Can citizen science data guide the surveillance of invasive plants? A model-based test with Acacia trees in Portugal
Can citizen science data guide the surveillance of invasive plants? A model-based test with Acacia trees in Portugal
复制标题
DOI：
10.1007/s10530-019-01962-6
复制
发表时间：
2019-06-01
影响因子：
2.9
通讯作者：
Vicente, Joana Raquel
中科院分区：
文献类型：
--
作者：
de Sa, Nuno Cesar;Marchante, Helia;Vicente, Joana Raquel
随着入侵外来植物(IAP)的迅速扩张，准确和及时的分布数据对于成功的管理越来越关键。然而，研究人员/技术人员要获得所有国际行动方案和领土的数据并非易事。在这种情况下，公民科学平台收集的数据可以成为一个有用的工具，补充专业数据。我们假设，结合公民收集的IAP数据和研究人员收集的数据可以提高物种分布模型(SDMS)的准确性，并优化监测工作。为了测试这一点，我们从公民科学平台(Inventoras.pt)和三种广泛存在于葡萄牙的入侵相思物种的研究人员那里收集了数据，并生成了三个不同的数据集：研究人员、公民和研究人员加公民。我们使用集成方法(Biood2 R Package)模拟物种的潜在分布，以测试不同数据集对结果模型精度、物种分布的选定环境驱动因素和预测的空间分布的影响。所有的SDMS都获得了很高的精度，其中用研究人员的数据训练的模型获得了最高的值。然而，用公民数据训练的模型在所有情况下都极大地增加了预测的空间分布。对三个模型的空间投影进行了进一步的比较和排序，以确定每个物种的最高监测优先区域，即模型之间高度一致但缺乏发生数据的区域。这些结果可以用来指导未来的公民和研究人员的监测工作。
With the rapid expansion of invasive alien plants (IAPs), accurate and timely distribution data is increasingly critical to successful management. However, it is not easy for researchers/technicians to obtain data for all IAPs and territories. In this context, data collected by Citizen Science Platforms can be a useful tool, complementing professional data. We hypothesize that combining IAP data collected by citizens and data collected by researchers can improve the accuracy of species distribution models (SDMs) and optimize surveillance efforts. To test this, we gathered data from a Citizen Science Platform (Invasoras.pt) and from researchers on three invasive Acacia species widespread in Portugal and generated three different datasets: researchers, citizens, and researchers plus citizens. We modelled the potential distribution of the species using an ensemble approach (biomod2 R package) to test the effect of the different datasets on the resulting model accuracy, the selected environmental drivers of species distribution and the predicted spatial distribution. All SDMs obtained very high accuracy, with the highest values being obtained in the models trained with researchers' data. Nevertheless, models trained with citizen data vastly increased the predicted spatial distribution in all cases. The spatial projections of the three models were further compared and ranked to identify the areas of highest surveillance priority for each species, i.e., areas with high agreement between the models but where occurrence data is lacking. These results can be used to guide future surveillance efforts both for citizens and researchers.