Data and code from Rodkey et al. (2024): "Sod farms drive habitat selection of a migratory grassland shorebird during a critical stopover period"

The data and code in this repository was used to fit the final models for the submitted manuscript "Sod farms drive habitat selection of a migratory grassland shorebird" by Rodkey and coauthors. Abstract: Migratory shorebirds are one of the fastest declining groups of North American avifauna. Yet, relatively little is known about how these species select habitat during migration. We explored the habitat selection of Buff-breasted Sandpipers (Calidris subruficollis) during spring and fall migration through the Texas Coastal Plain, a major stopover region for this species. Using tracking data from 118 birds compiled over 4 years, we found Buff-breasted Sandpipers selected intensively managed crops such as sod and short-stature crop fields, but generally avoided rangeland and areas near trees and shrubs. This work supports prior studies that also indicate the importance of short-stature vegetation for this species. Use of sod and corn varied by season, with birds preferring sod in spring, and avoiding corn when it is tall, but selecting for corn in fall after harvest. This dependence on cropland in the Texas Coastal Plain is contrary to habitat use observed in other parts of their non-breeding range, where rangelands are used extensively. The species' almost complete reliance on a highly specialized crop, sod, at this critical stopover site raises concerns about potential exposure to contaminants as well as questions about whether current management practices are providing suitable conditions for migratory grassland birds. Data and code for model fitting bbsa_sodfarm_selection.zipFollowing files are used to fit final models for the Texas study and the range-wide study. script_tx is the script used to fit models on Texas dataset. script_tx.R script_rw is the script used to fit models on range-wide dataset. script_rw.R data_tx is the datafile for containing model input data for the Texas models. data_tx.csv data_rw is the datafile containing model input data for the rangewide models. data_rw.csv Columns in the datafiles for both Texas and rangewide models (data_tx and data_rw) are organized as follows: ID - unique identifier for tracked birds year - year in which bird was tracked season - season in which bird was tracked (spring or fall) fromTexas - (rangewide dataset only) whether or not the tracked bird was captured in the study region, the Texas Coastal Plain (T = bird was caught within study region; F = bird was caught outside of study region) night- (Texas dataset only) whether or not the location fell within daytime or nightime hours (y = nightime, n = daytime) droad - distance of location in meters from nearest road dwoody - distance of location in meters from nearest woody cover corn_prop250 - proportion of corn/sorghum landcover type within a 250m buffer around location cotton_prop500 - proportion of cotton/soybean landcover type within a 500m buffer around location herb_prop250 - proportion of herbaceous (i.e., grassy) landcover type within a 250m buffer around location rice_prop1000 - proportion of rice landcover type within a 1000m buffer around location sod_prop250 - proportion of sod landcover type within a 250m buffer around location presence - identifying used versus available locations (1 = used, as in, an actual GPS location of tracked bird; 0 = available, as in, a randomly generated location within the bird's area of use) Tracking data bbsa_txStudy_movebank_20240731.csv Tracking data used for Texas study. Lotek PinPoint GPS data were imported and decoded via Movebank and relevant columns were selected. Columns are organized as follows: timestamp - time of GPS location in UTC (format = YYYY-MM-DD HH:MM:SS.SSS) location.long - longitude of GPS location in decimal degrees, projected in the WGS 84 datum location.lat - latitude of GPS location in decimal degrees, projected in the WGS 84 datum lotek.crc.status.text - value of Lotek's cyclical redundancy check (OK = good data, OK(corrected) = message was corrected with Lotek's error correction algorithm, Fail = data has been corrupted on transmission, and therefore cannot be trusted) sensor.type - type of sensor, all rows are "gps" tag.local.identifier - Argos ID of the attached Lotek PinPoint GPS transmitter individual.local.identifier - unique individual identifier, in this case the USGS-issued band number attached to each tracked bird individual.taxon.canonical.name - species taxonomical name, all rows are "Calidris subruficollis" sex - sex of tracked bird (M = male, F = female) age - age of tracke bird (AHY = after-hatch-year, HY = hatch-year) Sod farm layer sodLayer_tgcp.tifRaster layer representing sod farms in the Western Gulf Coastal Plain ecoregion of Texas. Resolution of 30m with values from 0 - 1, with 1 representing sod farms. Sod farms were identified via visual inspection, cross-referencing the U.S. Department of Agriculture's Cropland Data Layer with Google Earth high resolution imagery from 2022. This was a modification of preliminary data on sod farm distribution in the region compiled by B. Ortego, S. Gifford and J. Lyons.本仓库中的数据与代码，用于拟合Rodkey及其合著者提交的题为《草皮农场驱动迁徙草原滨鸟的生境选择》的手稿的最终模型。 摘要：迁徙滨鸟是北美鸟类中衰退速度最快的类群之一。然而，目前对这些物种在迁徙过程中如何选择生境的了解仍相对较少。本研究针对通过德克萨斯沿海平原（该物种的主要中途停歇地）进行春季和秋季迁徙的黄胸滨鹬（*Calidris subruficollis*，Buff-breasted Sandpiper）的生境选择展开探索。我们利用4年间收集的118只个体的跟踪数据，发现黄胸滨鹬会优先选择集约管理的作物生境，例如草皮田和矮秆农田，但通常会避开牧场以及邻近树木和灌丛的区域。本研究结果与此前的研究结论一致，均表明矮秆植被对该物种的重要性。对草皮田和玉米田的利用存在季节差异：春季鸟类偏好草皮田，而当玉米长高时则会避开，但在秋季收获后会选择玉米田。这种对德克萨斯沿海平原农田的依赖，与该物种在非繁殖范围其他区域的生境利用模式相悖——在那些区域，牧场被广泛使用。该物种在这一关键中途停歇地几乎完全依赖高度特化的作物生境（草皮田），这引发了人们对其可能接触污染物的担忧，同时也提出了疑问：当前的管理措施是否能为迁徙草原鸟类提供适宜的生存条件。 ### 模型拟合所需数据与代码 `bbsa_sodfarm_selection.zip` 以下文件用于拟合德克萨斯研究与全分布范围研究的最终模型： - `script_tx.R`：用于拟合德克萨斯数据集模型的脚本 - `script_rw.R`：用于拟合全分布范围数据集模型的脚本 - `data_tx.csv`：包含德克萨斯模型所需模型输入数据的数据文件 - `data_rw.csv`：包含全分布范围模型所需模型输入数据的数据文件 德克萨斯与全分布范围模型的数据文件（`data_tx`与`data_rw`）的列组织方式如下： 1. `ID`：跟踪个体的唯一标识符 2. `year`：个体被跟踪的年份 3. `season`：个体被跟踪的季节（春季或秋季） 4. `fromTexas`：（仅全分布范围数据集）跟踪个体是否在研究区域——德克萨斯沿海平原——内被捕获（T = 在研究区域内捕获；F = 在研究区域外捕获） 5. `night`：（仅德克萨斯数据集）该点位是否处于夜间时段（y = 夜间，n = 日间） 6. `droad`：点位到最近道路的距离（单位：米） 7. `dwoody`：点位到最近木本植被覆盖区的距离（单位：米） 8. `corn_prop250`：点位周边250米缓冲区范围内玉米/高粱土地覆盖类型的占比 9. `cotton_prop500`：点位周边500米缓冲区范围内棉花/大豆土地覆盖类型的占比 10. `herb_prop250`：点位周边250米缓冲区范围内草本（即草本丛生）土地覆盖类型的占比 11. `rice_prop1000`：点位周边1000米缓冲区范围内水稻土地覆盖类型的占比 12. `sod_prop250`：点位周边250米缓冲区范围内草皮田土地覆盖类型的占比 13. `presence`：区分使用点位与可用点位（1 = 使用点位，即跟踪个体的实际GPS（Global Positioning System）定位；0 = 可用点位，即个体活动范围内随机生成的点位） ### 跟踪数据 `bbsa_txStudy_movebank_20240731.csv` 本文件为德克萨斯研究所用的跟踪数据。Lotek PinPoint GPS（Global Positioning System）数据通过Movebank平台导入并解码，并筛选出相关列。列组织方式如下： 1. `timestamp`：GPS定位的时间（UTC格式，格式为YYYY-MM-DD HH:MM:SS.SSS） 2. `location.long`：GPS定位的经度（十进制度格式，投影采用WGS 84（World Geodetic System 1984）基准面） 3. `location.lat`：GPS定位的纬度（十进制度格式，投影采用WGS 84（World Geodetic System 1984）基准面） 4. `lotek.crc.status.text`：Lotek的循环冗余校验（CRC，Cyclical Redundancy Check）状态值（OK = 数据完好；OK(corrected) = 消息通过Lotek的错误校正算法完成修正；Fail = 数据在传输过程中损坏，不可信） 5. `sensor.type`：传感器类型，所有行均为“gps” 6. `tag.local.identifier`：所附Lotek PinPoint GPS发射器的Argos ID 7. `individual.local.identifier`：个体唯一标识符，本研究中为美国地质调查局（USGS, United States Geological Survey）为每只跟踪个体配发的环志编号 8. `individual.taxon.canonical.name`：物种分类学名称，所有行均为“Calidris subruficollis” 9. `sex`：跟踪个体的性别（M = 雄性，F = 雌性） 10. `age`：跟踪个体的年龄（AHY = 成鸟（After Hatch Year）；HY = 幼鸟（Hatch Year）） ### 草皮农场图层 `sodLayer_tgcp.tif` 该栅格图层代表德克萨斯州西部墨西哥湾沿海平原生态区内的草皮农场分布。分辨率为30米，取值范围为0至1，其中1代表草皮田。草皮农场通过目视解译识别，并结合美国农业部（USDA, United States Department of Agriculture）的作物数据层与2022年谷歌地球高分辨率影像进行交叉验证。本数据集是对B. Ortego、S. Gifford与J. Lyons编制的该区域草皮农场分布初步数据的修改版本。