Mapping nectar-rich pollinator floral resources using airborne multispectral imagery.
Mapping nectar-rich pollinator floral resources using airborne multispectral imagery.
复制标题
DOI：
10.1016/j.jenvman.2022.114942
复制
发表时间：
2022-04
影响因子：
8.7
通讯作者：
S. Barnsley;A. Lovett;L. Dicks
中科院分区：
文献类型：
--
作者：
S. Barnsley;A. Lovett;L. Dicks
野生传粉者的数量是已知的正相关的景观水平上的花丰富的栖息地的数量。增加花卉资源在花蜜相对贫乏的农业系统中可能特别有益，对资源的时间和空间可用性有一个基线了解可以实现有针对性的生境管理。极高分辨率遥感有可能促进精细尺度的栖息地内传粉者觅食资源的精确测绘，从而允许识别和解决空间和时间差距，改善传粉者数量的预测，并使传粉者保护措施的远程监测成为可能。我们表明，3厘米和7厘米的空间分辨率的多光谱航空图像可以用来分类五个花蜜丰富的开花植物物种（李刺，山楂monogyna，悬钩子fruticosus，Silene dioicaandCentaureanigra）使用最大似然分类算法。2019年，我们分别获取了3月、5月和7月的3厘米和7厘米图像。在3 cm和7 cm分辨率下，每个月的总体准确度均超过90（范围92.32%-98.72%），支持先前的研究，即更高的空间分辨率不一定会导致更高的准确度，随着像素可变性的增加。剩下的挑战包括确定哪些共同在可见光范围内具有相似颜色的开花物种可以在分类中彼此区分，并且从分类中量化花单元密度，使得花蜜糖供应可以被计算了尽管如此，我们提供了一个原型的方法来映射传粉者觅食资源在农业方面，这可以扩展到其他花蜜丰富的物种。为开发遥感管道奠定了基础，该管道可以提供关于全年不同时间点富含花蜜的开花植物物种的可用性的宝贵数据。
Wild pollinator numbers are known to be positively associated with amounts of flower-rich habitat at landscape level. Increasing floral resources can be particularly beneficial in relatively nectar-poor agricultural systems and having a baseline understanding of the temporal and spatial availability of resources can allow targeted habitat management. Very high-resolution remote sensing has potential to facilitate accurate mapping of fine-scale, within-habitat pollinator foraging resources, thereby allowing spatial and temporal gaps to be identified and addressed, improving predictions of pollinator numbers, and enabling remote monitoring of pollinator conservation measures.Concentrating on hedgerow and flower-rich field margins in a UK agricultural landscape, we showed that multispectral airborne imagery with 3 cm and 7 cm spatial resolutions can be used to classify five nectar-rich flowering plant species (Prunus spinosa, Crataegus monogyna, Rubus fruticosus, Silene dioicaandCentaurea nigra) using a maximum likelihood classification algorithm. In 2019, we separately acquired 3 cm and 7 cm imagery for the months of March, May and July, respectively. Overall accuracies were above 90% for each month at both 3 cm and 7 cm resolutions (range 92.32%–98.72%), supporting previous research that suggests higher spatial resolutions do not necessarily lead to higher accuracies, as pixel variability is increased.Remaining challenges include determining which co-flowering species of similar colours in the visible range can be distinguished from one another within classifications and quantifying floral unit density from classifications so that the nectar sugar supply can be calculated. Nonetheless, we provided a prototype approach for mapping pollinator foraging resources in an agricultural context, which can be extended to other nectar-rich species. The foundation is set for developing a remote sensing pipeline that can provide valuable data on the availability of nectar-rich flowering plant species at different time-points throughout the year.