Habitat fragmentation and loss are causing biodiversity declines across the globe. As biodiversity is unevenly distributed, with many hotspots located in the tropics, conserving and protecting these areas is important to preserve as many species as possible. Chapter 2 presents an overview of the Ecology of the Atlantic Forest, a highly fragmented biodiversity hotspot. A major driver of habitat fragmentation is agriculture, and in the tropics coffee is major cash crop. Developing methods to monitor biodiversity effectively without labour intensive surveys can help us understand how communities are using fragmented landscapes and better inform management practices that promote biodiversity. Acoustic monitoring offers a promising set of tools to remotely monitor biodiversity. Developments in machine learning offer automatic species detection and classification in certain taxa. Chapters 3 and 4 use acoustic monitoring surveys conducted on fragmented landscapes in the Atlantic Forest to quantify bird and bat communities in forest and coffee matrix, respectively. Chapter 3 shows that acoustic composition can reflect local avian communities. Chapter 4 applies a convolutional neural network (CNN) optimised on UK bat calls to a Brazilian bat dataset to estimate bat diversity and show how bats preferentially use coffee habitats. In addition to monitoring biodiversity, monitoring microclimate forms a key part of climate smart agriculture for climate change mitigation. Coffee agriculture is limited to the tropics, overlapping with biodiverse regions, but is threatened by climate change. This presents a challenge to countries strongly reliant on coffee exports such as Brazil and Nicaragua. Chapter 5 uses data from microclimate weather stations in Nicaragua to demonstrate that sun-coffee management is vulnerable to supraoptimal temperature exposure regardless of local forest cover or elevation.