58 



Fishery Bulletin 90(1). 1992 



Table 2 



Environmental measurements compiled for each 5° latitude-longitude block.* 



1 Sea Current (N.. Winter)— Average northern component (in knots) of the surface water current in winter (Innis et al. 1979; their 

 fig. 2.2). 



2 Sea Current (W., Winter)— Average western component (in knots) of the surface water current in winter (Innis et al. 1979; fig. 2.3). 



3 Water Depth— Average sea depth (in m) (Bartholomew 1975; fig. 122). 



4 Solar Insolation (Jan.)— Average incoming solar radiation for January (in gm. ■ cal/cm'; Brunt 1934; table 2). 



5 Solar Insolation (Annual)— Average annual incoming solar radiation in gm. ■ cal/cm-; Brunt 1934; table 2). 



6 Sea Surface Temp. (Jan.)— Average January sea surface temperature (in °C; Robinson 1976: fig. 2 north of 5°S; Wyrtki 1974: 

 fig. 2 south of 5°S). 



7 Sea Surface Temp. (July)— Average July sea surface temperature (in °C; Robinson 1976: fig. 74 north of 5°S; Wyrtki 1974: fig. 

 8 south of 5°S). 



8 Sea Surface Temp. (Ann. Var.)— Average annual sea surface temperature variation (in °C; Robinson 1976: fig. 148 north of 5°S; 

 Wyrtki 1974: fig. 26 south of 5°S). 



9 Oxygen Min. Layer (Depth)— Annual mean depth (in m) of the absolute oxygen minimum surface with respect to the vertical (Levitus 

 1982: fig. 52). 



10 Surface Salinity— Average salinity ("Ain) of surface sea water (Levitus 1982: microfiche F-02, frames 2-5). 



1 1 Thermocline Depth (Winter)— Mean depths (in m) to the top of the thermocline for January, February, and March (Robinson 1976: 

 figs. 12, 24, and 36 north of 5°S; Cromwell 1958: fig. la south of 5°S). 



12 ThermocHne Depth (Summer)— Mean depths (in m) to the top of the thermocline for July, August, and September (Robinson 1976: 

 figs. 84, 96, and 108 north of 5°S; Cromwell 1958: fig. Ic south of 5°S). 



13 Surface Dissolved Oxygen— Annual mean dissolved oxygen (mL/L) of surface sea water (Levitus 1982: microfiche F-03, frames 2-5). 



'Abbreviations: Ann. Var. = Annual variation; Jan. = January; Min. = Minimum; N. = North; Temp. = Temperature; W. = West. 



As an example of the Mantel procedure, consider the 

 25 blocks for which two or more specimens were 

 available (Fig. 1). The geographic distances (in nautical 

 miles) between each pair of the 25 blocks (300 pairs 

 total) are computed. We then obtain the mean value 

 for a given morphological character for each block; con- 

 sider a character with large mean values in northern 

 blocks, a gradual change as one proceeds south, and 

 the smallest means in the most southerly blocks. We 

 calculate the absolute character difference for each pair 

 of blocks (300 difference values); in general, for this 

 hypothetical case, close blocks geographically exhibit 

 small differences in character means, while blocks far 

 apart (e.g., a northern and a southern block) have the 

 largest morphological differences. We and the Mantel 

 test would identify this morphological character as hav- 

 ing a strong regional pattern. We also compare recip- 

 rocals of geographic distance for each block pair with 

 corresponding morphological differences; this approach 

 indicates whether, in general, geographically close 

 blocks also are similar morphologically (a case of local 

 geographic patterning). The examplar morphological 

 character, thus, would be identified as displaying a 

 strong local pattern (in addition to the strong regional 

 pattern). In general, a character showing a regional 

 pattern (as we have defined it) also will exhibit a local 

 pattern, but the reverse is not necessarily true. For in- 

 stance, if the morphological character was large in both 

 the north and south, was small for blocks in the middle, 



and had gradual changes between adjacent blocks, it 

 would have a strong local pattern but no regional pat- 

 tern (because many distant blocks are nearly identical 

 morphologically). Detailed computational examples of 

 the Mantel test can be found in Douglas and Endler 

 (1982), Schnell et al. (1985b), and Manley (1985). 



We also computed matrix correlations (Sneath and 

 Sokal 1973) between character differences and the 

 associated geographic distances or reciprocals of dis- 

 tances between localities. The significance of these 

 coefficients cannot, however, be tested in the conven- 

 tional way, because all pairs of localities were used and 

 these are not statistically independent. However, the 

 resulting values are useful as descriptive statistics in- 

 dicating the degree of association of difference values. 



Morphological-environmental covariation 



Relatively little is knowm about the relationship (if any) 

 of geographic variation in morphological characteristics 

 of S. longirostris to differences in the environment. 

 Therefore, as an initial exploratory analysis of covari- 

 ation, we have calculated product-moment correla- 

 tions of block means for morphological characters with 

 environmental variables. Data were available for 13 

 environmental variables for the eastern tropical Pacific 

 Ocean (Table 2). We also used UPGMA to summarize 

 associations among these environmental variables 

 for 51 blocks with specimens of S. longirostris or 



