In order to understand evolutionary-ecological processes and make decisions concerning wildlife management (i.e., conservation and monitoring, according to Caughley (1994)), the ability to estimate abundances of wild animal species can prove imperative (Verdade et al., 2014). However, methods of data collection that involve direct interaction with wild animals can be invasive and pose risks to both wildlife and humans (Verdade et al., 2013). In this thesis, we propose methodologies that may be used to estimate animal abundances using different types of data, with an emphasis on data whose collection is relatively low-effort, cost-effective and poses the least risk of danger to the animal and observer. The ability to use these data to estimate abundance may allow for the establishment of large-scale wildlife monitoring programs. First we present a multivariate extension to the N-mixture model proposed by Royle (2004). This extension allows for the estimation of abundances for multiple species simultaneously, while also estimating the correlation between species abundances. This model is further extended to allow for data collected over long time periods through the addition of a first-order autoregressive term on the abundance. This model is then extended further to allow for the use of zero-inflated data by considering a hurdle-Poisson distribution for the latent abundances. We then provide an overview of various N-mixture models, aimed at introducing practitioners unfamiliar with statistics to this methodology. We demonstrate a Bayesian implementation of some of these models to estimate foraging bee abundance, with R code provided to allow model implementation by any interested practitioners. Finally we examine a scenario in which data is not composed of observations of individuals, but rather observations of animal vestiges (i.e., traces that an animal leaves behind as it moves through the environment). Here we present a novel modelling framework, the triple Poisson model, that allows for the estimation of animal abundance using vestige data, even when only very scarce data is available.