close

Вход

Забыли?

вход по аккаунту

?

Systematics of the Saguinus oedipus group of the bare-face tamarins Evidence from facial morphology.

код для вставкиСкачать
AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 89:73-84 (1992)
Systematics of the Saguinus oedipus Group of the Bare-Face
Tamarins: Evidence From Facial Morphology
ALLEN J. MOORE AWD JAMES M. CHEVERUD
Department of Entomology, University of Kentucky, Lexington, Kentuck.y
40546-0091(A.J.M.): Dewartment of Anatomv and Neurobiolom.
Washington University School of kedicine, St. Louis, Missouz63110
(A.J.M., J.M.C.)
KEY WORDS
Saguinus geoffroyi, S a g u ims leucopus, Saguinus
fuscicollis, Conservation
The systematics of the Saguinus oedipus group within the
ABSTRACT
bare-face tamarins remains open to question. Hershkovitz (Living New World
Monkeys (Platyrrhini), Vol. 1. Chicago: University of Chicago Press, 1977)
places the cotton-top and rufus-naped tamarins as subspecies of Saguinus
oedipus (S. 0. oedipus and S. o. geoffroyi, respectively). In contrast, several
other authors have argued that these two taxa should be considered separate
species (S. oedipus and S. geoffroyi). Phylogenetic relationships within the
group are also disputed. Resolving these different interpretations has been
difficult in part because no study of this group has included an objective
measure of expected levels of specific vs. subspecific variation. We used facial
measurements from 179 adult crania to address the systematics of this group
and included a related species that is known to include multiple subspecies.
Our sample included three taxa from the S. oedipus group of the bare-face
tamarins (S. oedipus, S. geoffroyi, and S. leucopus) and six subspecies from
the related hairy-face tamarin species S. fuscicollis. Comparisons to S. leucopus provided a relative measure of species-level differences. Analyses that
included S. fuscicollis provided a measure of subspecific variation. There was
no evidence of facial sexual dimorphism in any of these taxa. A variety of
multivariate statistical analyses including discriminant function and cluster
analysis suggest that S. oedipus and S. geoffroyi differ morphologically a t a
level consistent with species-level distinctions. The extent of differences between these taxa is large. The differences in their facial morphology was on
the order of differences between S. oedipus or S. geoflroyi and S. leucopus
rather than the extent of variation among S. fuscicollis subspecies. Furthermore, a comparison of collecting localities revealed that the variation we
observed among S. oedipus and S. geoffroyi was not clinal but presented a
large morphological discontinuity at the boundary between taxa. Our analyses also suggested that S. leucopus is more similar to S. oedipus than is either
to S. geoffroyi. Finally, it may be that there are some distinct species within
the S. fuscicollis group. However, this hypothesis, along with other phylogenetic relationships suggested by this study, will require more data and further study. o 1992 Wiiey-Liss, Inc.
Hershkovitz’ (1977) classification is perhaps the most comprehensive single treatment of the svstematics of the tamarins and
other callitrichid New World monkeys. This
work has been suggested as the standard
and exreference for this
tremely differentiated group (Mittermeier
0 1992 WILEY-LISS, INC.
and Coimbra-Filho, 1981),and Hershkovitz’
classification is widely utilized. However, as
~Received March 12,1991, accepted February 18,1992
Allen J Moore’s present address IS Department of Entomology,
S-225 Agricultural Science Center North, University of Kentucky, Lexington, ICY 40546-0091
74
A.J. MOORE AND J.M. CHEVERUD
is expected with such an ambitious work,
many details and pertinent issues remain
contentious. One of the areas of controversy
involves the status and relationships of taxa
within the Saguinus oedipus group of bareface tamarins (S. oedipus, s. geoffroyi, and
S. leucopus).
Thorington (1976) and Mittermeier and
Coimbra-Filho (1981) have questioned
Hershkovitz' (1977) classification of the cotton-top and rufus-naped tamarins (S. oedipus oedipus and S. oedipus geoffroyi sensu
Hershkovitz). Hershkovitz argued that
these are two subspecies, while Thorington
and Mittermeier and Coimbra-Filho have
suggested that the level of differences in
geographic distribution, behavior, and appearance indicate the need to consider separate species status. Earlier classifications,
such as Hill's (1957), also generally listed
these taxa as separate species. Thorington
(1976) further suggested that S. leucopus,
the third member of the S. oedipus group
within the bare-face tamarins, is more similar to the cotton-top tamarin than the cotton-top is to the rufus-naped tamarin. If the
relationships proposed by Thorington are
correct, which are necessarily opposite those
proposed by Hershkovitz, it would require
separate species status for S.geoffroyi since
S. leucopus is widely agreed to be a legitimate species.
The question of taxonomic rank takes on
added importance in this group given the
highly endangered status of at least the cotton-top tamarin (I.U.C.N., 1972, 1976; Hernandez-Camacho and Cooper, 1976; Wolfheim, 1983; Mittermeier, 1987). The status
of the rufus-naped tamarin is somewhat unclear (Hernandez-Camacho and Cooper,
19761, in part because it is often subsumed
nominally under its Colombian relative (see,
e.g., Wolfheim, 1983). Nonetheless, it too is
probably endangered given the rate of disappearance of its native habitat (HernandezCamacho and Cooper, 1976). Taxonomic distinctions can have major implications in
legislative policies and management practices (Daugherty et al., 1990; May, 1990).
Thus it is critical that we address and clarify
the taxonomy of endangered groups before
they are lost.
This study addresses the systematics of
the S. oedipus group within the bare-face
tamarins (S. oedipus, S. geoffroyi, and S.
leucopus). We investigate the level of morphological differentiation between these
three taxa using facial measurements. Our
investigation consists of three detailed comparisons. First, we consider the level of morphological distinction between the cottontop oedipus and the rufus-naped geoffroyi
relative to a third closely related species, S.
leucopus. If oedipus and geoffroyi are subspecies we expect that they are much more
similar to one another than either is to S.
Leucopus. We note that, even if oedipus and
geofjroyi are as morphologically distant
from one another than either is from S. leucopus, they still may be sister taxa but
should then be considered sister species
rather than sister subspecies. We are making a phenetic comparison without regard to
primitive vs. derived similarity, although
phenetic comparisons often perform as well
as cladistic comparisons in simulation analyses of phylogenetic reconstruction (Sokal,
1985).
Second, we compared the level of distinction between the cotton-top and rufus-naped
tamarins to the level found among recognized subspecies of S. fuscicollis. S. fuscicollis is a member of the hairy-face group of
tamarins (Hershkovitz, 1977) and was the
subject of a previous systematic study utilizing variation in facial morphology (Cheverud and Moore, 1990). S. fuscicollis contains a large number of subspecies. If the
oedipus and geoffroyi populations should be
accorded only subspecific status, then the
level of morphological difference between
these two taxa should be similar to the level
found among S.fuscicollis subspecies for the
same characters. Using S. fuscicollis subspecies as a metric for subspecific variation
assumes that facial differences are equally
likely to arise in each group. It is not necessary t o assume that the differences facial
morphology be of the same kind among S.
fuscicollis subspecies and bare-face tamarins, only that the degree of difference is
similar for a given level of taxonomic distinction. Consideration of other morphological
systems could yield other results.
Finally, if S. oedipus and S. geoffroyi are
subspecies, we expect some intergradation
SYSTEMATICS OF BARE-FACE T W I N S
75
in morphology between them along the subspecific boundary. Geographically adjacent
oedipus and geoffroyi populations should be
more similar to one another than random
pairs of oedipus and geofroyi populations.
MATERIALS AND METHODS
To address the systematics of the bareface tamarins, we studied a variety of Saguinus samples from several sources. Specimens of S . oedipus, S. geoffroyi, and S.
fuscicollis were made available to us from
the Field Museum of Natural History
(FMNH). The Oak Ridge Associated Universities' Marmoset Research Center-University of Tennessee tamarin collection (ORAU)
provided access to wild-born and -raised S.
oedipus and S. fuscicollis. Specimens of S.
leucopus were from the National Museum of
Natural History (NMNH) collection. Hershkovitz (1977) provides details of the specimens and the collection localities for the material available in the FMNH or the NMNH,
and Gengozian (1969) discusses the history
of animals imported into ORAU. The individuals from the FMNH sample used in the
geographic distance analyses were collected
from five localities distributed along a line
running from the northeast to the south and
west in northern Colombia (Appendix A).
The samples of rufus-naped tamarins were
collected from four localities, one in northern Colombia quite close to the southwestern oedipus localities, two in Panama, and i l
one from the Pacific coast of southern Colombia. Further details of these collection
localities and the sample sizes are provided
in Appendix A.
Data were collected on a total of 179 specimens from the three bare-face tamarin taxa
and six subspecies of S. fuscicollis. Sample
sizes for individual taxa are presented in
Table 1. Poorly represented (N < 5) S. fuscicollis subspecies from the FMNH collection (S. f - fuscicollis, S. f. tripartitus, S. f.
melanoleucus)that were included in our previous study (Cheverud and Moore, 1990)are
not included here. All animals in this study
were determined to be adult by completed
dental eruption and closure of the sphenooccipital suture.
I
1
I
76
A.J. MOORE AND J.M. CHEVERUD
As in our previous study (Cheverud and
Moore, 1990), we quantified morphological
differences among the groups based on 11
facial measurements (Table 1). Measurements were chosen that provided broad coverage of the face. Facial measurements were
taken from the left side of the skull using a
digital caliper with direct computer input.
Statistical procedures were performed on residuals reflecting “size-free’’data (below).Unless noted otherwise, all analyses were performed using SYSTAT 5.0 (Wilhnson, 1990).
We corrected for the potentially confounding influence of size and effects correlated
with size by regressing each of the 11 variables on a size variable created by taking the
average of the 11lengths. These regressions
included the entire sample. The model I
least-squares regression equation used was
Length (xi) = constant
+ @(sizei)+ ei,
where x, is the original measurement for individual i, size, is the average of the 11
lengths for individual i, and e, is the residual. Least-squares regression was used because we wished to control for the statistical
effects of size in our analyses. All subsequent analyses were performed on the residuals (e,) derived from this regression. Distributions of the variables were checked by
performing a Lilliefors test, which tests for
normality without assuming a particular
mean or standard deviation for the distribution (Wilkinson, 19901. The “size-free”residuals were normally distributed after this
transformation.
We use size-free data since changes in size
are often considered to be of less taxonomic
importance than other aspects of morphology. Analyses using raw data, long-transformed data, or other methods of size correction gave qualitatively similar results.
Size-related variation was of relatively minor importance in this analysis. Only 38%of
the variance in the total sample was attributable to the first principal component. Our
analysis of the size-free residuals controlled
for the minor yet obvious differences in size
among the taxa.
A previous study of these taxa by Hanihara and Natori (1988) suggested that there
may be some sexual dimarphism in the den-
tition of the bare-face tamarins. Other studies of these and related tamarins have found
little or no dimorphism in other cranial
characters (references in Hershkovitz, 1977;
Hanihara and Natori, 1988). We failed to
find sexual dimorphism in our previous
study of these facial measurements in S.fuscicollis subspecies (Cheverud and Moore,
1990).In this study we tested for differences
between the taxa, the sexes and an interaction between the sexes and taxa using multivariate analysis of variance (MANOVA).
This analysis was performed on both raw
and size-free data and provided both multivariate and univariate tests. For consistency, we have provided the results of the
analysis using size-free data. The results obtained from uncorrected data were the same
as those reported below. Four different
Saguinus taxa (S. oedipus, S. geoffroyi, S.
leucopus, and S. fuscicollis)were considered
in the analysis of sexual dimorphism.
Standard multivariate statistical methods were used to quantify the differences
among the taxa. Multivariate discriminant
function analysis (DFA; Klecka, 1980) was
performed on the 11size-free morphological
variables using nine different taxa (S.oedipus, S. geoffroyi, S. leucopus, and six S. fuscicollis subspecies) as the groups. The relationship between specific discriminant
functions and morphological measurements
was determined by examining the correlations between function scores and variables.
The relationship between discriminant
function scores and taxa was investigated by
examining the mean discriminant function
scores for each of the nine taxa. Significant
pairwise differences among taxa were obtained form a Tukey HSD multiple comparisons matrix of pairwise comparison probabilities. Finally, individuals were classified
on the basis of morphology into most likely
taxa in a posthoc fashion.
A morphological distance matrix based on
Mahalanobis D2 values was also computed
from the discriminant function analysis.
The relative magnitude of morphological
distance among the taxa was considered a
measure of their morphological distinctiveness. Also, a phenogram based on a cluster
analysis of the morphological distances between taxa was generated using a complete
77
SYSTEMATICS OF BARE-FACE T W I N S
linkage or farthest neighbor method. This
method sequentially links groups using the
most distant relationship between the members of different clusters as a measure of
group similarity (Sneath and Sokal, 1973).
A second discriminant function analysis
was performed using only the cotton-top and
rufus-naped specimens collected from
known localities. The nine collection localities form the groups for the analysis and
Mahalanobis D2 values were obtained from
the mean discriminant function scores for
each locality. The geographical distribution
of the function scores was evaluated by fitting a step curve to the distribution of the
first discriminant function over latitude and
longitude (SYGRAF’H; Wilkinson, 1990). D2
values were also compared to pairwise matrices of taxon membership (1if pair of localities are from the same taxon and 0 otherwise) and a one-dimensional geographical
distance based on the distances of collection
localities to the northeast or northwest of
the 0 4 oedipus locality. This distance is the
pairwise distance between localities along
a line running from the northeast from
locality 01 through 0 2 and 0 3 to 0 4 and
then northwest from 0 4 through 05, G6,
G7, and G8. This one-dimensional distance
has greater biological meaning and is therefore preferable to two-dimensional straight
line distances between localities because
straight line distances cross the Caribbean
Sea while this distance follows the coastline. Only the G9 site on the Pacific coast
of Colombia does not fall on this V-shaped
line.
Hypotheses involving matrix comparisons
were tested using quadratic assignment procedures (QAP; Dow and Cheverud, 1985;
Dow et al., 1987; Hubert, 1987; Cheverud et
al., 1989), sometimes referred to as Mantel’s
test (Dietz, 1983; Smouse et al., 1986). Quadratic assignment is a procedure that estimates appropriate probabilities for tests of
the null hypothesis of no structural similarity between a pair of matrices. Structural
similarity is measured here by the Pearson
product moment correlation between matrix
elements. Because elements of distance matrices are not independent of each other, the
number of degrees of freedom represented
by the matrix elements is unknown. Thus
standard tests of statistical significance
would be inappropriate. In our analyses, the
statistical significance of matrix correlations was determined by comparing the observed correlation to a distribution of matrix
correlations obtained by 500 randomizations. Partial correlations and associated
tests of significance were also calculated using QAP (Dow et al., 1987). Cheverud and
Moore (1990) provide a detailed description
of QAP and the use of this procedure in testing hypotheses relating morphology, geography. and taxon membership in the related 3.
fuscicollis tamarins.
RESULTS
A MANOVA was performed on the 11
measurements using sex, taxon, and sex by
taxon as the independent variables. One
hundred seventy-six individuals were analyzed, since three specimens were of unknown sex. We found no multivariate
(Wilk’s A = 0.978, df = 11;158, P = 0.979)
or univariate (all P > 0.35) differences between the sexes. In addition, there was no
significant multivariate sex by taxon interaction (Wilk‘s A = 0.801, df = 33;466,
P = 0.316). There was a single significant
sex by taxon univariate difference for
FM - MT ( P = 0.036); all other univariate
tests were highly nonsignificant (all
P > 0.12). This one significant result most
likely represents a Type I error, given the
lack of multivariate significance and its failure to reach Bonferroni levels of significance
( P < 0.05/11). In contrast t o the results for
sex, there were highly significant multivariate differences among the taxa (Wilk’s
A = 0.117, df = 33;466, P < 0.001). Statistically significant univariate differences
among taxa were observed in nine of the 11
measurements using the conservative sequential Bonferroni levels of significance
(Rice, 1989). Given that we found little or no
multivariate or univariate sex differences or
interaction between sex and taxon, all future analyses were performed without regard to sex.
Discriminant function analysis was performed using the nine taxa described above
as groups. Overall, the DFA was highly significant (Wilk‘s A = 0.012, df = 88;1058,
78
A.J. MOORE AND J.M. CHEVERUD
TABLE 2. Correlations between the size corrected facial variables and discriminant f u n c t h s , canonical correlations
derived for the nine t m a l
Discriminant function
-
Measurement
IS
- PM
IS - NA
IS - PNS
BR - NA
NA - FM
FM-PT
FM - ZS
FM - MT
ZI --- PM
ZI -- zs
ZI - MT
Canonical correlations
___
~
1
2
3
4
5
6
0.206
0.454
0.734
0.210
0.102
0.442
0.380
- 0.033
0.288
0.219
0.397
0.762
0.130
- 0.031
0.085
- 0.283
0.608
- 0.128
-- 0.145
- 0.550
0.445
0.479
- 0.334
0.662
0.253
- 0.326
- 0.269
0.459
0.196
0.085
- 0.554
- 0.143
0.206
0.271
0.367
0.629
0.188
- 0.149
0.385
- 0.306
0.006
- 0.001
- 0.357
- 0.058
0.330
- 0.340
0.562
0.497
0.577
0.577
0.144
0.212
- 0.128
0.432
- 0.278
- 0.205
- 0.258
- 0.175
- 0.050
0.398
.-__
0.056
- 0.010
0.089
- 0.152
- 0.282
0.237
0.005
0.543
- 0.380
- 0.024
- - 0.088
0.914
.______
-
-
-
For the first seven axes, the probabilities that different taxa differ along the specified axis are P ~<0.001;for axis 8, P
P < 0.001). The first seven roots were also
highly significant ( P < 0.001); root 8 was
not significant ( P = 0.11). Correlations between discriminant function scores and facial measurements are presented in Table 2.
Based on the correlations between the variables and the factors, the first factor contrasts posterior facial height (FM - MT)
with the anterior projection of the maxillary
alveolus (ZI - PM) and orbital width
(NA - FM). The second factor contrasts rnedial with lateral measurements (IS - NA,
IS - PNS VS.FM - PT, FM - ZS, ZI - PM,
ZI - MT). The third factor contrasts anterior and posterior face (NA - FM, ZI - PM,
ZI - ZS from FM - MT, ZI - MT). The remaining factors do not describe anatomically simple contrasts.
Significant pairwise differences among
taxa suggest that factors 1 , 2 , and 3 separate
the four species (Fig. 1)from one another,
while factor 4 separates S. f. fuscus from all
the other taxa and factors 5 , 6, 7, and 8 discriminate among the S. fuscicollis subspecies. Factor 1ordinates the species from low
to high scores; S. geoffroyi, S. oedipus, S.
leucopus, S. f. fuscus, and then the remaining S. fuscicollis subspecies. Factor 2 distinguishes S. geoffroyi (low scores) from S. leucopus and S. oedipus (high scores). In
addition, S. f. fuscus (low scores) is separated from S. leucopus, S. oedipus, S. f. nigrifrons, and S. f. weddelli (high scores) by
this factor. Factor 3 separates S. leucopus
and S. f. weddelli (low scores) from the rest
o f the groiips. Fact,or 4 separates S. f- fusrus
7
0.223
0.237
0.190
0.433
- 0.445
0.128
- 0.105
- 0.469
0.109
0.340
0.041
-
n 376
=
8__
0.232
0.384
- 0.386
0.126
0.089
- 0.502
0.110
-- 0.226
0.353
- 0.130
0.341
-
n 907
0.11
(low score) from the other S. fuscicollis subspecies. The remaining factors confirm our
earlier analyses regarding subspecific differences among S. fuscicollis (Cheverud and
Moore, 1990).
Posthoc classifications from the DFA are
presented in Table 3. The DFA correctly
classified 96% of the specimens with regard
to membership in the four species. Most errors in classification involved the S.fuscicollis subspecies; few errors occurred within
the bare-face group. One specimen each
from S.geoffroyi, S. oedipus, and S. leucopus
was misclassified as S. fuscicollis. Two S.
oedipus were misclassified as S. geoffroyi
and two as S. leucopus, while one S. leucopus was misclassified a s S. oedipus. One fuscicollis ( S . f. weddelli) specimen was misclassified for species (as S. leucopus). Only
72% of the S. fuscicollis were correctly classified among the S. fuscicollis subspecies, a
result similar to our previous findings with
this taxon (Cheverud and Moore, 1990).
The morphological distance (D2)matrix is
given in Table 4 and the tree diagram based
on the complete linkage method is presented
in Figure 2. The morphological distance between S. oedipus and S. geoffroyi (D2 =
14.47) is greater than that between S. oedipus and S. leucopus (D" = 12.661, as illustrated in the cluster analysis by the primary
linkage of the latter pair of taxa. Furthermore, the distance between S. oedipus and
S. geoffroyi is much greater than the average distance among subspecies ofS. fuscicollis (average D2 = 9.30).The average differ-
SYSTEMATICS OF BARE-FACE T W I N S
79
Fig. 1. Three-dimensional scatterplot for the first three discriminant function scores generated from
the DFA of the bare-face tamarin (5'. oedipus, S. geoffroyi, and S. leucopus) and the saddle-back tamarin
(S.fuscicollis weddelli, S. f lagonatus, S . f. leucogenys, S. f. nigrifrons, S. f illigeri, and S. f : fuscus) taxa.
Points are identified by species or subspecies names next to the spikes.
TABLE 3. Posthoc classification of tava based on, discriminant function of size corrected facial riariables
--
Taxa
__
1)S. geofroyi
2 ) S. oedipus
3: s.leucopus
4) s.f. fuseus
5) S. f. nigrifrons
6) S. f: illigeri
7 )S. f. leucogenys
8 ) S. f. lagonatus
9) S. f. weddelli
Total
1)
2)
3)
~
25
2
0
0
0
0
0
0
0
27
0
31
1
0
0
0
0
0
0
32
0
2
16
0
0
0
0
0
1
21
Number classified
5)___.......__6)
-..
4)
0
0
0
13
1
0
0
0
0
14
0
0
0
0
14
3
1
2
2
22
0
0
0
0
3
16
0
4
0
23
.
7)
8)
1
0
0
1
1
0
4
1
0
8
0
0
0
0
0
2
0
14
2
18
9)-.----____-.___
Total
0
1
1
0
1
2
0
26
1
22
13
179
8
14
3ci
2(0
14
20
23
5
Percent
correct
96.29
8fi 1 q
so.(y;>
92.0%
70.0%
69.6%
80.0%
63.6%
61.56)
ence among S. fuscicollis subspecies is the remaining five subspecies (average D2 =
inflated by the distant relationship of S. f. 6.86) is less than one-half of the S. geoffuscus to the other subspecies (average D2 = froyi-S. oedipus difference.
A further result from the analysis of mor14.18; also Cheverud and Moore, 1990).This
distant relationship of the S. f. fuscus sub- phological distance indicates that S. leucospecies to the other s. fuscicollis subspecies pus is intermediate in morphology between
is illustrated in the cluster analysis by its S. fuscicollis and the other members of the
joining the other subspecies at the same S. oedipus group of the bare-face tamarins.
level as S. geoffroyi joins with S. leucopus It is morphologically least distant from S. fi
and S. oedipus. The average distance among lagonotus and S. fi weddelli, although, over-
A.J. MOORE AND J.M. CHEVERUD
80
TABLE 4. Morphological distance matrix based on Mahalonobus 02 computed from the discriminant function analysis
of the nine taxa'
1) S. geoffroyi
2) S. oedipus
31 S. leucopus
41 s. f. fuscus
5) 5'. f. nigrifrons
6) S. f. illigeri
7) S. f. Zeucogenys
8) S. f. lagonatus
9) S. f. weddelli
14.747
19.922
27.465
40.752
38.231
39.014
29.990
38.456
12.656
22.487
22.723
25.199
34.302
21.331
23.298
14.635
12.850
15.842
18.742
9.902
8.622
13.331
14.003
14.363
12.283
16.946
5.148
7.113
6.536
5.701
7.677
4.356
8.030
9.261
8.770
6.035
Taxa include separate subspecies for Saguinus fuscicollis.
0
0
In
-5 /G9*
,
-4
-3
1
-2
,
I-'
-1
0
1
I
:
2
3
4
FIRST FUNCTION SCORE
~
oI I I
Fig. 3. Plot of locality means for first and second
discriminant function scores generated from the DFA of
the 11facial measurements at the nine collection localities of the cotton-top is. oedipus: 01-05) and ruius;
specimens represented in
naped (5.g ~ u f f r ~ y iG&GYi
the FMNH. Longitude and latitude locations for each
site are provided in Appendix A.
geoffroyi was highly significant (Wilk's A =
0.009; df = 88; 245; P < 0.001), with the
first two discriminant functions statistically
significant. As is shown in Figure 3, the first
function separates the cotton-top (0-1from
the rufus-naped (G-1 collection sites, while
the second function serves primarily to distinguish the Colombian Pacific coast geofall, it is slightly more similar to S. oedipus froyi locality (G9) from the others. This result is significant even though it is based on
than to S. fuscicollis.
The discriminant function analysis among only three individuals from the G9 site. Figthe nine collecting sites for S. oedipus and S. ure 4 is a morphological topography of the
Fig. 2. Tree diagram based on a cluster analysis of
the distances between the taxa generated using a complete linkage or farthest neighbor method. Morphological distances were calculated from 11 facial measurements. Taxa are identified by species or subspecies
names.
SYSTEMATICS OF BARE-FACE T W I N S
81
3
Fig. 4. Morphological topography of first discriminant function score across latitude and longitude,
fitted by a step function. Other continuous functions produce nearly identical fits to the surface. Collection sites for the FMNH specimens of S. oedipus (01-05) and S. geoffroyi (G6-G9) analyzed here are
provided in Appendix A.
first discriminant function scores for each phological distance (m) and taxon membersite relative to latitude and longitude. There ship (t) is extremely strong (rm,t= -0.955,
is no sign of intergradation among the oedi- P = 0.0221, followed by the correlation of
pus and geofroyi groups; instead, there is a morphological and geographical (g) distance
= 0.598, P = .002). To determine
sharp morphological boundary coinciding (rm,g
with the taxon boundary. The members of whether a morphological cline exists indethe geoffroyi collection site G6, which is geo- pendent of taxon membership, as we would
graphically adjacent to oedipus groups 0 4 expect among intergrading subspecies, we
and 05, are as morphologically different calculated the partial matrix correlation of
from the oedipus groups as are any other morphology with geography controlling for
taxon membership. This partial correlation
geofroyi population.
= 0.577, P = 0.1661,
These results are statistically confirmed was not significant (rm,g.t
by analyses of matrix similarity using the although the partial correlation of morpholeight collection sites in northern Colombia ogy with taxon membership controlling for
and Panama. The G9 site was not included geography remained quite strong and sta= -0.953, P =
due to its morphological and geographic dis- tistically significant (rm,t.g
tinctiveness. The correlation between mor- 0.022).
82
A.J. MOORE AND J.M. CHEVERUD
DISCUSSION
Our results suggest that S. oedipus, the
cotton-top tamarin, and S.geoffroyi, the rufus-naped tamarin [Moynihan, 1970; also
known as Geoflroyi’s tamarin (Hershkovitz,
197711, should be classified as separate species. DFA correctly classified the bare-face
tamarins but was much less successful in
classifying subspecies of saddle-back tamarins. On the basis of facial morphology, the
morphological difference between these two
species is similar to the distance between
either and S. leucopus and is much greater
than the average difference among S. fuscicollis Subspecies. A species-level distinction is consistent with Thorington’s (1976)
and Mittermeier and Coimbra-Filho’s(1981)
judgement. Natori (1988) and Natori and
Hanihara (1988) also proposed a specieslevel classification based on cranial measurements, although these studies lacked an
adequate measure of subspecific differences
for comparison. Classifying these taxa as
different species is also consistent with the
level of differences observed in behavioral
studies of S. oedipus and S. geoffroyi and of
S. leucopus (Moynihan, 1970; see also studies by Dawson, 1977; Neyman, 1977).
Our geographical analysis of S. oedipus
and S. geofjcroyi specimens from specific localities provides new evidence that strongly
confirms the morphological separation of
the taxa. There is no evidence for intergradation along the taxon boundary, as might
be expected among subspecies. Instead,
there is a sharp, steep morphological boundary between the taxa, consistent with species-level distinction. One surprising result
of this analysis is the morphological distinctiveness of the Colombian Pacific Coast locality (G9) of S.geoffroyi (see Figs. 3 and 4).
These three specimens are nearly as different from the other s.geoffroyi as S. geoffroyi
is from S. oedipus. Note, however, that the
morphological difference displayed within
S. geoffroyi is quite distinct from the interspecific difference and geoffroyi forms a coherent species (Fig. 4). This is supported by
the fact that excluding the G9 sample from
our analyses does not alter any of our interpretations or conclusions. This large-scale
morphological diversity within S. geoffroyi
needs to be confirmed (or refuted) by further
analysis of much larger samples. There are
apparently geographically intermediate
populations (Hershkovitz, 1977) but these
were not available to us.
Our analyses further suggest that S. leucopus is slightly more similar t o S. oedipus
than either is to S. geoffroyi. This result is
consistent with the suggestion made by
Thorington (1976) based on his examination
of museum skins. S. oedipus and S. leucopus
occur in northwestern Colombia and are in
close contact over much of their range. In
contrast, S.geoffroyi occurs in northwestern
Colombia and Panama, and may not have
any contact with either S. oedipus or S. leucopus (see maps in Hernandez-Camacho
and Cooper, 1976; Hershkovitz, 1977; see
also Wolfheim, 1983). However, this arrangement contradicts the classification of
Hershkovitz (1977:753-754) and the cladistic analysis proposed by Natori (1988). Resolution of the phylogenetic relationships of
these tamarins will likely depend on more
sensitive measures, most likely using
mtDNA or other molecular markers.
Including the S. fuscicollis subspecies in
this study yielded several benefits. In addition to providing a measure of subspecific
variation, our study suggests relationships
among the closely related bare-face and saddle-back tamarins. On the basis of these preliminary analyses, S. leucopus appears to be
more similar to the S.fuscicollis group than
are the other members of the bare-face tamarins (Fig. 1). s. leucopus appears to be
equally similar to S.f. fuscus, the remaining
S. fuscicollis subspecies, and the other barefaced tamarins. Among the s. fuscicollis
subspecies analyzed here, S. f. fuscus is the
most northern race and is one of the most
primitive or nearest to the hypothetical ancestor of the tamarins (Hershkovitz, 1977;
Cheverud and Moore, 1990). Hershkovitz
(1977:735,749-754) also asserts that S. leucopus is the most primitive of the bare-face
tamarins. Thus it may be that S. fi fuscus
and S. leucopus share primitive characters
leading to morphological similarity.
Within the saddle-back tamarins, S.f.fuscus may deserve separate species status
(Fig. 1). The morphological difference between this taxon and the other S. fuscicollis
SYSTEMATICS OF BARE-FACE TAMARINS
subspecies is of the same order as the species differences within the S. oedipus group.
In part, the distinctiveness of S. f. fuscus is
due to the exclusion of poorly represented
taxa, such as S. f. fuscicollis, from this
study. In our previous study of saddle-back
tamarins (Cheverud and Moore, 19901, we
found that S. f. fuscicollis linked S. f. fuscus
to the other subspecies. These two subspecies are the most primitive with regard to
coat color in Hershkovitz’ (1977) northern
and central races of saddle-back tamarins,
so their similarity mav be due to qharecl
primitive morphology.
Our current study therefore lends some
support to the suggestion of Thorington
(1988) that we should reexamine the relationships among the S. fuscicollis subspecies. As Thorington (1988) suggests, there
may be several very closely related species
within the S.fuscicollis group. However, our
previous analysis of the saddle-back tamarin showed that the taxa included within
this species are generally morphologically
very similar (Hershkovitz, 1977; Cheverud
and Moore, 1990), so tests of this hypothesis
will require larger samples of the taxa implicated as different a t the specific level and
more sensitive phylogenetic data and methods.
In summary, we find that S. oedipus and
S. geoffroyi should be accorded separate species status because they are as morphologically different from one another as either is
from the closely related S. leucopus. They
are also much more differentiated than are
subspecies of S. fuscicollis. Other issues
raised by our results, especially the phylogenetic relationships of the taxa, will have to
be resolved with techniques more powerful
than those utilized here or in previous studies. However, the advances in molecular systematics suggest that this can be accomplished. Given the endangered status of
most of these taxa (I.U.C.N., 1972, 19761,
and the importance of taxonomic relationships in conservation biology (Daugherty et
al., 1990; May, 19901, we look forward to
further systematic work in this area.
ACKNOWLEDGMENTS
We thank Bruce Patterson (Field Museum
of Natural History), Richard Thorington
83
(National Museum of Natural History), and
Suzette Tardif and Fred Smith (Marmoset
Research Center of the Oak Ridge Associated Universities and University of Tennessee), for access to their collections. P. Hershkovitz, R. Thorington, and R. Sussman
provided helpful discussions on tamarin systematics, and W. Jungers, L. Kohn, E. Routman, and C. Jaquish provided helpful suggestions and discussions. This research was
supported by NSF grants BSR-8906041 to
J.M.C. and a NSF Postdoctoral Fellowship
in Esvironmenta! Biology (BSR-8821275)to
A.J.M. During the preparation of this paper,
A.J.M. was supported by NSF grant BSR9022012.
LITERATURE CITED
Cheverud JM,and Moore AJ (1990) Subspecific morphological variation in the saddle-back tamarin (Suguinus fuscicollis). Am. J. Primatol. 21:l-15.
Cheverud JM, Wagner G, and Dow MM (1989) Methods
for the comparative analysis of variation patterns.
Syst. ZOO^., 38t201-213.
Daugherty CH, Cree A, Hay JM, and Thompson MB
(1990) Neglected taxonomy and continuing extinctions of tuatara (Sphenodon).Nature 347:177-179.
Dawson GA (1977) Composition and stability of social
groups of the tamarin, Saguinus oedipus geoffroyi, in
Panama: Ecological and behavioral implications. In
DG Kleiman (ed.): The Biology and Conservation of
the Callitrichidae. Washington, DC: Smithsonian Institution Press, pp. 23-37.
Dietz E J (1983) Permutation tests for association between two distance matrices. Syst. Zool. 32r21-26.
Dow MM, and Cbeverud J M (1985) Comparison of distance matrices in studies of population structure and
genetic microdifferentiation: Quadratic assignment.
Am. J. Phys. Anthropol. 68t367-374.
Dow MM, Cheverud JM, and Friedlaaender J (1987)
Partial correlation of distance matrices in studies of
population structure.Am. J . Phys. Anthropol. 72343352.
Gengozian N (1969) Marmosets: Their potential in experimental medicine. Ann. NY Acad. Sci. 162:336362.
Hanihara T, and Natori M (1988) Numerical analysis of
sexual dimorphism in Saguinus dentition. Primates
29~2245-2254.
Hernandez-Camacho J, and Cooper RW (1976) The nonhuman primates of Colombia. In RW Thorington and
PG Heltne (eds.): Neotropical Primates: Field Studies
and Conservation. Washington, DC: National Academy of Sciences, pp. 35-69.
Hershkovitz P (1977) Living New World Monkeys
(Platyrrhini), Vol. 1. Chicago: University of Chicago
Press.
Hill WCO (1957) Primates: Comparative Anatomy and
A.J. MOORE AND J.M. CHEVERUD
84
Taxonomy 3. Pithecoidea, Platyrrhini. Edinburgh:
Edinburgh University Press.
Hubert L (1987) Assignment Methods in Combinatorial
Data Analysis. New York Dekker Press.
I.U.C.N. (1972) Red Data Book, Mammalia, Volume 1.
Morges, Switzerland IUCN Publications.
I.U.C.N. (1976) Red Data Book: Update Sheets. Morges,
Switzerland: IUCN Publications.
Klecka WR (1980)Discriminant Analysis. Beverly Hills,
C A Sage Publications.
May RM (1990)Taxonomy as destiny. Nature 347:129130.
Mittermeier RA (1987) Framework for primate conservation in the Neotropical region. In CW Marsh and RA
Mittermeier (eds.): Primate Conservation in the Tropical Rain Forest. Ny: Alan R Liss, pp. 305-320.
Mittermeier RA, and Coimbra-Filho AF (1981) Systematics: Species and subspecies. In AF Coimbra-Filho
and RA Mittermeier (eds.): Ecology and Behavior of
Neotropical Primates. Rio de Janero: Academia
Brasileira de Ciencias, pp. 29-109.
Moynihan M (1970) Some behavior patterns of platyrrhine monkeys 11. Saguinus geoffroyi and some other
tamarins. Smithsonian Contrib. Zool. 28:1-77.
Natori M (1988) A cladistic analysis of interspecific relationships of Saguinus. Primates 29:263-276.
Natori M, and Hanihara T (1988)An analysis of interspecific relationships of Saguinus based on cranial
measurements. Primates 29:255-262.
Neyman PF (1977) Aspects of the ecology and social
organization of free-ranging cotton-top tamarins
(Saguinus oedipus) and the conservation status of the
species. In DG Kleiman (ed.): The Biology and Conservation of the Callitrichidae. Washington, DC: Smithsonian Institution Press, pp. 39-71.
Rice WR (1989) Analyzing tables of statistical tests. Evolution 43:223-225.
Smouse P, Long J , and Sokal RR (1986) Multiple regressions and correlation extensions of the Mantel test of
matrix correspondence. Syst. Zool. 35~627-632.
Sneath PHA, and Sokal PH (1973) Numerical Taxonomy. San Francisco: WH Freeman Press.
Sokal RR (1985) The continuing search for order. Am.
Nat. 126729-749.
Tnorington K i 2r 113i6j The sysLarrtatics uf N a w f u I M
monkeys. In: First Inter-American Conference on
Conservation and Utilization of American Nonhuman
Primates in Biomedical Research. Washington, DC:
Pan American Health Organization Scientific Publication No. 317, pp. 8-18.
Thorington RW Jr (1988) Taxonomic status of Saguinus
tripartitus (Milne-Edwards, 1878) Am. J . Primatol.
15:367-371.
Wilkinson L (1990)SYSTAT: The System for Statistics.
Evanston, IL: SYSTAT Inc.
Wolfheim J H (1983) Primates of the World: Distribution, Abundance, and Conservation. Seattle: University of Washington Press.
APPENDIX A. Collection localities for the Saguinus oedipus and Saguinus geoffroyi samples used in the geographic
distance analysis (see also Hershkouitz, 1977)'
Location
Saguinus oedipus
Colombia:
Bolivar: San Juan,
Nepomuceno
Bolivar: Las Campanas,
Symbol
Latitude N,
Longitude W
N
01
9"58',75"04'
7
02
9"30',75"21'
03
8"17',75"41'
04
05
8"00',76"44'
7"51',76"17'
G6
8"01',77"07'
16
G7
G8
8"59',79"33'
9"06',79"37'
3
G9
5"13',77"05'
3
Coloso
Cordoba: Catival, upper
Rio San Jorge
Antioquia: Rio Curulao
Cordoba: Socorre, Rio Sinu
Saguinus geoffroyi
Northern Colombia:
Cboco: Unguia
Panama Canal Zone:
Albrook Air Force Base
Madden Forest
Pacific coast of Colombia
Choco: Ria Sando
3
S. oedipus localities are numbered 01-05, reflecting a northeast to southwest transect. S.geoffroyi localities are numbered GGG9, reflecting a
transect from northern Colombia to Panama to the Pacific coast of southern Colombia.
Документ
Категория
Без категории
Просмотров
6
Размер файла
1 316 Кб
Теги
systematic, morphology, bare, facer, facial, saguinus, group, evidence, oedipus, tamarins
1/--страниц
Пожаловаться на содержимое документа