Data from: Ecological niche modeling for a cultivated plant species: a case study on taro (Colocasia esculenta) in Hawai‘i

Under the threat of ongoing and projected climate change, communities in the Pacific Islands face challenges of adapting culture and lifestyle to accommodate a changing landscape. Few models can effectively predict how biocultural livelihoods might be impacted. Here, we examine how environmental and anthropogenic factors influence an ecological niche model (ENM) for the realized niche of cultivated taro (Colocasia esculenta) in Hawai‘i. We created and tuned two sets of ENMs: one using only environmental variables, and one using both environmental and cultural characteristics of Hawa‘i. These models were projected under two different Intergovernmental Panel on Climate Change (IPCC) Representative Concentration Pathways (RCPs) for 2070. Models were selected and evaluated using average omission rate and area under the receiver operating characteristic curve (AUC). We compared optimal model predictions by comparing the percentage of taro plots predicted present and measured ENM overlap using Schoener’s D-statistic. The model including only environmental variables consisted of 19 Worldclim bioclimatic variables, in addition to slope, altitude, distance to perennial streams, soil evaporation, and soil moisture. The optimal model with environmental variables plus anthropogenic features also included a road density variable (which we assumed as a proxy for urbanization) and a variable indicating agricultural lands of importance to the State of Hawai‘i. The model including anthropogenic features performed better than the environment-only model based on omission rate, AUC, and review of spatial projections. The two models also differed in spatial projections for taro under anticipated future climate change. Our results demonstrate how ENMs including anthropogenic features can predict which areas might be best suited to plant cultivated species in the future, and how these areas could change under various climate projections. These predictions might inform biocultural conservation priorities and initiatives. In addition, we discuss the incongruences that arise when traditional ENM theory is applied to species whose distribution has been significantly impacted by human intervention, particularly at a local scale relevant to biocultural conservation initiatives.

在持续发生且可预见的气候变化威胁下，太平洋岛屿社区面临着调整文化与生活方式以适配不断变化的生存环境的挑战。目前鲜有模型能够有效预测生物文化生计可能受到的影响。本研究以夏威夷地区栽培芋（Colocasia esculenta）的实际生态位（realized niche）为研究对象，探讨环境与人为因素对生态位模型（ecological niche model, ENM）的影响。我们构建并调优了两组生态位模型：一组仅使用环境变量，另一组同时结合夏威夷的环境与文化特征。上述模型基于2070年的两种政府间气候变化专门委员会（Intergovernmental Panel on Climate Change, IPCC）典型浓度路径（Representative Concentration Pathways, RCPs）进行未来气候情景投影。模型通过平均遗漏率与受试者工作特征曲线下面积（area under the receiver operating characteristic curve, AUC）完成筛选与评估。我们通过对比预测适宜种植芋类的地块占比，以及采用舍纳D统计量（Schoener’s D-statistic）计算的生态位模型重叠度，来比较最优模型的预测结果。仅使用环境变量的模型包含19个Worldclim生物气候变量，此外还纳入了坡度、海拔、距常年溪流的距离、土壤蒸发量与土壤含水量。加入人为特征的最优模型还额外包含了道路密度变量（我们将其作为城市化程度的替代指标）以及一项代表夏威夷州重要农业用地的变量。基于平均遗漏率、AUC值以及空间投影结果的综合评估，纳入人为特征的模型表现优于仅使用环境变量的模型。在未来预期的气候变化情景下，两组模型的空间投影结果也存在显著差异。本研究结果证实，纳入人为特征的生态位模型能够预测未来最适宜栽培作物的种植区域，以及在不同气候投影情景下这些区域的变化趋势。此类预测可为生物文化保护优先级制定与相关举措提供决策参考。此外，本研究还讨论了当传统生态位模型理论应用于分布受人类活动显著影响的物种时所出现的不一致性，尤其是在与生物文化保护举措相关的局域尺度上。