Coastal Plant Communities of Latin America

possible.

Part I

The Benthic Algal Community

1

Vegetation Ecology of Rocky Shores

A.R.O. Chapman¹,     Department of Biology, Dalhousie University, Halifax, Nova Scotia, Canada

Publisher Summary

The distribution boundaries of algal zones on rocky shores are not coincident; there is a high degree of correspondence between community units recognized by major species or physiognomies and units recognized by multivariate analysis. Benthic ecologists who study rocky shores have provided many qualitative models for community regulation. This chapter discusses the vegetation ecology of rocky shores and describes the effects of disturbance, competition, predation, environmental stress, and recruitment through a single graphical model. It also reviews the disturbance on rocky shores. The sources of disturbance are manifold. They include the physical effects of wave force, ice scouring, moving rocks, and sand. There are biological disturbances including predation and parasitism. With so many forces—each operating at variable intensities and scales of space and time—there is so much spatial and temporal variation in the creation of new sites for colonization. There is no suitable classification of disturbances into categories, except at the coarsest level.

I. Vegetation Structure

A. Classes of Vegetation

B. Vegetation and Environment

II. Vegetation Dynamics

A. Community Level Site Availability

B. Species Availability

C. Species Performance

III. Integrating Concepts

A. Maintenance of Species Diversity

B. Successional Models

C. Stability

IV. Final Comment

References

I Vegetation Structure

There appear to be two central objectives of marine benthic vegetation science. The first of these is concerned with recognition of multispecies classes of vegetation. Implicit in this search for vegetation classes is the belief that there are recurring combinations of species. That is, species are not randomly distributed with respect to one another. The simple recognition of classes of co-occurring species does not imply that the co-occurrence depends on interactions among the constituent species. This is another, and perhaps, contentious issue in vegetation science.

The second major objective of vegetation science is to determine the relationship between vegetation and environment. It is important to know how multispecies assemblages are distributed in relation to environmental gradients or discontinuities. Often this has practical implications. For example, when studying the effects of pollution, it is important to see how the vegetation classes, or samples of the total vegetation, are distributed in relation to the pollution gradient. It is possible to do this on a species by species basis, but the number of dimensions quickly expands beyond human capability. Multivariate vegetation analyses reduce the number of dimensions to a few classes or (if ordination is used) to only 2 or 3 axes on a Cartesian plane. Moreover, there are emergent vegetation properties (such as species diversity) that are not evident in single species analyses.

A Classes of Vegetation

The zonation of marine vegetation into distinct bands running parallel to the shore has received much attention (Lewis, 1964; Pérès, 1982; Stephenson and Stephenson, 1972). The zones are described by certain indicator species. That is, the total species composition is not considered in the establishment of community classes. Single species indication of zones is of little concern here; of much more interest is the relationship between these zones and the total species complement. This will be a matter for some discussion.

The co-occurrence of species distribution boundaries on the shore gradient is central to a concept of marine vegetation developed by Doty (1946, 1957). Underwood (1978) tested whether or not the upslope or downslope boundaries were randomly dispersed. He found no significant departure from random dispersion of boundaries on any shores examined.

Dale (1984) used a different approach to test whether the downslope boundary of one species was followed by the upslope boundary of a second species significantly often. He found that the number of contiguities of downslope and upslope boundaries is not significantly greater than the expected number on a transect.

These studies by Underwood (1978) and Dale (1984) certainly suggest that the boundaries of indicator species–defined zones are not representative of boundary formation by all of the constituent species within the community. However, vegetation science is not concerned solely with the presence or absence of species at vegetation boundaries, but also with their abundance. Instead of asking whether or not the presences or absences of species are significantly correlated, it is interesting to ask whether abundances of species are related.

One approach to the examination of relationships among abundances is phytosociology. A major objective of this approach is the recognition of classes of vegetation, taking into account cover abundance values. In its pure form, the Braun-Blanquet phytosociology system is very formal and proceeds by the following steps:

1. phytocoenoses or associations are recognized by intuition or experience;

2. vegetation is sampled in units of vegetation that are as homogeneous as possible;

3. a raw data matrix table is constructed, in which columns are samples and the rows are species. The cells of the matrix contain the cover estimates for each species; and

4. the table is differentiated so that similar samples come to lie close together in the matrix.

In the final stages, the phytocoenoses are diagnosed by character species, which are always present in the recognized vegetation unit, but are absent or very rare in all others (Boudouresque, 1971c).

Most of the phytocoenoses (or biocoenoses) recognized in the western Mediterranean correspond to zones characterized by dominant indicator species. For example, at Port du Niel, Boudouresque (1971a) diagnosed seven biocoenoses, which correspond exactly to seven zones on the shore gradient. Another five biocoenoses were distributed along gradients of wave exposure and illumination.

Lindstrom and Foreman (1978) used phytosociological methods to classify intertidal and subtidal vegetation on islands close to Vancouver, Canada. In a departure from orthodoxy, these authors did not use subjective sampling. Instead, they placed quadrats systematically on randomly located transects. No distinct community units comparable to vegetation components recognized by other procedures emerged from the analyses. The authors felt that random sampling might have been a contributory factor in this result. In other words, the samples were not taken in preconceived units of homogeneous vegetation.

In general, the subjective sampling and circular logic processes that afflict orthodox phytosociological procedures have little to recommend them in the recognition of multivariate vegetation classes. Instead, we may turn to numerical approaches that were readily adopted by early practitioners of marine phytosociology (Russell, 1972, 1973). Numerical approaches recognize classes of vegetation on the basis of species composition and abundance in randomly collected samples. It is the samples that are classified.

Since the early 1970s, several publications have reported the existence of multivariate classes of vegetation that correspond to previously conceived units of zonation. Russell (1972, 1973) used a monothetic, divisive association analysis that differentiated the littoral fringe, eulittoral, and sublittoral zones. The analytical technique has shortcomings, and random sampling would have been more appropriate than the systematic approach used. Nevertheless, Bolton (1981), using an agglomerative, polythetic technique, was again able to demonstrate the multivariate identity of the shore zones recognized by Lewis (1964). A similar result was obtained by Lambert and Steinke (1986), working on South African shores, but Lindstrom and Foreman (1978) obtained no differentiation of multivariate classes of intertidal vegetation using either monothetic or polythetic classification. However, Lindstrom and Foreman (1978) did not sample above the mid-intertidal level, where a major vegetation discontinuity occurs (the litus line).

Multivariate classification of subtidal vegetation frequently substantiates the existence of community assemblages previously recognized by zones of dominants. Thus Hoek et al. (1975) were able to distinguish, by agglomerative clustering, seven coral reef communities that could be recognized by dominant or codominant species. Five of the seven communities occurred as zones on a depth gradient. In temperate waters, Neushul (1967), Prentice and Kain (1976), and Lindstrom and Foreman (1978) readily distinguished multivariate classes corresponding to preconceived vegetation units.

From these studies, it is clear that, although the distribution boundaries of algal zones on rocky shores are not usually coincident, there is a high degree of correspondence between community units recognized by major species or physiognomies and units recognized by multivariate analysis.

B Vegetation and Environment

It is, perhaps, more important to determine the relationship between vegetation and environment than to look at the internal structure of communities in the abstract. In this light, the analysis by Ang (1986) is especially interesting. Systematic quadrat sampling provided data on the distribution of Philippine coral reef algae. Environmental parameters were also measured (water motion, salinity, temperature, soluble phosphate, daylength, and tidal variation). The relationship between multivariate vegetation structure and environment was examined by canonical correlation analysis in which the principal component axes of the vegetation data matrix and the principal components of the environmental data matrix were ordinate in the same coordinate frame (Fig. 1.1). The two clusters represent samples collected at different times of the year (Cluster I corresponds to samples collected when water movement, tides, and temperatures were high; Cluster II samples were collected when water movement, tides, and temperatures were low). This approach is very promising because it allows reduction of the number of community and environmental factor dimensions to two or three, providing readily interpretable presentations. It is a useful alternative to multiple regression analysis where data do not fit the assumptions of inferential statistics.

Figure 1.1 Ordination of Philippine coral reef algal samples to show relationship between vegetation data and environmental data. Cluster I samples were collected when water movement, tides, and temperatures were high. Cluster II samples were collected when water movement, tides, and temperatures were low. Modified after Ang (1986).

Unfortunately, principal components analysis makes unrealistic assumptions about linear relationships between species distributions and environmental gradients. This shortcoming has been overcome by more recent algorithms, such as detrended canonical correspondence analysis used by Kautsky and van der Maarel (1990). These authors analyzed Baltic benthic communities in relation to various environmental parameters. The ordination diagram showed few discontinuities in sample distribution, but revealed trends in relation to water depth, substratum type, and water movement. Surprisingly, the authors concluded that the correlations between observed patterns of biotic distributions and environmental vectors indicate that zonation patterns in the Baltic Sea are to a large extent ruled by abiotic factors. This conclusion can be asserted only by experimental analysis, and this was certainly lacking.

Ordination has become a widespread procedure for the analysis of marine vegetation structure in relation to environment (e.g., Boudouresque, 1970, 1971b; Cinelli et al., 1977, 1979; Velimirov et al., 1977; John et al., 1977; Devinny, 1978; Russell, 1980; Field et al., 1980; Tittley and Shaw, 1980; Yoneshigue, 1985; Lambert and Steinke, 1986; Kooistra et al., 1989). Indeed, the availability of ordination programs for microcomputers guarantees their increasing popularity and perhaps their misuse through overlooking the assumptions inherent in each procedure.

Although ordination is, by far, the most appropriate procedure for relating trends in vegetation structure to environmental variation, it is also possible to use a classificatory approach. Littler and Murray (1975) used hierarchical cluster analysis to relate multivariate vegetation units to a gradient of sewage impact. In a later study, Seapy and Littler (1982) examined the relationship between community classes derived from cluster analysis and the tidal gradient on a shore. Although some of the discontinuities among vegetation classes were clearly related to tidal fluctuations, others were not, and no general picture emerges from this study. A more consistent relationship between community structure and sand inundation was obtained in a further study done on San Nicolas Island, California (Littler et al., 1983). A community dominated by long-lived mussels was characteristic of rock pinnacles free from the effects of sand inundation. On sand-scoured surfaces, community units dominated by (1) opportunistic green algae, (2) perennial sea grasses, or (3) large sea anemones were evident.

Boudouresque (1970) specifically classified groups of algal species as indicators of environmental conditions (illumination, wave exposure, depth, nitrogen richness). Nine supergroups were recognized, e.g., (1) SSB: the biocoenosis of sciaphilous species sciaphilous on wave exposed shores; (2) SCsl: biocoenosis of sciaphilous species on wave-sheltered shores. The classification has been modified in various ways (Boudouresque and Cinelli, 1971, 1976; Cinelli et al., 1977), but is afflicted by a circular logic. Groups of species are found in particular environments, which are then indicated by the species combinations.

II Vegetation Dynamics

The individual plants and shoots that are the constituent parts of marine vegetation are not immortal. Plants die either through intrinsic senescence processes (e.g., Leathesia difformis, Chapman and Goudey, 1983) or through the action of external forces. When death occurs or biomass is removed, space is made available for colonization, and a dynamic process of succession ensues. The characteristics of the process by which the vegetation gap is re-occupied depends on (1) species availability and (2) species performances (Pickett and McDonnell, 1989). The hierarchy of causes of vegetation dynamics are shown in Figure 1.2. These causes of vegetation dynamics operate interactively, so it is perhaps of little value to treat each independently. Thus, for example, species availability determines, to a large extent, whether or not space competition is an important determinant of community structure (Gaines and Roughgarden, 1985; Roughgarden, 1986; Menge and Sutherland, 1987). It is, however, extraordinarily difficult to present an integrated account because there are so many gaps in our understanding. For this reason, a partially fragmented account of vegetation dynamics is presented. This is followed by an introduction to some integrating concepts.

Figure 1.2 Partial hierarchy of causes of vegetation dynamics. Not all factors at the third level of hierachy are included in text body. Modified after Picket and McDonnell (1989).

A Community Level Site Availability

Pickett and McDonnell (1989) point out that the opening of sites by disturbance and species availability (Fig. 2) defines the boundaries in which differential species performances can act. Disturbance can be defined as the total or partial loss of biomass of resident organisms attributable to extrinsic causes (Chapman and Johnson, 1990). Intrinsic natural senescence is not included in this definition, but has certainly been confounded, in the past, by the action of extrinsic agents of disturbance (e.g., Dethier, 1984).

Disturbance on rocky shores has been reviewed comprehensively by Sousa (1985). The sources of disturbance are manifold. They include the obvious physical effects of wave force, ice scouring, moving rocks, and sand. In addition, there are biological disturbances including predation and parasitism. With so many forces, each operating at variable intensities and scales of space and time, it is hardly surprising that there is so much spatial and temporal variation in the creation of new sites for colonization. Indeed, there is no suitable classification of disturbances into categories, except at the coarsest level. An important distinction should be made between catastrophic and disastrous disturbance (Harper, 1977; Paine, 1979). Disasters occur with sufficient frequency that they may be expected to occur within the generation time of the organisms subjected to the disturbance. Such disasters leave relevant genetic and evolutionary memories in succeeding generations. Catastrophes are infrequent, extreme events of little selective consequence to fitness in succeeding generations. In terms of vegetation structure, Paine (1979) was able to show that the maintenance of stands of Postelsia palmaeformis requires frequent, predictable disturbance. Low disturbance or catastrophic disturbance regimes lead to local extinction of stands of this species. Disturbance in the habitat in which P. palmaeformis occurs removes mussels, which are competitively superior to this small kelp.

This example illustrates a general concept in benthic ecology—the nonequilibrium organization of communities. According to this concept, disturbances are often so frequent that they prevent any species from monopolizing resources to the exclusion of competitively inferior forms. Disturbance regimes are said to dictate the diversity of species (Connell, 1978; Roughgarden, 1986; Menge and Sutherland, 1987) by regulating resource availability and, also, by physically removing disturbance susceptible species.

Sousa (1985) proposed that disturbances be classified into two categories: Type 1, in which the disturbance opens a site surrounded by sites occupied by resident organisms; and Type 2, in which the open patch is isolated spatially from occupied sites. This distinction is thought to have important implications for the way in which cleared patches are colonized and for regional population dynamics (Sousa 1985). Type 1 patches can be colonized by water-borne propagules or by vegetative encroachment from surrounding vegetation. Type 2 patches can be colonized only by water-borne propagules. Space in both patch types can be filled by vegetative growth of algal fragments surviving disturbance. In fact, regrowth from fragments is very common, even after near-catastrophic ice scour (personal observation, 1987).

Because disturbances are spatially and temporally variable, they are said to be the major cause of mosaic structure in marine communities (Dethier, 1984; Sousa, 1984). After disturbance, there is a recovery period through which there are quantitative and qualitative changes in vegetation composition. When different patches are at different stages of recovery, a patchwork mosaic of community structure occurs (Sousa 1984). Additional variability in patch composition is introduced by patch size and location, so that a highly heterogeneous assemblage may be predicted.

B Species Availability

As indicated above, patches are colonized by either water-borne propagules or vegetative regrowth from surviving fragments. The study of propagule availability has developed into a sub-discipline now known as supply-side ecology (Underwood and Fairweather, 1989). The supply-side ecology of seaweeds is in its infancy. Most information available for marine organisms relates to barnacles (e.g., Connell, 1985; Gaines and Roughgarden, 1985). Barnacles and many other sessile invertebrates have rather long-lived larval stages (4–6 weeks), whereas seaweeds tend to have short-lived propagules that are difficult to identify because they are so small and undifferentiated.

Early work indicated that the dispersal shadows of seaweeds were quite small, i.e., on the scale of meters (Sundene, 1962; Anderson and North, 1966; Dayton, 1973; Deysher and Norton, 1982). However, other studies show that at least some species of seaweeds are able to disperse over long distances (Amsler and Searles, 1980; Zechman and Mathieson, 1985; Reed et al., 1988). The absolute ability to disperse is perhaps of less relevance than is a more quantitative measure of dispersal. It is important to determine the numbers of propagules dispersed to various distances from a parent source. In this respect, the study of Reed et al. (1988) is remarkable. These authors measured recruitment densities at various distances from parental stocks. Interestingly, recruitment of filamentous brown algae at 400 m was about 50% of that at 0 m from the parental zoospore source. In contrast, recruitment by two kelp species declined drastically with distance. This difference in performance may be related to phototactic capability. Spores of filamentous brown algae have positive phototaxis, which may keep them in suspension, ensuring wide dispersal. Kelp spores are not phototactic (advanced members of the Laminariales lack eyespots), and this may reduce suspension times. Reed et al. (1988) convincingly demonstrated long distance dispersal by spores as opposed to dispersal by fertile adult plants.

When initially sterile substratum was made available following the appearance of the volcanic island of Surtsey in 1963, it took 2 years for the first algal colonizers to appear (Jónsson and Gunnarsson, 1982). In contrast, when a severe ice scour (in 1987) removed most of the intertidal vegetation on exposed headlands over tens of kilometers near Halifax in eastern Canada, a full canopy of many perennial species reappeared within about 8 months. In the latter cases, regeneration was clearly from surviving fragments (personal observation, 1987). Ecological succession rarely involves colonization of sterilized substrata, and it seems likely that vegetative regeneration is a more important process in marine vegetation than is the settlement of water-borne propagules. This should be an area of extensive investigation in the future.

C Species Performance

1 Environmental Constraints and Physiological Tolerance

Stress can be distinguished from disturbance as defined above. Stress is a physiological constraint on production imposed by resource limitation or environmental conditions at the extremes of physiological tolerance. Seaweeds are truly marine, photoautotrophic, and mostly sessile plants. This combination of characteristics limits their distribution to a narrow fringe of the ocean perimeter. The fringe is narrow because light decreases dramatically with depth in coastal waters (Lüning, 1981), and because conditions become increasingly terrestrial with increasing height on the shore gradient. Hence, light and desiccation (and/or freezing) are primary stress factors limiting the performance of seaweeds composing marine vegetation.

At the lowest depth limit of algal distribution, the light available is 0.05–0.001% of surface values (Lüning, 1985). At this limit, only a few crustose species occur. The deepest-growing kelp vegetation occurs at 0.6 to 1.2% of surface light, and a closed kelp canopy occurs at 4% of surface