GPS radio-collars for spatial dynamics research

了解生物如何利用空间上的异质资源，以及它们如何在景观马赛克中生活、繁殖、分散和相互作用一直是生态学的目标。我提议扩展我对塞伦盖蒂生态系统中食物-网相互作用的实地研究，以测试食草动物和食肉动物的运动和空间相互作用的预测模型。从长远来看，我计划将这些运动模型嵌入到食物-网相互作用的空间显式模型中，以测试适应性运动行为对捕食者和猎物的种群动态和生存能力的影响。在高度流动的大型生物体上收集这类运动数据的唯一现实方法是使用GPS无线电发射器。该仪器使用卫星全球定位系统技术，以频繁的间隔精确地三角测量生物的空间位置。这种空间信息在项圈的记忆库中存储很长一段时间，然后使用远程遥测链接或通过从研究动物身上检索项圈来下载信息。运动轨迹的统计预测将基于状态空间运动模型，使用有偏见的随机行走的混合，其参数是使用我的研究小组帮助开创的蒙特卡洛马尔可夫链方法估计的。每个备选假设确定了运动模型中包含的不同参数组合，并使用信息论方法来评估相互竞争的模型。在每种情况下，使用GPS无线电项圈收集的数据将被用来对有偏向的随机行走模型(转弯角度和步长分布)进行参数化，空间上明确的参数由当地牧场条件、生境覆盖以及当地食草动物或食肉动物密度施加。

Understanding how organisms use spatially heterogeneous resources and how they live, reproduce, disperse and interact in landscape mosaics have long been goals in ecology. I am proposing to extend my field studies of food-web interactions in the Serengeti ecosystem to test predictive models of movement and spatial interactions by both herbivores and carnivores. In the longer term, I plan to imbed these movement models into spatially-explicit models of food-web interactions to test the impact of adaptive movement behavior on population dynamics and viability of both predators and prey. The only realistic way to gather movement data of this sort on highly mobile, large organisms is to use GPS radio-transmitters. This instrumentation uses satellite global positioning system technology to precisely triangulate spatial position of organisms at frequent intervals. This spatial information is stored for extended periods in a memory bank in the collar, then the information is downloaded either using a remote telemetry link or by retrieving the collar from the study animal. Statistical prediction of movement trajectories will be based on state-space movement models, using mixtures of biased random walks, the parameters of which are estimated using Monte Carlo Markov Chain methods which my research group helped pioneer. Each alternate hypothesis identifies a different mix of parameters incorporated into the movement model, with information-theoretic methods used to assess competing models. In each case, data gathered using GPS radio-collars will be used to parameterize biased random walk models (turn angles and step length distributions), with spatially explicit parameters imposed by local range conditions, habitat cover, and local herbivore or carnivore densities.