Trends and processes in Large Marine Ecosystems fisheries
Author(s)
Conti, Lorenza
Date Issued
February 10, 2011
Type
Doctoral Thesis
Abstract
This study aimed at exploring the dynamics of large marine ecosystems (LMEs) fisheries and the relationships between catches and potential drivers, taking into account both ecological and economic processes. Data analyses were based on fisheries landings time series provided by the Sea Around Us Project (SAUP) and time series of environmental and economic variables from different sources. Artificial neural networks (ANNs) have been extensively applied both for modeling and data analysis purposes.
A general introduction is provided in Chapter I, where the state-of-the-art in modern fisheries science is described, together with a general framework within which the aims and scopes of the thesis are presented.
In Chapter II the relationship between primary productivity (PP) and fish yield was investigated through PP estimates based on satellite observations and reported fisheries landings from 14 selected LMEs. Correspondence analysis performed on yield data outlined the role played by different trophic levels (TLs) in LMEs catches. PP temporal variability was significantly and positively correlated to average trophic level of catches (TLc) so that high yields in less variable ecosystems were characterized by low TLc. The percentage primary productivity required to sustain catches (%PPR) and TLc were used to assess fisheries impact on ecosystems and the PP model choice emerged as a critical step affecting the assessment of exploitation levels. A more conservative estimation of PP could contribute to a more precautionary approach to fisheries management where high levels of exploitation are more easily attained (Conti & Scardi, 2010).
In Chapter III the regional dynamics of industrial fisheries within the LMEs boundaries were investigated by means of an historical-descriptive approach. An unsupervised artificial neural network (Self Organizing Map, SOM) was proposed as a tool for analyzing changes in composition of fisheries landings over five decades in 51 LMEs. From the analysis of LMEs fishing histories a broad distinction emerged between two main types of fisheries, based on different exploitation strategies: (1) pelagic dominated landings showed stable or cyclic compositions, mainly driven by intrinsic oscillation in exploited populations of target species and are located in LMEs sharing specific features (e.g. productive regimes, upwelling) and (2) trawl fisheries seemed to be more affected by economic drivers (e.g. investments in fishing gears and new technologies), and tended to cluster in the northern hemisphere, where fishing pressures and investments have been historically higher. It could be further suggested that northern hemisphere LMEs are also generally characterized by wider continental shelves, which represent a critical feature for demersal exploitation.
In Chapter IV fisheries yields and landings composition by functional groups were predicted for 51 LMEs by means of an artificial neural network. Input variables selected for the two models described both ecological and economic features of each LME, and no assumptions on inter-variables relationships were made. The neural network provided accurate estimates of total fisheries yields and catches composition, overcoming the restrictive assumptions imposed by linear models. New insights into underlying processes governing fisheries harvests were provided by the sensitivity analysis carried out on the two models. Both economic and ecological predictors were strictly linked to fisheries landings. Catches composition seemed to be influenced by intrinsic ecosystems dynamics, while total yields seemed to be mainly driven by latitude and factors related to the economy of the fishing countries.
General conclusions and future perspectives are provided in Chapter V.
Additional information
Dottorato di ricerca in Ecologia e gestione delle risorse biologiche
File(s)![Thumbnail Image]()
Name
lconti_tesid.pdf
Size
3.67 MB
Format
Adobe PDF
Checksum (MD5)
1227937ecacb4ceb2ed9c9bc9bce1d39