Occasional Papers
Museum of Texas Tech University
Number 207
20 March 2001
Gleasonian Pattern of Mammalian Distribution at a
Macrogeographical Scale in Texas
James G. Owen
Gleasonian assemblages (Gleason, 1926;
Whittaker, 1975) are populations of species that chance
and environmental conditions have brought together
at the same time and place. Biotic exchange among
members of such assemblages is facultative rather than
obligatory. Membership in a suite of such species may
change, as environmental conditions change. The type
of community structure predicted by Gleason’s thesis
may be interpreted as a consequence of the
Hutchinsonian n-dimensional niche (Hutchinson, 1958;
Brown, 1995), where spatially autocorrelated resource
levels and distributional centroids change continuously
across the landscape (Brown, 1984). This pattern
focuses on the idiosyncratic, individualistic properties
of communities (Brown, 1995). It is very pervasive
in nature and has been demonstrated over a wide range
of scales, in contemporary (Brown and Kurzius, 1987;
Gleason, 1926; Grossman et al., 1982; Mac Arthur,
1972; Whittaker, 1975) and fossil communities, and
vertebrate (Wolfe, 1981; Graham, at al., 1996) and
invertebrate groups (Buzas and Culver, 1994).
Cements 1 (1916) concept of community organi¬
zation is the polar antithesis of Gleason’s. Clements’
(1916) view focuses on emergent properties, com¬
mon to the members of a community (Brown, 1995),
It holds that biotic assemblages comprise an integrated
whole, in which members are significantly interdepen¬
dent upon each other. Such species occur together
necessarily, or at least at a considerably higher fre¬
quency than one would expect by chance alone.
Clemensian communities, in the same region, may dif¬
fer from each other because of temporal asynchr onies
during the early stages of their development. Never¬
theless, all pass through a series of changes leading to
convergence upon a final state, that has been likened
to a super-organism (Clements, 1916; Brown and
Lomolino, 1998). Many investigators have challenged
the Clemensian view and it has been slowly super-
ceded by Gleason’s interpretation (Brown, 1984;
Whittaker, 1975).
Both of these views may be applied to patterns
of species importance along environmental gradients
and lead to testable hypotheses. Clements’ view pre¬
dicts discrete associations of species, separated by
relatively short ecotones; Gleason’s predicts that spe¬
cies importance values will be distributed along a gradu¬
ally changing continuum, without sharp steps. Mea¬
surements along an environmental gradient should serve
to test the hypothesis of Clements versus Gleason.
The question is which pattern comes closest to repre¬
senting nature? As with many strongly opposing mod¬
els both contain elements of truth. Importantly, the
2
Occasional Papers, Museum of Texas Tech University
conclusions to which one is led may be methodology
and scale dependent (Brown and Lomolino, 1998),
In this paper I explore patterns of geographical
association, within the conceptual framework of the
Clemensian and Gleasonian models, of mammalian
species on a regional scale. A high degree of geo¬
graphical separation, into associated groups, of mem¬
bers of a taxon would suggest a Clemensian pattern.
A high degree of geographical mixing among the ele¬
ments of a taxon would suggest a Gleasonian pattern.
The area treated is the state of Texas and the mamma¬
lian species treated are the class Mammalia n = 141
species, and the orders Chiroptera n = 31, Rodentia
n - 62, and Carnivora n = 29.
While there is taxonomic unity within these cat¬
egories, they embrace great ecological variation. Eco¬
logical interactions among populations of some of the
species subsumed within these taxonomic categories
probably do not occur enough to be significant. This
study focuses on the distributional patterns, at a re¬
gional scale, that would be predicted by the models of
Clements and Gleason. It does not imply species in¬
teractions, at a local scale, as an etiological basis for
such patterns.
Methods and Materials
The state of Texas occupies about 69.2 x 10 4
km 2 in the south-central region of the U.S.A. Since
the states' size, shape, and orientation were determined
by political considerations, it may be considered to be
an arbitrary sample for the study of mammalian distri¬
butions. I prepared maps of the geographical distribu¬
tions, within Texas, of 141 species of mammals native
to the state. I used all known records of marginal
specimens as a guide to map preparation. All maps
consisted of a system of quadrats, each representing
63.9 km on a side (Owen, 1990). I recorded a species
as either present or absent within each quadrat. Such
maps have often been used as a basis for making in¬
ferences about the biology of the distributions of mam¬
malian species (Simpson, 1964; Pagle, et ah 1991).
These data are functionally equivalent to a full census.
Results
Distribution of species among quadrats with
respect to sympatric species
In this paper, sympatric means geographical sym-
patry on a regional scale. A necessary and sufficient
condition for sympatry, between a given suite of spe¬
cies, is that they occur together in the same quadrat,
for at least one quadrat, somewhere within their col¬
lective ranges. It does imply close proximity at the
local community level, although this may often be the
case for many suites of species. This variable is sensi¬
tive to the size, shape, and orientation of ranges with
respect of each other. Figure 1 shows the number of
quadrats, as a function of the number of species per
quadrat. This plot represents the correspondence be¬
tween area and species richness.
1 fitted the data to Poisson distributions. These
distributions are used as null models that represent the
number of quadrats that would be inhabited by differ¬
ent numbers of species, expected on the basis of a
random distribution of species among quadrats. The
modes of the Poisson curves are located to the right,
with respect to the empirical data for Mammalia,
Chiroptera, and Rodentia; each of their modal values
is low. These taxa have considerably more area allo¬
cated to species-poor classes than one would expect
on the basis of chance alone. The Poisson curve for
Carnivora is located a little to the left of the data; its
modal value is very low and the entire curve is rela¬
tively flat. Carnivores have much more area allocated
to species-rich classes than one would predict on the
basis of chance. Not surprisingly, none of the Pois¬
son curves were significant using a chi-square test of
goodness-of-fit.
Owen— Gleasonian Pattern of Mammalian Distribution in Texas
3
Number of species
Figure 1. Number of quadrats, as a function of the number of species per quadrat Dashed cutvcs are
Poisson distributions fitted to the data. Solid curve for Carnivora is a positive binomial distribution.
Because the data resemble lognormal distribu¬
tions, I also fitted this model; none was significant.
The empirical data have higher, more peaked modes,
than a two-parameter lognormal distribution. The data
for Carnivora, but not the other taxa, had a significant
fit to a positive binomial distribution 16.22 =A3 S .
The fit to a positive binomial suggests that carnivores
may have a regular or uniform distribution, a condi¬
tion that is unusual in field or geographical ecology
(Elliott, 1977; Ludwig and Reynolds, 1988)
Figure 2 presents patterns exhibited by the num¬
ber of species when they are plotted as a function of
the number of species, with which they are sympat-
ric. Many species are geographically sympatric with
a wide range of other species, varying from few to
many. The data have much spread but there is an
overall tendency for many species to be geographi¬
cally sympatric with many other species. The straight
lines represent regressions, none of which were sig¬
nificant. The power of these tests was so low as to
give little confidence in failure to reject the hypothesis
of a slope equal to zero.
Combinations of geographically
sympatric species
I use the word combination to mean the set of
species that is present within a given quadrat. Distinct
combinations of sympatric species can differ from each
other with respect to either the identity of their con¬
stituent species or with respect to the number of spe¬
cies of which a combination is comprised, or both. If
two combinations differ in size, then they necessarily
differ in species composition, but the converse is not
necessarily true,
A plot of the number of species against the num¬
ber of different combinations, with which said spe¬
cies are combination members, is illustrated in figure
3. Each of these regressions was significant (P < 0,05),
4
Occasional Papers, Museum of Texas Tech University
</>
<D
O
0)
Cl
to
Number of sympatric species
Figure 2, Number of species, as a function of the total number of species with which
they are geographically sympatric.
except the one for carnivores* The power was low
for each test, except Rodentia. There is a general ten¬
dency for species of Mammalia, Chiroptera, and Ro¬
dentia to be members of combinations of small size.
In the plot for Carnivora there is a slight increase from
left to right in the regression line, due to the influence
of the large datum on the extreme right. In sum, many
species exist as a part of many different, but often
small (except Carnivora), species combinations.
Figure 4 shows the frequency distribution of the
number of combinations of species as a function of
the number of times each combination occurred. The
data for Chiroptera, Rodentia, and Carnivora are fitted
Owen— Gleasonian Pattern of Mammalian Distribution in Texas
5
25
20
15
10
5
0
25
20
15
</>
® 10
o
& 5
“ o
2 25
® 20
§ 15
Z 10
5
0
25
20
15
10
5
0
-
Mammalia
nflnfln nn nn nTTfm
-
1 1Mn ^
Chiroptera
fl~~— _n
—
Rodentia
.eta:
nlTFrnTT^ Fh=^— -FFi-i — n n
Carnivora
_ 11 l.l ri [I l.t 1,1 [) rp 1,1 1j
1 11 1 11 [ M M { 1 1 1 M
i? <? -\ N
Number of different species combinations
Figure 3. Frequency distribution of the number of species, as a function of the number
of different combinations in which they occurred.
to power functions; each curve is highly significant
(P <0.001). The lowest possible value for this vari¬
able is one. This value would be the case if all quad¬
rats were completely homogeneous i.e., if every spe¬
cies were sympatric with every other species through¬
out the entire state. Because each quadrat is charac¬
terized by one and only one combination, the maxi¬
mum possible number of combinations is 189. This
value would obtain if each of the quadrats had a unique
assemblage.
Mammalia had 184 different combinations, which
is not significantly different from the maximum of 189
(binomial test, P = 0.72), Ninety seven percent of
these combinations were observed only once. Ro¬
dents had 164 combinations, which also was not sig¬
nificantly different from the maximum possible value
of 189 (P = 0.069). Most of these combinations
(89.6%) also occurred only once. Bats had 73 combi¬
nations. This value was significantly less than the
6
Occasional Papers, Museum of Texas Tech University
0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16
Number of times species combination was observed
Figure 4. Frequency distribution of the number of combinations of species, as a function of the number of times each
combination was observed.
maximum number possible (P < 0*001). Of these
53.4% occurred once* Lastly carnivores occurred in
82 species combinations, which also is significantly
less than the maximum possible (P < 0.001). Of these
59.8% occurred only once. It is noteworthy that bats,
which can disperse by flying, had a lower number of
combinations than any other taxon. The upshot of
this figure is that most combinations of sympatric spe¬
cies occur only once and that most of the others oc¬
cur only two or three times.
The frequency distribution (Fig. 5) of the num¬
ber of species belonging to different combinations is
illustrated as a function of the number of species per
quadrat. I contrasted these data with a positive bino¬
mial distribution having its probability of “success”
set equal to P = 0.5. This distribution is symmetrical
by construction and approaches a normal distribution,
as sample size increases without bound. This approach
amounts to the null hypothesis that different combina¬
tions are equally likely. The procedure includes some
combinations that were not observed because one or
more of the component species do not have overlap¬
ping geographical ranges. This model is neutral in the
sense that it assumes that no species interactions or
extrinsic factors influence distribution. As such I did
Owen— Gleasonian Pattern of Mammalian Distribution in Texas
7
CD
_Q
£
3
O
CD
Q- 30
if)
o 25
20
15
10
5
0
40 50 60 70 80 90 4 6 8 10 12 14 16 18 20 22 24
Number of species per quadrat
Figure 5. Frequency distribution of the number of species pertaining to different combinations, as a function of the
number of species per quadrat (black dots). Circles represent the expected number of species pertaining to different
combinations, based on the null hypothesis that each different combination is equally likely.
not expect the data to conform to this model, but rather
was interested in the manner and degree in which lack
of conformity might be expressed.
I used a two-sided normal approximation to the
Mann-Whitney test to assess the null hypothesis that
the empirical and theoretical data belong to the same
underlying population. The alternative is that they dif¬
fer in location. This test was highly significant (P <
0.001) for each taxon. Lack of fit seems to be attrib¬
utable to both the locations of the modes and to their
degree of skewness. The empirical modes of all mam¬
mals, rodents, and bats are strongly shifted to the left
of their corresponding null modes. Their distributions
have strong positive skew. They are characterized by
an abundance of species-poor classes, i.e. classes that
are composed of a relatively low number of species.
Interestingly, the distribution of carnivores dif¬
fers from the other groups in several ways. Carni¬
vores are not positive skewed but rather seem to be
slightly negatively skewed (Fig. 5). They have more
area allocated to high species richness classes, with
respect to the other three taxa, and more species com¬
binations that are composed of a larger number of spe¬
cies. Overall they are more nearly symmetrical, their
8
Occasional Papers, Museum of Texas Tech University
mode is displaced a little to the right of the null mode
and, except for a single datum point, the empirical dis¬
tribution for carnivores is completely enclosed within
the density of its corresponding null distribution.
Impressionistically, carnivore assembly seems to be
closer to the null model than to that of the other three
taxa.
Discussion
The data presented in this paper, together with
findings referenced from the literature, suggest that a
mosaic-like pattern is characteristic of biotic assem¬
blages across a large range of different taxonomic,
temporal, and spatial scales. The spatial scales are
known to range from local, homogeneous habitat
patches to subcontinental regions. The taxonomic
scales range from feeding guilds of mice to an entire
class and the temporal scales span millions of years.
Two exceptional patterns are emergent among
Carnivora: (1) High species-richness is distributed over
a greater area for carnivores than it is for all mam¬
mals, rodents, or bats; (2) geographically, carnivores
are grouped together in such a way as to produce com¬
binations of species that are species-rich relative to
the other taxa. Both of these phenomena may be ex¬
plained as consequences of the larger geographical
ranges of carnivores. Larger geographical ranges
should overlap more with each other, yielding numeri¬
cally higher combinations of species. This effect is
reinforced by the presence of high carnivore richness
in central Texas (Owen, 1988). The accumulation of
carnivore species in central Texas and its gradual de¬
cline outwards to the east and west, gives carnivores
more opportunity to acquire numerically high combi¬
nations, while remaining within the study area. This
exceptional pattern for carnivores does not obviate their
inclusion within a Gleasonian frame of reference. The
same pattern and its explanation would seem probable
on a continental scale. North American carnivores
have considerably larger ranges than several other or¬
ders of North American mammals (Pagle, et al., 1991).
The Gleasonian pattern of spatial association
found in this study is not completely antithetical to
non-randomness. Two non-random patterns are iden¬
tified: (1) The lack of fit of the data to Poisson curves
(Fig. 1) and; (2) the lack of fit to binomial curves (Fig.
5). This means that one or more of the basic postu¬
lates of these distributions is seriously viol ated. A two-
dimensional Poisson or binomial process assumes,
among other things, that each quadrat or sampling unit
has equal probability of being inhabited by a given spe¬
cies. In particular, this postulate seems likely to have
been violated by the Texas data. The state is not ho¬
mogeneous with respect to its resource content for
mammals. Some areas offer more favorable condi¬
tions and so accumulate more species than others. This
is consonant with the widely varying qualitative fea¬
tures of different habitat types in Texas (Gould, 1975).
Limit relationships between the Poisson and binomial
distributions imply that, for large sample sizes, and
small probability of success for each species pres¬
ence-absence event, it makes little practical difference
which of the two distributions one uses (Hogg and
Craig, 1978; Boswell, et al., 1979).
Each of the patterns in the data, suggests a
patchy, mosaic-like geographical assemblage with little
internal homogeneity. At a regional scale species num¬
ber, degree of sympatry, and combinational size, oc¬
cur in constantly changing spatial configurations. This
change is probably produced by responses to environ¬
mental conditions that are species-specific, rather than
assembly-specific. The pattern corresponds to a
Gleasonian interpretation of mammalian distribution in
Texas at a statewide scale. Quadrats cannot (except
arbitrarily) be organized to form internally cohesive
biogeographical assemblages. This is the case in spite
of the ease with which certain areas of the state can
be superficially distinguished on the basis of their veg¬
etation physiognomy. Knowledge of the species com¬
position of one quadrat does not permit the prediction
of the species composition of other quadrats, at the
ordinal and class levels of taxonomy.
Owen— Gleasonian Pattern of Mammalian Distribution in Texas
9
Fossil faunas have demonstrated great age of
conditions that fit the Gleasonian hypothesis of spe¬
cies associations. Buzas and Culver (1994) studied
foramini feral communities, from Cenozoic shelf de¬
posits of the North American Atlantic Coastal Plain.
These communities exhibited little unity, during nearly
55 million years of successive oceanic transgressions
and regressions. Their results indicated a lack of local
community unity. Graham et ah, (1996) analyzed fos¬
sil mammalian communities at 2,945 localities in the
United States. They documented geographical range
shifts of individual species, at different times, in dif¬
ferent directions, and at different rates, in response to
late Quaternary environmental fluctuations. Such abun¬
dant data suggest that the pattern is real. Such long
expanses of time suggest that the Gleasonian pattern
is not an epi-phenomenon, but rather an integral part
of the ecological milieu through geological time.
That a mosaic-like pattern holds up for taxa at
the class and ordinal level is perhaps to be expected.
Members of the same class or order exhibit a wide
range of form and function. The rub is that Brown
and Kurzius (1987) found that patterns, qualitatively
very similar to the ones described in this paper, char¬
acterize members of a well defined desert, rodent feed¬
ing guild, where perhaps it was not to be expected.
Acknowledgments
The data analyses and interpretations used in this
paper were inspired by a similar study, at a different
scale and taxonomic level, by Brown and Kurzius
(1987). I thank James Brown for comments on an
early draft of this manuscript. I also thank two anony¬
mous reviews for comments.
10
Occasional Papers, Museum of Texas Tech University
Literature Otted
Boswell, M. T., Ord, J. K., and G. P. Patil. 1979.
Chance mechanisms underlying statistical dis¬
tributions. Pp. 3-156, in J. K. Ord, G. P. Patil,
and C Taillie, editors, Statistical distributions
in ecological work. International Co-operative
Publishing House, Fairland. Maryland.
Brown, J. H. 1984. On the relationship between abun¬
dance and distribution of species. American
Naturalist, 124:255-279.
Brown, J. H. 1995, Macroecology. University of
Chicago Press.
Brown, J. H., and M. A, Kurzius. 1987. Composition
of desert rodent faunas: combinations of co-
existi ng species. Ann a 1 es Zool ogi ci Fenn ici,
24:227-237.
Brown, J. H., and M. V. Lomolino. 1998. Biogeogra¬
phy, second edition. Sinauer Associates,
Sunderland, Massachusetts,
Buzas, M. A., and S. J. Culver. 1994. Species pool
and dynamics of marine paleocommunities.
Science, 264:1439-1441.
Clements, F. E. 1916. Plant succession: an analysis
of the development of vegetation. Carnegie
Institute of Washington, Washington, D. C.,
242:1-512.
Dice, L. R. 1943. The biotic provinces of North
America, University of Michigan Press, Ann
Arbor.
Elliott, J. M, 1977. Statistical analysis of samples of
benthic invertebrates. Freshwater biological
association scientific publication number 25.
Gleason, H. A. 1926. The individualistic concept of
plant associations. Bulletin Torry Botanical
Club, 53:7-26.
Gould, F. W. 1975. Texas plants: a checklist and
ecological summary. Texas Agricultural Ex¬
periment Station, College Station, Texas,
585:1-121.
Graham, R. W. et ah, 1996. Spatial response of mam¬
mals to late Quaternary environmental fluc¬
tuations. Science, 272:1601-1606,
Grossman G. D., P. B. Moyle, and J. O, Whitaker, Jr.
1982. Stocasticity in structural and functional
characteristics of an Indiana stream fish as¬
semblage: a test of community theory. Ameri¬
can Naturalist, 120:423-454.
Hogg, R. V., and A. T. Craig. 1978. Introduction to
mathematical statistics, fourth edition.
Macmillan, New York.
Hutchinson, G. E. 1958. Concluding Remarks. Cold
Spring Harbor Symposium on Quantitative
Biology, 22:415-427.
Ludwig, J. A,, and J. F. Reynolds. 1988. Statistical
ecology: a primer on methods and comput¬
ing. John Wiley, New York.
Mac Arthur, R. H. 1972. Geographical ecology: pat¬
terns in the distribution of species. Harper
and Row, New York.
Owen, J. G. 1988. On productivity as a predictor of
rodent and carnivore diversity. Ecology,
69:1161-1165.
Owen, J. G. 1990. Patterns of mammalian species
richness in relation to temperature, produc¬
tivity, and variance in elevation. Journal of
Mammalogy, 71:1-13.
Pagle, M. D., R. M. May, and A. R. Collie. 1991.
Ecological aspects of the geographical distri¬
bution and diversity of mammalian species.
American Naturalist, 137:791-815.
Rosenzweig, M. L. 1995. Species diversity in time
and space. Cambridge University Press, Great
Britain.
Webb, W. L. 1950. Biogeographic regions of Texas
and Oklahoma. Ecology, 31:426-433.
Whittaker, R. H. 1975. Communities and ecosys¬
tems, second edition. Macmillan, New York.
Wolfe, J. A. 1981. Vicarance biogeography of an-
giosperms in relation to paleobotanical data.
Pp. 413-427, in G. Nelson and D, E. Rosen
editors, Vicarance biogeography: a critique.
Columbia University Press, New York.
Owen—- Gleasonian Pattern of Mammalian Distribution in Texas
11
Address of author:
JAMES G. OWEN
Universidad Salvadoreha 'Alberto Masferrer",
Apartado Postal 2053, San Salvador, El Salvador
E-mail: jgowen@salnet.net
Publications of the Museum of Texas Tech University
[t was through the efforts of Horn Professor J Knox Jones, as director of Academic Publications, that
Texas Tech University initiated several publications series including the Occasional Papers of the Museum. This
and future editions in the series are a memorial to his dedication to excellence in academic publications. Profes¬
sor Jones enjoyed editing scientific publications and served the scientific community as an editor for the Journal
of Mammalogy, Evolution, The Texas Journal of Science, Occasional Papers of the Museum, and Special
Publications of the Museum. It is with special fondness that we remember Dr. J Knox Jones.
Institutional subscriptions are available through the Museum of Texas Tech University, attn: NSRL Publica¬
tions Secretary, Box 43191, Lubbock, TX 79409-3191. Individuals may also purchase separate numbers of the
Occasional Papers directly from the Museum of Texas Tech University.
ISSN 0149-175X
Museum of Texas Tech University, Lubbock, TX 79409-3191