Impacts of invasive Opuntia cacti on wild mammals in Kenya

Abstract

In this thesis, I explored the impacts of invasive plants on animal behaviour, using the invasion of Opuntia cacti in Laikipia County, Kenya, as a specific case study. In the opening chapter, I introduced the topic of biological invasions, addressing essential background material and identifying key knowledge gaps.
In the second chapter, I focused on the impacts of invasive plants on animal behaviour, an important – yet neglected – topic. I synthesised the disparate literature on invasive plants’ behavioural impacts within a novel mechanistic framework, revealing that invasive plants can cause profound behavioural changes in native animals, with ecological consequences at multiple scales. I also found that environmental context played an important role in moderating how an invader’s modes of impact translate into behavioural changes in native species, and how these behavioural changes then generate ecological impacts. Finally, I identified priority research questions relating to the behavioural impacts of invasive plants.
Invasive plants’ behavioural impacts can manifest as changes to the occurrence patterns of native animals. In Chapter 3, I used simulations to explore model selection in occupancy models, which are a powerful tool for studying the patterns and drivers of occurrence. Specifically, I investigated the consequences of collider bias – a type of confounding that can arise when adding explanatory variables to a model – for model selection using the Akaike Information Criterion (AIC) and Schwarz Criterion (or Bayesian Information Criterion, BIC). I found that the effect of collider bias, and consequently the inferential and predictive accuracy of the AIC/BIC-best model, depended on whether the collider bias was present in the occupancy or detection data-generating process. My findings illustrate the importance of distinguishing between inference and prediction in ecological modelling and have more general implications for the use of information criteria in all linear modelling approaches.
In Chapter 4, I applied the mechanistic framework from Chapter 2 and the modelling conclusions from Chapter 3 to the problem of understanding Opuntia’s behavioural impacts in Laikipia County. Specifically, I used camera traps to explore the effects of Opuntia on occupancy and activity for eight key mammal species. I found that the effects of Opuntia varied among mammal species and depended on the spatial scale of the Opuntia cover covariate. These findings have important implications for the conservation of endangered mammal species in the region, the future spread of Opuntia through seed dispersal, and interactions between wildlife and local communities.
In Chapter 5, I addressed key knowledge gaps pertaining to Opuntia’s biotic interactions with native animals. First, I quantified the relationship between height and fruiting in O. engelmannii and O. stricta, finding that height was positively related to fruiting for both species, and that the relationship was stronger for O. engelmannii than for O. stricta. I also found that local habitat variables were related to height and/or fruiting in both Opuntia species. Second, I documented the interactions between animals and Opuntia using camera traps. In so doing, I confirmed the importance of interactions that were previously thought to be important, while also highlighting interactions which have previously received little attention in the published scientific literature.