CIRCUMSCRIPTION OF ECHINOCEREUS ARIZONICUS SUBSP. ARIZONICUS: PHENETIC ANALYSIS OF MORPHOLOGICAL CHARACTERS IN SECTION TRIGLOCHIDIATUS (CACTACEAE), PART II

2006·7.9 MB·English

by  Marc A Baker

Madrono, Vol. 53, No. 4, pp. 388 399, 2006
CIRCUMSCRIPTION OF ECHINOCEREUS ARIZONICUS SUBSP. ARIZONICUS:
PHENETIC ANALYSIS OF MORPHOLOGICAL CHARACTERS IN SECTION
TRIGLOCHIDIATUS (CACTACEAE), PART
II
Marc A. Baker
College of Liberal Arts and Sciences, School of Life Sciences,
Arizona State University, P.O. Box 874501, Tempe, Arizona 85287-4501
[email protected]
Abstract
A multivariate analysis was performed for populations of Echinocereus (section Triglochidiatus)
to facilitate the taxonomic circumscription of E. arizonicus subsp. arizonicus. Twenty-one
morphological characters for 16 populations evidenced the validity of at least two subspecific
taxa within E. arizonicus: E. arizonicus subsp. arizonicus and E. arizonicus subsp. nigrihorridispiuus.
Principle components analysis indicated that stem characters were most diagnostic in defining two
distinct groups of populations, each including the type locality of one of the two subspecies.
Unweighted pair group method with arithmetic mean (UPGMA) clustered populations of the two
subspecies apart from one another and from those of the outgroup, E. triglochidiatus subsp.
mojavensis. For most measured characters, means differed significantly between the two subspecies.
Discriminant analysis correctly classified 97.0% for individuals ofE. arizonicus subsp. arizonicus and
94.7% for individuals of E. arizonicus subsp. nigrihorridispinus^ compared to an overall 97.8%
correct classification of individuals for all perfect-flowered taxa of section Triglochidiatus
investigated.
Key Words: Echinocereus arizonicus, section Triglochidiatus, multivariate analysis, morphological
characters.
Taxonomy within section Triglochidiatus H. tus, and E. yavapaiensis M. A. Baker (Hoffman
Bravo ofEchinocereus Engelm. in F. A. Wislize- 1992; Blum et al. 1998; Zimmerman & Parfitt
nus has vacillated dramatically over the past few 2003; Baker 2006). The primary rationale for
decades. The present study evaluates the circum- splitting E. triglochidiatus into several species is
scription of E. arizonicus Rose ex Orcutt subsp. the occurrence of polyploidy correlated with
arizonicus using multivariate techniques to com- morphology, geographic distribution, and floral
pare morphological variation within populations dimorphism (Table 1). Echinocereus arizonicus
to that among populations. It is the second represents smooth-spined, diploid, perfect-flow-
portion ofan ongoingphenetic analysis ofsection ered populations from the Sonoran-Chihuahuan
Triglochidiatus, the first of which addressed the Desert interface; E. triglochidiatus represents
recently discovered species, E. yavapaiensis M. A. papillate- [E. triglochidiatus subsp. mojavensis
Baker, and summarized the current knowledge of (Engelm. & Bigelow) W. Blum & Michael Lange]
polyploidy within the section (Baker 2006). The or angular-spined {E. triglochidiatus subsp. tri-
taxonomic status of E. arizonicus subsp. arizoni- glochidiatus), diploid, perfect-flowered, popula-
cus, which is federally listed as Endangered, has tions from the Mojave Desert, California east to
importance for conservation efforts because of northern New Mexico and north into Utah and
threats from mining and other human related Colorado; E. coccineus represents smooth to
impacts. papillate-spined, tetraploid, florally dimorphic
Until recently, the most widely recognized populations from southern Utah south into the
treatment of Echinocereus was by Benson mountains of Arizona and from southern Colo-
(1982), who grouped all of the red-flowered rado south into Texas and southern Chihuahua;
populations within the United States under E. santaritensis W. Blum & Rutow represents
a single species, E. triglochidiatus Engelm. in F. tetraploid, perfect-flowered, populations from
A. Wislizenus. Since Benson's treatment, other southern Arizona, and E. yavapaiensis hexaploid,
authors reported on the section (Taylor 1985; florally dimorphic, populations from central
Ferguson 1989) but did not use biosystematic Arizona. Several taxa endemic to Mexico remain
approaches. Recent taxonomic, cytological, and poorly understood. For a synopsis of chromo-
floral investigations have led specialists to sepa- some numbers in Echinocereus, section Triglochi-
rate Benson's North American E. triglochidiatus diatus, see Cota and Philbrick (1994) and Baker
into at least five species, E. arizonicus, E. (2006).
coccineus Engelm. in F. A. Wislizenus, E. Nomenclature herein follows that of Blum et
santaritensis W. Blum & Rutow, E. triglochidia- al. (1998). Three taxa are recognized within E.
2006] BAKER: CIRCUMSCRIPTION OF ECHINOCEREUS ARIZONICUS SUBSP. ARIZONICUS 389
Table 1. Current Taxonomic Statusof Echinocereus, Section Triglochidiatus for Populations in
THE United Statesas Recognized by Zimmerman & Parfitt (2003) and Baker (2006).
Status in the current
Taxon Spine surface Ploidy level Flowers multivariate analysis
E. arizonicus smooth-spined diploid perfect Analyzed
E. triglochidiatus subsp. papillate-spined diploid perfect Analyzed as an outgroup
mojavensis
E. triglochidiatus subsp. angular-spined diploid perfect Analyzed asan outgroup
triglochidiatus
E. santaritensis smooth-spined tetraploid perfect Analyzed asan outgroup
E. coccineus smooth to papillate-spined tetraploid dimorphic Not included, see Baker
2006
E. yavapaiensis smooth-spined hexaploid dimorphic Not included, see Baker
2006
arizonicus, E. arizonicus subsp. arizonicus, E. Methods
arizonicus subsp. nigrihorridispinus W. Blum &
Rutow, and E. arizonicus subsp. matudae (Bravo- Twenty one continuous characters (Table 2)
Hollis) Rutow. According to Blum et al. (1998), were measured for at least 30 mature individuals
E. arizonicus subsp. arizonicus occurs in Cochise, from six populations of E. arizonicus, two of E.
Gila, Graham, Pima, and Pinal Counties, Ar- triglochidiatus subsp. triglochidiatus, four of E.
izona, and is characterized as having 7-13 radial triglochidiatus subsp. mojavensis, and three of
spines and 1^central spines per areole and stems E. santaritensis (Fig. 1, Table 3). Because of
with 8-11 ribs; E. arizonicus subsp. nigrihorridis- permitting constraints, stem characters only were
pinus occurs in areas southeast ofthe distribution measured for the only known population of E.
of E. arizonicus subsp. arizonicus, in southern arizonicus subsp. matudae (Bravo-Hollis) Rutow.
Arizona, southwestern New Mexico, and north- For statistical purposes, the assumption was
ern Chihuahua and is characterized as having 10- made that all individuals measured within each
14 radial spines and 3-8 central spines per areole population belonged within a single taxon.
and stems with 10-13 ribs; E. arizonicus subsp. Although it is possible that individuals of more
matudae occurs northwestern Chihuahua and is than one taxon occurred within the vicinity of
characterized as having 7 11 radial spines and 1 any one study population, there were no loca-
4 central spines per areole and stems with 6-8 tions where this was apparent from the physiog-
ribs. nomy ofthe individuals present.
Table 2. Listof Characters UsedintheAnalysis. Except for STEML, three measurementswere made for
each character per individual. The youngest fully mature areoles were chosen for stem measurements.
Character Explanation
STEML Length oflongest stem from ground level
STEMDIA Average stem diameter
NRIBS Average number ofstem ribs (costae)
DBTWARLS Average distance between stem areoles
NCENTRALS Average number ofcentral spines per stem areole
NRADIALS Average number ofradial spines per stem areole
LRADIALS Average length ofradial spines per stem areole
LCENTRALS Average length ofcentral spines per stem areole
THKNCNTR
Greatest thickness ofcentral spine (just above the swollen base) per stem areole
FLRL Flower length from base ofpericarp to tip oflongest tepal
FLRWIDTH Flower width from tepal tip to tepal tip
OUTSDIA Outside diameter ofthroat ofpericarp at constrictionjust above ovary
AXIALL Axial length ofpericarpal pith between ovary and base ofpericarp
LTOUPPER Length ofpericarp from base to uppermost areole
NTEPALS Number oftepals, excluding those with areoles at their base
STYLEDIA Diameter ofstyle at midpoint.
SPINEL Length oflongest within uppermost areole ofpericarp
STYLEL Length ofstyle from its base to bottom ofstigma
NCTRYDIA Diameter ofnectar chamber
OVARYL
Length ofovary
STAMENL
Length ofstamens collectively
.
MADRONO
390 [Vol. 53
-115 -llA -113 -112 -111 -IID -109 -IDS -1D7 -IDG
Fig. 1. Locationsofstudy sites forpopulationsofEchmocereussection Triglochidiatiis. Numbers next to symbols
refer to those ofpopulations in Table 1
MANOVA
Principle component analysis (PCA) (Systat?, (SPSSIO) was used to test the
SPPS Software Inc. 2000) was used to assess the assumptions of multivariate statistics and to test
taxonomic values of characters and to assign the significance of characters among taxa. Data
populations of Arizona and New Mexico popula- were transformed, as necessary, to meet multi-
tions ofE. arizonicusto either E. arizonicus subsp. variate assumptions. Not all variables met
arizonicus, or E. arizonicus subsp. nigrihorridispi- homogeneity of variance assumptions after
nus. This was done by comparing individuals transformations. Also, the assumption of homo-
within unknown populations to those ofthe two geneity of covariance matrices could not be met
type localities. In order to assess the taxonomic (Box's M test, P < 0.001). The Box's M test,
value of groups of variables (primarily stem however, is generally too strict with the large
characters vs. flower characters), PCA was first sample sizes generally necessary for multivariate
ANOVA
performed on all characters ofpopulations ofall applications of (Tabachnick 2001).
threeperfect-floweredspecies,andthenperformed Discriminant analysis (DA; SPSSIO) was used to
usingflowercharactersand stemcharactersalone. testforthecorrectclassificationofindividualswithin
Varimax rotation was used to improve the in- their respective taxa. Because of the small sample
terpretability ofthe scatter diagrams. Thecluster- size, individuals of E. arizonicus subsp. matudae
ingalgorithm unweightedpairgroupmethodwith werenot included intheanalysis. Nineindividuals
arithmetic mean (UPGMA) was performed with were identified as multivariate outliers (Mahala-
NTSYS® 2.1e (Rohlf 2000) to compare phenetic nobis distance-squares from group means with P
distance among populations of E. arizonicus and < 0.001) and were deleted from the analysis. A
the outgroup, E. triglochidiatussubsp. mojavensis. permit (TE-844147) for collecting flowers from
Thisoutgroupwasselectedbecauseitwastheonly individualsofE. arizonicuswasissuedbytheU. S.
other diploid taxon for which at least three Fish and Wildlife Service, Southwest Regional
populations had been measured. Office, Albuqerque, New Mexico.
I
2006] BAKER: CIRCUMSCRIPTION OF ECHINOCEREUS ARIZONICUS SUBSP. ARIZONICUS 391
03 u
^ o
li 'N-l rf\^xx^ji rty\^ rKy^.o^r^<-oi OyKrs Oisn '-^ ••-^N^n-^i ^rino
Uo ^^^^^^^^o^^q CXQ^!":-;:^; C^3q^oXq ^^^ 02
i
o o o o
in o o in in
OmN m
in «
(^1
On
en
in
cn1
(N
in
00 o od OoN ^m b—I ^ O(NN n
ON ON
b in b fN in
b ho\ 0o0 r-H b ooo
oON OIs TT O in o'(d
m
in rn in
d d
ON o00 00 (N m o min O1^ ON
(N in
o
rn CO rn m m m (N in in in min min
Uo N £
< ^
N
< u
^ 5
N (fl si U6 5= o:
0 ^
0 o
c 2 ^ ^ uc CQ g k. = §o ^O
'a ^ U 5
U « 5
w 2 Z U
a
X) X3
C/5
§ 52 -SS2 ^ § CsO ^~~s ^O SS^ ?O!^^ ii
O o O
.M -ij .IJ 5>)' f .bp^tTj .bJ!)
G Q Q Q Q G ~ G C
ki kj kj k] ki k) ki ki kj k^ ki ki
O -7
.
MADRONO
392 [Vol. 53
Results of central spines, and stem diameter (Table 5).
The highest loadings in the second component
Principle Components Analysis (PCA) were for the distance between areole, thickness of
central spine, and stem diameter.
Flower characters dominated the first two Descriptive statistics, based on the defined
components of the PCA analysis that included populations, are presented in Table 6. Those for
all characters (Table 4). When components one outgroups are presented in Table 7.
and two were plotted together (not shown),
however, there was a poor grouping of individ-
uals within their respective taxa. Component Unweighted Pair Group Method with Arithmetic
Mean (UPGMA)
three, which also explained a large percent ofthe
total variance, had a stronger loading for stem UPGMA
characters. When components one and threewere Results from a analysis of stem
plotted together, individuals of the three taxa characters for Echinocereus arizonicus with E.
began to resolve but there remained a great deal triglochidiatus subsp. mojavensis as an outgroup
of overlap (Fig. 2). For the PCA using only indicated that all populations were placed cor-
flower characters, none of the components rectly within their respective taxa (Fig. 4). As
adequately grouped individuals into species. expected, the outgroup (populations 11-14) was
As with the interspecific analysis, stem char- placed with the greatest phenetic distance with
acters discriminated individuals among taxa ofE. respect to populations within E. arizonicus.
arizonicus better than flower characters. A Populations of both E. arizonicus subsp. arizoni-
scatterplot of the first two components of PCA cus (populations 1-3) and E. arizonicus subsp.
defined two groups that maintained the identity nigrihorridispinus (populations 4-6) were distinct-
of individuals within populations (Fig. 3). One ly grouped and the single population of E.
group was defined by populations one, two, and arizonicus subsp. matudae (population 7) placed
three, and the other group by populations four, basally to either ofthe other two subspecies.
five, and six. The first group contained the type
locality for E. arizonicus subsp. arizonicus (pop- Multivariate Analysis ofVariance (MANOVA)
ulation one) and the second for E. arizonicus
subsp. nigrihorridispinus (population six). The Individuals of E. arizonicus subsp. arizonicus,
highest loadings in the first component were for as a group, possessed seven characters with
length ofboth central and radial spines, number distinct means: stem length, stem diameter,
Table 4. Component Loadings for all Charactersof PCA with Varimax Rotation of Populations
OF E. ARIZONICUS (EXCLUDING E. ARIZONICUSSVBSP. MATUDAE), E. SANTARITENSIS, AND E. TRIGLOCHIDIATUS.
For translation ofcharacter acronyms, see Table 1
Component
1 2 3 4 5
STYLEL 0.9 0.0 -0.1 0.0 0.0
FLRL 0.9 -0.2 0.0 -0.0 0.2
STAMENL 0.8 0.1 0.1 -0.1 -0.3
LTOUPPER 0.8 0.2 -0.1 -0.2 0.3
NRADIALS 0.0 0.9 0.0 -0.1 0.0
NCENTRALS 0.1 0.9 0.0 -0.1 -0.1
DBTWARLS 0.1 -0.8 0.1 0.3 -0.1
NRIBS -0.0 0.8 0.0 0.1 0.3
OVARYL 0.4 -0.5 0.1 -0.1 0.3
NCTRYDIA -0.1 0.0 0.9 -0.1 -0.1
OUTSDIA
0.0 0.1 0.8 0.1 0.1
STYLEDIA 0.0 -0.0 0.6 -0.0 0.0
LCENTRALS -0.1 0.1 -0.0 0.9 -0.1
LRADIALS -0.1 -0.4 -0.0 0.8 0.2
AXIALL 0.4 0.1 -0.2 -0.2 0.7
NTEPALS 0.0 0.0 0.3 0.3 0.6
STEMDIA 0.1 -0.1 0.2 -0.2 0.0
STEML -0.1 0.3 0.1 0.1 0.2
THKNCNTR -0.0 -0.1 0.2 0.2 -0.4
FLRWIDTH 0.5 -0.0 0.3 0.2 0.2
SPINEL 0.2 0.5 0.1 0.4 -0.0
Percent oftotal variance explained
17.0 17.8 10.3 9.8 7.2
2006] BAKER: CIRCUMSCRIPTION OF ECHINOCEREUS ARIZONICUS SUBSP. ARIZONICUS 393
8
E. arizonicus
E. santaritensis
E. triglochidiatus
CO
c • T
0
c
o
Q-
E
o
O
T
-2 T
T
-4
-6 -4 -2
Component
1
Fig. 2. Scatterplotofcomponents 1 and 3forPCAwithvarimax rotationusingallcharactersforindividualsofE.
arizonicus, E. santaritensis, and E. triglochidiatus.
Population
® • A • 01
• ® 02
A O 03
A A 04
A A 05
A 06
c •©• i®m^ ^•^ ^ A
(CD A
o t
Q.
A
E
o
O
-1 ® • • ^ A A
• A^ AA A O ^ A
A A
A
-2 A
1 0 1
Component
1
Fig. 3. Scatter plot of first two components of PCA with varimax rotation including stem characters for
populations ofE. arizonicus, excluding the one population ofE. arizonicus ssp. nuitudac.
MADRONO
394 [Vol. 53
Table Component Loadings for Stem resolution among the species sampled, there were
5.
Characters of PCA with Varimax Rotation of significant differences in means for several
Populationsof E. arizonicus. Excludingthatof morphological characters between E. arizonicus
e. arizonicussvbsp. matudae. subsp. arizonicus and E. arizonicus subsp. nigri-
horridispinus. Morphological differences coupled
Component
with allopatric geographic distribution have long
1 2 3 been considered basic criteria for the recognition
NCENTRALS 0.8 -0.1 -0.1 of two separate subspecies (Stebbins 1950;
LCENTRALS 0.8 -0.3 0.4 Lawrence 1951). Most ofthe known geographical
NRADIALS 0.8 0.0 0.0 distribution for E. arizonicus is represented by E.
LRADIALS 0.7 -0.3 0.4 arizonicus subsp. nigrihorridispinus, whichwasthe
DRIBS 0.6 -0.1 -0.3 most variable subspecific taxon within the species
STEML -0.0 0.8 -0.2 in terms of the diagnostic stem characters.
SDTBETMWDAIRALS -0.4 0.8 0.2 Individuals of E. arizonicus subsp. arizonicus,
THKNCNTR -00..00 00..17 00..58 wmhoirceh vhaarvieablae lwiimitthedregsepoegcrtaphtoicmroasntge,flwoewreer
Percent oftotal variance explained
characters.
30.7 20.7 16.6
Populations of E. arizonicus subsp. arizonicus
differed significantly from those of E. arizonicus
subsp. nigrihorridispinus in the means of most
distance between areoles, number of central characters measured (Table 8). However, as
spines, number of radial spines, length of shown by PCA, flower characters, as a group,
pericarp, and style diameter (Table 8). Individu- were not generally diagnostic within the perfect-
als of E. arizonicus subsp. nigrihorridispinus, as flowered taxa as a whole and no single flower
a group, possessed ten characters with distinct character appeared to be useful in separating
means: stem length, number of ribs, distance populations of E. arizonicus subsp. arizonicus
between areoles, number of centrals, number of from those of E. arizonicus subsp. nigrihorridis-
radials, length of radials, flower length, pericarp pinus. Although stem characters were shown to
spine length, style length, and ovary length be more diagnostic than flower characters, the
(Table 8). former tend to be more affected by factors ofage
and environment. Stem diameter, for example,
should be avoided as a key character because of
Discriminant Analysis
its correlation with available water. Although age
Discriminant analysis correctly classified is generally a critical factor in stem length, stem
97.8% ofthe original grouped cases that included length is mostly determinate within section
all of the study taxa (Table 9). Individuals of E. Triglochidiatus because of its cespitose habit. In
arizonicus subsp. arizonicus were classified cor- addition, the effects ofage were minimized in the
rectly 97.0%, with 3.0% incorrectly classified as present study by the measurement ofonlymature
E. arizonicussubsp. nigrihorridispinus. Individuals individuals. Diagnostic characters that are less
of E. arizonicus subsp. nigrihorridispinus were affected by age and environment are number of
correctly classified 94.7%, with 2.1% incorrectly ribs, number and length of central spines, and
classified as E. arizonicus subsp. arizonicus and number and length of radial spines. Although
3.2% as E. santaritensis. A single individual (1%) even these characters may be affected by etiola-
of E. santaritensis was misclassified as E. tion, no individuals occurring in deep shade were
arizonicus subsp. nigrihorridispinus and one as included in the present study.
E. triglochidiatus subsp. mojavensis Within E. The geographic ranges, as defined herein, ofE.
.
triglochidiatus, only a single individual was arizonicus subsp. arizonicus and E. arizonicus
misclassified, which was between the two sub- subsp. nigrihorridispinus differs from that of
species. All between group correlations were Blum et al. (1998) in that populations of E.
highly significant (Table 10). The jackknifed arizonicus subsp. arizonicus are more restricted.
classification matrix showed a one percent re- Data herein were not sufficient to adequately
duction in correct classification among each evaluate the taxonomic circumscription of E.
group. arizonicus subsp. matudae. Although individuals
within the single known population possessed the
Discussion fewest ribs, greatestdistance between areoles, and
fewest central and radial spines in comparison to
Phenetic analysis presented herein supports the those of the other two subspecies, additional
recognition of at least two infraspecific taxa populations of E. arizonicus subsp. matudae,
within E. arizonicus: E. arizonicus subsp. arizoni- should be sought and measured in order to
cus and E. arizonicus subsp. nigrihorridispinus. properly address morphological variation
Although PCA shows incomplete interspecific throughout its range.
2006] BAKER: CIRCUMSCRIPTION OF ECHINOCEREUS ARIZONICUS SUBSP. ARIZONICUS 395
00O^n^_r-OrJO--;^rnO'^'Nt;MD--;'^_OOO--;
^00—I—|^voo^^OO^OOLr)OOO^O^Or^o
m ^ ^ — n ^
I r<-i
O O
r<-,
ON ON ri
o
in r-
00 iyS ^ —h' O ^ ^ ri
'OONi^^r-^t^oooo-^O^^^—'Oorirjoo^Or*"j
MO o O H O c<j o in uri --h' ri o ri r^i o ri ^
ri^M3inorii^inooir)ONOov-)Ttmvor^O(^oooo
in^ON—''^Ooooorio-^
ri —I ^^T^f vo'vf-^ in
m ^ r^i r<-j in r<-i
00 r^ ON
ONO^rnONrninrj i—n O'O^rjrJOrlONm—lOO'^
—H o ri o ri r-' o ON —< iri ^ r<-i o in
MD^oq^ON—^TtooNirn~;'sfoooinONr~-'vO'*^
o o
OriN o6 o6 -H rj ON r-i^r'iO ^r-^^r^^*-(^Nnrj-iONrrii rn
oo d oo in ri f^—' 0r0o m rn r*~> MD ON od m mrj iinn in r<-) ro<-)
r^ in ON in rj >n in in
o
rj m
d r- ri r-
in ON in ON in mrn 00 in ri
ri ri
o o o o o o o O o O o o o O O O o O o O O
2^!>0- a:
J J < W ^
a:
MADRONO
396 [Vol. 53
O^N d doo — d ^ oo o ON in (rN<^SO>'s-diori-udo^r'-^OdN^O^N
^
^ >n -H r-i
o6 0\ 0\ r<{ 0\ \6 dr-^rioNrnTtri--<dodr<-i'sdod
OOOOOOr<-)OOinr^(^>ntr)inoNr^O>nir)in
oooooor-r^r<-) OOroOOin—Hi^inoqr^O
^ — ^
' \o < >s6 r--" ' r<-i
00
O o orj (-~- n O ri o(\ nrj
in ON ri
OO 00 d q dON d 00 d 00 r) OiNn q ri ON d
>sd rj ON ri od rn od
ri ri
q q «r) q r- q d do q O q q r- r- q
ON od \6 00 ON ri in in ri
in 00 in ri in ri in in
o
o q o in
d d
•n in od in in m ri rmi
"n ri
m m m m m m
rn rj r)
1^ r- r--
do dON d d md ri drJ m ri iinn drj ON riin 0d0 rqi ro<q-i
OqN q 00 q dON d in d dON roi q O in OONN r0i0 iinn
ri ri
1/5 o
o OoN q in roi OOn ri rj od d q iqn mriin ri ir-n d
ri 00 OO rj in in in
o
q OO r- od q d in o q q m in r- in in
>n 00 od ri ri
rj r) ri
o o n
mri ri ri m ri rj mrj mrj r4 r4 rj rmj ri r) rj mri
en
<
3
w d1 hwj
D
O Owh
^ Q Z O < J z
;^ hJ hJ
—
2006] BAKER: CIRCUMSCRIPTION OF ECHINOCEREUS ARIZONICUS SUBSP. ARIZONICUS 397
"~1
0.21 0.56 0.91 1.26 1.61
Coefficient
UPGMA
Fig. 4. Phenogram from including populations ofE. arizonicus and E. triglochidiatus ssp. mojavensis.
Matrix composed ofmean values for each character.
Similarly, data were not adequate for the for E. triglochidiatus ssp. triglochidiatus. Further-
UPGMA
assessment of the taxonomic circumscription of more, suggests that E. triglochidiatus
the two subspecies ofE. triglochidiatus, primarily subsp. mojavensismay becomposed ofmore than
because ofthe lack ofdata from the type locality one taxonomically definable group, a western
Table 8. Selected Homogeneous Subsets from MANOVA Significance (Duncans Multiple Range
Test) where P > 0.02.; Subset A hasthe Smallest Mean Valueand E hasthe Largest. For all subsets
containinga singletaxon,/? = 1.000. 1 = E. arizonicussubsp. arizonicus, 2 = E. arizonicussubsp. nigrihorridispinus,
3 = E. santaritensis, 4 = E. triglochidiatus subsp. mojavensis, 5 = E. triglochidiatus subsp. triglochidiatus. See
Tables 5 and 6 for specific means.
Subset
A B C D E
STEML 5 3 4 2 1
STEMDIA
3 4, 2 2, 5 1
NRIBS 5 1, 4 2 3
DBTWARLS
3 2 1 4 5
NCENTRALS
5 4 1 3 2
NRADIALS 5 4 1 3 2
LRADIALS 1 3 2 5 4
LCENTRALS
1 5, 3 4, 2
THKNCNTR
FLRL 23 4, 3, 41 2, 55 5, 1
OUTSDIA 3 5, 4 2, 1
LTOUPPER
4, 5, 2 1 3
STYLEDIA 3, 4 5, 2 1
SPINEL 5 4 3, 1 2
OSTVYALREYLL 2 4, 1 5 3
STAMENL 2 3 1, 4 5
4 2, 3 1, 5