MODELS TO EVALUATE HEADSTARTING AS A MANAGEMENT TOOL FOR LONG-LIVED TURTLES

Most turtle species suffer high mortality in their first year, have a long juvenile period, and can live for decades once they reach adultho od. Conservationists have implemented a number of recovery plans for t hreatened turtle populations, including experimental ''headstart'' pro grams. Headstarting involves the captive rearing of hatchlings from eg gs collected in the wild. The hatchlings are held for several months t o help them avoid high mortality in their first year. It is hoped that these turtles survive and grow like wild turtles after release. The p urpose of our study was to evaluate headstarting as a management tool for threatened turtle populations. We critically examined the populati on-level effects of headstarting with a series of deterministic matrix models for yellow mud turtles (Kinosternon flavescens), a ''non-threa tened,'' well-studied species, and endangered Kemp's ridley sea turtle s (Lepidochelys kempi). We show that management efforts focused exclus ively on improving survival in the first year of life are unlikely to be effective for long-lived species such as turtles. Population projec tions for both turtles predict that headstarting can augment increasin g populations when adult survival is returned to or maintained at high levels, provided that headstarted juveniles are as vigorous as wild t urtles. However, when subadult and adult survival is reduced, headstar ting cannot compensate for losses in later stages. Proportional sensit ivity (elasticity) analyses of stage-based matrix models indicated tha t annual survival rates for subadult and adult turtles are most critic al; small decreases in the survival of older turtles can quickly overc ome any potential benefits of headstarting. In general, the biological benefits of headstarting programs may be overestimated for turtles, a nd a careful examination of stage-specific mortality sources, demograp hy, and life history can guide us toward more effective management str ategies.