ABSTRACT

Global Positioning System (GPS) technology was used to track movements of extensively farmed red deer hinds over the calving period on two New Zealand high-country farms, during the summers of 2006/07, 2007/08 (Haycocks Station) and 2008/09 (White Rock Station). The goal was to analyse positional data around the expected birthing period to define patterns of hind movement that are potentially descriptive of parturient behaviour. A Hidden Markov model (HMM) was used to identify the time and site of calving based on four pre-defined stages of behaviour described for red deer hinds. These stages are: Late pregnancy (State 1), birth-site searching (State 2), birthing (State 3) and post-birthing (State 4). Previous research on red deer hinds over calving on an intensive lowland farm showed a close alignment between actual calving behaviour and those behaviours inferred from the algorithms developed in the HMM. Hinds in the high-country study conformed to the four-stage model of positional movement patterns. Inferred birthing dates ranged from 8 Nov to 3 Dec. The length of time hinds spent searching for a birthing site (State 2) was highly variable and averaged 16 (SE 3.9) hours, with an average 4-fold increase in distance between consecutive GPS recordings relative to that occurring earlier in late pregnancy (State 1), while in the birthing stage (State 3) there was a 40% reduction in distance between recordings compared with State 1. Although actual incidences of birthing were not observable in this study, it is concluded that definable behavioural characteristics of red deer hinds over the calving period can be discerned from GPS positional data. A key observation of this study was the between-hind variability of duration of these innate behaviours that indicates behavioural plasticity in the face of intraspecific competition for resources or perception of threat within extensive farm environments.

摘要

在 2006/07、2007/08（海科克斯站）和 2008/09 的夏季，全球定位系统 (GPS) 技术被用于跟踪两个新西兰高地农场在产犊期间广泛养殖的红鹿母鹿的运动（白石站）。目标是分析预期分娩期间的位置数据，以定义可能描述分娩行为的后部运动模式。隐马尔可夫模型 (HMM) 用于根据为马鹿后腿描述的四个预先定义的行为阶段来识别产犊的时间和地点。这些阶段是：妊娠晚期（状态 1）、出生地点搜索（状态 2）、分娩（状态 3）和分娩后（状态 4）。先前关于在密集低地农场产犊的马鹿后腿的研究表明，实际产犊行为与从 HMM 中开发的算法推断出的行为密切相关。高地研究中的海因斯符合位置运动模式的四阶段模型。推断的分娩日期范围为 11 月 8 日至 12 月 3 日。母猪寻找分娩地点（状态 2）的时间长度变化很大，平均为 16 (SE 3.9) 小时，连续 GPS 之间的距离平均增加 4 倍记录相对于妊娠晚期（状态 1）早期发生的记录，而在分娩阶段（状态 3）与状态 1 相比，记录之间的距离减少了 40%。尽管在本研究中无法观察到实际的分娩发生率，得出的结论是，可以从 GPS 位置数据中辨别出马鹿后腿在产犊期间的可定义行为特征。本研究的一个关键观察结果是这些先天行为持续时间的中间变异性，这表明面对种内资源竞争或广泛农场环境中的威胁感知时行为可塑性。