Introduction

Riprap is the rocky rubble used to build jetties, breakwaters, and armored shorelines. Roughly 30% of the southern California shoreline supports some form of riprap, while 29% of the shoreline is natural rocky substrate (Clark et al. 2002). Given that riprap approximately equals the natural rocky coastline in length, the ecological contribution of these structures to marine ecosystems could be profound. Despite the extensive scientific literature devoted to the rocky intertidal of southern California, only a handful of ecological studies focus on riprap (Reish 1964; Rader 1998; Davis et al. 2002). Even if current ecological knowledge is sufficient to understand processes on anthropogenic structures, any large scale or biogeographic study on rocky intertidal or shallow rocky benthic communities must account for riprap as possible habitat.

The ecological study of urban marine ecosystems is very much in its infancy (Glasby and Connell 1999; Bulleri 2006). What little research has been conducted on marine riprap has revealed somewhat conflicting results and the potential for biogeographic variation certainly exists (Table 1). Some studies report no differences in community structure or diversity between riprap and natural rocky habitats (Chapman 2006; Clynick 2006), while others have found clear differences for only some taxa (Bulleri and Chapman 2004; Moschella et al. 2005; Osborn 2005). Some studies have found that riprap supports lower diversity compared to that typical of natural rocky habitats (Moschella et al. 2005), while others suggest it supports more (Clynick 2006). Recent studies also suggest that anthropogenic structures favor invasive species over native ones (Wasson et al. 2005; Glasby et al. 2007; Tyrrell and Byers 2007). Riprap, for example, facilitated the establishment and spread of invasive Codium fragile (Suringar) in the Mediterranean (Bulleri and Airoldi 2005; Bulleri et al. 2006). Given the ubiquity of riprap in southern California the potential for invasive species to establish new populations in the region could be high.

Table 1 Summary of previous studies comparing riprap to natural rock
Full size table

The design and fabrication of riprap structures include several features that influence physical factors important to rocky intertidal organisms. One of the most important of these is wave energy (Southward and Orton 1954; Denny and Wethey 2001; Jonsson et al. 2006). The slope, angle from shore, and water depth on the weather side of riprap structures may all enhance wave forces. In addition, structures built away from the shoreline have a lee side that creates an area of relatively low wave energy which is not common on exposed coasts. Rock type, area, and age are also factors that may influence marine communities on riprap.

The first steps in understanding the ecological roles of riprap in urban marine environments is to document the communities present on riprap and test for differences between riprap and natural rocky habitats (Niemela 1999). In this study, I compare the diversity and community structure of exposed rocky intertidal communities on four riprap and four natural sites in southern California. I ask the following questions: (1) on average, does diversity or community composition differ between intertidal communities on riprap and natural rocky habitats in southern California, (2) if so, which organisms contribute to those differences, (3) which physical factors are contributing to these differences, and (4) do riprap habitats support higher abundances of invasive species than natural habitats?

Materials and methods

Field sites and data collection

I surveyed intertidal communities at four exposed riprap and four exposed natural rocky boulder sites in southern California, USA between December 2004 and March 2006 (Fig. 1; Table 2). Each site was sampled only once due to limited resources and personnel, and therefore I cannot address temporal variation in any of the ecological patterns discussed here. However, the sites were sampled in a haphazard order, which eliminated any temporal bias between the two habitat types. The four riprap sites consisted of the northern jetty at the entrance to Mission Bay (MB), the outer breakwater at Dana Pt. Harbor (DPB), the northern jetty at the entrance to Newport Bay (NP), and the northwestern breakwater at San Pedro (SP). The four natural rocky sites included the Scripps Intertidal Reserve (SIO), Dana Pt. State Reserve (DPR), Corona del Mar (CDM) and Pt. Fermin (PF). Each of these sites were chosen because they are accessible, on the open coast and exposed to ambient wave conditions, and are reasonably large in area. In addition the natural sites are steep, uneven, and have large boulders. A general description of these temperate rocky intertidal communities is illustrated in several classic publications (Abbott and Hollenberg 1976; Morris et al. 1980; Ricketts et al. 1985).

Fig. 1
figure 1

The southern California coastline with sites labeled. Riprap sites include Mission Bay (MB), Dana Pt. Breakwater (DPB), Newport (NP), and San Pedro Breakwater (SP). The natural rocky sites include Scripps Institute of Oceanography (SIO), Dana Pt. State Reserve (DPR), Corona del Mar (CDM), and Pt. Fermin (PF)

Full size image
Table 2 A summary of the riprap and natural rocky sites used in this study
Full size table

Within each site, a minimum of five vertical transects were laid haphazardly in locations chosen to represent ambient wave exposure. Because the slopes of each site varied (and thus also the transect length, Table 2), I used biological boundaries to define the endpoints of transects. Each transect began in the upper intertidal at the upper limit of Chthamalus fissus Darwin and ended in the mid to low intertidal at the upper limit of Eisenia arborea Areschoug, and were only laid on the exposed side of the structures. Along each transect eight 0.25 × 0.25-m PVC quadrats were placed at evenly spaced intervals only on the tops or outward facing sides of boulders. Quadrats were moved laterally (i.e., at the same tidal height) when necessary to avoid sampling very deep crevices or when they fell between boulders. This sampling scheme was designed to ensure consistency among sites and to avoid a bias by sampling microhabitats present only at some sites (e.g., tidepools, Davis et al. 2002; Murray et al. 2006) or excessive variation by sampling many microhabitats. This sampling scheme also ensures equal effort at the same relative tidal heights. Note that in this scheme the entire vertical transect, rather than individual quadrats, represents a sample of the intertidal community (Davis et al. 2002; Murray et al. 2006).

To quantify community composition I identified all species larger than 3–5 mm within each quadrat in situ. I counted all mobile organisms, except those that were too fast (e.g., Pachygrapsus crassipes Randall) or too cryptic (e.g., Idotea spp.). Some taxa were extremely difficult to identify to species in the field (e.g., Polysiphonia spp.). To be conservative, I lumped all similar species that could not be identified consistently throughout the study (Oliver and Beattie 1996). This included all limpets (which were generally small and notoriously difficult to identify) with the exception of Lottia gigantea Sowerby, which was consistently distinguished from the other limpets.

I took photos using a digital camera positioned directly above each quadrat (Foster et al. 1991). In the lab each photo was cropped to include only the quadrat and analyzed using Image J (Rasband 2005). I quantified percent cover by projecting 25 dots randomly onto each photo and assigning a value of 4% to each organism that occurred under each dot. Since the transect is the sample of the community, percent cover and counts of each species were summed within each transect.

Physical data

Differences in many physical factors between habitats or among sites may help explain variation in the biological communities. I obtained data on several oceanic variables from the Coastal Data Information Program (CDIP) operated by the Scripps Institution of Oceanography, which maintains an archive of data collected from buoys throughout southern California (http://www.cdip.ucsd.edu 2006; Table 2). In addition to the CDIP data I measured slope, the distance from the upper intertidal to the lower intertidal (i.e., transect length), area of the entire rocky intertidal site, the distance from shore that a 1-m-high wave would break, the width of the surf zone, and latitude. Slope, distance from upper intertidal to lower intertidal, distance from shore of breaking waves, and surf zone width are all variables that affect or are indicative of the amount of force a wave will impart upon intertidal organisms (Denny and Wethey 2001). Diversity is known to vary with both area and latitude (e.g., Pianka 1966; Valentine 1966; MacArthur and Wilson 1967). To measure slope I used a protractor placed on each quadrat and averaged the angles from the horizontal within each transect. The surf zone is defined as the area between the depth at which waves break and the shoreline. I estimated the surf zone width at the point of each transect by using linear measuring tools in Google Earth (http://www.earth.google.com) to measure the distance between the point at which waves began to break and the shoreline in aerial images of each site taken near the same time on the same day. I also measured area using Google Earth. Using the rule of thumb that a wave breaks at a depth approximately equal to its height, the distance from shore that a 1-m wave would break at each site was estimated from NOAA navigational charts (Denny and Wethey 2001).

Analyses

I used two different measures of diversity: richness (i.e., number of species) and Simpson’s Diversity Index (1-D) (Magurran 2004). The reciprocal form of Simpson’s Diversity Index is considered more robust, especially at smaller sample sizes, than the more popular Shannon–Wiener Index (Lande 1996; Magurran 2004). Diversity measures across sites and across the two habitat types (riprap and natural rock) were compared using a nested analysis of variance (ANOVA) with Site (MB, SIO, DBW, DPR, CDM, NP, SP, PF) as a random factor nested within Habitat (Natural or Riprap). Data were log transformed as necessary to achieve equal variances.

Because my data consist of both counts and percent cover they are on different scales. Combining them may not be meaningful or difficult to interpret in a multivariate analysis without a severe transformation (Clarke 1993; Anderson and Underwood 1994). In order to avoid assumptions about weighting mobile species differently than sessile species, I used a presence/absence transformation. However, this removes all information concerning abundance. Therefore, I also performed analyses on untransformed percent cover of sessile species and counts of mobile species separately. This approach is more powerful than a single analysis using both types of data with (or without) a transformation because differences can be investigated in different parts of the assemblage. When comparisons were made using presence/absence data, including richness, sessile species that were observed in quadrats, but that did not occur under a dot while quantifying percent cover, are included as present. Otherwise, in comparisons involving percent cover data, these relatively non-abundant species were excluded.

To test for differences in community composition I used a Two-way nested analysis of similarities (ANOSIM) performed on Bray–Curtis dissimilarities (Clarke 1993; McCune and Grace 2002). Except when using presence/absence, Bray–Curtis values were calculated using non-transformed and non-standardized data. I used non-Metric Multidimensional Scaling (MDS) to ordinate the data for visual comparisons.

I identified species contributing cumulatively to 75% of the dissimilarities using SIMPER. SIMPER (similarity percentages) is a routine in the Primer package that identifies the average contribution of individual species to the dissimilarity between groups, and thus helps to identify which species may be contributing large differences between groups (Clarke 1993).

I used ANOVA (as described above) to test species identified by SIMPER for differences in abundance between riprap and natural rock.

To identify which environmental variables might explain community patterns, I conducted a Mantel test using the BIO-ENV procedure in Primer (Clarke 1993). The biological similarity matrix was generated using Bray–Curtis dissimilarities on non-standardized presence/absence data. The physical similarity matrix was generated using Normalized Euclidean Distance on untransformed data obtained from CDIP and the field (Table 2). The Spearman rank correlation was calculated and permutation was used to compare the two matrices.

Diversity and univariate statistics were computed in R and multivariate analyses conducted in Primer (Clarke 1993; Team 2005).

Results

I recorded a total of 89 species, 71 of which were on natural rock and 67 on riprap (Table 3). Natural rock did not significantly differ from riprap in total species richness (F1,6 = 1.076, P = 0.306) nor sessile species richness (53 and 57 species, respectively; F1,6 = 1.681, P = 0.203; Fig. 2). However, natural rock did differ from riprap in mobile species richness (18 and 10, respectively, F1,6 = 36.527, P = <0.001).

Table 3 Summary of species found in Natural Rocky (NR) intertidal and Riprap (RR) intertidal habitats with the number of transects each species occurred in each habitat type
Full size table
Fig. 2
figure 2

Box-and-whisker plots of Simpson and Richness values within each site. Sessile and Mobile species have been analyzed separately, and then combined and analyzed together. Shaded bars indicate natural rocky sites and open bars indicate riprap sites

Full size image

Simpson’s Diversity Index yielded the same results; no significant difference between riprap and natural rock for all species together (t6 = −0.0153, P = 0.988), nor for sessile species (F1,6 = 1.651, P = 0.207). But significantly greater diversity on natural rock existed for mobile species (F1,6 = 9.718, P = 0.003).

Of the sessile species, 13 were recorded exclusively on natural rock and 17 only on riprap (Table 3). Sessile species included the following: 49 algae, 1 plant, 2 anemones, 5 barnacles, 5 bryozoans, 4 bivalves, 1 gastropod, 2 polychaetes, and 1 sponge. All ten mobile species found on riprap were shared with natural rock, whereas the additional eight on natural rock were exclusive to that habitat (Table 3). Of the mobile fauna 12 species were gastropods, 5 were chitons, and 1 species was an echinoderm. The eight mobile species found in natural rocky habitats but not on riprap account for only 33 individuals out of 7,498 counted. These data certainly underestimate the species richness in both habitats since several taxa were extremely difficult to identify to species level (e.g., Gelidium spp., Polysiphonia spp., and small juvenile limpets).

Several species were found in only one of the habitats. Natural rock harbored a total of 21 species that were not found on riprap. Riprap hosted 16 not found on natural rock (Table 3). Several of these species occurred in relatively low abundance, being found in only one or two transects. Two notable exceptions were Caulacanthus ustulatus (Mertens ex Turner) Kützing (an invasive alga) and Acanthinucella spirata (Blainville) (a predatory gastropod), which were the only two species found in more than three transects in one habitat and in none on the other (20 and 9 transects, respectively).

Results were consistent for all comparisons of community structure using ANOSIM whether considering only sessile, only mobile, or all species together. Significant differences existed among sites (<0.001), but not between natural rock and riprap as habitats (P > 0.1; Table 4). Significant R values between sites suggest that one or more sites differ from one or some of the others, regardless of whether it consists of natural rock or riprap. The non-significant R values at the habitat level suggest that at least some of the sites that are different from each other belong to the same habitat group (i.e., Riprap or Natural Rock), and that differences between the habitats are minimal in comparison. MDS ordinations support these conclusions to a certain extent (Fig. 3). A natural rocky site might be much more similar to some riprap sites than to other natural rocky sites (indicated in the MDS by the distance between symbols). This is particularly evident among sessile species. However, the tendency of riprap or natural rocky sites to fall more towards one side or the other of the MDS indicates there may be some differences not detected by ANOSIM at the habitat level (see Sect. “Discussion”).

Table 4 Results of nested 2-way ANOSIM tests for community differences
Full size table
Fig. 3
figure 3

MDS ordinations of riprap and natural rocky sites: sessile species only, mobile species only, and all species with a presence absence transformation. Filled symbols are riprap sites (filled diamond Mission Bay, filled square Dana Pt. Breakwater, filled circle Newport, filled inverted triangle San Pedro). Open symbols are natural rocky sites (open triangle Scripps Institute of Oceanography, open diamond Dana Pt. State Reserve, open square Corona del Mar, open circle Pt. Fermin)

Full size image

SIMPER identified four mobile taxa that contributed cumulatively to over 90% of the dissimilarity among mobile species: limpets (not including L. gigantea), the snails Littorina keenae Rosewater and Littorina scutulata Gould, and L. gigantea (Table 5). The latter three differed significantly in abundance between riprap and natural rock. L. gigantea occurred in greater abundance on riprap (F1,6 = 49.351, P = ≤0.001). But the L. keenae and L. scutulata occurred in greater abundance on natural rock (F1,6 = 25.730, P ≤ 0.001; F1,6 = 58.613, P ≤ 0.001, respectively). Limpets were not significantly different between the two habitats (F1,6 = 0.589, P = 0.447). It should be noted that L. keenae generally occurred above the beginning of transects on both riprap and natural rock, and therefore its true abundance was certainly underestimated. This is because on the riprap structures spray reaches the uppermost parts, allowing L. keenae to inhabit these rocks. Therefore, it was impossible to include the upper limit of L. keenae within the uppermost part of the transects.

Table 5 Species identified by SIMPER as contributing to the dissimilarities between natural rock and riprap for mobile species only
Full size table

The Mantel test identified three variables; significant wave height, the width of the surf zone, and length of the transect that together best explained the distribution of all species between sites (ρ = 0.716). A permutation test rejected the null hypothesis that ρ was equal to 0 (P = 0.003). Transect length acts as a proxy for the linear distance (but not the tidal height) between high and low tide because the endpoints were determined biologically. This latter variable effectively determines the area over which the energy contained in a wave is spread when it hits the shore. The width of the surf zone determines how much wave energy is lost before reaching the intertidal zone. All three variables suggest that variation in wave forces experienced by the organisms in the rocky intertidal explain their distribution among sites and between habitats.

Discussion

Taken together, these results suggest that the diversity and community composition of intertidal organisms are similar on open coast riprap and natural rock in southern California. Although some differences existed between individual sites, the variation among natural rocky sites or among riprap sites was greater than the variation between the different habitat types as a whole (Table 4; Fig. 3). This result, in essence, is due to different spatial scales. One might expect smaller scale variation between sites. But each site still shares characteristics of the region, and at larger scales, that is all the natural rock versus all the riprap, significant differences did not exist. One caveat to these observations is that replication at the habitat level (four sites for riprap and natural rock, respectively) is not very high and there is a good chance differences between habitats may not be detectable without including more sites. The MDS ordinations (Fig. 3) suggest differences may appear given more replication. Although some riprap and natural sites appear very similar (e.g., sessile species from Mission Bay and Pt. Fermin overlap), the tendency of the two habitats to fall more towards either side of the ordinations is evidence that differences between riprap and natural rock were not detected by ANOSIM. This is particularly true among mobile species, for which diversity was significantly different between the two habitats.

Approximately 75% of the species in this study were sessile. No differences were found among them. Therefore, it is understandable why the differences between riprap and natural rock disappear when considering all the species together for diversity and community structure. The mobile species are simply too few when considering the entire assemblage.

A handful of mobile organisms, encountered relatively infrequently in both habitats, caused the difference in mobile species richness. Only 33 individuals out of 7,498 make up the eight species, found exclusively on natural rock. In addition to the eight mobile species missing from riprap, there were 13 more sessile species found only on natural rock and sixteen found only on riprap. Of these species (41.5% of the total species in this study), most were encountered in only one or two transects. Why were these species so uncommon? These species may simply not be abundant on the tops or sides of boulders, preferring flat benches or tide pools (e.g., Tegula funebralis (Adams), an intertidal snail) (Ricketts et al. 1985).

Another reason why these species were uncommon, and a potential weakness of this study, may be an inability to detect less common species in these habitats. Sampling designs using quadrats are known to under sample rare species (Murray et al. 2006). One piece of evidence suggesting the less common species were missed is that several of the species found only on natural rock in this study are quite abundant on the sheltered side of the breakwaters where wave exposure is dramatically reduced (see below). In addition, the abundance of some gastropods, such as Mexicanthina lugubris (Sowerby) were noticeably higher in cracks and crevices, and even the interstitial space within the breakwater (personal observation), than the outer faces of boulders. These microhabitats were not sampled here, and were beyond the scope of this study. In fact, the only species in this study, mobile or sessile, that has not been observed on both riprap and natural rock at some point in time by the author or in a published study (e.g., Rader 1998) is Macron lividus (Adams), a predatory snail. These observations suggest the differences in mobile and sessile species not shared between the two habitats in this study are due to the sampling method used to detect them. Thus, the distribution of rare species on riprap structures, especially in relation to wave exposure, certainly deserves further scrutiny. The conclusion that riprap does not sustain rare or uncommon species is premature (Chapman 2003).

Chapman (2003) found a similar pattern on seawalls, although she did not quantify abundance, so only richness could be evaluated. In her study there were no differences between seawalls and natural rocky reefs for algae, or for all taxa considered together. However, she did find differences in the number of animal species, and mobile animals in particular. Chapman attributes this difference to the lack of suitable micro-habitats, such as crevices or tide pools on the seawalls, although none of the alternative hypotheses, such as differences in wave exposure, were tested. Clearly there are many other micro-habitats on riprap structures which lack the smooth simplicity of seawalls. As suggested above, several micro-habitats on riprap likely contain species that also occurred in natural rocky habitats. Still, it is interesting that the same general pattern for mobile species was also found for seawalls, another anthropogenic habitat.

These results differed from patterns described on many other riprap structures (but not all, Table 1) as well as studies in terrestrial urban areas that show differences in diversity and community structure in anthropogenically modified habitats (Rebele 1994; Eversham et al. 1996; McDonnell et al. 1997). This suggests that there is regional variation in the ability of riprap to support marine communities.

There may be several reasons why riprap and natural rocky intertidal areas were found to be similar in this study but not in others (Table 1), apart from a low statistical power. Osborn (2005) conducted her study on the Endocladia/Balanus communities restricted to the upper intertidal zone in central California. Variation in physical and biological factors defines upper and lower intertidal zones and it is not surprising that results from the upper intertidal may differ from those from the lower intertidal. Chapman and Bulleri (2003) found just such a situation on seawalls, where intertidal communities displayed differences at high and mid-shore levels, but not at lower levels. Another reason may be age. All the riprap sites in this study were several decades old, the youngest being the Dana Pt. Breakwater (DBW), 38 years old at the time of sampling (Bottin 1988). In studies where age was found to influence diversity all the sites were relatively young, less than 20 years, and most less than 10 (Sammarco et al. 2004; Osborn 2005; Pinn et al. 2005). It seems reasonable that when constructing new riprap habitats that provide large amount of initial bare space there may be several years, even decades, of succession (Moore 1939; Reish 1964). In addition, for some regions the amount of natural rocky habitat is relatively diminutive in comparison to soft bottom communities and therefore dispersal to new rocky habitat may be limited (e.g., Bacchiocchi and Airoldi 2003). In contrast, southern California has a well developed and diverse rocky flora and fauna that is probably able to colonize new structures immediately and thoroughly, both intertidally and subtidally. All these possibilities suggest that regional variation in habitat distribution, species pool, history, and other physical factors may be very important for the ability of riprap to support local marine communities.

Evidence suggests much of the variation is caused by differences in wave exposure, which is known to be of profound influence on rocky intertidal communities (Ricketts et al. 1985; Denny and Wethey 2001; Davis et al. 2002; Denny et al. 2004). The Mantel test identified three variables that together best explain the distribution of species observed: mean significant wave height, surf zone width, and the transect length. All three of these factors influence the wave forces experienced by intertidal organisms (Denny and Wethey 2001; Smith 2003; Bucharth and Hughes 2006). The higher the wave height the greater the force exerted on rocky intertidal organisms. Mean significant wave height was often measured from the same offshore buoy for riprap and rocky intertidal sites (Table 2). Wave height data is simply not available on a finer spatial scale at the shoreline. The surf zone was usually narrower at riprap sites, which means that waves approach much closer to shore before breaking, and in turn deliver more kinetic energy to the intertidal zone. The transect length was indicative of the steepness of the shore, which determines the area over which a wave imparts kinetic energy as it hits the shoreline. These distances were always shorter on riprap, which means more energy imparted on a smaller area. The net result is that organisms may be experiencing greater wave energy on riprap sites than on nearby natural sites under similar conditions. Indeed, the importance of wave forces have been documented in several other studies on riprap (Southward and Orton 1954; Davis et al. 2002; Bacchiocchi and Airoldi 2003; Bulleri et al. 2006; Jonsson et al. 2006).

Larger wave forces could prevent some of the species from holding onto the rock at riprap sites. Most of the mobile species not found on riprap are high-spired gastropods, which are known to be more susceptible to wave forces (Vermeij 1993). Except for the chitons Lepidochitona hartwegii (Carpenter) and Nuttallina spp., the mobile species occurring in greater abundance on natural rock are also high-spired (Table 3). This also explains why L. gigantea, a limpet with a low profile, was found in greater abundances on riprap.

The wave climate, described by the data from the offshore buoys, is similar throughout the study region. It seems likely that variation in waves is more strongly influenced by the morphology of the riprap structures themselves than differences in wave climate. Several artifacts of construction enhance the forces generated by waves as they collide with breakwaters (Bucharth and Hughes 2006). For example, riprap structures are generally very steep, commonly with slopes of 30° or more (Bottin 1988). Also, since jetties and breakwaters are built extending out from shore, the water in front of them is usually deeper, which allows a wave to approach much closer before friction with the bottom slows the wave down. Indeed, waves frequently break directly onto the intertidal zone on riprap structures, whereas in natural areas waves frequently break before reaching shore (personal observation). In the only other ecological study to investigate rocky intertidal communities on riprap in southern California, Davis et al. (2002) concluded that the riprap in San Diego Bay facilitated communities more similar to those in exposed conditions on the outer coast largely due to the manner in which waves collided with the riprap structures. Wave exposure clearly plays a dominant role structuring communities in riprap habitats, just as it does in natural ones.

Only a few of the species found in either habitat were invasive (Table 3). Caulacanthus ustulatus (Rhodophyta), a species not recorded in the region before 1990, occurred in moderate abundance on natural rock but not at all on riprap (Zuccarello et al. 2002; Murray et al. 2005). The invasive bryozoans Bugula neritina (Linnaeus) and Watersipora subtorquata (d’Orbigny) occurred uncommonly only on natural rock, while Mytilus galloprovincialis Lamarck occurred infrequently in both habitats. Some studies have found that anthropogenic structures facilitate invasions (Bulleri and Airoldi 2005; Wasson et al. 2005; Glasby et al. 2007; Bulleri et al. 2006; Tyrrell and Byers 2007). That is clearly not the case for the open coast southern California riprap structures. However, the rocky intertidal environment on the west coast of North America, especially in wave exposed conditions, is known to be sparsely invaded (Maloney et al. 2006). Furthermore, C. ustulatus and M. galloprovincialis have been observed on riprap inside protected bays (Becker et al. 2007, personal observation). It is very likely that riprap in protected bays and estuaries harbor larger numbers of non-natives. These observations imply that the anthropogenic origin of many hard substrates by itself is not sufficient to explain a preponderance of invasive species in some marine habitats, and that an interaction with wave exposure may be involved.

An important aspect of coastal riprap not addressed in this study is that breakwaters create protected stretches of rocky shoreline on their lee side. If wave exposure does influence invasive species in rocky intertidal environments, then the protected sides of some riprap structures may allow more of them to exist in close proximity to the exposed coastal habitats (Bulleri and Airoldi 2005; Martin et al. 2005; Bulleri et al. 2006).

Given that riprap rivals the natural rocky coastline in length (Clark et al. 2002), it certainly has the potential to be a major habitat for marine communities in southern California, whether similar to natural rock or not. Some species clearly thrive on riprap. For example, L. gigantea, a species susceptible to human harvesting that significantly modifies spatial patterns in the upper rocky intertidal (Pombo and Escofet 1996; Lindberg et al. 1998), was five times more abundant on riprap than on natural rock (data not shown). Riprap structures are also known to attract and support a variety of fish and have been reported as extremely good lobster diving and sport fishing sites (Chapman 1963; Davis et al. 1982; Kovach 1996). An interesting conservation application of riprap might be as marine reserves or other management tools. Unlike many terrestrial habitats, humans can be excluded from riprap without compromising its intended anthropogenic purpose, that is, absorbing wave energy.

There are many questions that deserve scientific scrutiny concerning the ecological role of riprap in marine environments. For example, whether or not organisms living on riprap structures are contributing reproductively to local populations is vital. If they are, then populations on riprap structures may be considered ecological resources and should be monitored. If not, then they likely act as demographic sinks. In southern California where much of the riprap has been constructed over soft bottoms, it probably enhances the abundances of organisms living on rocky substrates, at the expense of the soft-bottom communities (Davis et al. 1982). Furthermore, if riprap populations are propagule sources then they could heavily influence the connectivity and genetic structures of populations by altering the distance between suitable habitats and thereby facilitating migration (Becker et al. 2007). This question is very important for conservation efforts since, at present, monitoring studies in southern California typically ignore riprap (J. Engle, personal communication).

One especially important aspect of riprap that seems to have never been investigated ecologically is the interstitial space. When large boulders are piled on top of each other there is naturally a great volume of space in between them. In fact, engineers have found this “pore” space to have a strong influence on the stability of the structure and its ability to absorb wave energy (Bucharth and Hughes 2006). It seems probable that pore space has a strong biological influence as well. All of the space on boulders inside the riprap structures is potential habitat. Most riprap structures are permeable to some extent and benthic organisms colonize every available centimeter of space (personal observation). This pore habitat, which is in essence three dimensional, is vast and likely greatly exceeds the benthic habitats measured here and in other studies, which has been essentially two dimensional in nature. Physical conditions are very different in this pore space and may approximate a cave environment. Such caves occur on the west coast of North America, but not to a large extent (e.g., Secord and Muller-Parker 2005). But riprap is common in the United States and thus the pore habitat is also common (Smith 2003). Preliminary observations suggest a large number of filter-feeding organisms such as barnacles, sponges, bryozoans, hydroids, tunicates, and anemones thrive in this habitat. It may also provide shelter for a variety of species as juveniles (Binns and Remmick 1994). Finally, these organisms almost certainly influence the characteristics of the water column (Wilkinson et al. 1996).

There are many complex economic and social considerations to be accounted for when deciding how best to use anthropogenic structures in urban marine environments (Love et al. 2003; Airoldi et al. 2005; Moschella et al. 2005). While these aspects are beyond the scope of this study, they do underscore the necessity and urgency of studying the ecological importance of riprap and other anthropogenic structures in marine environments. Regardless of the answers to these questions, given the increasing human population along the coast (Forstall 1996), and the increasing sea level and storminess due to global climate change (Dean et al. 1987; McCarthy et al. 2001), it seems certain that riprap structures will increase in extent along the world’s coastlines.