Translocations are increasingly used for species recovery but are often performed without considering how such actions will impact both recipient and source populations. Imperiled species often occur in networks of sub-populations each contributing to connectivity, viability and persistence, and balancing the benefit of translocations to recipient populations against the risk to source populations will support broad conservation goals. Serial translocations of American martens (Martes americana) have resulted in a recovery network in the Great Lakes region, though an important sub-population in the Chequamegon National Forest (Wisconsin, USA) remains imperiled. These martens were sourced from a sub-population in the south of the Superior National Forest (Minnesota, USA); however, it remains unclear if this source can sustain additional removals. To assess viability of both populations under future translocation actions, we quantified vital rates and populated demographic matrix models to project both populations forward 50 years under 16 potential translocation programs. We assessed extinction risk and recovery potential and found that the source population of martens is resistant to declines under temporally dispersed schedules involving < 100 adult females. The recipient population, however, requires 50-100 adult females to reduce extinction risk, and even more to approach recovery. These findings call into question the population recovery goals, and suggest that a single source may be sufficient to reduce recipient extinction risk, but that multiple source populations are needed to promote recovery. Our work highlights the importance of considering both source and recipient populations, and provides an analytical decision-making framework to guide sustainable translocation programming.

In the SSF, we captured American martens (n=242) from November through February 2007-2015 (Erb et al., 2016). Individuals were fitted with Holohil systems LTD (Carp, ON, CA) MI-2 vhf collars (31g) transmitting at 40 pulses per minute, with a pulse mortality signal after 12 hours of inactivity, and an iButton temperature sensor to determine time of death. All animals were relocated every 7-10 days until death, collar loss, or the end of the study; all mortality events were investigated within 1-3 days of detecting a mortality signal.