Data from: Mixed effects of drought on species-level traits and plant composition in the United States mixed-grass prairie

Komatsu, Porensky, Reinhart, Wilcox, and Koerner conducted a randomized complete block design with five drought treatments replicated three times per block within grazing treatment paddocks (40.4×30.5 m), and with three blocks (80.8×61.0 m) at one site in Montana and one site in Wyoming. The three grazing treatments included a light, moderate, and heavy grazing regime. The drought treatments included five levels: 0% (control), 25%, 50%, 75%, and 99% rainfall reduction to mimic a range of different realistic drought scenarios, achieved with 3×4 m rain-out shelters erected over 2×2 m plots (Figure 1c). Within each grazing paddock, there were two control treatments (0% rainout). Rainfall reduction treatments were applied during the growing season from May to October in 2019 and 2020, except during a short period when cattle grazed in each paddock in July in WY and August in MT. Then from 2021-2023 no drought treatments were applied to any plots (recovery). Prior to the experimental setup, all cattle grazing treatment paddocks were grazed using conventional practices of moderate summer grazing. Three cattle grazing treatments were randomly assigned within each block. Cattle grazing treatments consisted of destock, stable, and heavy cattle grazing, where forage utilization was manipulated each year of the experiment at levels consistent with three common regional livestock drought management strategies. In destock cattle grazing paddocks, forage utilization was 50% during the drought in years 1 and 2 (2019, 2020, respectively) and 30% during recovery years 3 and 4 (2021, 2022, respectively). The stable cattle grazing treatment consisted of forage utilization at 50% during drought and recovery (2019-2022). The heavy cattle grazing treatment consisted of forage utilization at 50% during the first drought year (year 1, 2019), 70% during the second drought year (year 2, 2020) and the first year of recovery (year 3, 2021), and 50% during the second year of recovery (year 4, 2022). To manipulate cattle grazing, beef cattle (Bos taurus) herds were allowed into paddocks in July in WY and August in MT and grazed freely until appropriate forage utilization was met, measured using a visual obstruction pole (Robel et al., 1970).
Data were collected by Bloodworth, Komatsu, Porensky, Reinhart, Vaarre-Lamoureux, Wilcox, and Koerner. We measured foliar plant community composition using the pin-drop method in a designated 1 m2 subplot within each 2×2 m plot (Frost et al. 2023) during peak growing season in each year of the experiment (2019-2023). We also collected plant functional traits on each site’s top 90% of plant species based on the plant species composition (percent cover) data measured, as described above, in years prior to trait measurements (2019-2021). Functional trait data were collected once in 2022 on nine individuals of each species found in ambient rainfall conditions at each site. C3 graminoid and forb trait measurements were collected in May and C4 graminoid and shrub trait measurements were collected in late June. Traits included plant height, leaf thickness, leaf dry matter content (LDMC), leaf area, and specific leaf area (SLA). Plant height was measured to the tallest stretched vegetative point. Leaf thickness of the second fully expanded leaf from the top was measured using a micrometer caliper (0.25-0.01 mm United Scientific PMSET04 Precision Measuring Micrometer Caliper). Using the same leaf, LDMC was measured as the dry weight of the leaf (dried at 60°C for at least 1 week) divided by the wet weight of the leaf. Again, using the same leaf, we measured leaf area using ImageJ (Rasband 2021). SLA was calculated by dividing the leaf area by the dry weight of the leaf.
Using the same collection techniques we collected plant functional traits (plant height, leaf thickness, LDMC, leaf area, and SLA) on three common grasses and two common forbs at each site from areas within the experimental blocks that received ambient rainfall. The focal plant species in WY were Bouteloua gracilis, Koeleria macrantha, Pascopyrum smithii, Logfia arvensis, and Vicia americana. The focal plant species in MT were Bromus arvensis, Hesperostipa comata, Koeleria macrantha, Sphaeralcea coccinea, andTragopogon dubius. For each plant species, we collected nine individuals from across blocks during peak growing season (late June or early July) of 2019-2022. Leaf thickness was not collected in 2020. L. arvensis leaves were too small for the scale used and therefore measurements that used leaf weight (LDMC and SLA) are not reported for L. arvensis.
Yearly precipitation levels in this study were obtained from local precipitation gauges (within 2 miles of the site) for the WY site and from a NOAA weather station in Miles City, MT (within 15 miles of the site) for the MT site. Precipitation was calculated as the total precipitation from October of the previous year through the month when measurements were collected. In 2019, 2021, and 2022, measurements were collected in June. In 2020 due to delays because of the COVID-19 Pandemic measurements were collected in July.

Within Excel files:
sites are listed as FK (Fort Keogh Livestock Range and Research Laboratory in Miles City, Montana) and TB (Thunder Basin Ecoregion in Bill, Wyoming).
Block, paddock, and plot: randomized complete block design.
Slope: scale 1-6 with 1 being the flattest and 6 being the steepest slope.
Rainfall reduction & drought: 0, 25, 50, 75, or 99% rainfall reduction on plot
grazing category: L refers to low grazing utilization (30%), M refers to moderate grazing utilization (50%), H refers to heavy grazing utilization (70%). 5 letters denote 5 years of grazing treatment
grazing treatment: general treatment name (stable, heavy, restock)
livestock_util_2019, 2020, 2021: 30%, 50%, or 70% utilization
aerial_basal: Aerial or Basal relatives cover measurements (aerial refers to amount of vegetation at tallest height, basal refers to amount of vegetation near the surface of the ground).
Native_introduced: N=native, I=introduced
Annual_perennial: P=perennial, A=annual
Merged_Traits_FK_2022 and Merged_Traits_TB_2022: paddock is LG (equivalent to MLLMM grazing category), MG (equivalent to MMMMM), and HG (equivalent to HHMMM)

Komatsu、Porensky、Reinhart、Wilcox与Koerner在蒙大拿州与怀俄明州各一处试验站点，开展了随机完全区组设计（randomized complete block design）试验。放牧处理样地规格为40.4×30.5 m，每个区组内设置5种干旱处理并重复3次；共设3个区组，总面积为80.8×61.0 m。试验设置轻度、中度、重度3类放牧处理，以及5个降雨缩减梯度的干旱处理：0%（对照组）、25%、50%、75%与99%，以模拟一系列真实干旱情景，该处理通过搭建3×4 m的遮雨棚（rain-out shelters）覆盖2×2 m样方实现（图1c）。每个放牧样地内设置2个降雨缩减对照组（0%遮雨）。2019与2020年的生长季（5月至10月）实施降雨缩减处理，但怀俄明州站点于7月、蒙大拿州站点于8月进行肉牛放牧的短暂时段除外。2021-2023年期间，所有样方均不再施加干旱处理，进入恢复阶段。试验设置前，所有放牧处理样地均按照常规夏季中度放牧方式进行放牧。每个区组内随机分配3种放牧处理。

放牧处理分为休牧、稳定放牧与重度放牧，试验期间每年通过调控牧草利用率，匹配区域3种常见的家畜干旱管理策略。休牧样地在干旱第1、2年（分别为2019、2020年）的牧草利用率为50%，恢复第1、2年（分别为2021、2022年）的牧草利用率为30%。稳定放牧处理在干旱期与恢复期（2019-2022年）均维持50%的牧草利用率。重度放牧处理在首个干旱年（第1年，2019年）的牧草利用率为50%，第二个干旱年（第2年，2020年）及首个恢复年（第3年，2021年）的牧草利用率为70%，第二个恢复年（第4年，2022年）恢复至50%。为调控放牧强度，研究团队将肉牛（Bos taurus）群置于样地内：怀俄明州站点于7月、蒙大拿州站点于8月放入牛群，自由放牧至达到预设牧草利用率，利用率通过可视障碍杆（Robel等，1970）测定。

数据由Bloodworth、Komatsu、Porensky、Reinhart、Vaarre-Lamoureux、Wilcox与Koerner收集。试验期间（2019-2023年）的每个生长季高峰期，在每个2×2 m样方内指定的1 m²小样方中，采用点针法（pin-drop method）测定植物群落组成（Frost等，2023）。此外，基于试验前（2019-2021年）测定的植物物种组成（盖度百分比）数据，选取每个站点占总盖度90%的优势植物物种，采集其功能性状。功能性状数据于2022年采集一次，每个物种在自然降雨条件下的样方中采集9个个体。C3类禾草与杂草的性状测定于5月开展，C4类禾草与灌木的性状测定于6月下旬开展。测定的性状包括株高、叶厚度、叶片干物质含量（LDMC）、叶面积与比叶面积（SLA）。株高测定至植株伸展最高的营养点。叶厚度采用千分尺（United Scientific PMSET04精密测量千分尺，量程0.01-0.25 mm）测定植株顶部第二片完全展开的叶片。使用同一片叶片，将叶片在60℃下烘干至少1周后的干重除以鲜重，计算得到叶片干物质含量（LDMC）。同样使用该叶片，通过ImageJ软件（ImageJ，Rasband 2021）测定叶面积。比叶面积（SLA）通过叶面积除以叶片干重计算得到。

研究团队采用相同的采集方法，于2019-2022年的生长季高峰期（6月下旬或7月上旬），在每个站点试验区组内的自然降雨条件区域，采集3种常见禾草与2种常见杂草的功能性状（株高、叶厚度、LDMC、叶面积与SLA）。怀俄明州站点的目标物种为格兰马草（Bouteloua gracilis）、高山雀麦（Koeleria macrantha）、史密斯披碱草（Pascopyrum smithii）、田基黄（Logfia arvensis）与美国野豌豆（Vicia americana）；蒙大拿州站点的目标物种为野燕麦（Bromus arvensis）、大针茅（Hesperostipa comata）、高山雀麦（Koeleria macrantha）、猩红花葵（Sphaeralcea coccinea）与西洋婆罗门参（Tragopogon dubius）。每个物种从各区间采集9个个体，2020年未开展叶厚度测定。田基黄（Logfia arvensis）的叶片过小，无法适配所用测量尺度，因此未报告其基于叶片重量的性状（LDMC与SLA）。

本研究的年降水量数据来自站点周边2英里内的本地雨量计（怀俄明州站点），以及蒙大拿州迈尔斯城的美国国家海洋和大气管理局（NOAA）气象站（距离站点15英里内，蒙大拿州站点）。降水量计算时段为前一年10月至采样当月。2019、2021与2022年的采样时间为6月；2020年因COVID-19疫情导致延迟，采样时间改为7月。

Excel文件中：
站点分别标记为FK（Fort Keogh Livestock Range and Research Laboratory，蒙大拿州迈尔斯城）与TB（Thunder Basin Ecoregion，怀俄明州比尔市）。
区组、样地与样方：采用随机完全区组设计。
坡度：采用1-6级评分，1代表最平缓，6代表最陡峭。
降雨缩减与干旱：样方的降雨缩减比例为0、25、50、75或99%。
放牧类别：L代表低放牧利用率（30%），M代表中度放牧利用率（50%），H代表重度放牧利用率（70%）；5位字母分别代表5年的放牧处理序列。
放牧处理：通用处理名称（稳定放牧、重度放牧、休牧）。
livestock_util_2019、2020、2021：牧草利用率为30%、50%或70%。
aerial_basal：地上或基部盖度测定（地上盖度指植被最高处的覆盖量，基部盖度指地面附近植被的覆盖量）。
Native_introduced：N=本土物种，I=外来物种。
Annual_perennial：P=多年生植物，A=一年生植物。
Merged_Traits_FK_2022与Merged_Traits_TB_2022：样地分类为LG（对应MLLMM放牧类别）、MG（对应MMMMM）与HG（对应HHMMM）。