Satellite tags have revolutionized our understanding of marine animal movements. However, tags may stop transmitting for many reasons and little research has rigorously examined tag failure. Using a long-term, large-scale, multi-species dataset, we evaluated factors influencing tracking duration of satellite tags to inform study design for future tracking studies. We leveraged data on battery status transmitted with location data, recapture events, and number of transmission days to probabilistically quantify multiple potential causes of failure (i.e., battery failure, premature detachment, and tag damage/fouling). We used a combination of logistic regressions and an ordinary linear model including several predictor variables (i.e., tag type, battery life, species, sex, size, and foraging region). We examined subsets of data from 360 satellite tags encompassing 86,889 tracking days deployed on four species of marine turtles throughout the Gulf of Mexico, Caribbean, and Bahamas from 2008 to 2019. Only 4.1% of batteries died before failure due to other causes. We observed species-specific variation in how long tags remain attached: hawksbills retained 50% of their tags for 1649 days (95% CI 995–1800), loggerheads for 584 days (95% CI 400–690), and green turtles for 294 days (95% CI 198–450). Estimated tracking duration varied by foraging region (Caribbean: 385 days; Bahamas: 356; southern Gulf of Mexico [SGOM]: 276, northern Gulf of Mexico [NGOM]: 177). Additionally, we documented species-specific variation in estimated tracking duration among foraging regions. Based on sensor data, within the Gulf of Mexico, across species, we estimated that 50% of tags began to foul after 83 95% CI (70–120) days. The main factor that limited tracking duration was tag damage (i.e., fouling and/or antenna breakage). Turtles that spent most of their time in the Gulf of Mexico had shorter tracking durations than those in the Bahamas and Caribbean, with shortest durations observed in the NGOM. Additionally, tracking duration varied by species, likely as a result of behaviors that damage tags. This information will help researchers, tag companies, permitting agencies, and funders better predict expected tracking durations, improving study designs for imperiled marine turtles. Our results highlight the heterogeneity in telemetry device longevity, and we provide a framework for researchers to evaluate telemetry devices with respect to their study objectives.

中文翻译：

---

已实现的海龟卫星追踪持续时间的驱动程序

卫星标签彻底改变了我们对海洋动物运动的理解。但是，由于多种原因，标签可能会停止传输，很少有研究严格检查标签故障。使用长期的，大规模的，多物种的数据集，我们评估了影响卫星标签跟踪持续时间的因素，以为未来的跟踪研究提供信息。我们利用与位置数据，重新捕获事件和传输天数一起传输的电池状态数据来概率性地量化故障的多种潜在原因（即电池故障，过早脱离和标签损坏/结垢）。我们结合了逻辑回归和普通线性模型，其中包括几个预测变量（即标签类型，电池寿命，种类，性别，大小和觅食区域）。我们检查了2008年至2019年期间在360个卫星标签中收集的数据子集，其中包括在墨西哥湾，加勒比海和巴哈马群岛上的四种海龟上部署的86,889个跟踪日。只有4.1％的电池因其他原因而在失效之前死亡。我们观察到了特定种类的标签在标签上保留多长时间的变化：在1649天（95％CI 995–1800）中保留了其标签的50％，在584天（95％CI 400–690）中保存了，而海龟则保留了294天天（95％CI 198–450）。估计的追踪持续时间因觅食地区而异（加勒比：385天；巴哈马：356；墨西哥湾南部[SGOM]：276，墨西哥湾北部[NGOM]：177）。此外，我们在觅食区域之间的估计跟踪持续时间中记录了特定物种的变化。根据墨西哥湾内跨物种的传感器数据，我们估计83％95％CI（70-120）天后50％的标签开始变脏。限制跟踪持续时间的主要因素是标签损坏（即结垢和/或天线损坏）。与巴哈马和加勒比海地区相比，在墨西哥湾度过大部分时间的海龟的追踪时间较短，而在NGOM中观察到的时间最短。此外，跟踪持续时间因物种而异，可能是由于损坏标签的行为所致。这些信息将帮助研究人员，标签公司，许可机构和资助者更好地预测预期的跟踪持续时间，从而改善濒危海龟的研究设计。我们的研究结果突出了遥测设备寿命的异质性，并且为研究人员提供了一个框架，可根据他们的研究目标评估遥测设备。