BULLETIN OF MARINE SCIENCE, 37(1): 279-298, 1985
WHAT DO NATURAL REEFS TELL US ABOUT DESIGNING ARTIFICIAL REEFS IN SOUTHERN CALIFORNIA? Marion L. Patton, Robert S. Grove and Robert F. Harman ABSTRACT We investigated the relationship between fish abundances and easily-measured environmental variables on shallow subtidal shoals in Southern California. Our object was to develop a method for large-scale prediction ofthe fish fauna of natural and artificial shoals. In a 2-year period, 337 samples were taken at 85 survey sites located between Pt. Conception and the Mexican border. Substrate variables and cryptic, demersal fish were counted with transects. Other fish were counted with a visibility-proportioned free swim. Fish abundance and species number were usually saturating functions of environmental variables, i.e., fish abundances changed with environmental variable values when the values of these variables were low, but were constant at the maximum when environmental variable values were high. On hard substrates, the abundances of about half of the fish species were saturating functions of bottom relief; the abundances of most of the rest appeared unrelated to bottom relief. Species number and total fish abundance were saturating functions of bottom relief. Species number and total fish abundance were not clearly related to kelp density, algal turf density and rock area per square meter of sea surface on high relief, but were saturating functions of these variables on low relief. Since invertebrate prey are probably more abundant on sites with high relief, dense turf, dense kelp and much rock surface area, these results suggest that fish abundance is not always limited by food. These results also suggest that fish abundance is not always limited by shelter availability because the number of shelter interstices probably increases with relief and rock area, and because most fish were not significantly more abundant on breakwaters, which have many large interstices, than on high natural reefs. These data suggest that a high, complex artificial reef enriched with algal plantings might produce no more fish than a smaller, simpler, cheaper reef.
It is difficult to design an artificial reef to provide the maximum fishery enhancement per unit cost because it is not clear how substrate variables affect marine fish abundances. Builders of artificial reefs often assume that fish abun-
dance is limited by the resources their reefs provide. Natural populations are not always resource-limited, however, (Andrewartha, 1961) and their assumption may not always be true. Artificial reefs are often designed to have high values for variables which previous studies have related to fish abundance, e.g., bottom relief(Limbaugh, 1955; Quast, 1968a; 1968b; Feder et al., 1974; Ebeling et al., 1980); subparviosilvosa (algal turf) density (Ebeling, pers. comm.); kelp density (Coyer, 1979) and rock surface area (Ehrlich, 1975). In this study, we examine the relationship between the values of these variables and the abundances of supra-benthic and crypticdemersal fish, i.e., Type Band C fish (Sato, 1984) in the Southern California Bight. Fish abundance might also be limited by availability of shelter interstices (Alevison and Brooks, 1975; Randall, 1963; Ebeling et al., 1980). Shelter is difficult to define experimentally but there are probably more large interstices on artificial breakwaters than on natural mainland reefs because breakwaters are usually constructed of large boulders piled loosely to dissipate wave force as turbulence (Bascom, 1980). We could, therefore, attempt to learn whether such shelter was limiting on natural high-relief areas by comparing the fish abundances of breakwaters with those of high natural reefs. 279
280
BULLETIN OF MARINE SCIENCE, VOL. 37, NO.1,
1985
llBO
//1
123,4
--1979 ~1980
11
SITES SITES
7 B
34°
340
N
33°
1
33°
1190
1180
Figure 1. Site locations. Small arrows, 1979 sites; large arrows, 1980 sites. Numbers refer to site names in Table I. Dotted line separates southern and northern subregions.
METHODS
In 1979, 270 dives were made sampling fish abundances at 76 sites lying between Pt. Conception and the Mexican border (Fig. I, Table I). The coast was divided into II 32-km regions. Where possible, sites with high, intermediate and low relief were sampled at reef-base depths of6 and 14 m within each region. In site selection, relief was defined as mean boulder size or, for a solid reef, as the mean height of the vertical slope (Fig. 2B). Relief was> 1.5 m for high-relief sites, 1.5-0.5 m for intermediate-relief sites and <0.5 m for low-relief sites. The divers counted every fish observed except clinids, atherinids, blennies, pricklebacks and cottids (although we included Scorpaenichthys marmoratus). To eliminate rare and transient species we deleted, after the survey was completed, those fish which appeared in 10 or fewer samples and those which had a "fidelity" of 0.5 or less. Fidelity was the total number of samples in which fish appeared divided by the total number of samples taken on all the sites where the fish was observed in one sample or more. This reduced the 1979 species list to 25. We used a I x 30-m transect adjusted to the bottom contours (Fig. 2A) to count substrate variables and cryptic, demersal fish. Other fish were counted with a straight-line free swim. The diver swam at a speed of about II m/min and counted all the fish he could see. The product of the swim distance and visibility was held constant at 1,000. Visibility was the distance at which a 20 cm white globe, tethered I m above the bottom, disappeared as a result of turbidity or kelp interference. Sampling was done when visibility was greater than 4 m, i.e., largely during the clear-water summer-fall season. In 1980, we did a study to compare the effects of kelp and bottom relief on fish abundance wherein we divided the bight into four regions and studied, in each region, a high-relief kelp-free site, a highrelief kelp site, and a low-relief kelp site (Fig. I, Table I). Methods were similar to those of the first study except that we did a surface as well as a bottom free swim, counted atherinids and sampled 2 m of bottom for cryptic, demersal fish. Eight replicate samples, each including a free swim and 2 m2 of substrate sampled for cryptic, demersal fish, were taken per site. Substrate variables were measured with 12 transects of I x 5-m randomly placed on each site. In the 1980 study, only fish which appeared in seven or more samples and had a total abundance of 20 or more were included in the cluster and discriminant analyses. Trachurus symmetricus, which had a very low fidelity in 1979, was not counted in 1980. To aid speculation about the relationship between the abundance of fish-prey invertebrates and bottom relief, data were taken from a companion study of about 200 epimacrofaunal invertebrates at 75 Southern California Bight sites. Invertebrates were counted using a I x 30-m transect adjusted to
PATTON ET AL.: NATURAL REEFS AND ARTIFICIAL
A.
SURVEY
281
REEF DESIGN
TECHNIQUES
=-
~
.
··.·P.,S( 30
B
RELIEF
meters
surface
contour
distance
VARIABLES
Tap Tap
TOTAL
REEF
TOTAL REEF
HT.
C.
Free
I /
1/
II
'II
I
I
Area
IJ
I I
AREA
Distanc~~A =[1.5AI
Swim
Rack
7/ II
ROCK
tn
I
II'
I I
II I
I
I
I
I
I (
J
r-TII /
I
TT
:~9ceks:rierr;, =D~Jtance:A
Figure 2. A. Sampling technique. The dotted line is the free swim, the solid line is the transect. B. Relief variables as measured on a boulder (left) and a solid (right) reef. Rock height is the vertical height of the boulders and, on the solid reefs, the height of the vertical slope. On both reefs, total reef height is the vertical distance between the highest point on the reef (rock top) and flat level sand. C. Free swim rock area per square meter of sea surface is the square of the estimated I-meter contour distance. It is larger in rugose areas (top).
282
BULLETIN OF MARINE SCIENCE, VOL. 37, NO. I, 1985
Table I. Site name, number and depth for sites studied in 1979 and 1980 (site number refers to locations indicated in Fig. I) 1979 Sand sites
1979 Rock sites Site numbe'"
I 2 3 4 7 8 10 13 14 IS 16 17 23 25 27 30 32 43 51 59 62 64 67 69 70 72 76 79 80 81 82 83 84
1980 Sites
Site Site
Dum-
Site location/site depth (m)
ber
Pt. Conception/7, 13 Government Pt.l5 Cojo Anchorage/6, 12 Cojo Anchorage/6 (recent burial) Arroyo Quemodo/6 Tajigas/6 Naples Reef/6, II University Pt.l5 Mohawk Reef/6 Canby's Reef/16 Hazel Reef/8, 13 Carpinteria Reef/5 Bass Rock/7, 13 Deep Hole/6, 13 Trancas Canyon/I 3 Pt. Dume/6, 14 Big Kelp Reef/5, 12 Rocky Pt./6, 12 Pt. Vicente/6, 13 Bunker Pt./6, 15 Huntington Beach/6 Crystal Cove/6, 14 Halfway Rock/6, IS Salt Creek/6, 12 Dana Pt./6, 14 Dana Kelp/9 San Mateo Kelp/6, 15 San Onofre Kelp/I 2 Homo Canyon/7, 13 Las Pulgas Canyon/ 6,13 Batiquitos Kelp/13 New Hope Rock/8 Pt. Loma Kelp/I 3
4 7 8 10 II 13 17 24 27 28 39 41 47 60 61 62 77 78 82
Site location/site depth (m)
Cojo, Anchorage/6, 14 Arroyo Quemodo/ IS Tajigas/15 Naples/I 6 Ellwood Beach/5 University Pt.l5 Carpinteria Reef/IS Bass Rock/6, 14 Trancas Canyon/6 Zuma Beach/6 Manhattan Beach/6, 15 Flat Rock/I 6 Resort Cove/I 6 Anaheim Bay/5, 16 Bolsa Chica/ IS Huntington Beach/IS San Onofre Creek/I 3 San Onofre Nuclear Generating Plantl8 Batiquitos Kelp/6
number
Site location/site depth
A B C
Northern sites CoalOil/7 Goleta Bay/IS Mohawk Reef/6
D E F G
Central sites Deep Hole/13 Captain Watkins/ I0 Hole in the Wall/II Lechuza/9
H I
J K L M
Southern sites (Palos Verdes) Haggerty's Landing/8 Flat Rock Don't Drive There/5 Southern sites (San Onofre) Victor Hugo/IO San Mateo Rock/6 Batiquitos Kelp/I 2
.• Site numbers missing in sequence refer to a companion study of invertebrate sites.
bottom contours. At each site, three or four samples were taken at the same transect elevation above flat substrate (Patton and Harman, in prep.). Numerical Analysis Methods. -Site mean abundances, transformed by square root and then standardized by species mean, were used to calculate Bray-Curtis (Bray and Curtis, 1957) site similarities which were expressed as dissimilarities or "biological distances." All distance indices lose sensitivity in the longer distance ranges (Beals, 1973) because species abundance distributions across environmental gradients are usually non-monotonic and truncate (Swan, 1970). A corrective technique has been devised by Williamson (1978) and modified for use with quantitative data by Smith (1981) wherein longer distances are re-estimated using intermediate sites to give a better correspondence with actual biological differences. In the present study, distances greater than or equal to 0.6 were reestimated with the modified step-across procedure. Biological distances were analyzed with agglomerative hierarchical cluster analysis with a flexible sorting strategy (Beta = -0.25) (Lance and Williams, 1967). The result is a hierarchical structure called a dendrogram. Weighted discriminant analysis (Smith, 1976; 1979) was used to determine which environmental variables were correlated with the biological distances between the site groups appearing in the cluster analysis. Discriminant analysis (Green, 1976; 1979; Smith, 1976; 1979) entails defining a hypothetical
283
PATrON ET AL.: NATURAL REEFS AND ARTIFICIAL REEF DESIGN
Table 2. Discriminant analysis, 1979. Site position in the hypothetical discriminant space corresponds to the values of site environmental variables. The space is transformed and rotated to show a maximum separation of the biological groups defined in cluster analysis in a minimum number of dimensions. The new dimensions are called axes and are numbered according to the percent of between-group variance they account for. Discriminant coefficients reflect the importance of the various environmental variables on these axes (see Methods) Discriminant coefficients (coefficients of separate determination) Discriminant axes Environmental variables·
Total relief % vertical height Rock height Pt. Conception distance % unstable substrate Sewage (PCA axis I) Depth Total reef height Rock area Kelp density Algal turf density % Total between-group variance
Axis I
Axis 2
Axis 3
28.7 20.7 16.9 6.9 3.8 0.1 0.9 5.8 5.9 1.6 0.0 59.0
11.5 0.5 5.7 27.7 0.7 0.7 0.6 3.1 7.5 8.4 6.1 26.0
2.1 0.6 1.3 18.4 25.7 13.4 14.5 0.9 3.0 1.3 2.7 10.0
• Some variables with low discriminant coefficients omitted.
multidimensional space in which each dimension corresponds to one measured environmental variable. Site position in this space corresponds to the values of its environmental variables. This space is then transformed and rotated to show a maximal amount of group separation (the groups are based on the biota and are defined in cluster analysis) in a minimal number of newly-defined dimensions (called axes). The axes are ordered according to the amount of group separation they account for, i.e., the first axis shows the most group separation, the second axis the second most, etc. Environmental variables correlated with the separation of the biologically-defined groups will be correlated with the new axes. Coefficients of separate determination (Hope, 1969; Smith, 1976; 1979) are used to determine which variables are related to each axis, i.e., variables with high coefficients are closely related to group separation on the axis in question. Only group membership is used in the calculations of regular discriminant analysis; weighted discriminant analysis also uses the biological distance between a given site and the other group members, the internal biological similarity of the group as a whole and the biological distances between the groups. The weighted method is often more accurate and robust (Smith, 1976; 1979). In graphic analysis, site means were plotted against environmental variables using two methods. In the first method we divided the sites into categories depending on the value of the environmental variable in question and then calculated category means and standard errors of fish abundances and plotted them against the environmental variable. In the second method, site mean abundance and species number were plotted against environmental variables and a smoothed curve was fitted to the points using a computer-generated cubic spline function (SAS Institute, 1981). Pure-sand sites usually had very low fish abundances and were included only in the relief and kelp-density graphs. Environmental Variables.-Reliefwas measured as percent vertical substrate, mean rock height, and total reef height. Vertical substrate had a slope of70-120 degrees. Rock height was mean boulder size on boulder reefs and the height of the vertical slope on solid reefs. Rock heights greater than 1.5 m were recorded as 1.5 m. The greatest rock height encountered was 14 m at Pt. Conception. Total reef height was the height of the top of the reef above flat, level sand (Fig. 2B). A derived variable called "total relief' was proportional to total reef height for reefs shorter than 4 m, but constant at the maximum for reefs taller than 4 m. In the first study, turf was the percent cover of a11algae with stipes shorter than 10 cm. In the second study, turf also included foliose bryozoans. Kelp density was the number of Macrocystis sp. stipes longer than 2 m per square meter of sea bottom. The rock area per square meter of sea surface was estimated by squaring the bottom contour (Fig. 2C). In the first study, the bottom contour was estimated from percent vertical substrate and per cent unstable substrate. In the second study, fine brass chains were adjusted to follow the bottom contour.
284
BULLETIN OF MARINE SCIENCE, VOL. 37, NO.1,
ANISOTREMUS
002:
DAVIDSONII,
~ I -+ -- - -- --
•••••••••
+---- --
HALICHOERES 2 + I
I I
iI
1
r '..
.) ••'
o btc:.~·
SEMICINCTUS,O
.'
o +~c»..
HVPSVPOPS 2 +
'f
+---- ---
¢
!
o !c.caf -+----
.•..··,
OALll,
0
- .•.-------+----
.L ··T
0.5
t I
...•.
I
o a
o -+ QrJl[»'''''''O''. -- ----+- ----- .•.------
OXYJULIS 4 +
0
~ 1 C
~ 0
1I
,1, ••••••.•.••.
I 2 + I
,I.
't
••.• 'f
PUNCTIPINNIS,
OXVLEBIUS 1.0 +
0
~"""?"''''''''''~ 0~ge + 1
t
..'!for'
I
-+ -- --
PULCHER,
I 1
.
0
.Q
1 r? f······¢········ ..· f I.. I ..
~
o 1ci;.. - +--- ----
.
.
'T -+ --
PARALABRAX
--
--
-
I
----
+ -- ----
.;I/i,.'
f.
.
....
'! J"'f ·0 o 3c:.·0 - +----- -- + -- ---
---+
-.
f
0
..
·
-O
- - + - ---
-- - +
SCDRPAENICHTHVS MARMORATUS, 0
iCJt
·«J .....----+ O· ----
---
--- .•
+ ---
.......f o Jc..J.t::::::ri:·~··'''·''''¢ -. -- -- ---+ --- -- --+- --- ---+
-+--- -- -- +
DAMALICHTHYS 2 +
- .•- -- -- -- .•-- -- -- -+--- ----
0.4
••
9'
0
: L·~t::::::¢;;::+':··:::#
0.8
-+ >1.5
VACCA,
I I I
EMBIOTDCA
..,;;
0
1!I
.
o +ct'r'l'·D ·O ..··•• - -t - ---- - + -- -- - -- +- ----o 0.5 1.0
MVSTlNUS,
I
SEBASTES ATROVIRENS, N-O
::i/;~"++f I ,., ! .++•• t ...
1I
SEBASTES 2 +
1.
..~ . .
~~u"""Q"'" -+ ---
·..·..0
-+---- - -- + --
CARNATUS,
02..
o
0
."":'-~
" ••••
o
--
SEBASTES 4
...~
.'
---+
~~1"" "i'·.' ' ' ;:.' ..# 0.21
PHANERODON FURCATUS,N-O
--
- +-- -- -- -+--- ---- +-- - - ---+
0.4 + I
o ~~-----+-------+-----~~+
+ --
0
::1 J. .j.:]r
,0.
""
I'" ..,1,. o oar.Q··· .. 'r'
+- -- -----+
o ac:.t
- .• - ---
--
PICTUS,
+
0.51
o
NEBULIFER, N-O
.
i
---+ --- -- -- +- - -----+
SEMICOSSYPHUS 2 +
,
't
••"..
O~· - +- - - - ---+-----
+ 2 +
~
;:. .. '
.6
~
I
!c.c*~(..."f'"
CHROMIS
N-O
I I
-+-------+-------+-------+
.q
•..•..•
CALlFDRNICA,
....'
+
+
f ....'1',1
····c··
··c;+
+-------+
CHRVSOMELAtS,
I !
0.3
- + - ---
+ -------
- +- -- - -- - +-- - -- -- +-- -- ---+
- .•.
I I
W
U
';'
NIGRICANS,
.:'
NICHOLS II, N-D
.1..
N-O
I
f
---
~E:A+STES
0.
I ..,,,~..
.'
GIRELLA 2 +
CORVPHOPTERUS
f ::!
1.0 .•.
0
b
J
---+-------+-------+
LYTHRYPNUS
- +- -- --- -+ -- - ----
---+
~
L~~"""
005
N-O
..r ..
o
MEDIAL UNA CALIFORNIENSIS, 1.0 .•.
¢
CARYl.
'1k:.O··.'Q·····o
\••..
od' - .•.--- -- -- .•.-- -- - --+----
+
.I..~., ..Y
I
.
SERRANOIDES
HVPSURUS 2 +
N-O
I +. i~t::>¢·· ..·..~
I
RUBICUND US. 0
i
r
I .'"
- .•.-------+ --- ----+ -- -- --- .•. JACKSON!.
SEBASTES
::~:!!.;.:::~.~:L.~
lei"
o
1
,1,
N-O
+....g
EMBIOTOCA 2 + I
..•.•
---+-------
- + - - --
1
•.
CLATHRATUS,
:~.+*.....¢...
t
1
~
-+----- -- .•.
¢:.. ....,'fl, ••••••••••••
I
PARALABRAX 2 +
0
1985
o
LATERALlS,
i
+
t .... 0
f
+
.. "
LJ.+o·:::::~·"""··· ....··9 - +- ------
o
+ ----
0.5
-- -. - ------
1.0
+
>1.5
1.0 .•.
o do"Y -+ -- ----
o
-+ ---
0.5
--
>1.5
MEAN
NORTHERN SOUTHERN
o
• ~ --+------ - .•. 1.0
•
..·0.. N-O ROCK
HEIGHT,
M
OXVPHIL
SITES SITES IC
NON-OXYPHILIC
±SE
Figure 3. Log abundance of the most abundant study species plotted against mean rock height. Puresand sites have zero rock height. Mean rock heights larger than 1.5 m represented as 1.5 m. Sites were divided into categories depending on rock height. Category means and standard errors were calculated and plotted. The number of sites per category, i.e., sample number, are listed below in the order of
PATTON ET AL.: NATURAL
REEFS AND ARTIFICIAL
OXYPHILIC
TOTAL
• ••
w
u
z
z => CD
-
'+
5 Z ....I
W U Z
0 §and
1.0
0.5
z => CD
-1 - -.: -
••
•
••
285
REEF DESIGN
> 1.5
I/~ sand
0.5
>1.5
1.0
'+
NON.()XVPHILIC
5
NO.
SPECIES
20
Z ....I
•
•
•
en w
u
3 2
w
e;
• • • •
-
•• • •••
10
•• •• •
• •
•
•
I
ci z
;--® sand
0.5
1.0
>1.5
sand
MEAN
ROCK
0.5
1.0
>1.5
HT., M
Figure 4. Species number and In abundance of oxyphilic (See Fig. 3) fish, non-oxyphilic fish and all fish vs. mean rock height. Mean rock heights larger than 1.5 m represented as 1.5 m. Unless otherwise noted, large dots represent five superimposed data points. Small dots represent single data points. The straight-line distance between ends of the chains was measured and the result used to calculate rock area. The ratio of the contour distance to the straight-line distance was also included in discriminant analysis as "rugosity." Estimates of regional sewage pollution were obtained by summarizing, with a PCA analysis (Pimentel, 1979), data from Word and Mearns (1979) study of sediment contaminants. Geographical location was calculated as the distance from Pt. Conception ("Pt. Conception distance"). Percent unstable substrate was the proportion of sand, gravel and mud substrate. RESULTS
GraphiC Analysis-Relief - The abundances of about half of the fish studied, called "oxyphilic" (sharp-loving) fish, were "saturating functions" of rock height, i.e., fish abundance was proportional to rock height when rock height was less than about 0.75 m, but was constant at the maximum when rock height was greater than 0.75 m (Fig. 3). If pure-sand sites were neglected, the abundances of about half of the fish appeared unrelated to rock height. Only Paralabrax nebulifer
their mean rock heights, beginning with zero rock height. Northern site sample numbers were 14,4, 9, 3, and 8. Southern site sample numbers were 13, 8, 7, 5 and 6. For graphic analysis, fish were classed as oxyphilic (0) if their mean log abundance, calculated over all reefs with a mean rock height of more 0.5 m, was more than 100% greater than their mean log abundance on low-relief sites which were not pure sand.
286
BULLETIN OF MARINE SCIENCE, VOL. 37, NO. I, 1985
OXYPHILIC
TOTAL
3
3 w ()
..I ",If1.___ ....•..... -y
Z
,,
2
o 2 Z
::J III
,, ,, ,
w Z
o
Z 0
+ Z
o ...J
f
::J
LI...
••••••
••••••._
~
0
30
60
+ Cl
o
0 •.•..•
•.••••
o
30
NON-OXYPHILIC
Z
2
...J
•.•. _ 60
NO. SPECIES
+__------+-------t
.... f
15
L ..,....
L 1: -- ---
, ,, ,
SE
Cl
I
()
±
;------r/
1. 1:----
,, ,
w
I
() 10
I I
,, ,,
w
--
+--------"'I ..
I
Go
I
, I
I
I
o
f
Z
5
,,
, I
I
! 0u...
o
...•..
_
30
60 %
VERTICAL
0 •.•... 0
....•.. 30
....•.. _ 60
SUBSTRATE
Figure S. Species number and In abundance of oxyphilic (see Fig. 3) fish, non-oxyphilic fish and all fish vs. mean % vertical substrate. Sand sites have 0% vertical substrate. Category sample numbers (see Fig. 3) were 38, 11, 11,9 and 8.
appeared, at least on hard substrates in the north, to be more abundant on low relief. In graphic analysis, fish were classed as oxyphilic if their mean log abundance, calculated over all reefs with a mean rock height of more than 0.5 m, was more than 100% greater than their mean log abundance on lower reefs which were not pure sand. Species number and the total abundance of oxyphilic fish, non-oxyphilic fish and all fish appeared to be saturating functions of rock height (Fig. 4), percent vertical substrate (Fig. 5), and total reef height (Fig. 6). TUff - Fish abundance was less clearly related to percent turf cover than it was to relief (Fig. 7). Fish abundance and species number were not related to percent turf cover on high relief. Abundance and species number were saturating functions of turf density on low relief. At any given turf density, oxyphilic fish abundance, total fish abundance and species number was higher on high-relief sites.
Kelp. - Fish abundance and species number were not related to kelp density on high-relief sites (Fig. 8). On low-relief sites, they were saturating functions of kelp
PATTON ET AL.: NATURAL REEFS AND ARTIFICIAL
287
REEF DESIGN
OXYPHILIC
TOTAL
•
w u 2
o
2 ::::> CO
••
+"
z Z ..J
10
15
+
!O
NON·OXYPHILIC
2
NO. SPECIES
20
..J
••
•
M<-?'. •• • • •
I
~
•
,
'"U
••
w w
"-
'" ci 2
10
5
10
15
15
TOTAL REEF HT .• M
Figure 6. Species number and In abundance of oxyphilic (see Fig. 3) fish, non-oxyphilic fish and all fish vs. mean total reef height. Sand sites have 0 total reef height. Unless otherwise noted, large dots represent five superimposed data points. Small dots represent single data points.
density if pure-sand sites, some of which had dense kelp, were included. If puresand sites were excluded, fish abundance and species number were not clearly related to kelp density on low relief. abundance was not related to rock area on high relief sites (Fig. 9). On low-relief sites, fish abundance and species number were saturating functions of rock area. Species number may increase slowly with rock area on high relief sites.
RockArea.-Fish
Pt. Conception Distance, Sewage. - Figure 10 shows that some fish were more abundant in the south, that some were more abundant in the north and that some appeared to be evenly distributed. It also shows that some fish were less abundant off the Palos Verdes Peninsula, which lies 120 miles south of Pt. Conception, than they were either north or south of this region. Palos Verdes deep-water sediments were sewage-contaminated (Word and Mearns, 1979), and the low Palos Verdes abundances may be related to sewage contamination. Cluster Analysis. - Bottom relief and geographic region, which appeared important in graphic analysis, also appeared important in cluster analysis. The impoverished pure-sand sites were eliminated from cluster and discriminant analysis because they tended to dominate the results. The most distinct cluster was still that of sites with much unstable substrate, i.e., low-relief sites which were mostly sand or mostly low cobble which was populated by early-succession organisms and therefore probably periodically buried or disturbed (Fig. 11). The rest of the sites were divided into southern and northern clusters which were subdivided by relief.
288
BULLETIN OF MARINE SCIENCE, VOL. 37, NO. I, 1985
OXYPHILIC
fl··· ."
......
TO T • l
3.0
f
f+ ..·..·..·f
w
2.0
o z
'"
1...
Q
r
1.0
w 0 Z
'"
Q
Z
f"'1'
Z
J--+------------1
::l
'"
~ +
·.-HIGH
Z
---lOW
-
RELIEF RELIEF
Cl
o
-'
0LL... 40
0
''""
J
-----------1
~ 1.5
0
::l
+__
o
80
40
ft
+
z
---''80
SPECIES
NON-OXYPHILIC
~ 0 -'
15
f iJ·l-t ......
2.0
!;; - - --------------,
f·'l
·..f
1 f/TL----f----- ----------
oZ10 Ul W
o
1·0
W 11. Ul
0 •••••••
o
••..••• 40
---'__
5
0 ••••••
80
0
TURF.
•••••• 40
••••••••••• 80
% COVER
Figure 7. Species number and log abundance of oxyphilic (see Fig. 3) fish, non-oxyphilic fish and all fish versus algal turf density. High-relief sites are those with a mean rock height >0.7 m. Sand sites excluded. Category sample numbers (see Fig. 3) were, for high-relief sites, 4,5 and 10; for low-relief sites, 6, 10,8 and 7. I SE.
The low-relief southern sites were again divided into shallow and deep sites. The cluster oflow and relatively impoverished, southern sites included some northern sites, probably as a result of random error. Discriminant analysis indicated that all three relief variables, Pt. Conception distance, depth, percent unstable substrate and sewage were important predictors of between-site variations in fish fauna (Table 2). Turf, kelp density and rock area appeared relatively unimportant. Total relief, which was constant at the maximum on the higher reefs, was a better predictor than total reef height, again indicating that fish abundance is a saturating function of bottom relief. Cluster, discriminant and graphic analysis of the 1980 data corroborated our earlier conclusions. In the dendrogram (Fig. 12), the high- and low-relief sites were grouped in different clusters. The high-relief kelp and kelp-free sites from a given region were always nearest neighbors and samples from one such site pair (Palos Verdes) formed a mixed cluster. This indicated that relief and geographical region were better predictors than kelp density. Many environmental variables were identical for all samples at the same site; this made it necessary to perform discriminant analysis on the site means. Since discriminant analysis requires that the environmental variables be fewer than the number of samples, it was necessary in the second study, to perform discriminant analysis in stages, analyzing only a few variables at a time. All environmental
PATTON ET AL.: NATURAL
REEFS AND ARTIFICIAL
TOTAL
OXYPHILIC
.. i .. , ~ W
..r.
u z
·f
......¥
2
0 2 z ::J
en
fW+~~-----ft
w
.. ..
+
Z
1
" 0
o
•••••.•• __
"'--
0
-
" ..J
2
¥H ..
.........•.•....
15
'"~ :;: 10
I I I I
0-
'"
I
0 z
I
~ 2
KELP
RELIEF,
lOW
R ELI
E F,
INCl.SAND 51 T ES
NO
RELIEF,
SAND 51 T E S
4
5
SPECIES
~lul
?i*~~::;"'"-#
, ,
0""'-_ ••....•. __ 0
4
HIGH
lOW
•
I
-.-
! •
0
'-l' .....:¢.t···::"~·_::~~~······f
..•..
0
•••••• 4
NON-OXYPHILIC
o
--....
~
+ Z
.'
,:
...J
~ 0 •••.••• _
~.&~~_:__:~I
~
U Z
en
289
REEF DESIGN
"'-2
..••••.••.. 4
5TIPES/M2
Figure 8. Species number and log abundance of oxyphilic (see Fig. 3) fish, non-oxyphilic fish and all fish versus kelp density. High-relief sites are those with a mean rock height >0.4 m. Category sample numbers (see Fig. 3) were, for high-relief sites 17, 2, 4, and 3; for low-relief sites including sand sites 29, 4, 14 and 4; for low-relief sites excluding sand sites, 6, 4, II and 4. I SE.
variables that appeared important in any analysis were included in a final analysis (Table 3). This analysis indicated that, again, relief, geographical region depth
and sewage were the most important variables and that algal turf, kelp and rock area were relatively unimportant. Shelter Interstices. -Statistical comparison of four breakwater-stone reefs with 18 nearby high, natural reefs that were not piles of large boulders showed that the only fish significantly more abundant on breakwaters were Chromis punctipinnis, Damalichthys vacca and Phanerodonfurcatus; that species number, and the abundances of all fish, oxyphilic fish and non-oxyphilic fish were not significantly different on breakwaters; and that Oxylebius pictus was significantly less abundant on breakwaters (Table 4). DISCUSSION
There is evidence that fish abundance may be a saturating function of artificial reef height as well as of natural reef height. Fish abundance did not appear to be related to reef height on reefs higher than 1 m constructed in Puget Sound (Hueckel et al., 1983) and in Kuwait (McClure et al., 1984). Also, the traditional building block of Japanese artificial reefs is 1 m3 (Sato, pers. comm.), and reviews of the Japanese artificial reefliterature (Grove and Sonu, 1983; Mottet, 1981) indicate that, in water shallower than 40 m, the abundance of supra-benthic fish and of
290
BULLETIN OF MARINE SCIENCE, VOL. 37, NO. 1,1985
3.0
...
I
1.5
"
..' 1."! ........··f ','
;I.
w U Z
1/.i
.,,/ "t ."
..
o
,-
,.t----1",!
,'-
t/
+ Z
•••••
HIGH
<.?
-.-
L OW
o
±
-'
I
f'
..
•• 1.5
I
Z 0
t············!···f··········!
w U Z
lD
I
o
TOTAL
Z :J
if',
..
2.0
OXYPHILIC
0 •.•..•.
--'
RELIEF RELIEF
S E
.....•...•.
:J lD
+ ~
<.?
NON-OXYPHILIC
., SPECIES
2
15
o
-' m w
U
10
w 0-
m
o '-'o
---''--
.....•..._
0 •.•••.
'--
•....•..••
o
EST. ROCK AREA M'/M'SEA SURFACE
Figure 9. Species number and log abundance of oxyphilic (see Fig. 3) fish, non-oxyphilic fish and all fish versus rock area. High-relief sites are those with a mean rock height> O.7 m. Sand sites excluded. Category sample numbers (see Fig. 3) were, for high-relief sites, 3, 6, 4 and 7; for low-relief sites, 2, 13,8and8.
cryptic, demersal fish (Type Band C fish) (Sato, 1984) are not related to reef height. It is not clear what essential resources high bottom relief provides. The abundance of fish-prey invertebrates probably increases with increasing bottom relief. Many local fish depend heavily on substrate invertebrates for food at some stage of their life cycle (Hobson and Chess, 1976; Ellison et aI., 1979; Stephens and Zerba, 1981). This generalization probably includes not only fish that graze directly on substrate invertebrates, but also most local nocturnal planktivores because their prey are often organisms which, during the day, live on the substrate (Hobson and Chess, 1976). It also applies to local herbivores because algae-encrusting invertebrates are probably important to their nutrition (Feder et aI., 1974). Invertebrate abundance (Pequegnat, 1964; Patton, in prep.) and biomass density (Pequegnat, 1968) increase with elevation above the sand (Fig. 13), i.e., invertebrate abundance and biomass are not saturating functions of elevation. This suggests invertebrate abundance should be higher in areas of high relief, and that invertebrate abundance, unlike fish abundance, probably continues to increase with rock height above a rock height of 0.75 m. We can conclude that fish abundance is probably not limited by food on high-relief areas if we make three assumptions: first, that the abundance of fish-prey invertebrates and the abundance of the invertebrates of Figure 13 change with relief in the same way; second,
PATTON ET AL.: NATURAL
ANI$OTREMUS
DAVIDSONII
REEFS AND ARTIFICIAL
PARALABAAX
SEBASTES
CLATHRATUS
o.a.
: a
2/ , I I I
..f
0) HALICHOERES
JACKSON
EMBIOTOCA
TUS
I
2 •
..c,'C)""~'9' ---.- .....
MEDIALUNA
RUBICUNOUS
CALIFORNIENSIS
t f' •
...
OALLI
I
o.s!
~
•
...•
.'
SEBASTES
.+
.•
0.8·
.
..
1;
..•••..
g'!'
"0 .. ·0 -.-~....t·. .... - - -. - --. - --.
I I.
..
a to
GIRElLA
a!?·
"
..
"
NIGRICAN$
, ,
04~··1,·, 0.5.
2;
.
.~"r.
0.25.
?"f"'~
SCORPAENICHTHYS
CALIFOANICA
CHROM IS
PllNCTlPINNI
S
III
i(·:t.,~:·:~....••
"
2
j
'"
o
.,::··,i1
SEMICOSSYPHUS
PULCHER
!
~ ..~::..
NEBULIFER
0.51 ..2 .+.:/E? a ~
I
~~'~.::~.t"~
l·..• ?' !,::' ,
MVSTINU5
:'? .::'
1~"~
.
O·
TOXOTES
... .:)
OAMALICHTHYS
2 •
:0:.'.:) VACCA
"
,
2 •
~~:..¢:~+., ..
'i9'~"'O: ....~
f
SEBASTES
1
1.0.
i ..
2,
I I
O'
PARALABRAX
... , ...
SEBASTES
2'
1
o~'?
PICTUS
+).....
.f-oob·.:)
025~
!t
a
RHACOCHILUS
.,~' +~'..~ ...•••
2;
,,0
OXYLEBIUS
4 • I
MARMORATUS
0.5.
2~!.. +.:.~.t ." f ... I I
::: ." '.:)9 ','t:·\)..:)
a
,
f.'...•
0·.:)., .. .:) --.-.-, . CARNATUS
J..
I I I
4'
o~·:t··009
"""
SEBASTES
NICHOLSII
.r
OXYJULIS
4·
CHRYSOMELAS
,
I
CORYPHOPTERUS
.... , ....••21
1
I
1
+
o .0
1.0·
o
CARVI
-
LYTHRYPNUS
Z :J
.
d.
I 1.84 •
o!1r
"
HVP5URUS
I
05~""?999
c
..
-
I·'Y-.c?
1.0·II
"
: -"'~'p '.~
I
"88~:.,' y·,.tk.·
..l··.\
HYPSYPOPS
"'z
",
'.92;
I I
a
.
_,_ •••
o·
o·
'0
SEMICINC
'1
SERRANOIOES
,.~
..•. -I....
C) ...:).~.~
291
REEF DESIGN
~'"
a:
O~ ATROVIRENS
"
EMBIOTOCA
9!t LATERA
LIS
, ,
2·
....••
1~' I',
..•
o!o··t:·~t:·t
.
a
PHANEROOONFURCATUS ...•...
o
PT.
CONCEPTION
60
120
HIGH-RELIEF LOW-RELIEF
'80 SITES
±
SITES
SE
OISTANCE.lnaul.milesl
Figure 10. Species number and log abundance of the most abundant study species versus Pt. Conception distance. Arrows indicate species apparently less abundant off Palos Verdes. High-relief sites are those with a mean rock height >0.7 m. Pure-sand sites excluded. Category sample numbers (see Fig. 3) were, for high-relief sites 3, 4, 4, 2 and 7; for low-relief sites, 3, 6, 5, 4, and II.
292
BULLETIN OF MARINE SCIENCE, VOL. 37, NO. L, 1985
Shallow
low
~
rDeep
r--
Southern~
Hi gh
low
Northern~ High
Unstable Substrate
I
I 20 BIOLOGICAL
40
I
o
DISTANCE
Figure 11. Cluster analysis of 1979 site similarities calculated with a Bray-Curtis index from species abundances.
Table 3. Discriminant
analysis, 1980 (see Fig. 5) Discriminant coefficients (coefficients of separate determination) Discriminant axes
Environmental
variables
Total relief Rock height Pt. Conception distance Sewage (PCA axis 1) Depth Turf density Kelp density Rugosity % Total between-group variance
Axis I
Axis 2
25.7 23.5 22.2 13.6 9.8 0.2 2.1
36.6 27.4 16.1 0.5
2.8
4.3 20.0
68.0
3.9 8.3
2.9
PATTON ET AL.: NATURAL REEFS AND ARTIFICIAL
KElP-
SAN
ONOFRE
293
REEF DESIGN
FREE
KELP
PALOS VERDES
M I XED
SOUTH
KELP
I'ENTRAl
HIGH
-
RELI EF
-
KElP-
FREE
KELP-
FREE
NORTH
-
KELP
CENTRAL
NORTH lOWRELIEF
LOW-RELIEF
I
I
I
4
3
2
SOUTH,
PALOS
SOUTH,
SAN
VERDES
ONOFRE
1
o
BIOL. 01$1.
Figure 12. Cluster analysis of 1980 site similarities calculated with a Bray-Curtis index from species abundances.
294
BULLETIN OF MARINE SCIENCE. VOL. 37, NO. I. 1985
Table 4. Statistical tests: breakwaters versus natural reefs in northern and southern bight regions. Tests are for significant differences in fish abundances and species number between natural and breakwater reefs in the northern and southern halves of the bight. In ANOVA all tested variables but Paralabrax clathratus and Oxyjulis californica abundance and species number were transformed before, testing by log(n + I) to control heteroscedasticity (recommendation of Green, 1979). Transformation did not control heteroscedasticity in Chromis punctipinnis and Embiotocajacksoni so breakwater and non-breakwater abundances of these species were compared with the Mann-Whitney Utest (Sokal and Rolhlf, 1973). Several less common list species with heteroscedasticity uncontrollable by log transformation were omitted. ANOV A degrees of freedom 3,20
Tested
variable
Damalichthys vacca
Girella nigricans
Hypsypops rubicundus
Medialuna californiensis
Oxylebius pictus
Semicossyphus pulcher
Phanerodon furcatus
Rhacochilus toxotes
Scorpaenichthys marmoratus
Sebastes mystinus
Sebastes serranoides
Oxyjulis californica
Paralabrax clathratus
Total abundance
Oxyphilic abundance
Source
of variatlon
Reef type Region Interaction Reef type Region Interaction Reef type
F
4.87 0.07 0.99 0.96 1.98 1.87 1.87
Region
46.36
Interaction Reef type Region Interaction Reef type Region Interaction Reef type Region Interaction Reef type Region Interaction Reef type Region Interaction Reef type Region Interaction Reef type Region Interaction Reef type Region Interaction Reef type Region In teraction Reef type Region In teraction Reef type Region Interaction Reef type Region Interaction
1.13 0.09 1.58 0.07 7.40 0.12 0.01 1.75 9.60 1.31 20.21 5.33 5.33 3.57 0.37 0.0 0.86 0.02 0.42 0.63 1.24 0.40 1.08 0.52 0.35 1.00 1.99 0.56 0.00 0.00 0.08 2.03 7.56 1.82 2.67 8.05 1.04
Probability'
0.04>0.25 >0.25 0.25 0.18 0.19 0.19 <0.01" >0.25 >0.25 0.22 >0.25 0.01>0.25 >0.25 0.20 <0.01>0.25 <0.010.03 0.030.07 >0.25 >0.25 >0.25 >0.25 >0.25 >0.25 >0.25 >0.25 >0.25 >0.25 >0.25 >0.25 0.17 >0.25 >0.25 >0.25 >0.25 0.16 0.010.19 0.12 0.010.25
Reeftype with significantly higher abundance
Breakwater
Southern
Natural
Southern Breakwater Southern
Southern
Southern
PATrON
ET AL.: NATURAL REEFS AND ARTIFICIAL
295
REEF DESIGN
Table 4. Continued
Tested
variable
Non-oxyphilic abundance
Number species
Tested
variable
Chromis punctipinnis Embiotoca jacksoni
Source
of variation
Reef type Region Interaction Reef type Region Interaction NI
N2
18 18
6 6
F
0.13 1.75 0.27 0.03 0.91 4.63 U
84.5 72.0
Probability"
Reef type with significantly higher abundance
>0.25 0.20 >0.61 >0.25 >0.35 0.04*b Probability'
<0.05* >0.20
Reef type with signifihigher abundance
cantly
Breakwater
'"Significant. 'ANOYA. b Species number
was higher on breakwaters
in the south,
higher on natural reefs in the north.
Neither
difference was significant
(I-test,
Sokal and Rohlf, 1973). e
Mann-Whitney
V-test.
that fish are distributed throughout the year as they were when we took our summer-fall samples; third, that foraging or survival is not somehow more difficult for fish on high relief. The abundance of fish-prey invertebrates should also increase with increasing algal turf (Laur and Ebeling, 1983) and kelp density (Coyer, 1979; Bernstein and Jung, 1979) and some fish eat turf or kelp (Feder et aI., 1974). Sessile organisms are usually more abundant on rock substrate (Thorson, 1957), so food should also increase with rock area. However, fish abundance and, usually, species number, were only related to rock area and turf or kelp density on low relief and when the values of these variables themselves were low, i.e., when food was truly scarce. This again suggests that, on many sites, fish abundance is not limited by available food. Although species number appeared to increase with rock area on high natural reefs, species number was not significantly higher on breakwaters than it was on natural reefs. Since rock area is probably much higher on breakwaters, this suggests that species number is also a saturating function of rock area.
The relatively low abundances of some fish off the Palos Verdes Peninsula may not be related to sewage pollution. Many abundances were low off Palos Verdes because fish were fewer on 0.3-1.0-m rocks there. The bottom off Palos Verdes is paved with such rocks, but in other regions most such rocks are unusual features in a sea of sand. Possibly, some fish abundances may be saturating functions of the rock area within a given region. Similarly, Japanese workers have observed that the placement of more than 300 m2 of artificial reef per square kilometer of sea floor produces no further fishery enhancement (Sato, 1984). If, as is probable, the number of shelter interstices increases with relief, with rock area per meter of sea surface and with kelp density, then fish abundance is also a saturating function of shelter availability. Also, our measurements showed that only Chromis punctipinnis, Phanerodon furcatus and Damalichthys vacca were significantly more abundant on the breakwaters than on high natural reefs. This also suggests that most fish abundances are saturating functions of shelter availability, if it is assumed that the fish did not hide in the breakwater caves as the divers approached. In fact, most study species are either attracted by divers or ignore them (e.g., surf perch [Ebeling and Laur, 1985]). C. punctipinnis, however, nests and shelters in caves (Feder et aI., 1974) and may well be limited
296
BULLETIN OF MARINE SCIENCE, VOL. 37, NO.1, 1985
8
9 INVERTEBRATES
...... ••
~ 8 z C[
o
z :;) m C[
+•,.
•• •••
•• •
•• •• .... •• •...
•• • •• • •• • •
Z
7-Z
+
•
--•.. +
• ••
•
••
z Z 7 ...J
··
•
• • • ••
f
"
•••
•
• •
•••••
~ PATTON
6
'"
PEQUEGNAT
rI
5
6
o
2
3 TRANSECT
4 ELEV.,
5
M.
Figure 13. Abundance and biomass of substrate invertebrates versus transect elevation. Circles represent the abundance of epimacrofaunal invertebrate study species versus mean transect elevation above base of vertical slope. Pure-sand sites excluded (from Patton, in prep.). Squares represent invertebrate biomass versus substrate elevation above sand. Zero elevation is mixed rock and unstable substrate (from Pequegnat, 1968).
by shelter availability. By contrast, Damalichthys vacca does not utilize caves or interstices. It is one of the few fish we studied whose abundances were not saturating functions of rock height, and whose abundances increase with turf density on high relief (Patton, 1983) and which eats hard-shelled food (Feder et al., 1974; Ellison et al., 1979; Laur and Ebeling, 1983). Damalichthys vacca might be a food-limited species and we can speculate that more food was available on breakwaters (Ebeling, pers. comm.). Factors controlling the distribution of P. furcatus are unclear since this fish is commonly abundant in very disparate habitats (Patton, 1983). With the above exceptions, our data suggest that fish abundance in the bight is not always limited by food or shelter. This study does not reveal what variables limit fish abundance when food or shelter do not, but fishing pressure, predation and temperature may be important. Southern California is in a transitional zone and many of its fish are, therefore, at the fringes of their ranges (Hom and Allen, 1978), so small temperature changes might have a disproportionate biological
PATTON ET AL.: NATURAL REEFS AND ARTIFICIAL
REEF DESIGN
297
effect. Bight temperature variability does affect reproductive success in some fish (Horn, 1974). Stephens (1981) considers the effect of temperature on reproductive success to be the most important environmental variable controlling fish faunal changes in King Harbor, a high-relief site in the center of the bight. This might also be true for the rest of the bight. These results suggest that an artificial reef could easily be "over-engineered"; that is, a high, complex reef enriched with algal plantings might easily produce no more fish than a smaller, simpler, cheaper reef. A barge-load of boulders might well produce more fish if scattered than if heaped into a high pile because the boulders in the center of the pile would provide nothing useful to most fish. These conclusions are based on correlative data so experimental corroboration would not be redundant. Because information about fishery enhancement is as important as fishery enhancement itself, we would recommend that the next bight artificial reef project compare the fish abundance on low, simple reefs, consisting oflines or small clusters of 0.75 to 1.0 m boulders, with the fish abundances on high, complex reefs consisting of piles of the same boulders. Our observations suggest that, except for Chromis, the fish abundances on the two types of reefs would not be significantly different. ACKNOWLEDGMENTS We acknowledge the assistance of R. James, W. Sandlin, G. Jordan and S. Brown for diving partnership, of A. Ebeling for criticism of the manuscript and of J. S. Stephens, Jr. and J. Palmer for manuscript criticism, advice, and support. LITERATURE CITED A1evizon, W. S. and M. G. Brooks. 1975. The comparative structure of two western Atlantic reeffish assemblages. Bull. Mar. Sci. 25: 482-490. Andrewartha, H. G. 1961. Introduction to the study of animal populations. University of Chicago Press, Chicago and London. 281 pp. Bascom, W. 1980. Waves and beaches. Anchor/Doubleday, Garden City, New York. 366 pp. Beals, E. W. 1973. Ordination: mathematical elegance and ecological naivete. J. Ecol. 61: 23-35. Bernstein, B. B. and N. Jung. 1979. Selective pressures and coevolution in a kelp canopy community in Southern California. Ecol. Monogr. 49: 335-355. Bray, J. R. and J. T. Curtis. 1957. An ordination of the upland forest communities of southern Wisconsin. Ecol. Monogr. 27: 325-349. Coyer, J. 1979. The invertebrate assemblage associated with Macrocystis pyrifera and its utilization as a food source by kelp forest fishes. Ph.D. Thesis, University of Southern California. 364 pp. Ebeling, A. W., R. J. Larson and W. S. Alevizon. 1980. Habitat groups and island-mainland distribution ofke]p bed fishes off Santa Barbara, California. Pages 403-43] in D. M. Power, ed. The California Islands. Proc. of a Multidiscip]inary Symposium. Santa Barbara Museum of Natural History, Santa Barbara, Calif. -and D. R. Laur. 1985. The influence of plant cover on surfperch abundance at an offshore temperate reef. Environ. BioI. Fish. Ehr]ich, P. R. 1975. The population biology of coral reef fishes. Ann. Rev. Ecol. Syst. 6: 211-247. Ellison, J. P., C. Terry and J. S. Stephens, Jr. 1979. Food resource utilization among five species of embiotocids at King Harbor, California with preliminary estimates of caloric intake. Mar. BioI. 52: ]61-]69. Feder, H. M., C. H. Turner and C. Limbaugh. 1974. Observations on fishes associated with kelp beds in Southern California. Calif. Dept. Fish. Game, Fish. Bull. 160: 1-44. Green, P. E. 1976. Mathematical tools for applied multivariate analysis. Academic Press, San Francisco. 376 pp. Green, R. H. 1979. Sampling design and statistical methods for environmental biologists. John Wiley and Sons, New York. 257 pp. Grove, R. S. and C. J. Sonu. 1983. Review ofJapanese fishing reef technology. Southern California Edison Co., Research and Development Series, Rosemead, CA. 112 pp. Hobson, E. S. and 1. R. Chess. 1976. Trophic interactions among fishes and zooplankters near shore at Santa Catalina Island, California. U.S. Fish Bull. 74: 567-598.
298
BULLETIN OFMARINESCIENCE, VOL.37, NO. L,
1985
Hope, K. 1969. Methods of multivariate analysis, Gordon and Breach, New York. 288 pp. Horn, M. H. 1974. Fishes. Pages 11-1-11-123 in M. D. Dailey, B. Hill and N. Lansing, eds. A summary of knowledge of the Southern California coastal zone and offshore areas. Biological Environm., Vol. II. So. Calif. Ocean Stud. Consort. for Div. Mar., Minerals, Bur. Land Mgmt., U.S. Dept. Interior, Contract No. 08550-CT4-1. --and L. G. Allen. 1978. A distributional analysis of California coastal marine fishes. J. Biogeog. 5: 23-42. Hueckel, G. J., R. M. Buckley and B. L. Benson. 1983. The biological and fishery development on concrete habitat enhancement structures off Gedney Island in Puget Sound, Washington. Wash. Dept. Fish. Tech. Rept. No. 78. 67 pp. Lance, G. N. and W. T. Williams. 1967. A general theory of classificatory sorting strategies. I. Hierarchical Systems. Computer J. 9: 373-380. Laur, D. R. and A. W. Ebeling. 1983. Predator-prey relationships in surf perches. Env. BioI. Fish. 8: 217-229. Limbaugh, C. 1955. Fish life in the kelp beds and the effects of kelp harvesting. Univ. Calif. Inst. Mar. Res. IMR Ref. 55-59: 1-158. Mottet, M. G. 1981. Enhancement of the marine environment for fisheries and aquaculture in Japan. Wash. Fish. Tech. Rept. No. 69. 176 pp. Patton, M. P. 1983. Factors controlling the distribution and abundance of the subtidal macrofauna of the Southern California Bight. Part III. Fish: bight-wide study. So. Calif. Edison Co. Research and Development Series, Rosemead, CA. 45 pp.
Pequegnat, W. E, 1964. The epifauna of a California siltstone reef. Ecology 45: 272-283. ---. 1968. Distribution of epifauna biomass on a sublittoral rock-reef. Pacific Sci. 22: 37-40. Pimentel, R. A. 1979. Morphometries/The multivariate analysis of biological data. Kendall/Hunt, Dubuque, IA. 276 pp. Quast, J. C. 1968a. Some physical aspects of the inshore environment, particularly as it affects kelpbed fishes. Pages 25-34 in W. J. North and C. L. Hubbs, eds. Utilization of kelp-bed resources in Southern California, Calif. Dept. Fish and Game, Fish Bull. 139. ---. 1968b. Fish fauna of the rocky inshore zone. Calif. Dept. Fish and Game, Fish Bull. 139: 35-55. Randall, J. E. 1963. An analysis of the fish populations of artificial and natural reefs in the Virgin Islands. Carib. J. Sci. 3: 31-47. SAS Institute. 1981. SAS/Graph users guide. SAS Institute Inc., Cary, NC. 126 pp. Sato, O. 1984. Scientific rationales for fishing reef design. Bull. Mar. Sci. 37: 329-335. Smith, R. W. 1976. Numerical analysis of ecological survey data. Ph.D. Thesis, University of Southern California. 401 pp. ---. 1979. Discriminant analysis. Ecological analysis package, Tech. Rept. No.1. EcoAnalysis Inc., 114 Fox St., Ojai, CA. 51 pp. ---. 1981. The re-estimation of ecological distances with the step-across procedure. Ecological analysis package, Tech. Rept. No.2. EcoAnalysis Inc., 114 Fox St., Ojai, CA. 19 pp. Sokal, R. R. and F. J. Rohlf. 1973. Introduction to biostatistics. W. H. Freeman and Co., San Francisco. 368 pp. Stephens, J. S., Jr. 1981. Fish in Southern California: an evaluation of population ecology. Occidental College Magazine, Spring. Pp. 11-15. --and K. Zerba. 1981. Factors affecting fish diversity on a temperate reef. Env. BioI. Fish. 6: 111-121. Swan, J. M. A. 1970. An examination of some ordination problems by use of simulated vegetational data. Ecology 51: 89-102. Thorson, G. 1957. Bottom communities (sublittoral or shallow shelf). Geol. Soc. Am. Mem. 67: 461-543. Williamson, M. H. 1978. The ordination of incidence data. J. Ecol. 66: 911-920. Word, J. Q. and A. J. Mearns. 1979. 60-meter control survey off Southern California. South. Calif. Coastal Water Res. Proj. TM-299. 58 pp. DATEACCEPTED: December 18, 1984. ADDRESSES: (M.L.P.) VANTUNA Research Group, Department of Biology, Occidental, Los Angeles, CA 90041; (R.S.G.) Southern California Edison Company, Research and Development, P.O. Box 800, Rosemead, CA 91770; (R.F.H.) Department of Oceanography, University of Hawaii at Manoa, Honolulu, HI 96822.
