Figure 6.

(a) The change in abundance as a function of the RCI for the 16 species that occur in both the Pleistocene
storm and Recent Storm deposit. We calculated the change in abundance as maximum abundance minus minimum
abundance divided by maximum abundance. The positive correlation is highly significant (n = 16, r = 0.800, y =
0.37724 + 0.54946x, p < 0.0001) indicating that the species near the edge of their geographic ranges experienced
greater temporal variation in population size. (b) The change in abundance as a function of In (mean abundance) for
the same species as (a). The lack of a relationship between these variables (n = 16, y = 0.67742 - 1.6701e - 2x, r =
0.1260 p = 0.6417) suggests that temporal changes in abundance may have been more influenced by the distance
from the centre of the range than by abundance per se.

the abundances between those species that turned over (median abundance = 12, n = 51) and those that persisted
(median abundance = 28.25, n = 16; Mann-Whitney W = 649.5, p = 0.1223). The abundance for persistent species
was calculated as the mean abundance between Pale• Storm and Recent Storm. It appears as if species near the
edges of their geographic range were more likely to show local colonization or extinction and that such turnover was
more influenced by position in the geographic range than by local abundance per se.
Discussion

Inferring Pleistocene ranges from contemporary distributions

We first address two potential problems in our use of data on geographic ranges for contemporary molluscs to make
inferences about abundance and distribution during and since the Pleistocene. First, range boundaries at the time
that the Paleo samples were deposited might have been very different from those at present. Second, apparent
colonization and extinction events between the Paleo and Recent samples might have reflected major shifts in
ranges rather than local turnover of species.

Several kinds of evidence suggest that range boundaries of most of the species might have been similar at the times
when both the Paleo and Recent samples were deposited. First, of the 25 species in the Pleistocene sample that
have gone locally extinct, only six have modern distributions that do not overlap with our study site, implying a
different Pleistocene range. Furthermore, of these six species, only one (Corbula tenuis) has a contemporary range
limit more than 1 ° latitude from our site. Furthermore, this species exhibited a low abundance that then may have
been near the Pleistocene range boundary (see the Appendix). Only one of the 25 species that went locally extinct at
our study site appeared to have experienced large-range shifts.

Second, there was an almost equal number of colonization and extinction events between the Paleo and Recent
samples (Table 2). This, coupled with the fact that the vast majority of species are nearer the present northern than
southern limits of their ranges (44 and six, respectively), indicates that there have not been large, consistent shifts in
the ranges of many species. If the Paleo and Recent climates had been different, we would expect an unequal
distribution of colonization and extinction events, reflecting coordinated northward or southward shifts in the
ranges of multiple species. Our data suggest that the differences in species composition between the Pleistocene
interglacial and recent samples reflect colonization and extinction responses of species in reflection of their
individualistic requirements. Valentine and Jablonski (1993) suggest a similar explanation for changes in the
composition of fossil mollusc assemblages during the
Pleistocene and Holocene along the Pacific Coast of North America. The fact that the majority of species are nearer
their northern than southern range boundaries reflects the pronounced latitudinal gradient in shallow-water mollusc
species diversity in the Eastern Pacific as a whole (e.g. Jablonski and Valentine 1990) as well as within the Gulf of
California.

Third, if our predictions are sound but there had been major shifts in ranges between Paleo and Recent samples, we
would expect only the Recent assemblage to show the predicted triangular relationship between the abundance and
centre index clearly. The fact that both Paleo samples exhibit constraint patterns virtually indistinguishable from the
Recent one suggests that the Paleo
communities were subjected to the very similar environmental conditions.

We do not want to imply that the Pleistocene interglacial and contemporary range boundaries were necessarily
coincident. The large number of local extinctions and colonizations suggests that the environments were somewhat
different when the Pleistocene and Recent samples were deposited. Thus, it would not be surprising that there had
been some shifts in geographic range
boundaries. Other studies have found associations of Pleistocene interglacial fossil molluscs from the Pacific Coast of
North America that contain species whose contemporary geographic ranges do not overlap. These changes in
community composition and geographic ranges have been also interpreted as reflecting the individualistic responses
of species to somewhat different past and present environments (e.g. Valentine and Jablonski, 1993). Nevertheless,
the fact that position within the present geographic range seems to predict the constraints on past as well as present
abundance, suggests that proximity to the edge of the range indexes the existence of environmental conditions that
preclude high abundance.

Species extending beyond panamic province

We used the range centre index values as a simplifying means by which to estimate the relative latitudinal range
position of each species within our study site. The utility of this measure, as mentioned earlier, ignores other
potentially influential measures of range size (total length or width of range, volume of water mass occupied). Due
to the linear nature of the North and South
American Pacific coastline and narrow continental shelf, the ranges of molluscs can be simplified as linear (Jablonski
and Valentine, 1990). However, our sample site lies within the Gulf of California, approximately 40% of the species
found at our site extend beyond the Gulf of California and thus inhabit both the Gulf of California and Baja coasts
(see the Appendix). This
raises a potential problem. A species whose range includes the Gulf of California and also extends north beyond the
tip of Baja occupies more coastline and potential geographic area than a species whose range is restricted to the
Gulf. Even though they may both have the same northern most range terminus of the head of the Gulf, species
extending beyond the Gulf may potentially confound our analysis by exhibiting different abundance patterns.

We evaluated the robustness of our initial patterns to this effect by comparing the relationship between the relative
abundance and RCI for species extending beyond and those whose range is restricted to the Gulf of California and
the south (see the Appendix). Abundances and RC! values for each species were replotted and the randomization
analysis was repeated for both groups. Species extending beyond the Gulf of California (SEB) yielded abundance/per
cent relative abundance patterns indistinguishable from species restricted to the Gulf and the south (SRT) and to Figs
2b and 4a, (SEB, observed SS distance from constraint = 0.42402, p < 0.001; SRT, observed SS distance from
constraint = 0.66915, p = 0.005) This indicates that the relative latitudinal position as measured by the RCI is not
affected by this biogeographical feature and seems to index conditions dictating local abundance accurately.

Implications of relative position in the geographic range

We conclude that the location of a population within the geographic range of a species influences both the relative
abundance of that species within local and regional assemblages and the magnitude of temporal change in
abundance between long-separated periods of similar environmental conditions. In both cases, location within the
range does not permit accurate prediction of abundance, but it does constrain the range of possible variation.
Interestingly, location within the range appears to be a better predictor of long-term variation in abundance than
initial abundance.

The triangular relationship between relative position in the range and abundance appears to be yet another case
demonstrating the utility of a 'macroecological' approach that attempts to draw constraint envelopes to characterize
ranges of possible or probable variation (Fig. 2c; for other examples, see Brown and Maurer (1987, 1989)). The
constraint of location in the geographic range on maximal local and regional abundance appears to offer an example
of a connection between patterns and processes on local, regional and geographical scales (see Ricklefs, 1987;
Ricklefs and Schluter, 1993) and, thus, between ecology and biogeography and between contemporary and historical
phenomena.

The ability to demonstrate this constraint and to make this connection seems to depend on species-specific niche
requirements of the mollusc species that have not changed significantly in approximately 100 000 years. There is
much paleontological evidence of morphological stasis for periods of thousands to millions of years (e.g. Eldredge
and Gould, 1972; Stanley and Yang, 1987). We observed such stasis in these molluscs. All of the Pleistocene fossils
were indistinguishable from shells of contemporary species. Tull and B6hning-Gaese (1993) have documented virtual
stasis between the Paleo and Recent deposits at this same site in the predatory behavior of naticid and muricid
gastropods as evidenced by the patterns of drilling of shells of prey. Our
results suggest that many ecological relationships of these molluscs have changed very little in approximately 100
000 years (see also Marquet, 1993).

Connections between ecology, biogeography and paleontology

While the results of the present study cannot be taken as unequivocally validating Brown's (1984; modified in Brown
et al. (in press)) niche-based model for the regulation of abundance and distribution, they are consistent with that
model. All of the a priori predictions made on the basis of the model were supported empirically. A reformation of
Hutchinson's (1957) concept of a
multidimensional niche appears to have promise not only in understanding the individualistic pattern of distribution
and abundance exhibited by each species, but also in obtaining more general insights. Thus, although it is almost
tautological to say that each species has a unique set of niche requirements, nevertheless the majority of species
appear to show similar patterns of
variation in abundance over space (Brown et al., in press) and probably also in time.
Furthermore, the ways that the niche requirements and location within the geographic range interact to constrain
the local abundance of each species also constrain the relative abundances of different species within local and
regional assemblages. It is well known that communities contain a few common and many rare species (e.g. Preston,
1948, 1962a,b; MacArthur, 1957; Williams, 1964). The extensions of Brown's (1984) model and the empirical results
presented here suggest that the few common species that dominate local communities are populations near the
centres of their geographic ranges. The converse, however, is not true: locally rare species can represent populations
from anywhere in their species ranges.

Hanski (1982) has suggested that communities comprise what he calls core species, which are abundant within and
tend not to turn over between local assemblages - and what he calls satellite species, which are rare and frequently
turn over between even nearby sample sites. While our results do not show the bimodal segregation into core and
satellite species recognized by Hanski
(1982), they do suggest that those species which frequently co-exist in both space and time tend to be not only the
abundant species but also those near the centres of their geographic ranges. Thus, fundamental characteristics of
community structure, such as the relative abundance of individual species and the frequency of co-existence of
multiple species, are influenced by the location of the sample site with respect to the geographic ranges of the
constituent species. This has important
ecological and evolutionary implications. For example, the strength and predictability of interspecific interactions
and the opportunities for co-evolution should depend in part on the location of the species within their geographic
ranges.

Contemporary ecology is still struggling to reconcile the different perspectives on community organization debated
by Gleason (1917, 1926) and Clements (1916, 1949). On the one hand, there is abundant evidence that many
assemblages consist of species whose abundances and distributions exhibit highly individualistic patterns of spatial
and temporal variation (e.g. Whittaker, 1967). These appear to reflect the relationships between the unique niche
requirements
of each species and the special features of local environments. On the other hand, assemblages appear to exhibit
emergent patterns of structure and dynamics that are sometimes very general (e.g. Odum, 1969; Morse et al., 1988;
Brown and Maurer, 1989; Brown and Nicoletto, 1991; see also DiMichele, 1994). Many of these emergent properties
are exhibited in
statistical distributions that reflect fundamental constraints on the pattern of variation.

Both the Gleasonian individualism and the Clementsian emergent properties are exhibited by these molluscan
assemblages. The individualism is exemplified by the wide variation in abundances among contemporaneously co-
existing species, even those near the centres of their ranges. As above, we attribute much of this variation to
differences in the niches of these species, i.e. differences in trophic status, habitat specialization and biotic
interactions.

The Clementsian emergence is exhibited in the characteristic relationship between abundance and position within
the geographic range for the communities as a whole (Fig. 2c) and in the stability of this relationship over time. One
way to sort through the Gleasonian individualism in search of the general, emergent properties is to increase the
scale of ecological studies to include large numbers of species, geographic spatial scales and long time periods. That
is the approach
that we have taken here. We have studied a large, ecologically heterogeneous assemblage of molluscs, rather than a
single species or guild. We have analysed the abundances and distributions with respect to geographic space and
evolutionary time. We have discovered seemingly robust statistical patterns.