Ecological restoration is critical for recovering degraded ecosystems but is challenged by variable success and low predictability. Understanding which outcomes are more predictable and less variable following restoration can improve restoration effectiveness. Recent theory asserts that the predictability of outcomes would follow an order from most to least predictable from coarse to fine community properties (physical structure > taxonomic diversity > functional composition > taxonomic composition) and that predictability would increase with more severe environmental conditions constraining species establishment. We tested this “hierarchy of predictability” hypothesis by synthesizing outcomes along an aridity gradient with 11 grassland restoration projects across the United States. We used 1829 vegetation monitoring plots from 227 restoration treatments, spread across 52 sites. We fit generalized linear mixed-effects models to predict six indicators of restoration outcomes as a function of restoration characteristics (i.e., seed mixes, disturbance, management actions, time since restoration) and used variance explained by models and model residuals as proxies for restoration predictability. We did not find consistent support for our hypotheses. Physical structure was among the most predictable outcomes when the response variable was relative abundance of grasses, but unpredictable for total canopy cover. Similarly, one dimension of taxonomic composition related to species identities was unpredictable, but another dimension of taxonomic composition indicating whether exotic or native species dominated the community was highly predictable. Taxonomic diversity (i.e., species richness) and functional composition (i.e., mean trait values) were intermittently predictable. Predictability also did not increase consistently with aridity. The dimension of taxonomic composition related to the identity of species in restored communities was more predictable (i.e., smaller residuals) in more arid sites, but functional composition was less predictable (i.e., larger residuals), and other outcomes showed no significant trend. Restoration outcomes were most predictable when they related to variation in dominant species, while those responding to rare species were harder to predict, indicating a potential role of scale in restoration predictability. Overall, our results highlight additional factors that might influence restoration predictability and add support to the importance of continuous monitoring and active management beyond one-time seed addition for successful grassland restoration in the United States.

中文翻译：

---

测试不同环境严重程度梯度下草原恢复的可预测性层次

生态恢复对于恢复退化的生态系统至关重要，但面临着成功率参差不齐和可预测性低的挑战。了解修复后哪些结果更可预测且变化更小，可以提高修复效果。最近的理论认为，结果的可预测性将遵循从粗到细的群落属性（物理结构>分类多样性>功能组成>分类组成）从最可预测到最不可预测的顺序，并且随着限制物种建立的环境条件更加严格，可预测性会增加。我们通过综合美国 11 个草原恢复项目的干旱梯度结果来测试这种“可预测性层次”假设。我们使用了来自 227 个恢复处理的 1829 个植被监测样地，分布在 52 个地点。我们拟合广义线性混合效应模型来预测恢复结果的六个指标，作为恢复特征（即种子混合、干扰、管理行动、恢复以来的时间）的函数，并使用模型和模型残差解释的方差作为恢复可预测性的代理。我们没有找到对我们的假设的一致支持。当响应变量是草的相对丰度时，物理结构是最可预测的结果之一，但总冠层覆盖率是不可预测的。同样，与物种身份相关的分类组成的一个维度是不可预测的，但指示外来物种还是本土物种在群落中占主导地位的分类组成的另一个维度是高度可预测的。分类多样性（即物种丰富度）和功能组成（即平均性状值）是间歇性可预测的。可预测性也并没有随着干旱而持续增加。在更干旱的地区，与恢复群落中物种身份相关的分类组成的维度更容易预测（即残差较小），但功能组成的可预测性较差（即残差较大），并且其他结果没有显示出显着的趋势。当恢复结果与优势物种的变化相关时，恢复结果是最可预测的，而对稀有物种的反应则更难以预测，这表明规模在恢复可预测性中具有潜在作用。总体而言，我们的结果强调了可能影响恢复可预测性的其他因素，并为美国草地成功恢复除了一次性种子添加之外的持续监测和主动管理的重要性提供了支持。