生物多样性 ›› 2016, Vol. 24 ›› Issue (10): 1117-1128.DOI: 10.17520/biods.2016164

• • 上一篇    下一篇

基于生态位模型预测野生油茶的潜在分布

崔相艳1,2, 王文娟1,2, 杨小强1,2, 李述2, 秦声远1,2, 戎俊1,2,*()   

  1. 1 南昌大学生命科学研究院流域生态学研究所, 南昌大学生命科学学院, 南昌 330031
    2 南昌大学鄱阳湖环境与资源利用教育部重点实验室, 南昌 330031
  • 收稿日期:2016-06-21 接受日期:2016-08-25 出版日期:2016-10-20 发布日期:2016-11-10
  • 通讯作者: 戎俊
  • 基金资助:
    国家自然科学基金(31460072)和江西省“赣鄱英才555工程”项目

Potential distribution of wild Camellia oleifera based on ecological niche modeling

Xiangyan Cui1,2, Wenjuan Wang1,2, Xiaoqiang Yang1,2, Shu Li2, Shengyuan Qin1,2, Jun Rong1,2,*()   

  1. 1 Center for Watershed Ecology, Institute of Life Science, Nanchang University and School of Life Sciences, Nanchang University, Nanchang 330031
    2 Key Laboratory of Poyang Lake Environment and Resource Utilization, Ministry of Education, Nanchang University, Nanchang 330031
  • Received:2016-06-21 Accepted:2016-08-25 Online:2016-10-20 Published:2016-11-10
  • Contact: Rong Jun

摘要:

油茶(Camellia oleifera)是我国第一大木本油料作物, 野生油茶是油茶育种的宝贵遗传资源。本研究从中国数字植物标本馆(CVH, http://www.cvh.org.cn/)获得可靠的野生油茶分布点数据, 结合气象和土壤数据, 分别应用最大熵(MaxEnt)模型和规则集遗传算法(GARP)模型构建了野生油茶的生态位模型, 预测了野生油茶的潜在分布区, 并分析了影响野生油茶分布的主要环境变量。根据生态位模型预测的分布概率值, 对野生油茶的潜在分布区划分适生等级, 并与主要油茶产地的实际分布数据进行比较, 以验证适生等级划分的可靠性。结果表明, 两种模型的预测结果均能较好地反映油茶的分布情况。GARP模型预测的潜在分布区更广, 而MaxEnt模型的预测结果更精确。两种模型的预测结果均显示, 野生油茶的潜在分布区大部分位于中国, 但在中南半岛也有部分分布。MaxEnt模型预测的野生油茶在中国的潜在分布区与我国亚热带常绿阔叶林的分布区基本吻合, 高适生区主要可以分为3大区域: (1)东北-西南走向的武夷山脉及附近的群山区域; (2)东西走向的南岭山脉及附近的群山区域; (3)东北-西南走向的武陵山脉及附近的群山区域。MaxEnt模型分析显示, 影响野生油茶分布的主要环境变量是昼夜温差月均值、最干季降水量与最暖季降水量。油茶生长面积较大的地区绝大部分都位于MaxEnt模型预测的中、高适生区, 说明适生等级的划分较可靠。实地考察显示, 生态位模型的预测结果对于寻找野生油茶资源具有较高的参考价值。此外, 本研究也充分显示, 利用中国数字植物标本馆的植物分布数据, 结合相应的环境数据构建生态位模型, 有助于了解作物野生近缘种的地理分布。

关键词: 野生油茶, 地理分布, 生态位模型, 降水量, 温度, 最大熵模型, 规则集遗传算法模型

Abstract:

Camellia oleifera is the dominant woody oil crop in China, and wild C. oleifera is a valuable genetic resource for C. oleifera breeding. Using distribution data of wild C. oleifera from the Chinese Virtual Herbarium (CVH, http://www.cvh.org.cn/), together with climate and soil data, ecological niche models were constructed with MaxEnt and genetic algorithm for rule-set prediction (GARP) models to predict the potential distribution of wild C. oleifera, and the major environmental factors affecting the distribution of wild C. oleifera were analyzed. Based on the presence probability of wild C. oleifera predicted by the models, the distribution regions of wild C. oleifera were divided into different suitable growing categories, which were then compared with actual distribution data of major C. oleifera production fields to evaluate reliability. Results indicated that the predictions of both MaxEnt and GARP models represented the distributions of C. oleifera well. The potential distribution range predicted by the GARP model was wider, while that predicted by the MaxEnt model was more accurate. Predictions of both the MaxEnt and GARP models showed that the potential distribution regions of wild C. oleifera were located mainly in China and partly in the Indo-China Peninsula. According to predictions of the MaxEnt model, the potential distribution regions of wild C. oleifera in China were matched with the distribution regions of subtropical evergreen broad-leaved forests, and the highly suitable growing regions could be divided into three large regions: (1) northeastern-southwestern trending Wuyi Mountain and adjacent mountainous regions; (2) eastern-western trending Nanling Mountain and adjacent mountainous regions; (3) northeastern-southwestern trending Wuling Mountain and adjacent mountainous regions. The analysis of the MaxEnt model showed that the major environmental factors affecting the distribution of wild C. oleifera were mean monthly diurnal temperature range, precipitation during the driest quarter, and precipitation during the warmest quarter. The vast majority of the regions with large growing areas of C. oleifera were located in the medium to highly suitable growing regions predicted by the MaxEnt model, suggesting that the division of suitable growing regions was reliable. The field investigations showed that the model predictions had high reference values for finding wild C. oleifera resources. Additionally, the study shows that using the plant distribution data from CVH and related environmental data to construct an ecological niche model can help to understand the geographic distribution of crop wild relatives.

Key words: wild Camellia oleifera, geographic distribution, ecological niche model, precipitation, temperature, MaxEnt model, genetic algorithm for rule-set prediction (GARP) model