交通方式對路死動物屍體被偵測率的影響

How modes of transportation influence road carcass detection probability

讓民眾依調查樣線現況，使用不同交通方式進行道路沿線路殺調查，有助於提升參與意願以獲得足夠樣本數。然而交通方式的不同，對於路殺動物調查的偵測率卻有很大的影響。為此本研究針對步行、腳踏車、機車與汽車4 種交通方式對路死動物屍體調查偵測率進行試驗，同時評估行進速度、調查者經驗、道路型態、物種類群、路死動物屍體的面積大小對偵測率的影響。研究結果顯示步行可以調查到最多的屍體 74%，其次是腳踏車42% 與機車32%，而汽車僅有12%；經模式推估其中最主要的影響因子為交通方式與調查行進速度，其次是調查者經驗與道路類型。而影響每一個屍體被發現與否最關鍵的因子為動物屍體的面積大小與調查行進速度。
調查者維持低速行走可以偵測到最多的物種與數量，但考量公民科學調查的可行性與提高參與度，本研究建議可使用機車調查並以6km/hr 以下的時速行進，且完成整條3km 樣線的調查時間不低於30 min。培養具有經驗的志工以及維持固定樣區固定志工，也有助於提升調查偵測率，是研究團隊必需重視的工作之一。本研究結果將有助於校正臺灣野生動物路殺總量的估算，以及評估不同動物類群受路殺威脅情形。

Volunteers participated in a roadkill survey with the liberty of choosing different modes of transportation, including by foot, bike, scooter, or car, depending on the local line-transect status. This method helps increase participation rate and obtain a larger sample size, although the roadkill survey detection rate would be affected by the chosen mode of transportation. In this study, we tested the transportation influence on road carcass detection probability, and evaluated factors of roadkill survey detection probability, including the survey speed of observers, experience of observers, road types, species taxon, and area size of roadkill body. Surveys by foot showed the highest effectiveness (74%), bicycle (42%), scooter (32%) was second highest, and car (12%) resulted in the lowest detection rate. The modeling results show that transportation mode and observer moving speed were the primary factors influencing the carcass survey detection rate, and the experience of observers and road types were the minor factors. We also discovered that the area size of roadkill body and observer moving speed can significantly affect carcass detection probability.
Slow-walking observers could search for the most species and roadkill bodies. Considering the trade-off between feasibility and participation of citizen science projects, we suggest volunteers/observers search carcasses by scooter, keeping the moving speed below 6km/hr, and allowing more than 30 minutes of search time on the 3km line-transect. The research team needs to train experienced volunteers and maintain the fixed study site with the same observers to improve detection probability. This study result will be useful to correct sampling bias and estimate the abundance of animals killed on the roads in Taiwan. We could also assess the roadkill risk of different taxonomic animals.