of these events.

Detalles del artículo

Sección
Articles