Relating sardine recruitment in the Northern Benguela to satellite-derived sea surface height using a neural network pattern recognition approach

Processes of enrichment, concentration and retention are thought to be important for the successful recruitment of small pelagic fish in upwelling areas, but are difficult to measure. In this study, a novel approach is used to examine the role of spatio-temporal oceanographic variability on recruitment success of the Northern Benguela sardine Sardinops sagax. This approach applies a neural network pattern recognition technique, called a self-organising map (SOM), to a seven-year time series of satellite-derived sea level data. The Northern Benguela is characterised by quasi-perennial upwelling of cold, nutrient-rich water and is influenced by intrusions of warm, nutrient-poor Angola Current water from the north. In this paper, these processes are categorised in terms of their influence on recruitment success through the key ocean triad mechanisms of enrichment, concentration and retention. Moderate upwelling is seen as favourable for recruitment, whereas strong upwelling, weak upwelling and Angola Current intrusion appear detrimental to recruitment success. The SOM was used to identify characteristic patterns from sea level difference data and these were interpreted with the aid of sea surface temperature data. We found that the major oceanographic processes of upwelling and Angola Current intrusion dominated these patterns, allowing them to be partitioned into those representing recruitment favourable conditions and those representing adverse conditions for recruitment. A marginally significant relationship was found between the index of sardine recruitment and the frequency of recruitment favourable conditions (r super(2) = 0.61, p = 0.068, n = 6). Because larvae are vulnerable to environmental influences for a period of at least 50 days after spawning, the SOM was then used to identify windows of persistent favourable conditions lasting longer than 50 days, termed recruitment favourable periods (RFPs). The occurrence of RFPs was compared with back-calculated spawning dates for each cohort. Finally, a comparison of RFPs with the time of spawning and the index of recruitment showed that in years where there were 50 or more days of favourable conditions following spawning, good recruitment followed (Mann-Whitney U-test: p = 0.064, n = 6). These results show the value of the SOM technique for describing spatio-temporal variability in oceanographic processes. Variability in these processes appears to be an important factor influencing recruitment in the Northern Benguela sardine, although the available data time series is currently too short to be conclusive. Nonetheless, the analysis of satellite data, using a neural network pattern-recognition approach, provides a useful framework for investigating fisheries recruitment problems.

Pattern of benthic biomass size spectra from shallow waters in the East China Seas

Benthic biomass size spectra (BSS) and normalized biomass size spectra were constructed, and benthic secondary production was estimated by a size spectrum equation in the shallow waters in the East China Sea, ranging latitudinally from 40A degrees N to 29A degrees N. The BSS patterns were bimodal, two biomass peaks corresponding to meiofauna and macrofauna, respectively, separated by a trough of low biomass at 8-256 mu g individual dry weight which varied in position with median sediment particle size. The BSS also displayed bimodality within meiofauna size ranges, which in most stations was due to the relative proportions of nematodes and other meiofauna taxa. Re-analysis of data from sites in the UK, South Africa, and Antarctic showed a similar bimodality in the adult species body size distribution within the meiofauna size range. Macrofaunal production estimated by the size spectrum equation was very similar to the results of Brey90 empirical equation. However, these production values were much lower than those calculated by Brey01. Different individual dry-to-wet conversion ratios, temperature deviation, and macrofauna taxonomic composition might be responsible for the between-model differences. The macrofaunal P/B ratios calculated by this equation ranged from 0.3 to 3.4 which were in accordance with values from Northern Hemisphere mid-latitudes. Meiofaunal production estimates will need further empirical support.