Preservation of an Automated Feeding System to Enhance Nonhuman Primate Social Management

摘要 在过去的8年里，YNPRC野外工作站成功地维持了大规模的自动化饲养， 香港特区政府已推行一套系统，以加强对猕猴社群的照顾和管理。目前，有8种化合物 配备自动送料机，外壳约37%的SPF恒河猴。这种自动化喂养 系统提供了几个优点，超过标准的垃圾箱喂养的做法，包括减少食物浪费，改善 动物健康的临床监督，超重动物的体重管理，以及一种自动化方法， 进行群体普查。鉴于我们先前的成功，最近的研究工作集中在增加 喂养数据在非人灵长类动物（NHP）社会管理。事实上，在恒河猴社会中， 管理是利用有效的社会健康监测方法，以确定风险群体， 在发生重大战斗和伤害之前，社会不稳定。决定支配地位的相互作用 这些群体中的从属关系是稀疏的;因此，收集足够的行为数据， 明确地发现社会不稳定需要大量时间和人力资源。父母补助金（R24 OD 030035）旨在为NHP建立更有效的数据采集策略 通过开发基于喂养交互网络（FIN）的机器学习（ML）模型， 可以用来补充行为数据，帮助管理者识别有社会不稳定风险的群体。 虽然YNPRC现场站的自动化进料系统已成功运行了许多年， 多年来，我们最近遇到了意想不到的设备挑战。首先，我们最古老的电路板 制造商已停止生产进料装置。目前，两个大院配备了这些 旧馈线（4个单元/化合物）和一个单元的电路板不可用。其次，虚拟机 我们目前的数据服务器的操作系统不再受支持，无法接受升级。因此， 补充应用程序要求支持1）8个新的自动喂料机，以取代8个停产的装置，以及2） 升级后的服务器具有更快的处理器和更大的存储容量，沿着运行新的虚拟机 系统，以保持在YNPRC外地站的8化合物自动化喂养系统。 鉴于自动喂养数据是父母补助金目标的核心， 服务器和自动化饲养单元是至关重要的，这个项目的成功和它的潜力，以改善动物 福祉和管理效率，这反过来又促进了使用NHP进行高质量的科学研究。 此外，保留8种化合物的自动喂养系统将允许必要的灵活性， 选择最合适的社会群体进行研究（2组/年，共6组），考虑育种者的变化 雄性引进时间表、群体规模和稳定性以及家庭结构。使用最合适的社交 基于母基金中概述的标准的小组对于构建强大的FIN驱动的ML模型至关重要， 将对YNPRC和其他圈养NHP殖民地的NHP社会管理实践产生持续影响。

ABSTRACT For the past 8 years, the YNPRC Field Station has successfully maintained a large-scale automated feeding system to enhance the care and management of its rhesus macaque social groups. Currently, 8 compounds are equipped with automated feeders, housing approx. 37% of its SPF rhesus population. This automated feeding system offers several advantages over standard bin feeding practices, including reduced food waste, improved clinical oversight of animal health, weight management of overweight animals, and an automated method to conduct group census. Given our prior success, recent research efforts have focused on increasing the utility of feeding data in nonhuman primate (NHP) social management. Indeed, a common challenge in rhesus social management is the utilization of efficient social health surveillance methodology to identify groups at risk for social instability before the onset of significant fighting and wounding. The interactions that define dominance and affiliative relationships in these groups are sparce; thus, the gathering of sufficient behavioral data to unequivocally detect social instability requires considerable time and staff resources. The parent grant (R24 OD030035) of this supplement application seeks to establish a more efficient data acquisition strategy for NHP social management by developing feeding interaction network (FIN)-based machine learning (ML) models that can be used to supplement behavioral data and help managers identify groups at risk for social instability. Although the automated feeding system at the YNPRC Field Station has been operated successfully for many years, we have recently encountered unanticipated equipment challenges. First, the circuit boards of our oldest feeder units have been discontinued by the manufacturer. Currently, two compounds are equipped with these older feeders (4 units/compound) and one unit’s circuit board is non-operational. Secondly, the virtual machine operating system of our current data server is no longer supported and cannot receive upgrades. Thus, this supplement application requests support for 1) 8 new automated feeders to replace 8 discontinued units and 2) an upgraded server with a faster processor and large storage capacity along with a new virtual machine operating system, to preserve the 8-compound automated feeding system at the YNPRC Field Station. Given that automated feeding data is central to the aims of the parent grant, the purchase of an upgraded server and automated feeding units is critical to the success of this project and its potential to improve animal wellbeing and management efficiency, which in turn, fosters the conduct of high-quality science using NHPs. Additionally, the preservation of an 8-compound automated feeding system will allow necessary flexibility to select the most appropriate social groups to study (2 groups/yr, 6 groups total), considering changes in breeder male introduction schedules, group size and stability, and family structures. Use of the most appropriate social groups based on criteria outlined in the parent grant will be critical for building robust FIN-driven ML models that will exert a sustained impact on NHP social management practices at YNPRC and other captive NHP colonies.