Seasonal Distribution and Abundance of Tidepool Fishes at Six Locations within the Southern California Bight, 2004-2005

Southern California rocky shores differ from those in central and northern California in both community structure and oceanographic influences. In the rocky intertidal zone, the algal community north of Point Conception is primarily dominated by brown algae and large foliose red algae as opposed to more turf-forming species south of Point Conception (Murray and Bray 1993); thus, biomass of algal species greatly decreases in the southern California rocky intertidal (Murray and Bray 1993). While differences in rocky intertidal fish assemblages also likely exist between the northern and southern coasts, little is known about the community structure of fishes in the Southern California Bight. These fishes have remained largely overlooked primarily because their cryptic nature and ability to move make them difficult to sample. The coastline within the bight receives less northwesterly wind and wave energy than that to the north due to south facing shores in many areas and protection from the Channel Islands (Hickey 1993). Also, sea surface temperatures generally are warmer and sandy beaches are more prevalent in the south than on the central and northern coast. Is this change in both oceanographic influence and algal species composition accompanied by a change in the community structure of rocky intertidal fishes?

The Southern California Bight, extending from Point Conception to the Mexico border, provides a unique mix of warm-temperate and cool-temperate subtidal fish species influenced by the eddying effect of the California Current (Hickey 1993; Horn et al. 2006). Within the Bight cool waters from the north mix with warm waters from the south resulting in a transition zone between northern and southern fishes. Nearshore fishes associated with warm water generally are found in greater abundance to the south and those associated with cooler water are found primarily to the north. Also, localized temperature regimes and habitat play an important role in dictating species composition. While this same pattern may be true of intertidal fish species in general, few studies have assessed the distribution of intertidal fishes in the Southern California Bight (Crase 1992; Davis 2000; Craig and Pondella 2006).

Several studies have examined the abundance and distribution of rocky intertidal fishes on the California coast north of Point Conception (Graham 1970; Burgess 1978; Grossman 1982; Yoshiyama 1981; Yoshiyama et al. 1986; Shanks and Pfister 2009). Given that Point Conception is a known biogeographic barrier between northern and southern nearshore fish species (Horn and Allen 1978; Allen 1985; Horn et al. 2006) and that such a defined difference exists in oceanographic processes and intertidal community structure, it is surprising that rocky intertidal fish assemblages south of Point Conception remained largely overlooked until the mid-1990's (Crase 1992; Davis 2000, Craig and Pondella 2006).

Prochazka et al. (1999) summarized the biogeography of intertidal fishes found along the eastern Pacific coast. They divided the region into three subregions based on intertidal fish faunas: the first extended from Alaska south to Point Conception, the second from Point Conception to Ecuador, and the third extended along the South American coast. They described the first region as low in species richness (average of 15 species per site) compared to Indo-West Pacific assemblages and the second region as high in species richness with an average of 25 species per site. While the authors conceded that a lack of coverage of studies may impact the results of their assessment (only one southern California study was included in the second region (Cross 1982), their conclusions paint an inaccurate picture of southern California intertidal fish assemblages, i.e., one of high species richness.

Localized assessments of distribution and abundance of intertidal fishes in California (Horn and Martin 2006) reveal differences in familial dominance and in species richness between northern, central and southern California. The Cottidae (sculpins) dominate northern and southern assemblages, but the Stichaeidae and Pholidae comprise the greatest percentage of total catch in central California, and the Clinidae dominate northern Baja California. Species richness is greatest at northern sites (18 species) and central sites (17 species), and lowest in southern California (seven species). The authors suggest this may reflect the overall dominance of intertidal fish assemblages by cold-temperate species.

Patterns of species richness and familial associations identified along the California coast (Horn and Martin 2006) persist over varying temporal scales. Numerous studies conducted over periods less than three years (Grossman 1982; Stepien et al. 1991; Crase 1992; Gibson and Yoshiyama 1999) demonstrate predictability of intertidal ichthyofaunal assemblages, minus significant environmental perturbations such as an El Niño Southern Oscillation (ENSO) event (Davis 2000). Even comparisons of species composition and abundance over greater time scales of five years or more indicate consistency (Yoshiyama et al. 1986; Horn and Martin 2006) of dominant families and species though not necessarily abundances. Any changes detected likely are attributable to variability in spawning and recruitment success due to dynamic environmental conditions (Thomson and Lehner 1976; Gibson and Yoshiyama 1999; Davis and Levin 2002).

While a growing number of studies indicate intertidal fish spawning and recruitment can be impacted by changes in environmental conditions, it is not known how community structure might respond to climate change. With increasing sea surface temperatures, some species may exhibit northward shifts in abundance following their optimal thermal regime (Sagarin et al. 1999; Horn and Martin 2006), some might adapt to new environmental conditions, and still others may become extinct. Repeated sampling of intertidal fish assemblages has provided valuable information regarding temporal and spatial variation. Without continued sampling, any changes or shifts in distribution and abundance that would accompany increasing temperatures might go unnoticed. Additionally, shifts may occur on a small spatial scale thus current coastal coverage may provide inadequate baseline data with which to detect change.

Many aspects of the ecology, physiology, and distribution of intertidal fishes have been investigated (Thomson and Lehner 1976; Barton 1982; Grossman 1982; Davis 2000; White and Brown 2013; Richards 2011; Shanks and Pfister 2009; Martin 2014; Mandic et al. 2014; Knope et al. 2017; Rangel and Johnson 2019) including community structure, but all of these studies were limited to one or two sites in very close proximity. Yoshiyama et al. (1986) reviewed studies of intertidal fish assemblages in California from Cape Arago to Piedras Blancas, but similar numbers and coverage of studies do not exist for southern California. Therefore, to better understand rocky intertidal fish assemblages and how they vary over time and space, the current study was designed to cover multiple sites over an expanse of coastline in southern California.

The present study describes tidepool fish assemblages at six sites covering 150 km in the central portion of the Southern California Bight with all sites sampled in the same low tide series for each sampling period. Additional goals of this study were to: 1) compare variation in fish assemblages by latitude, season, and wave exposure, 2) identify associations between fish species and environmental variables, and 3) detect recruitment pulses for individual species.

The study area extended approximately 150 km along the coastline of the Southern California Bight from northern Los Angeles County to Orange County (Fig. 1). Sites were sampled quarterly for a period of 14 months (January 2004 to February 2005) with one overlapping season totaling five sampling seasons. Tidal heights in this area range from –0.6 to 2.1 m and low tides are semi-diurnal. Because of the timing of seasonal low tides, afternoon and evening tides were sampled during fall and winter while early morning tides were sampled during spring and summer. To eliminate variation among low tide series, all sites were sampled on days within the same series.

Six sites were selected based on distribution within the Bight and the number of discrete tidepools available within rocky benches. From north to south, sites were located at Leo Carrillo State Park, Malibu (N 34.0431, W 118.9370); Paradise Cove, Malibu (N 34.0120, W 118.7921); Resort Point, Palos Verdes (N 33.7662, W 118.4244); White's Point, San Pedro (N 33.7141, W 118.3163); Little Corona, Corona del Mar (N 33.5891, W 117.8685); and Shaw's Cove, Laguna Beach (N 33.5450, W 117.7993) (Fig. 1). Pools were selected based on the following criteria: 1) complete separation from the ocean, 2) stability through time, 3) ability to make a complete search, and 4) size. Because all sites had to be sampled in the same low tide series, the number of pools at each site (Leo Carrillo, n = 5; Paradise Cove, n = 5; Resort Point, n = 5; White's Point, n = 4; Little Corona, n = 4; Shaw's Cove, n = 5) was dictated by how many could be sampled in one low tide period.

For the purposes of this study, we determined that a capture and release method involving draining and collecting would reduce impacts of the collection process. The same set of pools was sampled with replacement for each successive date of collection. Water was removed from each pool using a gas-powered, centrifugal pump with a mesh covered intake hose or bailed by bucket. The area of the pool was searched thoroughly with fish being collected by hand or dipnet. If no fish were found after a five-minute period, searching ceased. Boulders were moved and over-turned as necessary. Fish were placed in buckets with fresh seawater then identified, individually measured (standard length) to the nearest millimeter and weighed as a species group (wet weight via Pesola spring scales) to the nearest 0.5 gram. Boulders and cobble that were removed to adequately search for fishes were returned to original locations. Pools then were refilled, rocks were replaced, and fishes were returned to the pool from which they were collected.

Before each pool was drained, temperature (electronic thermometer), salinity (refractometer), pH (pH meter), and dissolved oxygen concentrations (DO meter) were measured for all sampling dates. Also, sea urchins and sea stars were counted, mussels and anemones were quantified (none, single, few, some, many), and percent cover of boulders, upright macroalgae, surfgrass, cobble/shell hash, and sand were visually estimated. Due to availability of the instruments, dissolved oxygen and pH were measured for only three of the five sampling dates, January 2004, May 2004, and December 2004. July 2004 dissolved oxygen was measured, but not pH. Neither dissolved oxygen nor pH were measured February 2004. Surface area and volume were measured for each pool. Surface area was determined by drawing the perimeter of each pool on graph paper and scaling each square to a representative unit of length. Each drawing was scanned into a digital image file. ImageJ (NIH) software was then used to measure the surface area for each pool by taking the mean of three freehand selections. Ten random depth measurements were taken throughout each pool. Volume was estimated by multiplying surface area by mean pool depth. These physical characteristics of pools were measured once. Effects of erosion rates on volume and surface area for the duration of this one-year study were negligible (Emery 1946). While sand levels did vary in a few pools at Shaw's and Paradise Coves, variation was slight and considered less than the error of the measuring technique.

Tidal heights were measured once for each pool using standard surveying equipment. A scope was placed on a level tripod at least a few feet above the highest pool. The bottom end of a stadia rod was placed at the surface of a pool and a relative height measurement was taken from the scope. At the precise time of predicted low tide, a measurement was taken of ocean surface level by this same means to calculate the tidal height of the scope and thus the tidal height of each pool.

Sites also were categorized by wave exposure. While many coastal regions can be categorized by qualitative assessment, southern California rocky intertidal areas experience seemingly more consistent exposure. We used a model-based approach that predicted site-specific relative exposures using predicted inshore wave height generated from the Coastal Data Information Program by the Scripps Institution of Oceanography (cdip.ucsd.edu). Waves are measured by two types of instruments. Pressure sensors measured the height of the water column passing over them from fixed underwater locations. Buoys rode atop the sea surface and measured their own acceleration as they rose and fell with each passing wave crest and trough. Both types of data generate sea surface elevations. The predicted inshore wave height model uses two wave models. The first, called Wavewatch III, is a wind-wave generation and propagation model. The second is a spectral refraction-diffraction wave model for shallow water (10 m > depth > 300 m) which models the effects of bathymetry on waves as they travel from deep water to the coast. The model used to specifically predict inshore wave heights combines these two models to predict wave heights at 10 m depth. While the model does not predict breaking wave height, it does provide a general measure of how much wave energy is reaching a particular area on the coast relative to other locations. To ground-truth the outcome of the model, two factors were considered at each site. Presence or absence of Silvetia compressa was noted as this algal species is associated with areas protected from wave action (Murray and Bray 1993). Also, tidal heights of vertical zones of algae and sessile invertebrates tend to increase at wave-exposed sites (Murray and Bray 1993) due to higher spray and wave run-up. Thus, tidal heights of pools where fishes were found likely would increase as well.

One site, Resort Point, required sampling over two days during the January 2003 sampling period. Pools 1 and 5 were sampled 1/18/04 and pools 2, 3, and 4 on 1/22/04. The literature indicates fish abundances vary only slightly between low tides in the same low tide series (Thomson and Lehner 1976; Allen unpublished data), therefore, pools 1 and 5 were not resampled on the second day. This site was sampled in one day for all other seasons.

The northern most site lies within Leo Carrillo State Park in Malibu, California. The area was protected from collection of intertidal invertebrates by lifeguards and park rangers. Study pools were situated around the perimeter of a south-facing rock outcrop situated 2 m to 5 m above the surrounding sand. Pools were distributed from 0.8 m to 1.9 m above mean lower low water (MLLW). They were the highest with the smallest mean pool surface area of all six sites. Not only were there few alternate pools available that met the criteria for selection, but there were very few other tidepools in general. This site was exposed to high wave energy based on the model and field observations and lacked Silvetia compressa.

The Paradise Cove site was located approximately 16 km downcoast of Leo Carrillo. Access to this site was very limited as most adjacent land was privately owned, thus received few visitors. Sand levels fluctuated greatly in the Malibu area. Consequently, it was difficult to find five permanent pools in the 2.4 km expanse of beach downcoast of Point Dume. Observations indicated the study pools were the only permanent pools at this site. They were situated in a row along a narrow, low-lying rock outcrop perpendicular to shore and ranged from -0.3 m to 0.9 m above MLLW. The surrounding area was a series of low-lying rock ridges, boulder fields and sand beaches that extended from the access point at Paradise Cove up coast to Point Dume. The south-facing cove was the most protected from northwesterly wave energy of the six study sites with abundant Silvetia compressa and low tidepools.

Resort Point was situated approximately 61 km downcoast of Paradise Cove on the north side of Palos Verdes peninsula. The study site was at the bottom of a very steep cliff and required a walk across a cobble beach to the rocky point, thus it received very few visitors other than local surfers. Intertidal fishes were sampled at this site by Crase (1992) in 1983/84 using the same draining methodology described earlier. The five original study pools were relocated using photographic documentation and re-sampled in the current study. The study pools were on top of and adjacent to a large rocky bench that rose 2 m above the surrounding surge channel and boulder-covered tidal flat in which there were many tidepools. Tidal heights of study pools extended from 0.5 m to 1.8 m above MLLW. Palos Verdes peninsula had very few sandy beaches with the nearest sizable beach approximately 6 km upcoast of the study site. This west-facing site experienced the highest wave energy based on the model and field observations and had the largest pools of the six sites.

White's Point also was located on Palos Verdes peninsula but 10.5 km downcoast of Resort Point. The study site was on the lee side of the point in a southeast-facing cove. This area received little protection from collecting and was very accessible to the public. The site was best characterized as a rocky bench defined by a series of rock ridges. Adjacent to the bench was an expansive boulder field to the southeast. The pools were situated in surge channels formed between rock ridges with the exception of Pool 1 which was located in a small rocky outcrop approximately 60 m downcoast of Pool 2. Study pools were distributed from 0.1 m to 0.9 m above MLLW. Some tidepools occurred lower than the study pools and many were found around the point towards the windward side. White's Point experienced intermediate wave energy.

Little Corona was the next site 50 km downcoast of White's Point just south of Newport Harbor. This marine reserve was protected by the city of Corona del Mar and received many visitors. Study pools were located on a southwest-facing rocky bench that was situated between a small sandy beach up-coast and an expansive boulder field downcoast. This bench rose 1 m to 2 m above the surrounding tidal flat and the pools ranged from 0.2 m to 1.0 m above MLLW. Little Corona ranked as the second most hydrodynamically protected site of the six sites sampled.

The most southerly site was Shaw's Cove in Laguna Beach approximately 9.5 km downcoast of Little Corona. The city of Laguna Beach protected the area from collecting. A narrow rocky bench lined the up-coast side of this southwest-facing cove. The bench sat above a sandy beach on the downcoast end by 1 to 2 m. The rocky bench was a relatively flat table of rock that was intermittently scored by surge channels and tidepools. Silvetia compressa grew thickly on top of the bench. Pool 0 and 4 were depressions similar to surge channels while the others were situated on top of the bench. Pools extended from 0.03 m to 1.0 m above MLLW. Shaw's Cove received intermediate wave energy.

Species richness and species diversity (Shannon-Wiener Index, H′) were calculated for fish assemblages in each pool for each season, for each site for each season, and overall, for each site:

Multiple regression analysis was conducted to assess relationships of species diversity with latitude and wave exposure (mean significant wave height). Multiple regression also was used to evaluate changes in total fish abundance per pool with pool physical characteristics of surface area (SA), volume (Vol), depth, and surface area to volume ratio (SA:Vol). SA alone was significantly correlated with total abundance (r² = 0.34, F_4,23 = 4.012, p < 0.05). Therefore, species abundances were standardized to SA by calculating species densities (individuals/m²) by averaging the number of fish found in each pool per square meter of pool surface area. One-way Analysis of Variance (ANOVA) was used to test for seasonality of species diversity. We used a two-way ANOVA (STATISTICA 9., StatSoft Inc, Tulsa, OK) to test abundances and densities both for the three most common species (Clinocottus analis, Clinocottus recalvus, Girella nigricans) and for all fishes found by site and season. Abundances and densities were log-transformed to normalize the data. A Tukey post-hoc was then used to assess groupings of the results. Sea urchin and sea star counts were standardized to pool surface area (individuals/m²) then densities for each were assessed for variation by site using Kruskal-Wallis.

Principal Component Analysis (PCA) was used to identify relationships among 17 environmental variables recorded for each sampling unit. Redundant variables were then removed from the environmental variable matrix used in a later Canonical Correspondence Analysis (CCA) to improve the accuracy of the analysis (STATISTICA 9., StatSoft Inc, Tulsa, OK). Neither PCA nor CCA, which will be discussed later, are significance tests because the ordination scores are not independent of one another and are generally used to describe patterns with no p-values generated. Rather than reducing the number of environmental variables to a smaller number of synthetic variables, PCA was used to identify correlations that may exist among environmental variables and eliminate redundancy. The analysis was then grouped by site to suggest how sites corresponded with axes. For this analysis, environmental variable data were log-transformed to improve linearity.

To identify relationships among fish species and environmental variables, a Canonical Correspondence Analysis (CCA) was used. CCA constrains the ordination results of one matrix (in this case species densities) by a second matrix (environmental variables) to identify patterns that may not be observed by simply reviewing the data. This method was chosen over multiple regression because the data are non-parametric (zeroes prevent even transformed data from being normal). Species found once or that represented less than 2% of the total catch were eliminated from the analysis (Artedius lateralis, Kyphosus azureus, and Apodichthys fucorum). Environmental variables and species densities were averaged across seasons for each pool. Based on the correlation matrix presented by the PCA, redundant environmental variables were removed from the environmental variable matrix in the CCA. The remaining variables used in the CCA were latitude, pool tidal height, pool temperature, and percent cover of upright macroalgae, surfgrass, boulders, and cobble/shell hash. Surfgrass percent cover was winsorized (Quinn and Keough 2003) for one pool at Paradise Cove (PC01) to reduce impact of the outlier. Environmental variables with r-values greater than 0.350 were plotted.

To identify possible size stratification in pools of Clinocottus analis, mean standard length per pool was assessed using two methods. The first regressed mean standard length with tidal height of the pool. Tidal heights were also categorized as either high (0.9–1.9 m), mid (0.5–0.8 m) or low (–0.1–0.4 m). An ANOVA was used to again test mean standard length with tidal category.

Pool surface areas at each site varied greatly; Leo Carrillo, 0.5–1.7 m²; Paradise Cove, 0.7–3.0 m²; Resort Point, 4.6–25.3 m²; White's Point, 0.8–5.9 m²; Little Corona, 0.7–6.9 m²; and Shaw's Cove, 1.1–3.1 m² (Table 1). Resort Point pools had the highest average surface area of 13.10 m² while Leo Carrillo had the lowest at 0.95 m². Individual pool surface areas ranged from 0.46 m² for Pool 5 at Leo Carrillo to 25.29 m² for Pool 2 at Resort Point. Individual pool volumes and average depths also varied considerably from pool to pool (Table 1). Tidal heights of pools (Table 1) ranged from -0.03 m at Paradise Cove to +1.95 m at Leo Carrillo. Fish were found consistently in the highest pool at Leo Carrillo. Due to high wave energy at this site, this pool likely experiences some inundation of water even at high tides below +1.95 m as higher wave energy generally results in higher water levels and wave run-up at the shore (http://cdip.ucsd.edu). The lowest pool (–0.08 m) was only sampled once at Shaw's Cove (January 2004) because it was not isolated long enough to sample regularly.

From January 2004 to February 2005, pool temperatures ranged from a minimum of 13.3 °C at Paradise Cove in January 2004 to a maximum of 20.2 °C at Resort Point in July 2004. Maximum mean pool temperatures occurred during the summer in July 2005 and minimum during the winter in January 2004. No thermal stratification among the sites was evident. Sea surface temperatures for the Southern California Bight during the study period were measured by NOAA Buoy 46025 and monthly means ranged from a low of 14.2 °C in January 2004 to a high of 21.2 °C in July of 2004. Mean pool temperatures for each site typically were lower than the hourly sea surface temperature recorded by the buoy for the hour at which low tide occurred. The exception to this for most sites occurred during winter samplings when pools generally separated from the ocean later in the afternoon and thus had more daylight hours to warm.

Dissolved oxygen (DO) levels (Table 2) varied greatly from site to site and pool to pool with the maximum level of 17.08 mgO₂/L recorded at Resort Point in January 2004 and the minimum of 0.82 mgO₂/L at Leo Carrillo in May 2004. Mean site DO levels were highest in January 2004, lowest in May 2004 and July 2004, then increased December 2004. This trend corresponds with daytime and nighttime low tides; pools become supersaturated during the day and hypoxic at night (Martin and Bridges 1999). Interestingly, fishes occupied pools even with very low DO levels. If conditions become hypoxic, many intertidal fishes, including Clinocottus analis and Clinocottus recalvus, can emerge from the water to breathe air (Martin 1995; Martin and Bridges 1999; Martin 2014). Mean DO levels for each pool ranged from 2.7 to 8.6 mgO₂/L with an overall mean of 6.0 + 1.4 mgO₂/L. No levels were recorded February 2005 because a dissolved oxygen meter was not available.

Monthly mean maximum significant wave heights (m) (MSWH) were generated for each site (Fig. 2) for the period March 1998 to March 2005 using the model created by the Coastal Data Information Program allowing a comparison of wave energy between sites. Resort Point experienced the highest amount of wave energy with an overall mean height of 0.21 m while Paradise Cove was the most sheltered with a mean of 0.13 m. The same model was used to generate MSWH for Piedras Blancas (Blanchette unpublished data) on the central California coast. This site experienced high wave energy with an overall mean (1998 to 2002) of approximately 0.59 m. In comparison, the six southern California sites sampled in this study experienced much lower wave energy with the highest monthly mean of all sites being only 0.37 m at Resort Point. The outcome of the model agrees with field observations.

In the five sampling dates from January 2004 to February 2005, 3,043 fishes were captured weighing a total of 19.81 kg and belonging to ten species from six families (Table 3 and 4). Clinocottus analis dominated abundance in all seasons at 68.1% of the total catch followed by Girella nigricans at 10.5%, Clinocottus recalvus at 6.1%, and Hypsoblennius gilberti at 5.6% (Table 3). Clinocottus analis also dominated in biomass at 68.4% of the total biomass followed by Clinocottus recalvus at 12.2%, Girella nigricans at 10.9%, and Hypsoblennius gilberti at 4.7% (Table 4). These four species accounted for 90.3% of the total number of fishes captured and 96.2% of the total biomass. The overall low species richness and diversity values for total abundance (10, 1.16 respectively) (Table 5) reflect this predominance. The lowest species diversity was found at Leo Carrillo (0.56) while the highest was at the next site to the south Paradise Cove (1.30). The greatest number of species was found both overall and in one sampling season at Paradise Cove (8 and 7 species).

The effects of latitude, wave exposure, and date on species diversity (H′) for each site were examined. H′ decreased as wave exposure increased (r² = 0.29, F_3,26 = 3.620, p < 0.02) and increased with each successive date (r² = 0.29, F_3,26 = 3.620, p < 0.05). Latitude was not a significant factor (r² = 0.29, F_3,26 = 3.620, p > 0.05). H′ was also tested with sampling season to identify any seasonal trends, but this relationship was not significant (ANOVA, F_4,25 = 1.541, p > 0.05).

Overall mean densities (individuals/m²) for each site were highest for all sites January 2004 and lowest for most sites in May 2004 and July 2004. High densities did not persist for the second winter season sampled in February 2005 although some sites did experience a second peak in density in December 2004. Interestingly, fish abundance and density were not redundant metrics. Of the six sites sampled, abundance was highest at Resort Point where density was lowest (Fig. 3), and while abundance was the third lowest at Leo Carrillo, density was highest. Densities did vary by site (ANOVA, F_5,107 = 5.857, p < 0.001), but not by season (ANOVA, F_4,107 = 1.609, p > 0.05) or by season and site (ANOVA, F_20,107 = 0.472, p > 0.05). Densities were significantly higher at Leo Carrillo and White's Point (Tukey post hoc; p < 0.05). Overall mean density for each species at each site where they were found again indicates dominance by Clinocottus analis at 5.48/m² followed by Clinocottus recalvus at 4.52/m². Because overall densities were calculated for species only at the sites where they were captured, Girella nigricans and Hypsoblennius gilberti were replaced by Gobiesox rhessodon at 2.89/m² and Oligocottus snyderi at 2.04/m² in the top four most dense species.

No seasonal pattern was detected in either densities or abundances for the three most abundant species or total fishes found at each site (Abundances: Clinocottus analis, F_4,128 = 1.418, p > 0.05; Girella nigricans, F_4,79 = 1.710, p > 0.05; Clinocottus recalvus, F_4,44 = 0.967, p > 0.05; all fishes, F_4,64 = 0.793, p > 0.05. Densities: Clinocottus analis, F_4,128 = 1.670, p > 0.05; Girella nigricans, F_4,79 = 2.154, p > 0.05; Clinocottus recalvus, F_4,44 = 0.228, p > 0.05; all fishes, F_4,64 = 1.043, p > 0.05). A general seasonal trend of greater overall densities in the fall and winter was apparent (Fig. 3), but not significant (ANOVA, F_4,64 = 1.0, p > 0.05).

In five sampling dates, 379 fish were captured at Leo Carrillo weighing 4764.2 g and belonging to five species (Tables 3 and 4). Clinocottus analis dominated the assemblage at 78.6% of the total catch and 72.0% of the total biomass followed by Clinocottus recalvus at 20.6% and 27.4% respectively (Table 3 and 4). These two cottid species accounted for 99.2% of the total catch and 97.4% of the total biomass captured at this site. These percentages are reflected in low seasonal species diversity (H′) values ranging from 0.28 in February 2004 to 0.66 in May 2004 (Table 5). Leo Carrillo was the only site to experience higher seasonal H′ in spring and summer than in fall and winter and had the lowest overall species diversity of 0.56.

A total of 351 individuals were captured at Paradise Cove in five sampling periods weighing 1668.1 g and belonging to eight species. Clinocottus analis dominated in both abundance and biomass at 50.7% and 59.0% respectively. Clinocottus analis was followed in abundance by Oligocottus snyderi at 17.1%, Gibbonsia elegans at 16.8% and Hypsoblennius gilberti at 13.1%. In biomass, Clinocottus analis was followed by Gibbonsia elegans at 16.0%, Hypsoblennius gilberti at 13.8%, and O. snyderi at 9.7%. Greater species evenness is reflected in higher seasonal species diversity (H′) values ranging from 0.86 in May 2004 to 1.36 in December 2004. Paradise Cove had the highest overall H′ of the six sites (1.30).

The greatest number of individuals was captured at Resort Point probably due to the large size of the pools sampled. In the five sampling dates for this study, 1,084 fishes were captured weighing 7,508.9 g and belonging to six species. Clinocottus analis comprised 80.3% of the total catch and 71.8% of the total biomass followed by Clinocottus recalvus at 10.0% and 14.7%, Girella nigricans at 9.0% and 13.0%. These three species together comprised 99.3% of the total catch and 99.5% of the total biomass captured. The overall species diversity (H′) at Resort Point was 0.79 with the highest seasonal H′ in December 2004 at 0.96.

A total of 504 fishes were captured at White's Point weighing 2165.8 g and belonging to five species. Clinocottus analis dominated in both abundance and biomass at 65.5% and 71.3% respectively. Clincottus analis was followed in abundance by Gobiesox rhessodon at 27.0%, and Girella nigricans at 7.0%, and was followed in biomass by Girella nigricans at 19.3%, and Gobiesox rhessodon at 8.9%. These three species made up 99.5% of the total abundance and 99.5% of the biomass. Seasonal species diversity (H′) was greatest in December 2004 and the overall H′ was 0.85.

At Little Corona, 453 individuals were captured weighing 2406.8 g and belonging to five species. Again, Clinocottus analis dominated in both abundance (47%) and biomass (50.3%) followed by Hypsoblennius gilberti at 24.5% and 26.7%, and Girella nigricans at 24.7% and 21.1% respectively for both species. These three species comprised 96.2% of the total catch and 98.2% of the total biomass. Seasonal species diversity (H′) at Little Corona ranged from a low of 0.92 in May 2004 to a high of 1.32 in December 2004. Overall H′ was the second highest at 1.19.

The least number of fishes was captured at Shaw's Cove, 272 individuals weighing 1295.0 g and belonging to five species. Not surprisingly, Clinocottus analis again dominated this assemblage in abundance and biomass as well at 67.2% and 67.3% respectively. Girella nigricans followed at 25.7% and 17.7% respectively. These two species accounted for 92.9% of the total abundance and 85.0% of the total biomass. Seasonal species diversity (H′) plummeted to a low of 0.09 in May 2004 but recovered for a high of 1.10 in February 2005. Overall H′ was 0.87.

Principal Component Analysis determined Factor 1 (Axis 1) represented pool surface area (r = 0.77) and depth (r = 0.75) while Factor 2 (Axis 2) represented latitude (r = –0.80), pool temperature (r = 0.53) and associated biotic and abiotic factors (Fig. 4). These two factors accounted for 43% of the variation. Sea star density (r = –0.45) and surfgrass percent cover (r = –0.39) were positively correlated with latitude (notice r value of latitude) while pool temperature and cobble/shell hash percent cover (r = 0.47) were negatively correlated meaning northern sites tended to have more sea stars and surfgrass with cooler pool temperatures while southern sites tended to have pools characterized by cobble/shell hash bottoms and warmer pool temperatures.

Sampling units also were grouped by site and evaluated for environmental variable associations (Fig. 5). Resort Point (RP) had the largest, deepest pools of all six sites. White's Point (WP) and Leo Carrillo (LC) grouped about the same in regard to pool size and depth but, WP was characterized by cobble/shell hash substrate and warmer pool temperatures while LC had a greater density of sea stars. Paradise Cove (PC) also was characterized with sea stars and surfgrass. PC and Shaw's Cove (SC) were similar in regard to size and depth of pools, but SC was better described as greater cobble/shell hash percent cover and warmer pool temperatures.

Canonical Correspondence Analysis (CCA) yielded two significant axes in assessing relationships between fish species and environmental variables. Axis 1 was correlated positively with tidal height and negatively with surfgrass percent cover (Fig. 6). Gibbonsia elegans and Oligocottus snyderi were placed to the far left, reflecting their association with surfgrass and low pools. Axis 2 was correlated positively with latitude and negatively with temperature and percent cover of cobble/shell hash. Clinocottus recalvus and O. snyderi were associated with northern sites and cooler pool temperatures. Girella nigricans and Gobiesox rhessodon were associated with southern sites, warmer pool temperatures and cobble/shell hash bottoms. Clinocottus analis occurred in the center of both axes indicating it is found in all sampled pool types, temperatures and latitudes. All of these results follow associations observed in the field. These two axes explain 48.1% of the variance (26.7% and 21.4% respectively).

Size frequencies were assessed for the five most abundant species (Clinocottus analis, Girella nigricans, Clinocottus recalvus, Hypsoblennius gilberti, and Gobiesox rhessodon) to identify seasonal recruitment patterns. All individuals captured in one season for each species were grouped into 10 mm size classes (5 mm for Gobiesox rhessodon) (Fig. 7 and 8). Abundance of Clinocottus analis was highest in Jan 2004, but the greatest proportion of individuals less than 25 mm occurred in Dec 2004 and Feb 2005 (Fig. 7). All Girella nigricans captured were juveniles. The highest number of Girella nigricans and the largest proportion of individuals less than 35 mm were found in Dec 2004 (Fig. 7). Individual Clinocottus recalvus were rarely found smaller than 45 mm thus only one recruitment pulse was detected in Jan 2004 (Fig. 7). Abundance of Hypsoblennius gilberti was highest in July 2004. The greatest proportion of individuals less than 25 mm occurred in Jan 2004 (Fig. 8). Both abundance of Gobiesox rhessodon and proportion of individuals smaller than 22 mm peaked in Dec 2004 (Fig. 8).

Clinocottus analis was the only species found in great enough abundance to assess size frequencies for recruitment pulses at each site (Fig. 9). Interestingly, in January 2004, only the three southern sites experienced greater proportions of individuals smaller than 27 mm (SL). Recruitment occurred at only Shaw's Cove in July 2005 and only White's Point in December 2004. Individuals less than 27 mm were found at all sites during the February 2005 sampling period except Leo Carrillo. Mean standard length of Clinocottus analis was not correlated with pool tidal height (r² = 0.008, F_1,128 = 1.113, p > 0.2) nor was there a significant difference in standard length between tidal height categories (ANOVA, F_2,127 = 1.222, p > 0.2).

In the present study, ten species of fishes were captured in five samples at six sites in the central portion of the Southern California Bight. This is fewer than other rocky intertidal fish surveys done at the southern end of the bight and far fewer than surveys of central and northern California. A study conducted at False Point and Ocean Beach, San Diego (Davis 2000) collected 15 species in 18 samples, but only six were considered common while a second study at Cabrillo National Monument, San Diego (Craig and Pondella 2006) collected 12 species in four samples. These surveys reflect overall low species richness at southern California sites as opposed to species rich central and northern California rocky intertidal benches. Four sites surveyed in the Half Moon Bay area of northern California (Yoshiyama 1981) yielded 24 species representing nine families, and 29 species were collected in 15 samples at Dillon Beach north of the San Francisco Bay area (Grossman 1982). Repeated sampling at San Simeon from 1978 to 1983 yielded 28 species from ten families (L.G. Allen and M.H. Horn unpublished). Differences in species richness between this and other studies of southern California tidepools may be attributable to fewer samples of higher tidepools during this study, perhaps reduced availability of quality habitat at the sites selected or differences in collection methods. The more likely explanation may be that the two San Diego studies were conducted in an area now recognized as the warmer Californian province while our sites lie within the California Transition Zone (CTZ) of the Oregon Province (Briggs and Bowen 2012).

All studies conducted in California rocky intertidal pool habitats (Yoshiyama et al. 1986; Yoshiyama et al. 1987; Crase 1992; Davis 2000) using comparable methods find fish assemblages dominated by cottid species. This study is no exception. Cottidae comprised 74.2% of all fishes found and 80.6% of fish biomass. In fact, Cottidae dominated so strongly that species diversity was consistently low at all sites for all sampling dates.

Motivated by extreme El Niño Southern Oscillation (ENSO) effects, specifically elevated sea surface temperatures and sea levels, Davis (2000) compared two tidepool assemblages in San Diego over time. She concluded that temperature changes caused variations in abundances of particular species but showed little long-term effects. A reduction in the recruitment of Clinocottus analis, the dominant species in non-El Niño years, and an increase in abundance of warm-temperate species, such as Paraclinus integripinnis, produced greater species evenness during the ENSO event. In total abundance, Clinocottus analis was followed by Girella nigricans and Gobiesox rhessodon, both warm-temperate species. Craig and Pondella (2006) studied intertidal fishes at Cabrillo National Monument, San Diego in three “use zones” that varied in visitor access and thus human-induced disturbance. They found no significant difference among assemblages of the different zones. These sites were also dominated by Clinocottus analis as in Davis' (2000) study, however Gibbonsia elegans and the warm-temperate species Girella nigricans ranked second and third in abundance. Both studies described sites in San Diego, an area that receives influence from both the southern warm-temperate Californian biogeographic province and the northern cold-temperate Oregonian province (Stepien et. al. 1991). Intertidal fish species found in this area reflect an overlap of these two provinces thus species richness would be greater than sites located outside of this overlap along the central coast of the Southern California Bight.

Certain species were found in varying abundance in different areas of the study region. Girella nigricans was quite common at the four most southern sites, but only three individuals were found throughout the study at the two northern sites (Leo Carrillo and Paradise Cove). During site reconnaissance three Girella nigricans were observed in Pool 4 at Leo Carrillo, but none were found in that pool during sampling. Temperature likely influences this trend as the habitat sampled at the two northern sites is comparable to the southern sites. A large expanse of sandy beach extends along Santa Monica Bay separating the Malibu sites from the Palos Verdes peninsula. However, this is not likely a factor in larval dispersal as adults are not limited strictly to the intertidal. Another explanation might be that greater niche segregation occurred at the two northern sites which limited Girella nigricans to shallow ephemeral pools in a boulder/cobble area not sampled.

Clinocottus recalvus was found in abundance at two of the six sites, Resort Point and Leo Carrillo, and once at only one other site, Paradise Cove. The occurrence of Clinocottus recalvus at only northern sites may reflect its cool water affinity. However, Resort Point and Leo Carrillo are also the two most wave exposed sites which may indicate Clinocottus recalvus associates with high wave energy. Collections from two tidepools in Lunada Bay, the protected cove adjacent to Resort Point, during November 2003 (D. Pondella unpublished) found no Clinocottus recalvus. Obviously more collections are needed to test this hypothesis. Oligocottus snyderi was captured in abundance only at Paradise Cove, and one individual was captured at Leo Carrillo. This species is known to associate with surfgrass (Nakamura 1976) and cool water. While Leo Carrillo experiences cool water temperatures and has surfgrass as does Paradise Cove, none of the pools sampled actually contained surfgrass. Thus, this species may occur at Leo Carrillo, but only in surfgrass beds.

Rare and unusual species collected during this study included Apodichthys fucorum, Artedius lateralis, and Kyphosus azureus. According to Horn and Martin (2006), Apodichthys fucorum has never been recorded in tidepools in southern California. Only one individual was found throughout the study at Paradise Cove in a pool with surfgrass coverage. Artedius lateralis was also only captured once during this study at Paradise Cove. Cryptic fish studies indicate this species is common in the subtidal (coralline sculpin, Froeschke et al. 2005) which may indicate a preference for either the very low intertidal or subtidal environment. Kyphosus azureus was found in December 2004 during this study and in 1984 (Crase 1992). This warm-temperate species was also observed in low pools not sampled at Shaw's Cove.

Wave exposure of a site appears to exhibit the greatest influence on species diversity (H′) with diversity increasing at sites with lower wave energy. The three sites that experienced the lowest mean significant wave heights (Paradise Cove, Little Corona, and Shaw's Cove) also had the highest H′. These findings are contrary to Gibson (1972) whose exposed site had an H′ of 2.05 and a sheltered site of 1.52. Whether species diversity of fishes is affected directly by wave energy or indirectly by presence/absence of other organisms requires further investigation. Higher H′ did occur at Leo Carrillo during the summer sampling than other months. Because tidepools at this site are located on a bedrock outcrop that juts into the surf zone, they typically experience high wave action which drops off in July and August. Also, during the summer, tides are the highest of the year. This decrease in wave energy and/or increase in sea level might allow other species to make use of these pools.

Different results were obtained when assessing densities of fishes rather than abundances. This is likely more than an effect of standardizing for pool size. Resort Point is characterized by very large pools, of which there are many. Alternately, Leo Carrillo has very few small pools. Abundances standardized to pool size may be reflecting quality and availability of pool habitat found on the rocky bench at high tide rather than that found in pools during low tide. This raises the question, if a funneling effect does impact abundances of intertidal fishes found in tidepools, are assemblages better described by abundance or density. For example, between sites, if the quality and availability of habitat surrounding pools on the rocky bench is similar, i.e. habitat available at high tide, a given area can support a certain number of fish. If only a few pools are available at low tide, fishes must occupy pools in greater density than if many pools are available. Thus, rocky intertidal areas with many large pools registered high abundances due to the size of the pool, but low densities due to the availability of pools.

Gibbonsia elegans and Oligocottus snyderi exhibited a strong association with surfgrass and low pools. The three cottid species encountered (Clinocottus analis, Clinocottus recalvus, and Oligocottus snyderi) grouped together and were associated with northern sites as well as cooler pool temperatures. Girella nigricans and Gobiesox rhessodon related to southern sites, warmer pool temperatures and pools with cobble/shell hash bottoms. These associations reflect the general temperature preferences and distribution for these fish families. Cottidae is a cool-temperate family found in greatest richness in northern California while kyphosids (sea chubs) and gobiesocids (clingfishes) are characterized as warm-temperate families with their northern limits in southern to central California (Horn and Martin 2006).

Other species of the Cottidae are known to spend approximately 2 months in the plankton as larvae then recruit to the intertidal between 10 and 20 mm from spring to late summer (Pfister 1999). Davis and Levin (2002) detected recruitment pulses of Clinocottus analis in all of their quarterly samples, indicating this species recruited year-round during their study at two San Diego sites. However, the authors categorized all individuals under 35 mm as recruits. Size frequencies for Clinocottus analis at each site during the present study indicated recruitment (individuals less than 25 mm) occurred once a year in the early winter, although, recruitment occurred at only the three southern sites in January 2004 indicating regional variation. When all individuals captured for each season were combined, the greatest proportion of small individuals were found during the late fall and early winter sampling dates.

Small individuals of Girella nigricans and Gobiesox rhessodon were found during the December 2004 sampling date which also indicated peak recruitment in late fall or early winter. Hypsoblennius gilberti recruited most during the July 2004 sampling date. Patterns were less clear for Clinocottus recalvus. Too few small individuals were collected to indicate any possible recruitment seasonality. Because dynamic environmental conditions have been shown to impact recruitment success for intertidal fish species, greater replication of sampling seasons over several years is necessary to determine definitive patterns. Vertical size stratification of Clinocottus analis was not found during this study. This is likely due to the limited number of pools sampled at each site in a relatively narrow tidal height range rather than an actual reflection of size disbursement of individuals.

Given the highly variable environment that characterizes the rocky intertidal, it is interesting that patterns were detected at all. Results of this study demonstrate that assemblages of intertidal fishes are somewhat predictable. First, some species' northern and southern range limits are located in the Southern California Bight. Three species of cool-temperate cottids were found at the three northern sites while only Clinocottus analis was found at southern sites. The warm-temperate kyphosid, Girella nigricans, was found in very low abundance at the two northern-most sites but was quite common at sites to the south. Second, more species are found in greater evenness at sites with less wave energy. The three sites that experienced the lowest mean significant wave height also had the highest species diversities. Third, some species associate with particular environmental factors within a pool such as Gibbonsia elegans and Oligocottus snyderi with surfgrass. Fourth, the size and availability of pools at a given area will likely impact the number and density of a species differently. Lastly, rocky intertidal assemblages within the Southern California Bight will surely be dominated by Clinocottus analis. The central portion of the bight is less species-rich than either the northern end, which receives influence from the more species rich central California coast, or the southern end where the Oregonian and Californian biogeographic provinces overlap.

We would like to acknowledge the many people who have contributed to this project. To Drs. Karen Martin and Robert Carpenter who provided valuable input, and to Drs. Steven Dudgeon and Paul Wilson for statistical advice. Also, to Drs. Carol Blanchette, Daniel Pondella, and Michael Horn for sharing unpublished data. We are very grateful to the Coastal Data Information Program at Scripps Institution of Oceanography for providing the coefficients to generate predicted wave heights for the study sites and to Carol Blanchette and University of California, Santa Barbara for providing the predicted wave heights generated by the California Swell Model. Likewise, we are truly indebted to the many field volunteers without whom these data never would have been collected. They trudged on through many sleepless nights for the small pittance of beautiful seascapes lit by the moon. Finally, funding for this study was provided by the Nearshore Marine Fish Research Program and the Graduate Program at California State University, Northridge. All collections were conducted under the CDFG Scientific Collecting Permit #000032 (issued to L.G. Allen).