Conservation biology has guided multiple applications that relate to how humans interact with wildlife. The work of Caughley [1994] describes the core directions for conservation biology by introducing the main options for human intervention in nature. Caughley [1994] proposed the following management actions: (i) increase depleted populations; (ii) decrease excessive populations; (iii) establish maximum sustainable yields; (iv) carry out monitoring programs without additional actions over stable populations. In this thesis, we proposed new approaches to improve the state-of-the-art quantitative methods applied to challenges on two of Caughley’s management actions: 1) carry out monitoring programs without additional actions over stable populations and 2) decrease excessive populations. For the first management action, we explored the challenge of identifying avian species in audio-based monitoring systems using transfer learning and Convolution Neural Networks. We also explored the challenge of accurate insect classification and localisation in image-based monitoring systems focusing on species with agronomical and forensic importance. We proposed a framework for combining computer vision and machine learning to identify these species and explored solutions for this action in small and imbalanced datasets. Finally, for the second management action, we introduced two new statistical machine learning approaches to predict insect outbreaks and abundance to aid decision-making applied to Integrated Pest Management. We proposed a new statistical machine learning method, Pattern-Based Prediction (PBP), to predict insect outbreaks and a new approach combining statistical machine learning, causal analysis and time series embedding to guide the selection of climate time series and their lags to build statistical machine learning forecasting methods.