| 英文摘要 |
This proposal tempts to explore the feasibility of applying geostatistic techniques on squid stock assessment. The result revealed that the distributions of fishing effort was annually different, and at large the fishing started from 45oS on the slope, spread over the southern Patagonian Shelf subsequently depending on the squid concentration, and finally turn to north before finishing the season. The most dispersed fishing found in 1996 when most of the potential blocks were occupied, and the least dispersed years were 2001~2003. Observing from fortnight distribution pattern, the fishing sites spread out from very beginning, reaching a maximum to the seventh fortnight period (mid April). Within year, the individual values of CPUE frequencies were generally normally distributed. Time serial distribution of CPUE indicated that the first half of CPUE values of 1999 and 2000 significantly out numbered the rest of years. Spatial autocorrelation analysis exhibited that larger proximity caused correlation index of Moran’s I value down indicating spatial correlation was distance related. The relationship between annually CPUE and Moran’s I was binominal distributed so that the relationship could not be confirmed when Moran’s I less than 0.5. The high CPUE was related to low SST and high water temperature, and no CPUE relationships were confirm when proximity was greater than 9-cells units. The semi-variograms indicated that the small range of 1999, 2000 and 2002 indicating a clump distribution, while the large range of 1994 and 1997 exhibiting a bulk distribution. In conclusion, utilization of geostatistic technique on analyzing geo-referenced catch data is promising to get valuable stock assessment information. |