Safari Science: Testing the Feasability and Realibility of Tourists as Citizen Scientists in East African Protected Areas

Abstract

Wildlife populations continue to decline in East Africa, both in and around Protected Areas (PAs). High quality data on the density and distribution of wildlife are essential for effective conservation and management. However, methods that accurately capture data on wildlife locations at fine scales over large land areas have proven to be prohibitively difficult, expensive and time-consuming given the logistical and financial constraints of East African PAs. This study explores whether citizen science can be a reliable alternative to conventional methods of wildlife monitoring. Citizen science (CS) – also known as public participation in data gathering – has the potential to answer questions that have long plagued conservation scientists, as it has been shown to decrease monitoring costs while increasing public engagement in conservation issues. Despite the growing use of CS in ecological research, debates persist over the reliability of these datasets. An early and continuing concern of CS data is the effect of observer error, which if ignored can produce misleading ecological conclusions in modeled species-environment relationships. Through the implementation of a pilot program, I address the feasibility, reliability, and utility of CS for wildlife management in East African PAs. I ask (i) what problems and prospects arise when attempting to implement a CS program in East Africa? (ii) can the data generated from this method approximate the quality and quantity of data from more conventional sampling techniques?, and (iii) can CS data be useful for wildlife managers? To test these questions, this study uses two independent methods to gather spatial and demographic data on twenty-nine species within a private conservancy in southwestern Kenya. In method 1, I asked tourists to gather data on wildlife during their game drives. The novel use of mobile technology aided my ability to enter and manage data quickly, though this study is limited by a lack of volunteer interest and motivation. I report several observations regarding the potential for future CS programs in the region, finding a need for PA administration to take ownership over program and data management. In reality, the large amount of time needed to organize and implement CS programs on a daily basis severely restricts the potential for CS in small East African PAs, as lodge managers have little time, inclination, and resources to devote to these programs. Method 2 uses line transects, a conventional ecological sampling technique, providing a validation dataset for testing the reliability of CS data. I use generalized linear models (GLMs) to model the species-environment relationships for nine commonly reported species, as it is a common analytical technique used by ecologists and wildlife managers for ecological inference. I find SS data performed reasonably well for eight out of nine species, with accuracies ranging from 60-87%. However, these results are complicated by high uncertainty in model performance and ecological validity, thus limiting the usefulness of this data for management and conservation planning. I find GLMs are not the best methods for analyzing CS data as they lack the capacity to account for observer error, an important source of bias in CS datasets. When this error is sufficiently accounted for, CS data could be much more reliable for ecological inference by wildlife managers. In conclusion, I find citizen science to be a complex and difficult monitoring method, both logistically and analytically challenging. Though there is great potential for CS in East African PAs, substantial barriers exist and impede successful program establishment. I recommend future programs in East Africa carefully and realistically weigh their ability to address these obstacles before implementing a Safari Science approach to monitoring wildlife.