Latitudinal gradients in species richness in

Transcription

1 Journal of Animal Ecology 2009, 78, doi: /j x Latitudinal gradients in species richness in Blackwell Publishing Ltd assemblages of sessile animals in rocky intertidal zone: mechanisms determining scale-dependent variability Takehiro Okuda, 1,2 * Takashi Noda, 2 Tomoko Yamamoto, 3 Masakazu Hori 4 and Masahiro Nakaoka 5,6 1 Tohoku National Fisheries Research Institute, Fisheries Research Agency, Same-machi, Hachinohe, Aomori, , Japan; 2 Faculty of Environmental Earth Science, Hokkaido University, N10W5, Kita-ku, Sapporo, Hokkaido , Japan; 3 Faculty of Fisheries Sciences, Kagoshima University, Arata , Simoarata, Kagoshima , Japan; 4 National Research Institute of Fisheries and Environment of Inland Sea, Fisheries Research Agency, Maruishi , Hatsukaichi, Hiroshima , Japan; 5 Field Science Center for Northern Biosphere, Hokkaido University, N11W10, Kita-ku, Sapporo, Hokkaido , Japan; and 6 Graduate School of Science and Technology, Chiba University, Yayoi 1-33, Inage-ku, Chiba , Japan Summary 1. Although latitudinal gradients in species richness within a region are observed in a range of taxa and habitats, little is known about variability in its scale dependence or causal processes. The scale-dependent variability of latitudinal gradients in species richness can be affected by latitudinal differences in (i) the regional relative abundance distribution, and (ii) the degree of aggregated distribution (i.e., intraspecific aggregation and interspecific segregation; henceforth, the degree of aggregation) reflecting differences in ecological processes among regions, which are not mutually exclusive. 2. In rocky intertidal sessile animal assemblages along Japan s Pacific coast (between 31 N and 43 N), scale-dependent variability of the latitudinal gradient in species richness and its causal mechanisms were examined by explicitly incorporating three hierarchical spatial scales into the monitoring design: plots ( cm), shores (78 to 235 m), and regions (16 7 to 42 5 km). 3. To evaluate latitudinal differences in the degree of aggregation, the degree of intraspecific aggregation at each spatial scale in each region was examined using the standardized Morishita index. Furthermore, the observed species richness was compared with the species richness expected by random sampling from the regional species pool using randomization tests. 4. Latitudinal gradients in species richness were observed at all spatial scales, but the gradients became steadily more moderate with decreasing spatial scale. The slope of the relative abundance distribution decreased with decreasing latitude. 5. Tests of an index of intraspecific aggregation and randomization tests indicated that although species richness at smaller scales differed significantly from species richness expected based on a random distribution, the degree of aggregation did not vary with latitude. Although some ecological processes (possibly species sorting) may have played a role in determining species richness at small spatial scales, the importance of these processes did not vary with latitude. 6. Thus, scale-dependent variability in the latitudinal gradient of species richness appears to be explained mainly by latitudinal differences in the regional relative abundance distribution by imposing statistical constraint caused by decreasing grain size. Key-words: additive diversity components, α- and γ-diversity, hierarchically nested design, latitudinal diversity cline, occurrence-based randomization test *Correspondence author. okudy@affrc.go.jp 2008 The Authors. Journal compilation 2008 British Ecological Society

2 Latitudinal gradients in species richness 329 Fig. 1. Illustrations of scale-dependent variability in the latitudinal gradient in species richness derived by (a, b) latitudinal differences in relative abundance distribution and (c, d) latitudinal difference in the degree of aggregated distribution (henceforth, aggregation). (a) Regional relative abundance curve in low latitude and high latitude. The species rank represents the sequence from the most abundant species (a rank of 1) to the least abundant species (higher numbers). Dashed lines indicated the magnitude of the reduction in local species richness based on a random and passive sampling process at each spatial scale. Differences in colour and shape of symbols denote differences in latitude and spatial scale where the number of species is calculated, respectively; black means high latitude, white means low latitude, circle means large spatial scale, and square means small spatial scale. For example, the number of species at low latitude and at a large scale (i.e., white circle) is shown by the length of the black arrow. (b) The latitudinal gradient in local species richness at each spatial scale. Black and white objects correspond to the number of species in graph (a). (c) Latitudinal differences in the degree of aggregation. In situation (i), there is no regional difference in the degree of aggregation as a function of latitude. In situation (ii), the degree of aggregation increase with decreasing latitude. (d) The degree of modification of local species richness as a function of spatial scale. White and black arrows indicate the degree of modification of local species richness in situations (i) and (ii), respectively. Differences in the degree of modification of local species richness make the slope of the latitudinal gradient in local species richness gentler in situation (ii) than in situation (i). Introduction Latitudinal gradients in regional species richness are a commonly observed pattern in diverse taxonomic groups in a range of habitats (e.g., Rosenzweig 1995; Gaston & Blackburn 2000; Hillebrand 2004a). In general, although latitudinal gradients in species richness are clear and their slope is steep at very large spatial scales (e.g., a regional community: Cornell & Lawton 1992; Srivastava 1999), the gradients become less clear and have gentle slopes at smaller spatial scales (e.g., for a local community: Hubbell 1979; Clarke & Lidgard 2000; Hillebrand 2004a). Little is known, however, about how and why latitudinal gradients in species richness vary across multiple spatial scales. Two different processes may make the latitudinal gradient in species richness less clear and gentler at smaller spatial scales, and these processes are not mutually exclusive. First, the degree of modification of local species richness as a function of spatial scale (i.e., grain size) may be larger in regions where the slope of the regional relative abundance distribution is lower (Fig. 1a,b). Decreasing grain size imposes a simple statistical constraint that can explain reductions in species richness at smaller spatial scales (e.g., Williamson 1988; Rosenzweig 1995). Two previous studies indicated that the slope of the relative abundance curves as a function of scale becomes steeper at higher latitudes, which could be caused by evolutionary and biogeographical processes such as latitudinal changes in the speciation rate or in climatic factors (Hubbell 1979; Buzas & Culver 1999). Second, the degree of modification of local species richness as a function of spatial scale (grain size) may be larger in regions where the degree of aggregated distribution (i.e., intraspecific aggregation and interspecific segregation; henceforth, the degree of aggregation) is stronger (Fig. 1c,d). Persistent co-existence of competing species requires that intraspecific competition be greater than interspecific competition (e.g., Chesson 2000); such a situation may arise from aggregation (e.g., Chesson 2000). Increasing the degree of aggregation imposes a constraint that can explain the reduction in species richness at smaller spatial scales (e.g., Veech, Crist & Summerville 2003). Thus, the degree of aggregation would be stronger at lower latitudes. Aggregation can result from ecological processes such as species sorting via differences in the ability of species to perform under different environmental conditions, local dispersal limitations, and aggregative behaviour (Veech 2005). This suggests that the strength of aggregation may provide key insights into the ecological processes that determine community structure and species richness (Crist et al. 2003; Veech 2005).

3 330 T. Okuda et al. Fig. 2. An illustration of the hierarchically nested sampling design used in the study. Five rocky shores (black solid squares) were chosen for the census of intertidal organisms in each of the six regions (grey solid squares) along the Pacific coast of Japan between 31 N and 43 N. The withinhabitat species richness parameters (α 1, diversity within a plot; α 2, diversity within a shore; γ, diversity within a region) were calculated based on the hierarchical spatial arrangement of the study sites. An effective approach for investigating the patterns and causal processes responsible for scale-dependent variability in latitudinal gradients in species richness would be to compare these gradients at multiple spatial scales using a hierarchical sampling design (see Lande 1996; Crist et al. 2003; Noda 2004). The degree of statistical constraints on species richness that are derived from random and passive sampling processes depend on spatial scale (i.e., grain size; Fig. 1). Furthermore, the processes that determine community structure, which in turn results from intraspecific aggregation and interspecific segregation, change across spatial scales (e.g., Peterson & Parker 1998; Huston 1999). Russell et al. (2006) showed the existence of scale-dependent variability in the degree of community saturation, which indicate the influence of local ecological interactions on determining local species richness, by using a hierarchical sampling design in rocky intertidal communities along the West Coast of the USA between 33 N and 48 N. However, they did not focus on the scale-dependent variability in latitudinal patterns of species richness or on the underlying causal processes. In the present study, scale-dependent variability in latitudinal gradients in species richness and the underlying causal mechanisms for rocky intertidal sessile animals were examined by testing two predictions: (i) latitudinal differences in the regional relative abundance distribution cause scaledependent variability in the latitudinal gradient in species richness through statistical constraints that result from decreasing grain size; (ii) latitudinal differences in the degree of aggregation generate scale-dependent variability in the latitudinal gradients in species richness. To test these hypotheses, a field census was conducted along the Pacific coast of Japan between 31 N and 43 N that explicitly incorporated a hierarchy of spatial scales in the census design. Materials and methods CENSUS DESIGN A hierarchically nested sampling design (Noda 2004) was used to measure the scale-dependent variability of the latitudinal gradient in species richness of intertidal sessile animals. Six regions were chosen along the Pacific coast of Japan between 31 N and 43 N, with intervals between neighbouring regions ranging from 283 to 530 km (Fig. 2). Within each region, five shores were chosen at intervals of 4 to 25 km along the coastline. Within each shore, five census plots were established on rock walls at semi-exposed locations, with intervals between neighbouring plots ranging from 3 to 378 m (mean ± SD: 37 2 ± 49 0 m). The six regions reflect the spatial extent of regional communities of rocky intertidal sessile animals in Japan. Climate differs among the regions, but is relatively uniform within each region. Warm and cold currents also differ among the regions but not within a region (Okuda et al. 2004). In addition, Nakaoka et al. (2006) showed that

4 Latitudinal gradients in species richness 331 Table 1. Variables and spatial scales of the within-habitat species richness at different levels applied in this study Variable Definition (number of replicates within a region) Spatial scale Spatial extent (range) Surveyed area α 1 Within-plot diversity (n = 25) 1 m a 0 5 m 2 α 2 Within-shore diversity (n = 5) 129 m b (78 to 235 m) 2 5 m 2 γ Diversity within a region m c ( to m) 12 5 m 2 a Height of the plot; b mean (range) of regional mean distances between the two most distant plots within a shore; c mean (range) of distances between the two most distant plots within a region. the regional species composition estimated by summing local communities differed among regions but not within each region. More detailed descriptions of the study sites and their biogeographical features were provided by Okuda et al. (2004) and Nakaoka et al. (2006). The angles of the rock walls with respect to the vertical in the plots varied between 31 and 133 (mean ± SD: 73 2 ± 17 7 ). Although the slope varied widely across sites, 85% of the walls had slopes of between 50 and 100. To test whether the angle of the rock wall affected the latitudinal pattern of species richness, the best subset of variables (latitude, residual of the angles, and their interaction) for predicting species richness in a plot was determined by means of multiple regression based on Bayesian information criteria (BIC). To avoid collinearity between predictors, the analysis was conducted after centring both the latitude and the residual of the angle (Quinn & Keough 2002). In this analysis, only latitude was selected within the best subset of predictors (BIC = ). Therefore, variations in the angle of the rock wall do not appear to have created any significant bias in the present study. Each plot was 50 cm wide by 1 m high, with the mean tidal level located at the middle of the vertical range. This plot width is commonly used in investigations of local communities in rocky intertidal zones, and spatial variation in community structure has been shown to reflect the heterogeneity of biotic and abiotic conditions within this zone (e.g., Menge 1976; Navarrete 1996). The midpoints and corners of all plots were permanently marked with plastic or stainless steel anchors. Census plots were randomly selected from relatively steep rock slopes, thus some census plots contained cracks in the rock, but tide pools were not included in the sample. Sessile intertidal animals are affected by vertical environmental gradients (e.g., due to desiccation stress), and the abundance of each species varies greatly depending on changes in tidal level on scales ranging from tens of centimetres to several metres (Bertness et al. 2006). Therefore, each plot was divided vertically into 10 quadrats measuring 50 cm wide by 10 cm high, and species occurrence was surveyed in each quadrat. The proportion of the vertical range in the intertidal zone covered by the 1-m plots varied among regions, ranging from 73% at the Oshima Peninsula (a vertical extent of 137 cm between the mean high water and mean low water of the spring tides) to 41% at the Osumi Peninsula (a tidal extent of 242 cm). This proportion decreased with decreasing latitude. The presence or absence of all sessile animals that could be identified with the naked eye in the field (i.e., size > 2 mm) was determined within each quadrat during low tide in the summer (July and August 2003), early winter (November and December 2003), and spring (April and May 2004). A complete list of these animals is presented in Appendix. In addition, coverage by each sessile animal species was surveyed by counting the occurrence at 20 points per quadrat (i.e., 200 points per plot) positioned at intervals of 5 cm in both the vertical and the horizontal directions. SPECIES RICHNESS To accurately measure the seasonal variation in species richness in a local community, Moreno & Halffter (2001) recommended using either average species diversities over several seasons or the cumulative species diversities throughout several seasons. In this study, annual average species diversities were used to describe the species richness of each community, because the local community was defined as an assemblage that would interact within a single generation at the smallest spatial scale (Srivastava 1999). The α diversities (within-habitat diversities) were calculated at two spatial levels based on the hierarchical arrangement of the study sites (Fig. 2): α 1 diversity, the within-plot species richness, and α 2 diversity, the within-shore species richness. In addition, γ diversity, the within-habitat species richness at the broadest scale (within a region) was also calculated (Table 1). A regional community is generally defined as a community whose size is 100 or more times that of the local community and in which evolutionary and biogeographical processes are the dominant determinants of the community structure (e.g., Cornell & Lawton 1992; Srivastava 1999). The extent of the region defined in our study agrees with this definition (Table 1). The occurrence of each species at the shore and regional scales was surveyed by sampling a small portion of the entire area (Table 1). The α 1 diversity was calculated as the raw species richness, whereas the α 2 and γ diversities were estimated using a second-order jack-knife richness estimator (JACK2) following the method recommended by Brose, Martinez & Williams (2003). JACK2 was calculated as follows: S JACK2 Q1( 2m 3) Q2( m 2) = Sobs + + m mm ( 1) eqn 1 where S JACK2 is the estimated species richness (JACK2 estimator), S obs is the raw species richness, Q i is the number of species that occur in exactly i samples, and m is the total number of samples. REGIONAL RELATIVE ABUNDANCE DISTRIBUTION The relative abundance distribution of each species was calculated from the coverage of the plots by each sessile animal in each season, this value was applied at a regional scale in each season. Because the abundance of rare species cannot be estimated by the census procedure used in this study (i.e., the dot-counting method), the relative occurrence distributions, which were calculated from the presence or absence of each species of sessile animal at each plot in each season, 2

5 332 T. Okuda et al. were also compared as a surrogate for abundance of rare species (Buzas & Culver 1999). DEGREE OF AGGREGATION To examine the degree of aggregation, there are two useful classes of analysis: (i) estimation of intraspecific aggregation (Veech 2005) and (ii) a randomization test of species richness (e.g., Crist et al. 2003). Index of intraspecific aggregation The degree of intraspecific aggregation was examined using mean values of the standardized Morisita index (I MS : see Krebs 1999), which is robust with respect to variation in sample number and size (i.e., the abundance of a species in each sample). The index was calculated for each species at three spatial scales: (i) among the 10 quadrats within a plot (vertical), (ii) among the five plots within a shore (horizontal), and (iii) among the five shores within a region (horizontal). These calculations were based on the coverage data for all sessile animal species. Values of I MS range from 1 to 1, with 95% confidence limits at 0 5 and 0 5 (Krebs 1999). Values of I MS < 0 indicate intraspecific repulsion, whereas I MS > 0 indicates intraspecific aggregation (Krebs 1999; Veech 2005). The mean I MS of all species was calculated for each sample. In all cases, singletons (i.e., species found at only a single grid point) were eliminated from the calculation to avoid incorrect estimation of I MS (Veech 2005). Randomization test To reveal the degree of aggregation of a community, an indicator of the degree of deviation between observed (obs) and expected (exp) diversities at each spatial scale were calculated [obs/exp(α 1 ) and obs/ exp(α 2 )]. The expected diversities were obtained based on the null hypothesis that the presence or absence of each species in a quadrat at each tidal level is randomly determined from the hypothetical regional species pool according to the frequency of occurrence in all quadrats at the same tidal level in each region. The following is an example of the randomization routine. First, based on the observed occurrence of species A at tidal level 1 for the region as a whole, the presence or absence of species A was randomly determined at tidal level 1. By repeating this operation for all species observed in the region, a null quadrat community, in which intraspecific aggregation of all species did not differ from that in the real samples but in which the pattern of occurrence of species is random, was created for tidal level 1 (i.e., quadrat reassignments were independent among species). A null community within a plot was developed by repeating these operations from tidal level 1 to tidal level 10, and a null community within a shore was obtained by summing five null plot communities. Finally, expected diversities in each region were obtained by averaging values of α-diversity for the null communities in each season. The randomization routine was repeated 1000 times for each region and season to obtain means and 95% confidence intervals for obs/exp(α). In the randomization test, the regional species assemblage (instead of the shore-level assemblages) was treated as the species pool that is potentially able to exist in this community (Dupre 2000). The predominant environment in the shores that were studied was a rocky coastline, and the distances between pairs of shores ranged from 4 to 42 km. This suggests that dispersal limitations would rarely cause separation of the species pool over multiple generations for marine invertebrate larvae, whose dispersal distance ranges from tens of metres to several hundred kilometres (e.g., Kinlan & Gaines 2003). In this randomization test, species composition obtained from the five shores within each region were summed and used as a surrogate for the regional species pool, because a published species list is not available for each region. Advantages and drawbacks of each analysis The index of intraspecific aggregation can directly test the degree of intraspecific aggregation of each species by using the variation in the number of individuals of a given species among the communities. This index, however, cannot evaluate the degree of intraspecific aggregation of rare species (e.g., singletons) and interspecific segregation (e.g., Veech 2005). Randomization tests can examine the net effect on species richness of both intraspecific aggregation and interspecific segregation by using abundance or occurrence data (e.g., Crist et al. 2003). The results of randomization tests, however, are affected not only by the degree of nonrandom spatial distribution but also by the relative abundance distribution of regional species pool, and by the sampling effort; that is, the number of individuals within each sample (T. Hagino and T. Noda, unpublished). To fully reveal the degree of aggregation, both approaches should be applied together. Results PATTERN OF SPECIES RICHNESS Although a clear latitudinal gradient in γ-diversity was observed, the latitudinal gradients in the two α-diversities became progressively less clear (and the slopes of the gradients become gentler) with decreasing spatial scale (Fig. 3). A significant negative linear relationship between species richness and latitude was obtained (r 2 = 0 669). Marginally significant Fig. 3. Latitudinal trends in regional species richness and their spatial components as measured by three species diversity parameters for intertidal sessile animals along the Pacific coast of Japan. Vertical bars indicate the standard deviations of α 2 diversity (n = 5) and α 1 diversity (n = 25). The JACK2 estimators of within-habitat species richness at the regional and shore scales are represented by the following symbols: γ (species richness within a region) ( ), r 2 = 0 669, P = 0 047, y = x; α 2 (regional mean of species richness within a shore) ( ), r 2 = 0 626, P = 0 061, y = x. The raw data for the within-habitat species richness at a plot scale (α 1 diversity) are represented by ( ): r 2 = 0 567, P = 0 084, y = x.

6 Latitudinal gradients in species richness 333 Fig. 4. Differences in the regional relative abundance distributions and relative occurrence distributions among regions. (a) (c): The relative abundance distributions are based on abundance of each species calculated from coverage data. (d) (f ): The relative occurrence distributions are based on the presence or absence of each species. The species rank represents the sequence from the most abundant or occurrent species (a rank of 1) to the least abundant or occurrent species (higher numbers). In (c), there is no line for relative abundance in eastern Hokkaido because only one abundant species appeared in the coverage census in April and May (0 05 < P < 0 10) negative relationships were obtained between latitude and α 2 -diversity (r 2 = 0 626) and α 1 -diversity (r 2 = 0 567). The JACK2 estimators of species richness were strongly correlated with the raw species richness data at broadest spatial scale (γ, r = 0 998; α 2, r = 0 989; n = 6). REGIONAL RELATIVE ABUNDANCE DISTRIBUTION The slopes of the curves for relative abundance and relative occurrence tended to be steeper at higher latitudes (Fig. 4). The slopes of the curves for eastern Hokkaido and the Oshima Peninsula were steeper than those for the Osumi Peninsula and Kii Peninsula. Furthermore, the length of the curve, especially for species with a relative abundance of < 1% (rare species), tended to be longer at lower latitudes. INDEX OF INTRASPECIFIC AGGREGATION The mean I MS value for horizontal intraspecific aggregation was larger than the significance threshold for aggregation (I MS > 0 5) among plots within a shore in all regions except for the Oshima Peninsula and among shores within a region except for the Boso Peninsula (Table 2). On average, 73 and 78% of the assemblages were aggregated among plots within a shore (mean ± SD: 72 7 ± 25 9%) and among shores within a region (mean ± SD: 77 8 ± 27 3%), respectively. Although a trend of vertical intraspecific aggregation (0 < I MS < 0 5) was observed in all regions, this trend was not statistically significant (Table 2). On average, 37% of assemblages were vertically aggregated within a plot (mean ± SD: 37 2 ± 18 0%). The standardized values of intraspecific aggregation (I MS ) did not vary significantly with latitude at any spatial scale (plot level, r 2 = 0 211; shore level, r 2 < 0 001; regional level, r 2 = 0 001; Fig. 5). Vertical aggregation tended to increase with increasing latitude, but this tendency was not statistically significant (P = 0 359). RANDOMIZATION TEST Whereas the α 1 and α 2 diversities were significantly smaller than the corresponding expected values, the obs/exp(α) ratio did not vary significantly with latitude at both spatial scales (Fig. 6). The linear regressions for obs/exp(α) as a function of latitude were not significant (α 1, r 2 = 0 125; α 2, r 2 = 0 507), and in both regression equations, the slope of the regression were small (α 1, 0 003; α 2, 0 011). Discussion SCALE- DEPENDENT VARIABILITY IN THE LATITUDINAL GRADIENT IN SPECIES RICHNESS In rocky intertidal sessile animal assemblages along the Pacific coast of Japan, latitudinal gradients in species richness became progressively less clear and the slopes of the gradients became progressively gentler with decreasing spatial scales

7 334 T. Okuda et al. Table 2. Mean standardized values of Morisita index (I MS ) and the proportions of aggregated communities at each of the three spatial scales in each region. The 95% confidence limits for I MS are 0 5 and 0 5 Within a plot (vertical aggregation) Among plots within a shore (horizontal aggregation) Among shores within a region (horizontal aggregation) Region Mean I MS Proportion of Proportion of Proportion of aggregated communities a Mean I MS aggregated communities b Mean I MS aggregated communities c Osumi Peninsula Kii Peninsula Boso Peninsula Sanriku Coast Oshima Peninsula Eastern Hokkaido a Proportion of the within-plot communities whose mean I MS is statistically significant (I MS > 0 5); b proportion of within-shore communities whose mean I MS is statistically significant (I MS > 0 5); c proportion of within-region communities whose mean I MS is statistically significant (I MS > 0 5). Fig. 5. Latitudinal patterns in the mean standardized Morisita index (I MS ) at each spatial scale in each region. Dashed lines are 95% confidence limits for this index (i.e., I MS = 0 5). Values of I MS < 0 indicate intraspecific repulsion, whereas I MS > 0 indicates intraspecific aggregation. (a) Mean I MS within a plot (vertical aggregation): r 2 = 0 211, P = 0 359, y = x. (b) Mean I MS within a shore (horizontal aggregation): r 2 < 0 001, P = 0 993, y = ( x). (c) Mean I MS within a region (horizontal aggregation): r 2 = 0 001, P = 0 952, y = x. Vertical bars indicate standard deviations of the mean I MS within a plot (n = 25) and of the mean I MS within a shore (n = 5). Fig. 6. Latitudinal pattern in the ratio of observed (obs) to expected (exp) species richness (obs/exp). Solid lines are regression lines for each component of species richness. Vertical range bars are the 95% confidence interval of the ratio of observed to expected species richness, which represents the observed species richness divided by the lower and upper 95% confidence limits of expected species richness obtained from the randomization test. When a vertical range does not crosses the obs/exp = 1 line, observed species richness differs significantly from the expected value. (a) Obs/exp(α 1 ) (the ratio within a plot): r 2 = 0 125, P = 0 492, y = x. (b) Obs/exp(α 2 ) (the ratio within a shore): r 2 = 0 507, P = 0 112, y = x. (Fig. 3). These results concurred with previous studies that reported a clear latitudinal gradient in γ diversity but a weak latitudinal gradients in α diversity for various marine habitats around the world (Buzas & Culver 1999; Clarke & Lidgard 2000; Karlson, Cornell & Hughes 2004; Witman, Etter & Smith 2004). Furthermore, Hillebrand (2004b) used a metaanalysis of the results from 232 reports to demonstrate that latitudinal gradients in species richness in marine communities

8 Latitudinal gradients in species richness 335 were, on average, both stronger and steeper in regional data sets (γ diversity) than in local ones (α diversity), although latitudinal gradients in α diversity were significant. Thus, latitudinal gradients in species richness clearly exist at a regional scale, although the gradients are still significant at local scales (Hillebrand 2004b). It is likely that the spatial and temporal range of locality was arbitrary in the present study. In the spatial definition of locality, the census plots were each 1 m in height, and the proportion of the vertical tidal range covered by the census plots varied among regions, ranging from 41 to 73%. If equal proportions of the tidal range (e.g., 50% of this range) had been covered in each region, the latitudinal gradient in species richness could have been steeper than the present results. This is because the proportion of the tidal range covered by our census plots decreased with decreasing latitude, and latitude was negatively correlated with the species richness components at each spatial scale (regression slopes: α 1, 10 41; α 2, 25 32; γ, 46 57). Two temporal definitions of localities (seasonal average and cumulative) have been used to calculate local species richness in previous studies (e.g., Moreno & Halffter 2001). Both definitions explain a discrete aspect of the local community. Cumulative local species richness involves species that do not directly interact within a single generation but that instead interact indirectly over several generations. By contrast, seasonal average species richness does not reflect all interacting species, but only the species present in each season; this situation seems to correspond to the definition of a local community by Srivastava (1999) as a group of organisms in a small, environmentally homogeneous area in which all the species can encounter each other within some unit of ecological time (e.g., a single generation). Differences in these two temporal definitions of locality may not have seriously affected the results of the present study. This is because most sessile animals are present throughout the year in our study area. Furthermore, there were strong correlations between average and cumulative species richness at each spatial scale (α 1, r = 0 997; α 2, r = 0 989; γ, r = 0 973; n = 6). MECHANISMS THAT DETERMINE AGGREGATION AT SMALLER SPATIAL SCALES Significant aggregation was detected at the plot and shore levels. These aggregations may be caused by ecological processes such as dispersal limitation and species sorting via differences in the ability of a species to perform under different environmental conditions. Dispersal limitations may cause intraspecific aggregation at the shore level, because distances between pairs of shores in this study (range, 4 42 km) were probably sufficient to cause aggregated settlement patterns for marine invertebrate larvae, whose dispersal distances range from tens of metres to several hundred kilometres (e.g., Kinlan & Gaines 2003). However, a previous study did not detect a negative correlation between community similarity and geographical distance among shores (Nakaoka et al. 2006). This suggests that dispersal limitations may not play an important role in determining the species composition of the intertidal sessile animals in the present study. Species sorting via differences in ability of a species to perform under different environmental conditions may cause intraspecific aggregation among plots and shores. It is well known that the frequency of disturbance (e.g., Dayton 1971; Wootton 1998), intensity of predation (e.g., Menge et al. 1994; Navarrete 1996), and wave intensity (e.g., Hawkins et al. 1992; Harley & Helmuth 2003), which change within small spatial scales such as shore level, are important niche axes that determine community structure of rocky intertidal sessile animals. The amount of phytoplankton and nutrient availability, which change along larger spatial scales such as among shores and among regions, has a significant bottom-up effect on the rocky intertidal sessile animal assemblages (e.g., Menge et al. 1997). MECHANISMS THAT CAUSE SCALE- DEPENDENT VARIABILITY IN THE LATITUDINAL GRADIENT IN SPECIES RICHNESS Neither the mean I MS values nor the obs/exp(α) ratio showed the presence of a significant latitudinal gradient at any of the three spatial scales, indicating that the degree of aggregation did not vary with latitude. In contrast, the slopes of both the relative abundance and relative occurrence curves at a regional scale tended to be steeper at higher latitudes. Thus, the main mechanism responsible for decreasing steepness of the latitudinal gradient in species richness and making the existence of a gradient less clear at smaller spatial scales is not latitudinal changes in the degree of aggregation caused by ecological processes, but instead results from a latitudinal change in the slope of the regional relative abundance distribution. This raises the following question concerning latitudinal gradients in species richness: Why does the increasing size of the regional species pool at lower latitudes fails to cause stronger intraspecific aggregation? This question emerges naturally because the increment in the size of regional species pool may facilitate species interactions, consequently increasing species sorting at the local communities (e.g., Cornell & Lawton 1992; Hillebrand 2004a). The strength of a negative biological interaction that causes aggregation may not increase with increasing size of the regional species pool when mechanisms such as low productivity or frequent disturbance weaken the effect of interspecific competition, which would otherwise increase with increasing regional species richness (Huston 1979). In the present study, productivity (as measured using the phytoplankton concentration in the coastal water and the recruitment of larval barnacles) tended to increase with increasing latitude, whereas the disturbance frequency (calculated from the formation of bare space) did not vary with respect to latitude (T. Okuda, N. Ito, T. Yamamoto, unpublished). Conclusion In rocky intertidal sessile animal assemblages along the Pacific coast of Japan (between 31 N and 43 N), latitudinal

9 336 T. Okuda et al. gradients in species richness became less clear and their slopes became gentler at smaller spatial scales. It is likely that although some ecological processes (possibly species sorting) played a role in determining species richness at smaller spatial scales, the importance of these processes did not vary with latitude. Thus, scale-dependent variability in the latitudinal gradient in species richness appears to have been explained mainly by latitudinal differences in the regional relative abundance distribution as a result of statistical constraints related to the decreasing grain size at smaller spatial scales. To better understand the latitudinal gradients in species richness, researchers should further investigate latitudinal patterns in the regional relative abundance distributions and their underlying causal mechanisms, including a potentially higher speciation rate and lower extinction rate at lower latitudes (Hubbell 2001) and latitudinal differences in niche apportionment or the amount of available resources (Tokeshi 1999). Acknowledgements For providing access to laboratory facilities, we are grateful to the staff and students of the Akkeshi and Usujiri Marine Stations of Hokkaido University, the International Coastal Research Center of the Ocean Research Institute, The University of Tokyo, the Marine Biosystems Research Center of Chiba University, the Seto Marine Biological Laboratory of Kyoto University, and the Education and Research Center for Marine Environment and Resources of Kagoshima University. This study was made possible by the generous support and encouragement of local fishermen and fishery offices of the Fisherman s Cooperative Associations of Hokkaido, Iwate, Chiba, Wakayama, and Kagoshima Prefectures. We thank everyone who helped us with our fieldwork and data analyses, but particularly thank N. Kouchi for identification of the intertidal sessile animals, and Dr. D. Munroe for critically reading and checking the English text. This research was supported by a grant-in-aid from the Ministry of Education, Science, Culture and Sports, Japan (No ). References Bertness, M.D., Crain, C.M., Silliman, B.R., Bazterrica, M.C., Reyna, M.V., Hildago, F. & Farina, J.K. (2006) The community structure of Western Atlantic Patagonian rocky shores. Ecological Monographs, 76, Brose, U., Martinez, N.D. & Williams, R.J. (2003) Estimating species richness: sensitivity to sample coverage and insensitivity to spatial patterns. Ecology, 84, Buzas, M.A. & Culver, S.J. (1999) Understanding regional species diversity through the log series distribution of occurrences. Diversity and Distributions, 8, Chesson, P. (2000) Mechanisms of maintenance of species diversity. Annual Review of Ecology and Systematics, 31, Clarke, A. & Lidgard, S. (2000) Spatial patterns of diversity in the sea: bryozoan species richness in the North Atlantic. Journal of Animal Ecology, 69, Cornell, H.V. & Lawton, J.H. (1992) Species interactions, local and regional processes, and limits to the richness of ecological communities a theoretical perspective. Journal of Animal Ecology, 61, Crist, T.O., Veech, J.A., Gering, J.C. & Summerville, K.S. (2003) Partitioning species diversity across landscapes and regions: a hierarchical analysis of α, β and γ diversity. American Naturalist, 162, Dayton, P.K. (1971) Competition, disturbance, and community organization: the provision and subsequent utilization of space in a rocky intertidal community. Ecological Monographs, 41, Dupre, C. (2000) How to determine regional species pool: a study in two Swedish regions. Oikos, 89, Gaston, K.J. & Blackburn, T.M. (2000) Pattern and Process in Macroecology. Blackwell, Oxford, UK. Harley, C.D.G. & Helmuth, B.S.T. (2003) Local- and regional-scale effects of wave exposure, thermal stress, and absolute versus effective shore level on patterns of intertidal zonation. Limnology and Oceanography, 48, Hawkins, S.J., Hartnoll, R.G., Kain, J.M. & Norton, T.A. (1992). Plant-animal interactions on hard substrata in the North-east Atlantic. Plant-Animal Interaction in Marine Benthos. (eds D.M. John, S.J. Hawkins & J.H. Price), Vol. 46, pp Clarendon Press, Oxford, UK. Hillebrand, H. (2004a) On the generality of the latitudinal diversity gradient. American Naturalist, 163, Hillebrand, H. (2004b) Strength, slope and variability of marine latitudinal gradients. Marine Ecology Progress Series, 273, Hubbell, S.P. (1979) Tree dispersion, abundance, and diversity in a tropical dry forest. Science, 203, Hubbell, S.P. (2001) The unified neutral theory of biodiversity and biogeography. Princeton University Press, Princeton, NJ. Huston, M. (1979) A general hypothesis of species diversity. American Naturalist, 113, Huston, M.A. (1999) Local processes and regional patterns: appropriate scales for understanding variation in the diversity of plants and animals. Oikos, 86, Karlson, R.H., Cornell, H.V. & Hughes, T.P. (2004) Coral communities are regionally enriched along an oceanic biodiversity gradient. Nature, 429, Kinlan, B.P. & Gaines, S.D. (2003) Propagule dispersal in marine and terrestrial environments: a community perspective. Ecology, 84, Krebs, C.J. (1999) Ecological Methodology. 2nd edn. Benjamin Cummings Publishers, San Francisco, California. Lande, R. (1996) Statistics and partitioning of species diversity, and similarity among multiple communities. Oikos, 76, Menge, B.A. (1976) Organization of the New England rocky intertidal community: role of predation, competition and environmental heterogeneity. Ecological Monographs, 46, Menge, B.A., Berlow, E.L., Blanchette, C.A., Navarrete, S.A. & Yamada, S.B. (1994) The keystone species concept variation in interaction strength in a rocky intertidal habitat. Ecological Monographs, 64, Menge, B.A., Daley, B.A., Wheeler, P.A., Dahlhoff, E., Sanford, E. & Strub, P.T. (1997) Benthic-pelagic links and rocky intertidal communities: bottomup effects on top-down control? Proceedings of the National Academy of Sciences, USA, 94, Moreno, C.E. & Halffter, G. (2001) Spatial and temporal analysis of a, b and g diversities of bats in a fragmentation landscape. Biodiversity and Conservation, 10, Nakaoka, M., Ito, N., Yamamoto, T., Okuda, T. & Noda, T. (2006) Similarity of rocky intertidal assemblages along the Pacific coast of Japan: effects of spatial scales and geographic distance. Ecological Research, 21, Navarrete, S.A. (1996) Variable predation: effects of whelks on a mid-intertidal successional community. Ecological Monographs, 66, Noda, T. (2004) Spatial hierarchical approach in community ecology: a way beyond high context-dependency and low predictability in local phenomena. Population Ecology, 46, Okuda, T., Noda, T., Yamamoto, T., Ito, N. & Nakaoka, M. (2004) Latitudinal gradient of species diversity: multi-scale variability in rocky intertidal sessile assemblages along the Northwestern Pacific coast. Population Ecology, 46, Peterson, D.L. & Parker, V.T. (1998) Ecological Scale. Columbia University Press, New York. Quinn, G.P. & Keough, M.J. (2002) Experimental Design and Data Analysis for Biologists. Cambridge University Press, Cambridge, UK. Rosenzweig, M.L. (1995) Species Diversity in Space and Time. Cambridge University Press, Cambridge, UK. Russell, R., Wood, S.A., Allison, G. & Menge, B.A. (2006) Scale, environment, and trophic status: the context dependency of community saturation in rocky intertidal communities. American Naturalist, 167, E158 E170. Srivastava, D.S. (1999) Using local-regional richness plots to test for species saturation: pitfalls and potentials. Journal of Animal Ecology, 68, Tokeshi, M. (1999) Species Coexistence. Blackwell, London. Veech, J.A. (2005) Analyzing patterns of species diversity as departures from random expectations. Oikos, 108, Veech, J.A., Crist, T.O. & Summerville, K.S. (2003) Intraspecific aggregation decreases local species diversity of arthropods. Ecology, 84, Williamson, M. (1988). Relationship of species number to area, distance and other variables. Analytical Biogeography. (eds A.A. Myers & P.S. Giller), pp Chapman & Hall, London. Witman, J.D., Etter, R.J. & Smith, F. (2004) The relationship between regional and local species diversity in marine benthic communities: a global perspective. Proceedings of the National Academy of Sciences, USA, 101, Wootton, J.T. (1998) Effects of disturbance on species diversity: a multitrophic perspective. American Naturalist, 152, Received 20 August 2008; accepted 25 September 2008 Handling Editor: Andy Gonzalez

10 Latitudinal gradients in species richness 337 Supporting Information Additional supporting information may be found in the online version of this article. Appendix S1 List of all sessile animals observed in this study. + indicates the occurrence of each species. indicates the absence of each species Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting material, supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.