Land managers need reliable metrics for assessing the quality of restorations and natural areas and prioritizing management and conservation efforts. However, it can be difficult to select metrics that are robust to sampling methods and natural environmental differences among sites, while still providing relevant information regarding ecosystem changes or stressors. We collected herbaceous-layer vegetation data in wetlands and grasslands in four regions of the USA (the Midwest, subtropical Florida, arid southwest, and coastal New England) to determine if commonly used vegetation metrics (species richness, mean coefficient of conservatism [mean C], Floristic Quality Index [FQI], abundance-weighted mean C, and percent non-native species cover) were robust to environmental and methodological variables (region, site, observer, season, and year), and to determine adequate sample sizes for each metric. We constructed linear mixed effects models to determine the influence of these environmental and methodological variables on vegetation metrics and used metric accumulation curves to determine the effect of sample size on metric values. Species richness and FQI varied among regions, and year and observer effects were also highly supported in our models. Mean C was the metric most robust to sampling variables and stabilized at less sampling effort compared to other metrics. Assessment of mean C requires sampling a small number of quadrats (e.g. 20), but assessment of species richness or FQI requires more intensive sampling, particularly in species-rich sites. Based on our analysis, we recommend caution be used when comparing metric values among sites sampled in different regions, different years, or by different observers.

土地管理者需要可靠的指标来评估恢复和自然区域的质量，并确定管理和保护工作的优先级。然而，很难选择对采样方法和地点之间的自然环境差异具有鲁棒性的指标，同时仍能提供有关生态系统变化或压力因素的相关信息。我们收集了美国四个地区（中西部、亚热带佛罗里达、西南干旱和新英格兰沿海）湿地和草原的草本层植被数据，以确定是否常用的植被指标（物种丰富度、平均保守系数 [平均C ], 植物区系质量指数 [FQI], 丰度加权平均C和非本地物种覆盖率百分比）对环境和方法变量（地区、地点、观察者、季节和年份）具有稳健性，并为每个指标确定足够的样本量。我们构建了线性混合效应模型来确定这些环境和方法变量对植被度量的影响，并使用度量累积曲线来确定样本大小对度量值的影响。物种丰富度和 FQI 因地区而异，我们的模型也高度支持年份和观察者效应。平均C是对采样变量最稳健的指标，并且与其他指标相比，以较少的采样工作量稳定下来。平均 C 的评估需要采样少量的样方（例如 20 个），但物种丰富度或 FQI 的评估需要更密集的采样，特别是在物种丰富的地点。根据我们的分析，我们建议在比较不同地区、不同年份或不同观察者采样的站点之间的度量值时要谨慎。