Many forests exist in human-modified mosaic landscapes subject to logging, agriculture, and livestock grazing. These disturbances can reduce ecosystem resilience and ultimately result in crossing thresholds, leading to the transition of the ecosystem into new stable states with different plant communities. The interaction between these disturbances and forest succession is still poorly understood, despite the growing need to manage these human-modified landscapes for biodiversity and productivity. We studied abrupt changes in vegetation composition in a Central Mexican oak forest and explored how these changes could indicate the coexistence of various alternative stable states caused by agricultural and livestock legacies. We established 36 strip transects covering 9000 m2 in patches with various land use legacies where we sampled vegetation, soil characteristics, and human disturbance. We assessed future and past tendencies in vegetation spatial patterns by evaluating seedling composition and analyzed time series of NDVI signals recorded by LANDSAT images (1972–2019). We utilized NMDS, PERMANOVA, and ANOVA analyses to assess if stable states can be distinguished based on historical and current variations in vegetation composition and structure in patches with different management legacies. We then assess potential factors that may contribute to differences in plant community composition by fitting edaphic, topographic, and anthropogenic disturbance variables to the ordination analyses.

We identified three distinct plant communities: oak forest (dominated by late-successional Oak species), secondary shrubland (dominated by Vachellia-Dodonaea association), and a transitional forest-shrub community (a mixture of the first two communities). Time-series analyses and seedling composition provide evidence of long-term community segregation. Variation partition analysis demonstrated that anthropogenic disturbance (particularly livestock browsing), had a greater impact on species composition differentiation than environmental variables.

A state-and-transition model is proposed, where cattle browsing and agricultural legacies trigger the transition from oak forests to a stable secondary shrubland community. Three potential self-reinforcing and interacting feedbacks are suggested: (i) micro-environmental filtering, (ii) plant-trait mediated assembly process, and (iii) cascading effects on tree recruitment caused by cattle browsing. The proposed state-and-transition model explains forest successional dynamics in systems with multiple anthropogenic disturbances and provides a conceptual framework for understanding and guiding successional trajectories.

许多森林存在于人类改造的马赛克景观中，受到伐木、农业和牲畜放牧的影响。这些干扰会降低生态系统的恢复力，并最终导致跨越阈值，导致生态系统过渡到不同植物群落的新稳定状态。尽管人们越来越需要管理这些人类改变的景观以提高生物多样性和生产力，但这些干扰与森林演替之间的相互作用仍然知之甚少。我们研究了墨西哥中部橡树林植被组成的突变，并探讨了这些变化如何表明农业和畜牧业遗产造成的各种替代稳定状态的共存。我们在具有各种土地利用遗产的斑块中建立了 36 个占地 9000 平方米的带状样线，对植被、土壤特征和人类干扰进行了采样。我们通过评估幼苗组成并分析 LANDSAT 图像（1972-2019）记录的 NDVI 信号时间序列来评估植被空间格局的未来和过去趋势。我们利用 NMDS、PERMANOVA 和 ANOVA 分析来评估是否可以根据具有不同管理遗产的斑块中植被组成和结构的历史和当前变化来区分稳定状态。然后，我们通过将土壤、地形和人为干扰变量拟合到排序分析中，评估可能导致植物群落组成差异的潜在因素。

我们确定了三个不同的植物群落：橡树林（以晚期演替的橡树物种为主）、次生灌木丛（以Vachellia-Dodonaea群落为主）和过渡性森林-灌木群落（前两个群落的混合）。时间序列分析和幼苗组成提供了长期群落隔离的证据。变异分区分析表明，人为干扰（特别是牲畜啃食）对物种组成分化的影响比环境变量更大。

提出了一种状态与过渡模型，其中牛群觅食和农业遗产触发了从橡树林到稳定的次生灌木丛群落的过渡。提出了三种潜在的自我强化和相互作用的反馈：（i）微环境过滤，（ii）植物性状介导的组装过程，以及（iii）牛啃食引起的对树木补充的级联效应。所提出的状态与转变模型解释了具有多种人为干扰的系统中的森林演替动态，并为理解和指导演替轨迹提供了概念框架。