Comparison of Plankton Catch by Three Light-Trap Designs in

by RF Shaw · 2007 · Cited by 2 — The ichthyoplankton catch and zooplankton biomass estimates of three light-trap designs-cylindrical, quatrefoil, and rectangular-were compared over three

64 KB – 11 Pages

PAGE – 1 ============
Gulf of Mexico Science /ˆ5-#&Number 224˜˛ !ˆ˝˚˙##˘ˇ#!˘˜!!$˝!ˇ%Louisiana State University˜˜ Louisiana State University˜˜˝#University of Texas at Austin( ˙NOAA Fisheries!Old Dominion Universityet al./-3/,,/%%*4*/./2+4)=15*,”53-&/-3<*24/5()2&$$&3,(*4/--5.*49&$$&04&5,&8*$$*&.$&#54)/2*:&%*4,(*4/--5.*49-"4,&"3/.4"$4/3)5"2/-7&,-&%5˛&$/--&.%&*4"4*/.˚)"7/0&/,4;0+&2/,%9"2//+&2/-0"2*3,".+4/."4&*()42"&3*(./24)&25,&8*$/5,&8*$$*&.$˛&42*&6&2=15*,"53-&/-/3 PAGE - 2 ============ Gulf of Mexico Science, 2007(2), pp. 109-118 Comparison of Plankton Catch by Three Light-Trap Designs 1n the Northern Gulf of Mexico RICHARD F. SHAw, JosEPH S. CoPE, G. JoAN HoLT, ANDREAS R6PKE, SIMON R. THORROLD, jAMES G. DITTY, TALAT W. FAROOQI, AND JAY R. RooKER The ichthyoplankton catch and zooplankton biomass estimates of three light-trap designs-cylindrical, quatrefoil, and rectangular-were compared over three utive nights at an offshore petroleum platform in the northern Gulf of Mexico. The quatrefoil light trap had higher fish and zooplankton abundance estimates than the other two designs. Categorical analysis of the two abundant fish taxa, Opisthouema oglimmz and Anchoa spp., indicated that catch by the quatrefoil and rectangular traps was similar, capturing more larvae than juveniles and more 0. oglimtm than Anchoa spp. relative to cylindrical trap catch. Across all fish species, the quatrefoil captured a greater percentage of larvae. Other ontogenetic and species-specific differences were noted among the light-trap designs. Samples from vertical plankton tows mated larger size classes compared to light-trap catch. Light-trap catch per unit effort (CPUE) declined through the night, especially for the quatrefoil, and increased with depth. In contrast, the percentage of larvae captured across all traps increased through the night and decreased with depth, indicating that CPUE was related mostly to juvenile catch. The percentage of larvae also decreased with increasing water cunent speed. All planktonic sampling gears have biases in their collection of taxa and age classes. These sampling gears can be divided into two broad categories based upon their method of capture, active or passive (von Brandt, 1984), with some gear exhibiting characteristics of both (Rooker et a!., 1996). Active gears, like towed nets and pumps, force water through mesh, functionally sieving organisms from the water. In contrast, passive gears rely upon other factors (e.g., natural water currents to bring plankton to stationary nets or the behavior of photopositive organisms to voluntarily enter light traps), which minimize avoidance and extrusion associated with active gears (Barkley, 1972; Hernroth, 1987). However, the effective volume of water sampled by light traps is often difficult to estimate and is dependent on a suite of environmental factors, including current speed, turbidity, and the ambient light field (Anderson et a!., 2002; Lindquist and Shaw, 2005). Poorly understood species-and ontogenetic-specific behaviors also become increasingly important. Light-trap comparisons with active gears dominate comparative studies (Gregory and Powles, 1988), although comparisons with other passive gears are becoming more common (Choat et a!., 1993; Hernandez and Lindquist, 1999; Hernandez and Shaw, 2003). These recent comparative studies have recommended that passive gears should be used in conjunction with active gears to better encompass the full trum of species composition, size ranges, and developmental stages. Light-trap sampling gained popularity after Hungerford eta!. (1955) mentioned their ability to sample early life stages of fishes. Light traps are typically deployed in structurally complex habitats that preclude sampling with al gears (Kawaguchi et a!., 1986; Conrow et a!., 1990; Hernandez and Shaw, 2003), although they can also effectively sample pelagic ments (Thorrold, 1992; Jones, 2006). In tion, data are accumulating on the species-and stage-specific selectivity of light traps (Gregory and Powles, 1988; Choat et a!., 1993; Bickford and Schiel, 1999; Hernandez and Shaw, 2003). For example, the larger larval and juvenile stages of a number of taxonomic groupings such as pomacentrids, clupeiforms, scombrids, and rangids appear to always be well represented in light-trap samples (Doherty, 1987; Thorrold, 1993; Hernandez and Shaw, 2003). However, sampling efficiency among the various light-trap designs still remains relatively unknown nandez and Lindquist, 1999;Jones, 2006). This is partially a result of there being several light-trap designs and light sources available. Furthermore, a given design is often modified so that no standardized, design-specific template exists, making within-and across-gear or across-habitat comparisons difficult. In order to better evaluate research findings from light-trap studies and among light traps, a comparison of differences in catch rate and in taxonomic and ontogenetic composition, as well as physical factors affecting these parameters, is necessary (Lindquist and Shaw, 2005). The objective of this study is to © 2007 by the 1-'Iarine Environmental Sciences Consortium of Alabama 1Shaw et al.: Comparison of Plankton Catch by Three Light-Trap Designs in the NPublished by The Aquila Digital Community, 2007 PAGE - 3 ============ 110 GULF OF MEXICO SCIENCE, 2007, VOL. 25(2) explore such catch characteristics among three popular light-trap designs. METHODS Collection.-Sampling was conducted at Mobil's West Cameron 71D oil platform located off western Louisiana in 11 m of water (29°37'18"N, 93°l0'32"W). Because fish ment may be coupled with lunar cycles sford and Finn, 1997), samples were collected over three consecutive nights, 21 July-24 July 1994, during a full moon phase. Logistical and personnel constraints prevented us from tionally sampling during the new moon period. All sampling commenced and terminated at least 1 hr after sunset and before sunrise, respectively. Three light trap designs-a cylindrical trap (an acrylic model with the removable bottom catch cup replaced by a small conical plankton net with 235-!-lm mesh and cod end; Riley and Holt, 1993), a cloverleaf-shaped quatrefoil dez and Shaw, 2003), and a 3-chambered rectangular trap (modified from Doherty, 1987 and described by Ropke et a!., 1999)-were deployed consecutively along a guideline tached to the bottom within the platform structure. Sampling depths included 1-m face, all three nights) and 8.5-m depths (Night 3 only). A set of samples was defined as one 15-min sample by each of the three light traps. Trap order was randomized within each set. In addition, six vertical (8.5 m to surface; volume filtered = 2.4 m3) plankton tow samples [202-1-lm mesh, 60-cm diameter net, density or catch-per-unit-effort (CPUE) = fish · m - 3 and zooplankton dry weight biomass = mg · m - 3] were collected during Night 2. All fish in each sample were identified and measured. Each fish was assigned a life-history stage, larva or juvenile, as defined by Al1lstrom et a!. (1976), based on standard length. The plankton samples, ing fish, were then dried to constant weight (± 0.01 g) for an estimate of zooplankton biomass (Lovegrove, 1966). Water temperature (C) and salinity (ppt) were measured at sample depth with a Beckman Industrial electrodeless induction salinometer (Model RS5-3). Water current (em · s-1) was measured at sample depth with a vVhitney PVM-2 portable flow velocity meter. Suspended sediment load (dry weight in mg · liter-\ an approximate for turbidity, and percentage of organics within the suspended load were determined from filtered water ples taken at the surface (APHA, 1976). synthetically active radiation (PAR) at about 1-m depth and scalar (ambient) irracliance or refer-ence PAR (Eo PAR) at the platform sampling deck was measured (microeinsteins · m - 2 Ł s-I, where 1 1-1E · m - 2 · s -I = 51.2 lux or lumens · m - 2; Valiela, 1984) using a Biospherical ments PNF-300 profiling natural fluorometer. Trap designs.-The three light trap designs had unique physical characteristics (Table 1). All designs were similar in total height and in entrance gap width. All were constructed from transparent acrylic plastic and had four es, which were beveled toward the trap interior. All entrance slots were vertical in the cylindrical and quatrefoil traps, whereas the rectangular trap had two vertical and two horizontal slots. The rectangular trap had the greatest internal surface area and volume, and the quatrefoil had the greatest internal entrance area (i.e., the narrowest component of the entrance). The cylindrical trap was lowest in these three surements. Because the entire lateral surface of the quatrefoil is beveled into a four-leaf-clover design, the quatrefoil external entrance area (i.e., the widest component of the entrance) was nearly an order of magnitude larger than the other designs. The cylindrical and quatrefoil traps were single-chamber designs, whereas the rectangular trap consisted of three chambers, each separated by bevels that led progressively away from the trap entrance. The cylindrical and quatrefoil traps were both lit with the same bright halogen bulb. The rectangular trap was lit with three fluorescent bulbs, one in each chamber; the bulbs were turned on and off sequentially so that organisms were drawn toward and accumulated within the innermost chamber. The rectangular trap was specifically designed for remote sampling over a number of clays, which necessitated the use of an efficient, battery-powered light source; hence, fluorescent lights were used. Statistical analyses.-CPUE (number of fish · min-1), zooplankton dry weight biomass (mg · min -I), and percentage of larvae (number of larvae/total number of larval and juvenile fish in sample) for surface light-trap catches were analyzed with stepwise multiple regression cedures (SAS, 1989). Light-trap design was included as a qualitative variable using the quatrefoil as the reference design. Turbidity and current speed were entered as quantitative variables. Models also included night and hour, as well as interaction and orthogonal polynomial terms that could be important in explaining the response variance. Similar models were structed with data collected only during Night 3 to include the effect of depth (1 m vs 8.5 m) to 2Gulf of Mexico Science, Vol. 25 [2007], No. 2, Art. 2https://aquila.usm.edu/goms/vol25/iss2/2DOI: 10.18785/goms.2502.02 PAGE - 4 ============ SHAW ET AL.-PLANKTON CATCH BY THREE LIGHT-TRAP DESIGNS 111 TABLE l. Measurements and characteristics of the three light-trap designs used during the study. Trap height includes the conical plankton net with cod end at the bottom of the cylindrical and quatrefoil traps used to collect the draining sample when the traps are removed from the water. The cylindrical and quatrefoil traps used the same light source. Trap design Total trap height (em) Average entrance gap (em) Entrance surface area (cm2) Internal External Internal surface area (cm2 X 103) Internal volume (cm3 X 103) Net mesh size (l.tm) Cylindrical 101.5 0.8 57.6 864.0 8.6 35.0 235 Quatrefoil 144.8 1.0 177.8 5,597.1 13.0 69.0 202 Rectangular 129.5 1.3 100.0 640.0 17.1 107.7 500 Light source One 12-volt halogen fishing Three 6-watt, 6-volt, fluorescent tubes (Cool\1\Thite GE F6T5); 295 initial and 235 mean lumens output; 4,100 K color light (Brinkmann Starfire II); 250,000 candle power output = 2,677,824 lumens; 2,000-3,000 K color temperature Citation Riley and Holt (1993) investigate possible depth stratification or die! vertical migration (Richards et al., 1996); ity was omitted from the Night 3 analyses, because it was not measured at depth. Candidate models were evaluated with residual, influential, partial, and collinearity diagnostics (Belsley et al., 1980; Neter et al., 1996). The CPUE and zooplankton biomass estimates were natural log (loge) transformed before the analyses. age of larvae was transformed by the following equation: 1( . JX . (X+f) 2 arcsm V arcsm V , where x = the number oflarvae and n = the total number of fish in the sample. This tion is desirable if values contain many small and large proportions (Zar, 1984). To test for equal intercepts between the light-trap designs, a studentized t-test was calculated as t* = (bi -bj)/s{bi-bj), where s2{bi-b)= s2{bil + s2{bi) -2s{bi,bj); bi is the estimated intercept of trap i, s2{bd is the variance of bi, and s{bi ,bj) is the covariance between bi and bj (Neter et al., 1996). The Bray-Curtis similarity coefficient (Boesch, 1977) was calculated for each pair of light traps to evaluate the overlap of fish taxa caught by the traps. The coefficient of variation (CV = 100 X standard deviation/mean) for CPUE and mass was also calculated for each light trap to compare variability across trap designs. tion coefficients (r; Zar, 1984) were calculated between CPUE and biomass and between sus-temperature Hernandez Ropke et al. (1999) and Shaw (2003) pended solids and percentage of organics in the solids to determine the strength and direction of their relationships. Life history stage of dominant fish taxa caught by the different light traps was compared with a loglinear model (SAS, 1996). Catch by vertical plankton tows was not included in the analysis, because the net obtained only larval stages. This categorical analysis is specifically designed for count data and tends to be conservative. A parametric test was undesirable because of the low counts in some of our cells. This categorical (stage) model was preferred over a traditional length-frequency analysis because we were also interested in differences among other factor levels, i.e., trap design and species. The gency table consisted of marginal counts summed over all sets and included light-trap design, life history stage, and species as main effects. The appropriateness of the inclusion of interaction terms in the model was evaluated by comparison of the deviances generated by various models (Agresti, 1996). Only Opisthonema oglinum (Atlantic thread herring) and Anchoa spp. (anchovies) had sufficient numbers to compare life history stages among the light-trap designs. Although two species of Anchoa were identified, these were combined into a single Anchoa grouping for this analysis, resulting in a 2 X 2 X 3 contingency table. In the initial analysis, the cylindrical trap produced large adjusted Pearson residuals (3.40). One set taken at 8.5-m depth on Night 3 was excluded because of an unusually high catch of Anchoa juveniles by the cylindrical trap. 3Shaw et al.: Comparison of Plankton Catch by Three Light-Trap Designs in the NPublished by The Aquila Digital Community, 2007 PAGE - 5 ============ 112 GULF OF MEXICO SCIENCE, 2007, VOL. 25(2) TABLE 2. Mean values of physical parameters measured at 1-m depth over the three nights of the study, with the sample size and the standard deviation in parentheses. Ambient irradiance and photosynthetically active radiation (PAR) were measured on Night 1 only; ambient irradiance was measured at the platform's sampling deck. Night 1 Night 2 Night 3 Temperature (C) Salinity (ppt) 30.02 (18; 0.16) 21.23 (18; 0.26) 29.76 (24; 0.12) 25.13 (24; 0.35) 12 (72; 4) 9.95 (5; 3.45) 28.70 (5; 5.51) 29.91 (12; 0.09) 28.02 (12; 0.15) 4 (36; 2) 8.47 (4; 4.26) 40.53 (4; 7.69) Current speed (em · s -l) Suspended load (mg · liter-1) Percent organics Ambient irradiance (!1E · m -2 · s -I) PAR (!1E · m-2 Ł s-1) 20 (53; 5) 12.37 ( 4; 1.24) 29.59 (4; 0.53) 0.106 (6; 0.039) 0.056 (12; 0.043) Nearly half (22/53) of all Anchoa caught by the cylindrical trap were obtained in this one sample; all were Anchoa nasuta and all but one were juveniles. Elimination of this set resulted in a homogeneous association model that explained nearly all of the variance (deviance = 0.63, P > 0.25). Thus, results were interpreted through estimated conditional odds ratios (Agresti, 1996). REsuLTS Over the course of the 3-d study, surface salinity increased at the sample site, while current speed and suspended load decreased (Table 2). At the platform’s s;1mpling deck, the mean ambient irradiance was 0.106 f.!E · m-2 Ł s-1 ± 0.039 SD (5.43lumens · m-2 ± 2.01), whereas at about 1-m depth, the mean PAR was 0.056 f.!E · m-2 Ł s-1 ± 0.043 (2.87 lumens · m -2 ± 2.18). Twenty samples (300 min total sampling time) were collected by each light-trap design. A total of 477 larval and juvenile fish representing 13 families were identified (Table 3). The gular trap captured a total of three families. Nine families were captured by the cylindrical trap, with three families (Biennidae, Stromateidae, and Bothidae) collected exclusively by that design. Ten families were captured by the quatrefoil, with four families (Exocoetidae, Gobiidae, Microdesmidae, and Cynoglossidae) collected exclusively by that design. Individuals from families captured by only one light-trap design, however, were rare. OjJistlwnema oglinum was the most abundant species (39.2% of the total catch), followed by Anchoa spp. (26.0%). Anchoa included at least two species; of those which were identified to species, 31.4% were Anchoa hepsetus (striped anchovy) and 68.6% were A. nasuta (longnose anchovy). Harengula jagnana (scaled sardine) and Chloroscombrus c!nysurus (Atlantic bumper) were also common. These four coastal pelagic taxa accounted for over 86% of the total catch. The stepwise multiple regression on surface CPUE (Table 4A) and zooplankton biomass estimates (Table 4B) resulted in significant models. Quatrefoil CPUE and biomass estimates were significantly greater than the other designs, which were not significantly different from each other (t* = 0.9929, P = 0.1634 and t” = 0.5660, P = 0.2876). However, CPUE decreased for the quatrefoil throughout the night, while CPUE for the other designs remained relatively constant. A curvilinear decrease in CPUE was also noted over the three-night period. Although biomass mates were moderately and positively correlated with CPUE (r = 0.81), the variation in CPUE explained by biomass estimates was largely a linear function of other predictor variables in the model. The biomass model indicated that zooplankton biomass estimates decreased over the three-night period, perhaps due to changes in the physical environment (Table 2). Although the positive relationship between biomass mates and turbidity could be a sampling artifact, because our turbidity estimation technique (dry weight) also measured biomass, a low and negative correlation between suspended solids and percent organics in the solids (r = -0.39) suggested otherwise. The quatrefoil captured a significantly greater percentage of larvae than both the cylindrical and rectangular traps (Table 4C), whereas the percentage caught by the cylindrical and gular traps was not significantly different (t* = 0.6296, P = 0.2664). The percentage of total larvae collected increased throughout the night. However, the rate of increase was greatest on the second night and least on the third night as indicated by the significant quadratic night component. In addition, the percentage of larvae captured was inversely related to current speed. Four samples were collected by each light-trap design at both surface and 8.5-m depths during the third night. Only light-trap design and depth entered the CPUE stepwise multiple regression 4Gulf of Mexico Science, Vol. 25 [2007], No. 2, Art. 2https://aquila.usm.edu/goms/vol25/iss2/2DOI: 10.18785/goms.2502.02

PAGE – 6 ============
SHAW ET AL.-PLANKTON CATCH BY THREE LIGHT-TRAP DESIGNS 113 TAllLE 3. Fish taxa caught by three light-trap designs and vertical plankton net tows. Values indicate total number of fish caught and, within the parentheses, the number of positive catch samples (out of 20 light-trap samples and six net samples) and fish size range (mm), respectively. If only one fish was obtained, only the length of that fish is given. Dagger (t) indicates that fish were not measured clue to damage. For the two taxa that entered into our categorical statistical model, i.e., OjJisthonema og/inwn and Anchoa spp., the numbers within the brackets indicate number caught by stage: L = larvae andJ =juvenile. The Bray-Curtis similarity coefficients for taxa counts are given for each trap, where C = the coefficient between the cylindrical u·ap, Q = quatrefoil, and R = rectangular. Taxon Teleostei Clupeiformes Anchoa spp. Anchoa hejJsetus Anchoa nasuta Total Anchoa Anclwvie//a Harengu/a jaguana OjJisthonema oglinu 111 Synodus foetens HyjJorhamjJ!ms unifasciatus Perciformes (preflexion) Ch/oroscombms ch1)’Slll1/s Sciaenidae C)’Jwscion arenarius Ste//ifer /anceolatus Blenniclae Gobiidae M.icrodesmus longijJinnis Scomberomoms macula/us Pejlrilus alejJidotus Bothidae SymjJhums spp. Total taxa (Families) Total fish Mean fish/ sample (range) Mean CPUE (SD) Mean biomass (SD) Bray-Curtis coefficient CV CPUE/biomass CV indicates coefficient of variation. Cylindrical 2 (2; 4.0-7.5) 9 (6; 6.0-28.5) 42 (5; 18.3-30.5) [9 L: 22J] 1 (28.5) 11 (8; 16.0-27.0) 38 (12; 2.5-26.0) [4 L: 21 Jl 2 (2; 31.0-33.0) 1 (1.7) 2 (1 ; 1.8-2.0) 3 (1; 1.8-2.0) 1 (2.2) 1 (3.5) 1 (2.3) (9.5) 1 (3.0) 1 (2.3) 16 (9) 117 6 (0-37) 0.39 (0.58) 2.8 (2.6) 0.60 Q/0.56 R 148.8/92.6 Quatrefoil 35 (9; 1.7-10.3) 11 ( 4; 2.4-28) 16 (8; 21.0-38.7) [47 L: 14.J] 3 (3; 24.6-33.0) 36 (13; 13.9-32.5) 123 (19; 1.8-29.0) [62 L: 50 J] 4 (2; 30.3-35.5) 3 (3; 27.0-34.0) 19 (5; 0.9-1.7) 45 (12; 1.1-24.0) 2 (1; 1.5-1.6) 4 (2; 2.5-3.0) 3 (3; 3.0-3.5) 1 (2.5) 1 (18.0) 2 (2; 2.7-5.6) 2 (2; 2.7-3.6) 17 (10) 310 16 (2-37) 1.03 (0.63) 25.2 (12.7) 0.60 C/0.63 R 60.6/50.5 Rectangular 2 (2; t) 1(t) 1 (3.3) 7 (4; 10.5-21.0) 1 (27.5) [6 L: 1.J] 1 (25.4) 9 (5; 19.5-35.0) 26 (12; 2.2-24.0) [9 L: 9J] (1.5) 1 (2.3) 10 (3) 50 3 (0-12) 0.17 (0.20) 2.2 (2.3) 0.56 C/0.63 Q 118.6/104.6 Net 3(1;t) 22 (5; 1.8-9.0) l (9.5) [23 L: OJ] 11 (3; 2.5-13.0) [11 L: OJ] 6 (1; 1.3-1.8) 7 (4; 2.3-7.5) 1 (3.5) 1 (2.6) 8 (4) 52 9 (3-23) 3.61 (3.29) 120.7 (18.8) 91.3/15.6 model (Table 5). Overall, CPUE was significantly greater at depth. The quatrefoil CPUE was significantly greater than the other designs, which were not significantly different from each other ( t”‘ = 1.4120, P = 0.0871). Only light-trap design entered the zooplankton biomass model, resulting in an analysis of variance model (r2 = 0.53). Biom.ass estimates for the quatrefoil were significantly greater (P = 0.0004) than for the other two designs, which were not significantly different (t” = 0.0193, P = 0.4924). Percentage of larvae was marginally affected by depth (r2 = 0.14, P = 0.0728), with higher proportions of juveniles, relative to larvae, caught at 8.5 m. Thus, percentage of larvae tended to decrease with depth and to increase through the night (Table 4C), while CPUE exhibited the opposite trends (Tables 4A and 5). This indicates that CPUE was related to juvenile catch. The categorical analysis on 0. oglinum and Anchoa spp. revealed species-and stage-specific differences among light-trap designs. With the effect of trap design held constant, the estimated conditional odds ratio of capturing an 0. oglinum juvenile were 2.8 times that of capturing an Anchoajuvenile (Table 6). Larval and juvenile 0. oglinum were caught in approximately equal numbers (75 larvae : 80 juveniles); however, Anchoa larvae (62) were caught in greater numbers than juveniles (37). With species held 5Shaw et al.: Comparison of Plankton Catch by Three Light-Trap Designs in the NPublished by The Aquila Digital Community, 2007

PAGE – 8 ============
SHAW ET AL.-PLANKTON CATCH BY THREE LIGHT-TRAP DESIGNS 115 TABLE 5. Summary of the depth (l m vs 8.5 m) stepwise multiple regression analysis on natural log-transformed catch per unit effort, third night only (r2 = 0.68). See Table 4 for explanation. Source DF Jvlodel 3 Error 19 Variable Estimate Intercept 0.5378 Depth 0.3525 Cylindrical -0.4118 Rectangular -0.5792 not catch C. chi)’SUrus. Based on all fish taxa (Table 3), the catch between the quatrefoil and rectangular traps was the most similar ( 63%), and catch between the cylindrical and lar traps was the least similar (56%). DISCUSSION Light-trap design produced pronounced ferences in both CPUE and zooplankton biomass estimates. Similar results have been previously reported. Hernandez and Lindquist (1999) and Clavijo et al. (1996) reported that their chambered rectangular light trap was more efficient at capturing zooplankton and plankton than a two-chambered design. Meekan et al. (2001) tested two different sizes of rectangular traps and found that the smaller design captured more reef fish than the larger design. Mean CPUE by each light-trap design in this study falls within ranges reported in the literature. In contrast to catches in tropical habitats (e.g., Choat et al., 1993), the lar-trap catch during the present study is positioned in the lower portion of this range, possibly emphasizing the importance of ral and spatial differences among studies and of variations in light-trap design. Other subtle but important differences were found among the light-trap designs at netic and taxonomic levels. Although the cylin-ss p 2.1580 0.0001 0.9940 SE p (b 0) 0.0940 0.0001 0.0957 0.0016 0.1186 0.0025 0.1144 0.0001 drical and quatrefoil traps caught a similar number of taxa, the cylindrical trap tended to sample mostly juveniles, whereas the quatrefoil was more effective at sampling larvae. ing the two most abundant taxa, the cylindrical trap captured a greater percentage of Anchoa spp., whereas the quatrefoil and rectangular traps collected a greater percentage of 0. oglinum specimens. The rectangular trap captured no C. cht)>surus, and the quatrefoil caught the only juveniles. The rectangular trap collected only juvenile H. jaguana, whereas tl1e other two designs also obtained larval stages. Moreover, the similarity over all taxa was about 60% between traps. Several factors may have contributed to the differences in catch among the light-trap signs. The rounded shape of the cylindrical and quatrefoil traps may facilitate entry, because movement along the lateral external smface of the trap eventually leads to an entrance. The entire lateral surface of the quatrefoil collection chamber is curved inward to the entrances, providing more opportunities for an organism to orient itself with the entrance. If escapement is a function of random movements, then a trap with a large internal surface area or volume may retard escapement (Munro, 1974). However, the rectangular trap, which has both the largest total internal surface area and volume, had the lowest catch rate. Any retention advantage, however, TABLE 6. Results from the loglinear regression analysis on life history stage Q =juvenile, L =larva), species (A= Anchoaspp., 0 = Opisthonemn oglinum), and trap design (C =cylindrical, R =rectangular, Q =quatrefoil). For each parameter, the estimated conditional odds ratios (i’J), the Wale! x2 statistic, and the probability that 8 = l is given. To ease interpretation, odds ratios are calculated so that S> l. Palatlteter estimaled x’ Stage (J/L) * species (0/ A) 2.76 11.07 0.0009 Stage (J/L) * trap design (C/Q) 7.66 28.41 < 0.0001 Stage (J/L) * trap design (R/Q) 1.06 0.02 0.8907 Stage (J/L) * trap design (C/R) 7.20 13.05 0.0003 Species (A/0) * trap design (C/Q) 3.61 12.96 0.0003 Species (A/0) * trap design (Q/R) 1.38 0.45 0.5039 Species (A/0) * trap design (C/R) 4.98 8.38 0.0038 7Shaw et al.: Comparison of Plankton Catch by Three Light-Trap Designs in the NPublished by The Aquila Digital Community, 2007 PAGE - 9 ============ 116 GULF OF MEXICO SCIENCE, 2007, VOL. 25(2) may have been minimized by the short duration of the deployments (Fogarty and Addison, 1997). Organisms typically show a higher affinity toward greater light intensities and rnay respond differently to various wavelengths of light ted by different types of light bulbs (Kawamoto, 1959; Blaxter, 1968; Gehrke, 1994). In addition, the oil platform studied was lit for on-site and navigational safety reasons and, therefore, duced a competitive light field, which was compounded by a full moon phase. However, our on-site measurements of the ambient light field and PAR at 1-m depth were low, i.e., 5.43 and 2.87 lumens · m -2, respectively, compared to the lowest trap light source, i.e., 295 initial lumens with a mean of 235 lumens. In addition, our ambient irradiance and PAR measurements at a single platform were very similar to those measured by Keenan et al. (2007; mean ambient irradiance = 5.73 lumens · m - 2 ± 22.6 and 1-m depth PAR = 1.67 lumens · m-2 ± 1.28) for three of their platforms-Pr2, Gdk, and Yke-at the edge of the South Timbalier 151 complex, which consisted of a cluster of six fixed-leg platforms. Still, the rectangular trap's much dimmer fluorescent lights (235 vs 2,677,824 lumens) may have been at a distinct tage, although no statistically significant ences were found in total CPUE or biomass between the cylindrical and rectangular traps. Water turbidity further reduced the effective sampling radius of the light field for all designs. Lindquist and Shaw (2005) found that CPUE of larval and juvenile fishes decreased with ing water turbidity. Fisher and Bellwood (2002) used a light-trap design with baffles and showed that light intensity decreased rapidly with tance due to attenuation and increasing area of illumination. The variety of light-trap shapes, entrance configurations, and materials cast different reflected and scattered light fields. These differing light projections may influence the behavioral responses of organisms at trophic, ontogenetic, and species-specific levels greaves et al., 1993; Higgs and Fuiman, 1996; Hernandez and Lindquist, 1999; Lindquist et al., 2001). Consequently, each light trap may have also produced a different density gradient of photopositive zooplankton in response to the different light fields, which can influence the feeding reflex in fish (Keenan et al., 2003) and may also explain why some fish species that are not particularly photopositive are captured by light traps. If fish are attracted to zooplankton prey items that are, in turn, attracted to the light given off by the traps, then traps with higher zooplankton biomass estimates could have a higher CPUE. This is supported by the finding that both zooplankton biomass and CPUE decreased over our 3-d study. Therefore, ences in light intensity, light quality, light projection, and food density are likely to interact and to influence catch. One issue concerning light-trap comparisons with conventional towed gears that is not yet fully resolved is the effective size range of fish caught by light traps. Gregory and Powles (1988) found that light traps were more effective at sampling smaller larvae and larger juvenile fish than was a Miller high-speed sampler. Hernandez and Shaw (2003) found that light traps were very good at collecting larger size classes, but also ly overlapped with smaller size distributions from passive, nocturnal plankton net collections. Other studies have noted that light traps select for larger individuals only. In the present study, the vertical plankton tows, relative to the trap catches, did not collect larger size classes, indicating either nocturnal net avoidance or that these size classes were more rare and, therefore, not adequately sampled by the small volume of water filtered (2.4 m3). In addition, the tiveness of capturing larval fish varied among the light-trap designs tested. Gregory and Powles attributed their finding to extrusion of smaller larvae by the towed gear. Choat et al. (1993) and Brogan (1994) explained differences between studies as taxon-and habitat-specific variation. However, differences in the current regime among the study areas cannot be ignored, especially when dealing with passive planktonic gears and their sampling volumes. Gregory and Powles (1985) stated the flow in their study area was negligible, whereas Choat et al. (1993) cited a flow rate of 15 em· s-\ which is comparable to that observed during the present study (mean = 12 on · s -1) and to those generally observed in our area, i.e., 5-15 em · s - 1 (Shaw et al., 1985). During this study, some smaller, less-competent larvae may have been advected past the trap before they could enter. This hypothesis was proposed by Thorrold (1992) to explain the higher catch of drifting light traps relative to tethered traps. Further credence is given by the significant and negative relationship between current speed and percentage of larvae captured in the present study. However, Lindquist and Shaw (2005) found that not only clicl CPUE decrease with increasing current speed, but so did fish size, to the point where at the highest speeds, only preflexion larvae were collected, presumably by passive filtration. Even microscale currents immediately around or within each trap design may differ, acting to enhance or retard entry into or retention within the trap. Such small-scale turbulence immediately around light 8Gulf of Mexico Science, Vol. 25 [2007], No. 2, Art. 2https://aquila.usm.edu/goms/vol25/iss2/2DOI: 10.18785/goms.2502.02 PAGE - 10 ============ SHAW ET AL.-PLANKTON CATCH BY THREE LIGHT-TRAP DESIGNS 117 traps may lead to decreased CPUE at current speeds greater than 30 em · s -l (Lindquist and Shaw, 2005). For example, in this study, the cylindrical trap occasionally began to spin in the current while fishing, which would undoubtedly affect ingress. One statistical measure of precision or ducibility that allows for cmnparison of catch among the light-trap designs is the coefficient of variation (CV; Table 3), where lower values indicate higher precision and statistical power. The quatrefoil's CV for both zooplankton biomass and CPUE was nearly half that of the cylindrical and rectangular designs. The gular trap's CV for CPUE was slightly lower than the cylindrical trap's CV, but it was slightly higher for biomass. Given the quatrefoil's overall performance, it was selected for subsequent use in a number of platform-based studies dez and Shaw, 2003; Lindquist and Shaw, 2005; and studies cited within). The findings from this trap-comparison study are based on sampling at a structurally complex environment during a short-term period and may not apply to other environments. Confirmation at other locations over larger time scales may be necessary to enable researchers to objectively choose the most appropriate design for their particular environment, current regime, and planktonic community. AcKNOWLEDGMENTs Special thanks to Peter Doherty for constructive comments on an early draft of the manuscript. The Louisiana Sea Grant College Program (NOAA) funded this study. James Geagan and Brain Marx at the Department of Experimental Statistics, ana State University, provided statistical advice.Jim Lee of the Coastal Ecology Institute, Louisiana State University, analyzed water samples for suspended sediment and percent organic mates. Numerous student workers assisted in sample processing. Special thanks are given to Mobil, especially the crew of West Cameron 71D, for logistical support, as well as their hospitality. This is VIMS Contribution 2919. LITERATURE CITED AGRESTI, A. 1996. An introduction to categorical data analysis. John Wiley & Sons, Inc., New York. AHLSTROM, E. H., J. L. BUTLER, AND B. Y. SUMIDA. 1976. Pelagic stromateiod fishes (Pisces, Perciformes) of the eastern Pacific: kinds, distributions, and early life histories and observations on five of these from the northwest Atlantic. Bull. Mar. Sci. 26(3):285-402. ANDERSON, T. W., C. T. BARTELS, M.A. HIXON, E. BARTELS, M. H. CARR, AND J. M. SHENKER. 2002. Current velocity and catch efficiency in sampling settlement-stage larvae of coral reef fishes. Fish. Bull. 100:404-413. APHA [AMEIUCAN Punuc HEALTH AssociATION, &\IEIUCAN WATER VVORKS AssOCIATION, AND WATER POLLUTION CoNTROL FEDERATION]. 1976. Standard methods for the examination of water and wastewater. American Public Health Association, Washington, DC. BARKLEY, R. A. 1972. Selectivity of towed-net samplers. Fish. Bull. 70(3) :799-820. HELSLEY, D. A., E. KuH, AND R. E. WELSCH. 1980. Regression diagnostics . .John Wiley & Sons, Inc., New York. BLAXTER, J. H. S. 1968. Visual thresholds and spectral sensitivity of herring larvae . .J. Exp. Bioi. 48:39-53. BoESCH, D. F. 1977. Application of numerical tion in ecological investigations of water pollution. Special Scientific Report 77, VIMS (EPA-600/3-77-033). BROGAN, M. W. 1994. Two methods of sampling fish larvae over reefs: a comparison from the Gulf of California. Mar. Bioi. 118:33-44. CHOAT, J. H., R. H. DOHER1Y, B. A. KERIUGAN, AND J. M. LEIS. 1993. Larvae and pelagic young of coral reef fishes: comparison of tlwee towed nets, a purse seine, and two aggregation devices. Fish. Bull. 91:195-209. CrAVIJO, I. E., D. G. LINDQUIST, M. E. WHITTAKER, AND F.J. HERNANDEZ. 1996. Comparison of zooplankton from a mangrove lagoon and a coral reef: preliminary results fi·om light traps, p. 90. In: Proceedings of the 27th meeting of the of Marine Laboratories of the Caribbean . .J. F. Battey (eel.). University of the Virgin Islands, St. Thomas. CONROW, R., A. V. ZALE, AND R. W. GREGORY. 1990. Distributions and abundances of early life stages of fishes in a Florida lake dominated by aquatic macrophytes. Trans. Am. Fish. Soc. 119:521-528. DoHERIT, P. J. 1987. Light-traps: selective but useful devices for quantifYing the distributions and dances oflarval fishes. Bull. Mar. Sci. 41 (2) :423-431. FISHER, R., AND D. R. BELLWOOD. 2002. A light trap design for stratum-specific sampling of reef fish larvae. J. Exp. Mar. Bioi. Ecol. 269:27-37. FoGARIT, M. J., AND J. T. AnmsoN. 1997. Modelling capture processes in individual traps: entry, ment and soak time. ICES J. Mar. Sci. 54:193-205. GEHRKE, P. C. 1994. Influence of light intensity and wavelength on phototactic behaviour of larval silver perch Bidyanus bidyanus and golden perch 11iacquaria ambigua and the effectiveness of light traps. J. Fish Bioi. 44:741-751. GREGORY, R. S., AND P. M. PowLES. 1985. Chronology, distribution, and sizes of larval fish sampled by light traps in macrophytic Chemung Lake. Can . .J. Zoo!. 63:2569-2577. ---,AND---. 1988. Relative selectivities of ivlillcr high-speed samplers and light traps for collecting ichthyoplankton. Can. .J. Fish. Aquat. Sci. 45: 993-998. HARGREAVES, P.M., P.J. HERRING, AND J-I. GREENWAY. 1993. The response of tropical Atlantic decapod ceans to artificially lighted trawls. J. Plankton Res. 15(7):835-853. HERNANDEZ. JR., F. J., AND D. G. LINDQUIST. 1999. A comparison of two light-u·ap designs for sampling 9Shaw et al.: Comparison of Plankton Catch by Three Light-Trap Designs in the NPublished by The Aquila Digital Community, 2007 64 KB – 11 Pages