Naturally isolated montane forests in East Africa are hotspots of biodiversity, often characterised by high species endemism, and are fundamental contributors to water services. However, they are located in areas highly suitable for agriculture, making them a prime target for agricultural activities. The Eastern Arc Mountains (EAM) in Eastern Tanzania are within the target regions for agricultural development under the Southern Agricultural Growth Corridor of Tanzania (SAGCOT). However, forest monitoring initiatives that track long-term forest dynamics and the ecological impact of current agricultural development policies on forests, are lacking. Here, we use the STEF (Space-Time Extremes and Features) algorithm and Landsat time series to track forest disturbances (deforestation and degradation) and forest gains (regeneration) as spatio-temporal events over seventeen years (2001–2017) in the montane forests of the Mvomero District in Tanzania. We found that 27 % (∼ 20 487 ha) of montane forests were disturbed between 2001 and 2017, mainly led by deforestation (70 %). Small-scale crop farms with maize, banana, and cassava crops, were the most planted on deforested areas. Most disturbances occurred at lower elevation (lowland montane), but there was an increasing shift to higher elevations in recent years (2011–2017). Forest disturbances exclusively occurred at small spatial scales, a pattern similar to other forest montane landscapes in Africa, which lowers detection capabilities in global forest loss products. Our locally calibrated and validated deforestation map (Producer's accuracy = 80 %; User’s accuracy = 78 %) shows a gross underestimation of forest cover loss (>10 000 ha) by global forest loss products in these mountainous forest landscapes. Overall, we found few areas undergoing forest regeneration, with only 9 % of the disturbed forest regenerating over 17 years. Long-term conversion to cropland prevented regeneration in the lowlands, with regeneration mainly happening at higher elevations. However, the shift of deforestation and forest degradation to higher elevations may challenge high elevation regeneration trends, leaving the remaining blocks of montane forest in the Mvomero District at a risk of degradation and disappearance. Without effective forest conservation measures, market-driven agricultural development is likely to trigger an expansion of cropland at the expense of forests to meet the increased demand for the agricultural products promoted, with negative impact on biodiversity, carbon sequestration and water services.

东非天然孤立的山地森林是生物多样性的热点，通常特征是物种特有性高，是水服务的根本原因。但是，它们位于非常适合农业的地区，使其成为农业活动的主要目标。坦桑尼亚东部的南部农业增长走廊（SAGCOT）之下，坦桑尼亚东部的东部弧山（EAM）在农业发展的目标区域之内。但是，缺乏跟踪长期森林动态和当前农业发展政策对森林的生态影响的森林监测倡议。这里，我们使用STEF（时空极端和特征）算法和Landsat时间序列来跟踪山地森林17年（2001-2017年）中森林干扰（森林砍伐和退化）和森林收益（再生）作为时空事件。坦桑尼亚的Mvomero区。我们发现，在2001年至2017年之间，有27％（〜20487公顷）的山地森林受到干扰，主要是森林砍伐（70％）。种植玉米，香蕉和木薯的小规模农作物种植面积最大。大多数干扰发生在海拔较低的地方（低地山地），但近年来（2011-2017年）的趋势逐渐转向高海拔。森林干扰仅发生在较小的空间尺度上，与非洲其他森林山地景观类似，这降低了全球森林损失产品的检测能力。我们在当地校准和验证的森林砍伐图（生产者的准确度= 80％；用户的准确度= 78％）显示，在这些山区森林景观中，全球森林损失产品严重低估了森林覆盖率（> 10000公顷）。总体而言，我们发现很少有地区正在经历森林更新，只有9％的受干扰森林在17年内得以恢复。长期转为耕地阻止了低地的再生，再生主要发生在高海拔地区。但是，森林砍伐和森林退化向更高海拔的转移可能会挑战高海拔的更新趋势，从而使姆沃梅罗地区的山地森林剩余区块面临退化和消失的风险。