Geostatistical modelling of recreational fishing data: A fine-scale spatial analysis
Date of Award
Doctor of Philosophy
School of Science
Associate Professor Ute Mueller
Associate Professor Glenn Hyndes
Mrs Karina Ryan
The sustainability of recreational fisheries resources rely on effective management of the fishery which includes monitoring of any changes in the fishery. In order to facilitate the ongoing management of the recreational fishery, an understanding of the spatial dynamics of catch per unit effort (catch rate), fishing effort and species diversity is important for fishery managers to make area-specific decisions and to develop strategies for ecosystem based fisheries management. These indices are critical components of information used to inform on recreational fishing activities and evaluate the changes in the fishery resources.
Geostatistical techniques such as kriging can provide useful tools for characterising the spatial distributions of these indices. However, most recreational fishing data are highly skewed, zero-inflated and when expressed as ratios are impacted by the small number problem, which can influence estimates obtained from the traditional kriging. In addition, the use of recreational fishing data obtained through surveys may influence mapping and area-specific decisions as such data are associated with uncertainty.
In Western Australia, recreational fishing has a participation rate of approximately 30%. Data for this thesis were collected from boat-based recreational fishers through phone-diary surveys at spatial resolution that supports spatial analysis and mapping through geostatistical techniques. In this thesis, geostatistical modelling techniques were used to analyse those recreational fishing data in the West Coast Bioregion of Australia, with the development and evaluation of a data transformation approach that takes into account data characteristics and uncertainty. As a first step in the analysis, a suitable kriging estimator for recreational fishing data was determined. This was based on the application of ordinary, ordinary indicator and Poisson kriging estimators for seven aquatic species with different behaviours and distribution patterns. Some of these estimators can handle different distribution properties including high skewness, zero-inflation and small number problems. In general, the indicator kriging performed similarly across species with different life history characteristics and distribution patterns and provided accurate estimates of catch rates for most of those species. To evaluate the incorporation of measurement uncertainty, the study presents a soft indicator kriging approach that uses a logistic function transformation, which is combined with probability field simulation to determine the effect of measurement uncertainty on mapping and fishing area delineation. The results suggest that the incorporation of the measurement uncertainty improves the ability to draw valid conclusions about the estimation results, which may influence any decision regarding the delineation of areas with high catch rates for spatial management. The recreational fishing data used also provided the basis for studying the spatial patterns in species diversity in the entire fishery. The analysis revealed that species diversity, dominance and evenness display similar spatial patterns on a global scale.
The study highlighted the inherent spatial variability in catch rate, fishing effort and species diversity, illustrating areas with high values, or hotspots of these indices. This statistical-based modelling approach is important as it allows prediction of these indices in specific locations taking into account data characteristics and uncertainty. The estimated maps are important for supporting fishery resources management.
Access to this thesis is restricted to current ECU staff and students. Email request to firstname.lastname@example.org
Aidoo, E. (2016). Geostatistical modelling of recreational fishing data: A fine-scale spatial analysis. Retrieved from https://ro.ecu.edu.au/theses/1813