CONTRASTING PATTERNS OF FISH DENSITY AND ...

3 downloads 0 Views 754KB Size Report
Apr 29, 1988 - Sebastes atrovirens. Semicossyphus pulcher. Sphyraena argentea+. Trachurus symmetricus+. Triakis semifasciata. Urolophus hallen'.
BULLETIN OF MARINE SCIENCE, 44(2): 881-892, 1989

CONTRASTING PATTERNS OF FISH DENSITY AND ABUNDANCE AT AN ARTIFICIAL ROCK REEF AND A COBBLE-BOTTOM KELP FOREST Edward E. DeMartini, Dale A. Roberts and Todd W Anderson ABSTRACT We describe the results of an intensive, short-term (November 1986-January 1987) characterization of numerical and biomass abundances offishes at Pendleton Artificial Reef(PAR), a quarry rock reef off southern California. We emphasize area and density effects on fish abundance and discuss thcir implications for monitoring and resource management. PAR is compared with concurrent estimates of fish density and abundance within San Onofre Kelp bed (SDK), a cobble-bottom giant kelp (Macrocystis) forest, 5.5-km upcoast of PAR. An estimated 14,950 resident juvenile-adult fishes weighing 490 kg were present atop the 1.3 ha of rock modules and fringing sand-rock ecotones at PAR; biomass density thus was about 380 kg/ha. The biomass density of resident fishes in the kelp forest was about 325 kg/ha. Numerical density at SDK, however, was less than one-third (3,600 fish/hal of the density of rcsident fishes at PAR (11,500 fish/hal. Since the areal extent of the kelp-cobble habitat (-88 hal was nearly two orders of magnitude greater than the artificial reef, its standing stock of resident fishcs (about 29 MT) was almost 60 times greater than that at PAR. Our results have important implications for the design of reef mitigation studies. Evaluations of fish populations at reefs can be misleading when they are based on density data alone. It is obvious that any assessment of existing or potential biological mitigation provided by an introduced habitat should base its evaluation on abundance, not the density of organisms independent of the quantity of habitat. Another important topic for future research is the relation between fish density and the areal extent of introduced habitat.

Reef fish studies usually assess assemblage structure (Alevizon and Brooks, 1975; Jones and Thompson, 1978) and various types of density data and their patterns (Ebeling et aI., 1980; Kingett and Choat, 1981, and references therein). Rarely do assessments of reef fishes extend beyond density to provide estimates of abundance (Brock, 1954; Bardach, 1959; Randall, 1963; Russell, 1975; 1977; Quast, 1968; Larson and DeMartini, 1984), no doubt due to the time and cost necessary to quantify habitat. Density is an accurate measure of abundance in situations where amount of habitat is constant (Kock, 1982), but does not reflect abundance where the amount of habitat varies temporally or spatially (Jessee et a1., 1985). Direct estimates of abundance are required for evaluations of populations such as those used in mitigation studies where loss of organisms is a result of habitat loss or degradation (Schamberger and Farmer, 1978; Schamberger and Krohn, 1982). When density and abundance differ, these differences should be made explicit. An associated problem is the general lack of data on population size structure (age or length frequency distributions) of reef organisms. Lack of information on size structure precludes estimation of biomass abundance. Yet the standing stock biomass of organisms, particularly harvestable fishes on reefs, is the primary criterion used in monitoring and assessing the actual or potential resource value of reefs (Quast, 1968; Russell, 1977; Walton, 1979). Here we describe the results of an intensive, short-term (fall 1986) characterization of the abundance and population size structure of fishes at Pendleton Artificial Reef (PAR), 6 years after its construction. We emphasize differences between densities and abundances of fishes at PAR and contrast these patterns 881

882

BULLETIN OF MARINE SCIENCE, VOL. 44, NO.2, 1989

with analogous, concurrent estimates at the San Onofre Kelp (SOK) bed, a cobblebottom reef supporting a forest of giant kelp (Macrocyst is pynfera) located 5.5km upcoast of PAR. STUDY AREAS Pendleton Artificial Reef (PAR) was constructed in September 1980 from 10,000 tons of quarry rock at a depth of 13 m, 1.8-km offshore of San Diego County, California, U.S.A. (33°19'N, 117"31'w) (Carter et aI., 1985, p. 92-94). PAR consists of eight rockpiles or "modules" separated from one another by distances of 10-75 m (average 30 m) (Carter et aI., 1985, fig. 1; Jessee et aI., 1985). Jessee et al. (1985) describe the fish fauna at PAR 2-3 years following its construction, and earlier data are summarized by Grant et al. (1982). San Onofre Kelp (SOK) has occupied, on average, a l-km2 area of cobble bottom at depths of 818 m. During this study, SOK occupied approximately 90 ha of substrate, and the grand mean density of Macrocystis at SOK was about 3 plants/l00 m2• The average bottom depth sampled at SOK was 14.5 m. Larson and DeMartini (1984) describe in detail the environment and fish fauna at SOK. METHODS AND MATERIALS General Sampling Methods.-Fish densities at PAR were estimated directly from diver counts. At SOK, we used diver counts on bottom and "cinetransects" in the water column and in the surface canopy to record data (Ebeling et aI., 1980; Larson and DeMartini, 1984). Fish were scored as one of three size-specific maturity stages (juvenile, subadult, adult) on in situ diver counts and in film tallies. Stageswere recognized based on estimated total length (TL) in inches and known size-maturity relations (Larson and DeMartini, 1984). For fish that averaged -15 to 30 em (6 to 12 in), diver accuracy in estimating lengths was ±2.5 em (± I inch; E. DeMartini, unpubl.). Counts were made within belt transects whose dimensions were either fixed, or variable but estimable, depending on site and microhabitat. All data were collected during daylight (0800-1600 h) when average lateral underwater visibility exceeded 3 m (Larson and DeMartini, 1984). PAR. - Transects were stratified among benthic microhabitats (strata). All transects were 3-m wide by 1.5-m high. The length of transects varied among modules and strata but was constant over time at a particular station. "Crest" transects (on rocks atop modules; see Anderson et aI., 1989) ranged from 14-35 m in length, depending on module. "Slope" transects (sides of rock modules) likewise varied from 76-150 m in length; and "ecotone" transects (along the sand-rock interface) ranged from 97-207 m. For each of these three strata, one diver sampled a single transect per module on each survey. All species were tallied during a one-way swim except for crest transects; on the crest, divers tallied all species except Chrornis punctipinnis on an initial swim out along the module crest (Anderson et aI., 1989, fig. 1).On the return swim, only Chrornis was tallied. We also completed one "extralimital" count per module on each survey for select species whose vertical distribution sometimes partly extended> 1.5-m above the substrate. Divers estimated the numbers of each of three species (Anisotremus davidsonii. Cheilotrerna saturn urn, and Chrornis) that hovered in the lower water column above module slopes. These counts were added to the areal extrapolations of transect density when estimating abundance of the respective species. We sampled all strata within each of the eight modules at PAR on each survey. Three single-day surveys were completed (on 6 and 25 November 1986, and on 20 January 1987). A team offive divers was rotated among modules and strata over surveys in order to randomize the potential biases of observers.

SOK. -All bottom transects were 3-m wide by 1.5-m high by 75-m long. Dimensions of midwater (7.7-m depth) and surface canopy (3.I-m depth) transects depended on lateral underwater visibility (Larson and DeMartini, 1984). A total of six bottom transects, two midwater transects, and eight surface canopy transects were completed at a station on each sampling date. We used movie cameras and Super 8 film for cine recording (Ebeling ct aI., 1980). Swimming at a speed of 25 m/min, the 3-min cinetransects were about 75-m long. We sampled seven stations at SOK during October-December 1986. Surface canopy, midwater, and bottom microhabitats were sampled on a single-day survey at each station. All stations could not be sampled on the same day, however. The same five-diver team that conducted the PAR surveys completed 3-8 surveys at each of the seven stations at SOK. Three stations were located in regions of highest existing Macrocystis density (10-13 piants/IOO ml), two stations in regions of intermediate density (2-4 piants/IOO m2), and two stations in kelp-poor cobble « I piant/IOO m2). At both PAR and SOK, we characterized the population size structure of fishes in addition to

DEMARTINI ET AL.: FISH DENSITY-ABUNDANCE

883

RELATIONS AT TWO REEFS

Table I. Estimated surface area represented by each distinguishable microhabitat at (A) Pendleton Artificial Reef and (B) San Onofre Kelp bed (A) Pendleton Artificial Reef Area Microhabitat

Crest Slope Ecotone Total

(B) San Onofre Kelp bed

(m')

% of total

2,401 6,623 3,654 12,678

18.9 52.3 28.8 100.0

Microhabitat

High-density kelp Intermediate kelp Low-density kelp

Area (m')

% of

52,000 292,000 535,000 879,000

5.9 33.2 60.9 100.0

total

estimating their densities. Size structure was described during free-swims: Fishes were tallied by inchclass (TL) and species as randomly encountered. At PAR, each free swim was 20 min (10 min slopeecotone, plus 10 min crest) per module and survey. At SOK, the duration of observation was doubled to 40 min because of relatively low encounter rates. Bottom swims were 40 min. Canopy-midwater swims were divided into two 20-min (midwater, canopy) periods. A total 480 min were spent tallying fish lengths at PAR; 1,120 min were spent at SOK. Habitat Quantification. - The areal extent of rock habitat at PAR was estimated using bathymetric charts of the modules (provided by ECOSystems Management Associates, Inc.). We identified strata by depth contours; the areas enclosed within contours were estimated by planimetry (Table I). We estimated the areal extents of different kelp density regions within SOK based on downlooking sonar charts of kelp density (ECOSystems Management Associates, Inc.). We used the same method to estimate the areas enclosed within kelp density contours. In this manner, we described the total area of SOK by summing over all subareas of various kelp densities (Table I). Data Manipulations.-Data collected at PAR and SOK were processed in a similar fashion to allow comparisons between the two reefs. All transect counts were first converted to volumetric density and standardized (no./ I,000 m') based on known dimensions of the transects. Abundances were then calculated by multiplying density by the area of the corresponding microhabitat, followed by summing these subtotals over microhabitats. Microhabitats at SOK included the three vertical sampling strata (bottom, midwater, and surface canopy) as well as the three horizontal regions of differing Macrocystis density. Extents of the vertical sampling strata were defined by their thicknesses: canopy, 5.3 m; midwater, 7.7 m; and bottom, 1.5 m. At PAR the microhabitats evaluated were the crest, slope, and ecotone strata present on all eight of the modules. PAR strata had a constant (1.5-m) vertical thickness; relatively few fish occurred higher in the upper water column at PAR, and our cxtralimital counts (see General Methods- PAR) accounted for most of these. In all cases, numerical abundances were converted to biomass abundances by multiplying the numbers of individuals of each maturity stage and species by the mean body weight of that stage and species. Body weights were estimated by applying length-weight regressions to length-frequency data (Anderson et aI., 1989). Prior to statistical tests, numerical and biomass densities were expressed on an areal basis (numbers or kg per 1,000 m2) for all microhabitats pooled at each reef. Densities were calculated by dividing the abundance estimate by the total surface area of the respective reef. Areal-based densities were necessary because the kelp density regions at SOK differed in area, and because not all kelp bed stations were sampled on the same date. Variation in estimates was expressed as ± I SE of the estimate. At PAR, estimates were based on 2 degrees of freedom (the number of surveys minus one). Estimates at SOK were based on 32 degrees of freedom (the number of sampling dates minus one at each station, summed over all stations). Statistical Tests. - We compared densities and abundances between PAR and SOK with t-tests, corrected (as necessary) for unequal variances using Satterthwaite's approximation (Bailey, 1981, p. 185). We made formal comparisons for each species that was common and abundant (> I fish/I,OOOm2) at both reefs and for one compound taxon ("total fishes"). We also estimated and qualitatively evaluated the densities and abundances of several additional compound taxa, including total resident fishes. The following transient species (seasonal migrants that secondarily inhabit kelp beds: Feder et aI., 1974) were excluded from the resident category: Sarda chiliensis, Scomber japonicus, Sphyraena argentea, and Trachurus symmetricus. We also qualitatively compared two other compound taxa, "total fishes minus Chromis" (c. punctipinnis: a pomacentrid that numerically dominated fishes at PAR) and "total fishes minus Oxyjulis" (0. cali/ornica: a labrid that dominated fishes at SOK).

884

BULLETIN OF MARINE SCIENCE, VOL. 44, NO.2, 1989

PENDLETON ARTIFICIAL REEF

OTHER

SAN ONOFRE KELP BED

1% TRANSIENTS TRANSIENTS

RESIDENTS

t1)8%

NUMERICAL

OTHER RESIDENTS

1%

·C"ROM'.

DENSITY

18t!)

NUMERICAL ABUNDANCE

'55%

OXYJULIS

450/1,OOOM2

l,200/1,OOOM2

OTHER RESIDENTS

2rw6T\

25%

TRANSIENTS OXYJULIS

81%CHROMIS 15,100

FISH

397,000

FISH

2% TRANSIENTS

~'J'NMS~~~ R'~;~::T~

C"ROM'8

OTHER RESIDENTS

40KG/l,OOOM2

44K G/l.000M2

TRANSIENTS TRANSIENTS

BIOMASS ABUNDANCE

2% -::+::"16%CHROMIS

OXYJULIS

l'J

82% OTHER RESIDENTS 505

OTHER RESIDENTS

KG 38,900

KG

Figure I. Relative magnitudes offish densities and abundances (numbers and biomass) at Pendleton Artificial Reef (PAR) versus San Onofre Kelp bed (SOK) in fa111986. Estimates for total fishes are subdivided into transients and residents, and the latter category into Chromis punctipinnis (at PAR) or OXJjulis caliJarnica (at SOK) and other fishes. The pie diagrams are scaled proportionately only within each PAR-SOK contrast.

RESULTS

Numerical Densities. -At PAR, total densities averaged 1,200 fish/l,OOO m2, 99% of which were resident fishes (Table 2; Fig. 1). Average density dropped to 230/ 1,000 m2 if Chromis is excluded (Table 2; Fig. 1), The density of total fishes at SOK averaged only 450 fish/ 1,000 m2, and resident fishes represented 80% (Table 2; Fig. 1). Excluding Oxyjulis, the mean density of resident fishes was 115/1 ,000 m2 (Table 2; Fig. 1).

DEMARTINI ET AL.: FISH DENSITY·ABUNDANCE

RELATIONS AT TWO REEFS

885

Table 2. Estimated numerical densities for each of the 16 most numerous species of fishes (and total fishes) encountered at PAR in November 1986-January 1987 and the 22 most numerous fishes at SOK during October-December 1986. Values in parentheses are one standard error of the estimate. Key to PAR-SOK comparison: * = 0.05 > P> 0.01; ** = 0.01 > P> 0.001; *** = P < 0.001. The site having the greater density is labelled with the asterisk(s). Transient fishes are labelled with a "+" Numerical density (No.1 1.000 m') PAR

Anisotremus davidsonii Atherinops affinis Brachyistius frenatus Cheilotrema saturnum Chromis punctipinnis Embiotoca jacksoni Girella nigricans Ha/ichoeres semicinctus Heterostichus rostratus Hypsurus caryi Hypsypops rubicundus Medialuna califbrniensis Oxyjulis californica Paralabrax clathratus Paralabrax nebullfer Phanerodon furcatus Rhacochilus toxotes Rhacochilus vacca Sarda chiliensis+ Scomber japonicus+ Semicossyphus pulcher Sphyraena argentea+ Trachurus symmetricus+ Xenistius californiensis Total fishest

14.0 (7.29)t

o

o

15.0 (7.39)t 963.9 (280.1)t 30.2 (5.20)* 2.2 (0.33)* 50.9 (8.45)

o

o 11.4 (1.39)* 12.0 (4.96) 30.0 (2.93) 4.6 (0.75) 3.4 (2.42)

o

0.5 (0.14)* 2.4 (0.96) 0.2 (0.21)

o 38.3 (1.74)*

o 13.0 (13.0)

o 1,193 (266)*

SOK

P> 0.01; ** = 0.01 > P> 0.001; *** = P < O.OO\. The site with the greater abundance is labelled by asterisk(s). Transient fishes are labelled with a "+" Numerical abundance

PAR

Anisotremus davidsonii Atherinops affinis Brachyistius frenatus Cheilotrema saturnum Chromis punctipinnis Embiotoca jacksoni Girella nigricans Halichoeres semicinctus JIeterostichus rostratus Hypsurus caryi llypsypops rubicundus lvfedialuna californiensis Oxyjulis californica Paralahrax clathratus Paralabrax nebulifer Phanerodon furcatus Rhacochilus toxotes Rhacochilus vacca Sarda chiliensis+ Scomber japonicus+ Scorpaena guttata Sebastes atrovirens Semicossyphus pulcher Sphyraena argentea+ Trachurus symmetricus+ Triakis semifasciata Urolophus hallen' Xenistius californiensis Total

fishesf

(x 103)

178 (92)t

o o

191 (94)t 12,221 (3,551)t 382 (66)

27 (4) 645 (107)

o o 144 151 381 58 44

(18)* (63) (37) (10) (30)

o 6 (2) 30 (12)

o o

6 (2)

o 486 (22)

o 165 (166)

o o o 15.1 (3.4)

SOK

2 (2) 13,811 (5,870)* 1,587 (325)***

o

50 (23) 2,278 (387)*** 6 (4) 26,306 (5,524)*** 216 (45)*** 244 (138)

o

18,225 216,685 26,814 4,582 2,968 72 881 11,608 835 ]8 48 2,556 35,204 31,408 20 70

(4,374)*** (60,672)*** (4,558)*** (802)*** (905)** (43) (524) (11,318) (735) (18) (45) (320)** (15,842)* (15,233)* (18) (70)

56 (37) 396.7 (77.3)***

t Much larger value at PAR not meaningfully tested because of few degrees of freedom and extremely low power to detect differences. =!: Includes five other species at PAR and four other species at SOK.

At SOK, the estimated number of total fishes was 397,000; 318,000 of these were resident fishes (Table 3). Oxyjulis comprised over 68% of all residents at SOK (Table 3). Chromis occurred infrequently and in low numbers (Table 3). There were significantly greater numbers of eight major species at SOK, including the transient Trachurus (Table 3). Four resident species (plus Sphyraena) that were unique to SOK were obviously more numerous there than at PAR (Table 3), Biomass Densities. -At PAR, the biomass density of total fishes was 40 kg/I ,000 m2, equivalent to 400 kg/ha over 1.3 ha (Tables 1, 4; Fig. 1). Resident fishes comprised 99% of this total, and Chromis constituted 15% of all residents (Table 4; Fig. 1). Biomass density was greater at PAR for two major species, plus five other fishes that were either trivially present or unobserved at SOK (Table 4). At SOK, total biomass density was 44 kg/l,OOOm2, with residents contributing about 75% (Table 4; Fig. I). The density of all residents minus Oxyjulis was only 28 kg/l,OOO m2, because this species contributed nearly half of total resident biomass (Table 4; Fig. 1). Oxyjulis and Paralabrax clathratus were two major

DEMARTINI

ET AL.: FISH

DENSITY-ABUNDANCE

RELATIONS

AT TWO

887

REEFS

Table 4. Estimated biomass densities for each of the 16 most abundant species of fishes (and total fishes) encountered at PAR in November 1986-January 1987 and the 22 most abundant fishes at SOK during October-December 1986. Values in parentheses are one standard error of the estimate. Key to PAR-SOK comparison: * = 0.05> P> 0.01; ** = 0.01 > P> 0.001; *** = P < 0.001. The site with the greater density is labelled by asterisk(s). Transient fishes are labelled with a "+" Biomass

density

(kg/i,OOO

m')

SOK

PAR

Anisotremus davidsonii Atherinops alfinis Brachyistius frenatus Cheilotrema saturnum Chromis punctipinnis E mbiotoca jacksuni Girella nigricans Halichoeres semicinctus Heterostichus rostratus Hypsurus caryi Hypsypops rubicundus .Medialuna californiensis Oxyjulis californica Paralahrax clathratus Paralabrax nebullfer Phanerodon furcatus Rhacochilus toxotes Rhacochilus vacca Sarda chiliensis+ Scomber japonicus+ Semicossyphus pulcher Sphyraena argentea+ Trachurus symmetricus+ Xenistius californiensis

a

four other species

a

1.53 (0.75)t 6. I 9 (0.58)** 2.66 (0.24)*** 1.0 I (0.21)* 4.76 (0.78)