Seed removal can result in either seed predation or dispersal and is therefore an indicator of important ecosystem functions. To better understand how these variable processes are affected by environmental changes, rapid and reliable assessments with high temporal and spatial replication are needed. We address this need by testing the application of a rapid ecosystem function assessment (REFA) method to investigate seed removal in habitats with differing land use intensity. We recorded seed removal hourly over eight hours at 301 sites in five habitat types in three urban regions in Germany. We calculated seed removal rates per sampling period, traditionally used in seed removal studies, as well as instantaneous seed removal probabilities based on hourly sampling. Across regions, seed removal probabilities and rates were lowest in arable fields, a habitat type with high land-use intensity. Except in urban sealed areas, temperature had a negative effect on seed removal. Additional Cox hazard regressions suggest invertebrates as the main seed removing animals in arable fields, whereas vertebrates were likely removers in other habitat types. We confirm that seed removal is strongly negatively affected by human disturbance, indicating that the tested method is appropriate in different settings. We were able to recognize patterns in highly variable data and the method also has the advantages of low cost, high replication and high temporal resolution. However, there is a trade-off between the high temporal-resolution of instantaneous seed removal probabilities and the sampling effort, but adjustments in the standardized setup can be made depending on the study. To further utilize the extensive data collection in the REFA method, we propose to combine instantaneous seed removal probabilities, seed removal rates and Cox hazard regressions of seed removal to provide complementary information on the extent and temporal patterns of seed removal and indications about potential seed removing guilds. Keywords: Seed predation, seed dispersal, seed removal, granivore, REFA

去除种子可导致种子捕食或扩散，因此是重要的生态系统功能的指标。为了更好地理解这些可变过程如何受到环境变化的影响，需要对时间和空间高度重复的快速可靠评估。我们通过测试快速生态系统功能评估（REFA）方法在不同土地利用强度的生境中调查种子去除的方法来满足这一需求。我们在德国三个城市地区的五个栖息地类型的301个站点上，每小时记录了八个小时以上的种子清除情况。我们计算了传统上用于种子去除研究的每个采样周期的种子去除率，以及基于每小时采样的瞬时种子去除概率。在整个区域中，可耕田的种子去除率和去除率最低，高土地利用强度的栖息地类型。除城市密封地区外，温度对种子的去除有负面影响。另外的Cox危害回归表明，无脊椎动物是可耕地中主要的去除种子的动物，而脊椎动物可能是其他生境类型的去除者。我们确认种子的去除受到人为干扰的强烈负面影响，表明所测试的方法适合于不同的环境。我们能够识别出高度可变的数据中的模式，并且该方法还具有低成本，高复制性和高时间分辨率的优点。但是，瞬时种子去除概率的高时间分辨率与采样工作之间需要权衡取舍，但是可以根据研究对标准设置进行调整。为了进一步利用REFA方法中的大量数据收集，我们建议将瞬时种子去除概率，种子去除率和种子去除的Cox危害回归相结合，以提供有关种子去除的程度和时间模式的补充信息以及潜在的种子去除指示行会。关键词：种子捕食，种子传播，去除种子，食草动物，REFA