Threatened species are dependent on conservation, with strategic approaches underpinned by the principle of complementarity to find efficient solutions. But are paper-efficient designs always field-effective? After 51 years of monitoring a 56-km-long overlap of two turtle rookeries, there are contrasting population-recovery trends: Vulnerable loggerhead abundance is increasing; and Critically Endangered leatherback abundance is remaining constant, despite leatherback individuals having a higher reproductive output. This questions the efficacy of the conservation programme (annual monitoring and land–sea protection in a World Heritage Site). We use biotelemetry to test if the disparate recovery is biased by differences in detectability in the monitored section of the rookery, and if the reserve confers equal protection. The species' movement ecology contrasts strongly, with implications for nest-event detectability: ~ 66% of leatherback nesting is outside the monitoring area, compared to ~ 12% of loggerhead nesting. The marine reserve also strongly favours loggerheads at 95% protection, versus 25% protection for leatherbacks. We hypothesize that variability in leatherback movement ecology, and nest placement, is from ocean currents shaping their behaviour as hatchlings, potentially also determining the proportion of the population at risk of capture in pelagic fisheries. Efficient multi-species conservation strategies need to be carefully designed and adaptive to be effective.