714 



Fishery Bulletin 89(4). 1991 



The monthly sea turtle strandings and 

 shrimp fishing effort were separated into 

 two geographic zones: the upper coast 

 (subareas 17-18) and lower coast (sub- 

 areas 19-21). This was done because the 

 upper coast has a wider continental shelf 

 than the lower, so the distance a dead, 

 sick, or injured sea turtle would have to 

 travel from a particular depth interval to 

 the shore is greater on the upper coast 

 than on the lower (Fig. 1). Due to difficul- 

 ty of access, the small portion of subarea 

 17 east of the Mermentau River in south- 

 western Louisiana was not surveyed for 

 strandings, so approximately 86% of the 

 coastline of the upper coast zone was 

 surveyed for turtle strandings (Table 1). 

 For this reason, we included in the upper 

 coast zone only those turtle strandings 

 that occurred west of the Mermentau 

 River. We could not place a similar 

 boundary restriction on the fishing effort 

 data, so the eastern boundary of subarea 

 17 marked the eastern boundary of the 

 upper coast in this regard. However, 

 strandings and fishing effort were stan- 

 dardized to strandings per linear distance 

 of shoreline and to days fished per unit 

 area, respectively, so the exclusion of 

 strandings east of the Mermentau River 

 should have had little if any effect on our 

 results. 



Our analyses included fishing effort 

 from the six 5-fm intervals between and 

 30 fm (54.9 m) in shrimp statistical sub- 

 areas 17-21. We did not include effort 

 data beyond 30 fm, because only 6% of 

 the shrimping effort on the upper coast 

 and 8% on the lower coast occurred sea- 

 ward of 30 fm during 1986-89 (Table 2). 



Monthly sea turtle strandings within the upper and 

 lower coasts were standardized by dividing them by 

 distance of accessible shoreline (Table 1) in these two 

 zones, respectively, to obtain the monthly turtle strand- 

 ings per 100km (S). We used the amount of surface 

 area within shrimp statistical subareas and 5fm depth 

 intervals, as determined by Patella (1975), to standar- 

 dize monthly fishing effort within the upper and lower 

 coasts by depth interval. The surface area within a 

 depth interval was usually greatest nearshore and 

 decreased seaward in both zones (Fig. 1). For each 5-fm 

 depth interval, monthly fishing effort in the upper and 

 lower coast zones was divided by the surface area of 

 the geographic unit (zone x depth interval) within 

 which the effort occurred, to standardize effort to a 



Table 1 



Extent of accessible shoreline surveyed for sea turtle strandings during 1986-89 

 compared with total shoreline within the upper and lower coasts of north- 

 western Gulf of Mexico 1 . 



1 Derived from measurements made using dividers on National Ocean Service 



(NOAA) nautical charts. 

 -Figure 1; see also Kutkuhn (1962). 

 3 Only the accessible shoreline west of the Mermentau River, Louisiana, was 



surveyed for strandings in subarea 17. 



Table 2 



Distribution of shrimp fishing effort on the upper and lower coasts of north- 

 western Gulf of Mexico by depth, 1986-89. ' 



1 Adapted from data provided by Frank Patella, NMFS Galveston Lab., pers. 

 commun., June 1990. 



2 Shrimp statistical subareas 17 and 18 (Fig. 1; see also Kutkuhn 1962). 

 :l Shrimp statistical subareas 19-21 (Fig. 1; see also Kutkuhn 1962). 



measure of shrimping intensity. Standardized fishing 

 effort per unit area (E) was expressed as days fished 

 per 100km 2 . 



Product-moment correlation analysis requires that 

 the two variables have normal distributions. Neither 

 standardized strandings (S) nor fishing effort (E) were 

 normally distributed, as shown by large departures of 

 their skewness and kurtosis coefficients from zero 

 (Table 3). Therefore, we logarithmically transformed 

 both variables, after adding 1 to each value of S and 

 E (because some values were zero). The logarithmical- 

 ly transformed variables had skewness and kurtosis 

 coefficients closer to zero, thus approaching normal- 

 ity. For each of the 12 combinations of two geographic 

 zones and six depth intervals, product-moment correla- 



