Data from: Mapping foodscapes and sagebrush morphotypes with unmanned aerial systems for multiple herbivores

Context: The amount and composition of phytochemicals in forage plants influences habitat quality for wild herbivores. However, evaluating forage quality at fine resolutions across broad spatial extents (i.e., foodscapes) is challenging. Unmanned aerial systems (UAS) provide an avenue for bridging this gap in spatial scale. Objectives: We evaluated the potential for UAS technology to accurately predict nutritional quality of sagebrush (Artemisia spp.) across landscapes. We mapped seasonal forage quality across two sites in Idaho, USA, with different mixtures of species but similar structural morphotypes of sagebrush. Methods: We classified the sagebrush at both study sites using structural features of shrubs with object-based image analysis and machine learning and linked this classification to field measurements of phytochemicals to interpolate a foodscape for each phytochemical with regression kriging. We compared fine-scale landscape patterns of phytochemicals between sites and seasons. Results: Classification accuracy for morphotypes was high at both study sites (81–87%). Forage quality was highly variable both within and among sagebrush morphotypes. Coumarins were the most accurately mapped (r2=0.57–0.81), whereas monoterpenes were the most variable and least explained. Patches with higher crude protein were larger and more connected in summer than in winter. Conclusions: UAS allowed for a rapid collection of imagery for mapping foodscapes based on the phytochemical composition of sagebrush at fine scales but relatively broad extents. However, results suggest that a more advanced sensor (e.g., hyperspectral camera) is needed to map mixed species of sagebrush or to directly measure forage quality. Data Usage Notes: Spatial Reference: NAD83 UTM Zone 11N [Camas]/12N [Cedar Gulch] Patch Type Classifications: 25-cm resolution, classes are 3=on mound, 4=off mound, 5=dwarf. On-mound refers to mima mounds with deeper soils that pygmy rabbits use to dig their burrows and are dominated by big sagebrush (Artemisia tridentata), while off-mound refers to patches dominated by big sagebrush but not on mima mounds, while dwarf patches are dominated by short-statured sagebrush species (e.g., black sagebrush [A. nova], low sagebrush [A. arbuscula]). Maps of Phytochemistry: 25-cm resolution and were generated with regression kriging using the patch type layer and point values from leaf chemistry. Phytochemicals: Crude Protein, Coumarins, Total Monoterpenes, Chemical Diversity of Monoterpenes and two individual monoterpenes (1,8-cineole and camphor). If there was no spatial autocorrelation present in the semivariogram, then maps were not generated for that phytochemical.

背景：牧草植物中植物化学物质的含量与组成影响着野生动物栖息地的质量。然而，在广阔的空间范围内（即食物景观）对牧草质量进行精细分辨率的评估是一项挑战。无人机系统（UAS）为弥合这一空间尺度上的差距提供了途径。
目标：我们评估了无人机技术准确预测灌木状草本植物（Artemisia spp.）在景观中营养质量潜力的可能性。我们利用无人机技术在美国爱达荷州两个不同物种混合但结构形态相似的灌木状草本植物景观中，对季节性牧草质量进行了测绘。
方法：我们使用基于对象的光学图像分析和机器学习对两个研究站点中的灌木状草本植物进行了分类，并将此分类与植物化学物质的现场测量结果相连接，通过回归克里金法对每种植物化学物质进行食物景观的插值。我们比较了两个站点和不同季节中植物化学物质的精细景观模式。
结果：在两个研究站点中，形态类型的分类准确率均较高（81%–87%）。牧草质量在灌木状草本植物形态类型内部和之间均表现出高度可变性。香豆素是映射最准确的（r2=0.57–0.81），而单萜类化合物则是变化最大且解释最少的。夏季较冬季，粗蛋白含量较高的斑块更大且连通性更强。
结论：无人机允许在精细尺度但相对广泛的空间范围内，快速收集基于灌木状草本植物植物化学成分的食物景观测绘图像。然而，结果表明，需要更先进的传感器（例如，高光谱相机）来测绘灌木状草本植物的混合物种或直接测量牧草质量。
数据使用说明：空间参考：NAD83 UTM 区 11N [卡马斯]/12N [雪松沟]；斑块类型分类：25厘米分辨率，分类包括3=在土丘上，4=不在土丘上，5=矮生。在土丘上指的是具有较深土壤的米玛土丘，这些土丘是侏儒兔挖洞的地方，主要由大灌木状草本植物（Artemisia tridentata）组成，而不在土丘上指的是主要由大灌木状草本植物组成但不在米玛土丘上的斑块，而矮生斑块则主要由矮小灌木状草本植物物种（例如，黑灌木状草本植物 [A. nova]，矮灌木状草本植物 [A. arbuscula]）组成。植物化学物质地图：25厘米分辨率，通过使用斑块类型图层和叶化学的点值进行回归克里金法生成。植物化学物质：粗蛋白、香豆素、总单萜类化合物、单萜类化合物的化学多样性以及两种单萜类化合物（1,8-桉树脑和樟脑）。如果半变异函数中不存在空间自相关性，则不会为该植物化学物质生成地图。