Introduction
Declines of Amazonian bird species in remote locations have recently attracted a great deal of attention to the potential impacts of global change in the most extensively intact forest landscape on earth (Blake & Loiselle 2015; Stouffer et al. 2020). However, our understanding of these trends in Amazonian bird communities is based on only two studies (Blake & Loiselle 2015; Stouffer et al. 2020). Blake and Loiselle (2015) reported declines over an 8 year period from a remote lowland rainforest in Ecuador that may be explained by regional-wide precipitation patterns, although they hypothesize that these changes might also be explained by a series of strong La NiƱa events. More recently, Stouffer et al (2020) reported declines of terrestrial insectivores over >35 years in an isolated study site, near Manaus, Brazil. To broaden inference of trends across the region and across species, we evaluated the entire bird community at a site in southeastern Peru over a36-year interval. Our results contrast with previous finding and highlight the urgent need for long term surveys to better understand whether global change is driving widespread species declines in the lowland tropics.
Studies of ecological communities in remote locales absent other large-scale human impacts presents a unique opportunity to evaluate how biodiversity is affected by global change. Several long-term studies of neotropical tree communities have made theoretical and empirical advances in understanding long-term community dynamics (Rees et al. 2001; Swenson et al. 2012) . With regards to animal studies, several authors have documented long-term community change in the tropics - albeit for subsets of the community (Blake & Loiselle 2015; Brawn et al. 2017; Stouffer et al. 2020). What do these studies conclude in aggregate-across plant and animal groups, about how lowland tropical species are faring at locations with little direct human disturbance?
A major impediment to studying changes in tropical bird communities is deriving precise population estimates for the majority of species in the community; species-specific challenges in dealing with suites of species that require a variety of sampling techniques across different spatial scales, which complicates accurate abundance estimation (Robinsonet al. 2018). In 1982, Terborgh et al. (1990) undertook what became a seminal censusing effort and fully described for the first time, in great detail and precision, the structure and organization of the entire bird community present in a 97-ha plot of intact and mature Amazonian floodplain forest at Cocha Cashu Biological Station (Silmanet al. 2003). At the time, the study was monumental: the scientific community had barely learned to identify many bird species in South America, let alone how to quantify their abundances using sampling techniques designed for temperate or nearctic forest bird communities.
Here, after a 36 year interval, we re-censused the entire bird community on the 97-ha plot at Cocha Cashu originally surveyed by Terborghet al . (1990) , using the same methodology, in order to evaluate changes to a bird community in a remote Amazonian site that remains little-influenced by human activity. As in the original study, our approach allowed us to estimate population size for the entire bird community, providing a rich window into changes in the structure and organization of the community. We tested key hypotheses about which the local and regional processes are likely to drive patterns of community structure and organization. Previous studies such as those mentioned above have mostly attributed large changes in insectivores and decreased recruitment of understory species to regional changes in precipitation patterns. If these regional processes had also influenced the bird community at Cocha Cashu, we would expect to find changes, across entire suites of species (e.g. declines across all terrestrial insectivores as has been reported elsewhere (Van Klink et al. 2020; Wagner 2020). Alternatively, if local processes were the main determinants for community composition over the two time periods, we posited that changes in some species would be evident due to changes in specific habitats. For example, some changes may be attributed to a natural mass-mortality event of Guadua sp. bamboo patches found on or near the plot (Socolar et al. 2013). Thus, changes in abundance and persistence in the community would be evident mostly in species associated with bamboo and species associated with primary successional habitat influenced by the Manu river, with declines expected in both.
The focus here is to document patterns of change in a hyper-diverse community over time. Our results show that, by-and-large, the structure of the community has persisted virtually unchanged. This large-scale observational comparative study is a necessary first step to propose hypotheses regarding the role of fluctuating resources and biotic interactions such as competition and mutualism that may play a role in explaining the few changes we recorded. We end by stressing the importance of developing new analytical techniques and combining these with traditional census methods in order to generate more consistent long-term datasets for tropical bird communities.