摘要
对瓦屋山宝兴树蛙3个不同海拔分布(A:山下;B:山腰;C:山上)的异质种群鸣叫的时频参数进行分析,发现三地鸣叫各参数互相重叠,无明显界线.常规统计分析不能使三地的鸣叫聚类为各自的组.使用时延神经网络对三地宝兴树蛙鸣叫的时频参数进行有指导聚类,发现三种群鸣叫有地区性差异,并表现出由个体交流引起的山腰种群群体特征的不显著.推测差异可能是因为长期地理隔离所导致.基于神经网络的分析结果,对各种群间的个体交流情况进行了初步估计.
The time-domain and frequency-domain coefficients of three metapopulations （ low - A, middle - B, high - C） of Rhacophorus dugritei were analyzed on Mt. Wawu, Hongya, Sichuan. All the call characters of the three metapopulations were overlapped and they could not be classified with normal statistic method. Then, the time delayed neural network was used. When the three metapopulations＇ vocal parameters were clustered with time delayed neural network, the vocalization differences of the populations were found. The differences were possibly due to the long-time geographical isolation. It also showed that the characters of populations at the middle elevation were not distinct from those at low and high elevations possibly because of the individual migration from the other two places. Based on the result of the neural network, the individual migration among different populations was inferred.
出处
《应用与环境生物学报》
CAS
CSCD
北大核心
2006年第5期669-673,共5页
Chinese Journal of Applied and Environmental Biology
基金
中国科学院知识创新重要方向项目(No.KSCX2-1-03)资助~~
关键词
时延神经网络
宝兴树蛙
异质种群
鸣叫
聚类
time delayed neural network
metapopulation
Rhacophorus dugritei
call
clustering
