MORPHOLOGICAL VARIATION IN MALACOTHAMNUS FASCICULATUS (TORREY & A. GRAY) E. GREENE (MALVACEAE) AND RELATED SPECIES

11 Pages·1999·7.7 MB·English

by  Deborah L Benesh

Madrono, Vol. 46, No. 3, pp. 142-152, 1999
MORPHOLOGICAL VARIATION IN MALACOTHAMNUS FASCICULATUS
(TORREY & A. GRAY) E. GREENE (MALVACEAE) AND RELATED SPECIES
Deborah L. Benesh
Oklahoma Biological Survey, University of Oklahoma, Norman, OK 73019
Wayne J. Elisens
Department of Botany and Microbiology, University of Oklahoma,
OK
Norman, 73019
Abstract
Character polymorphism is widespread and species delimitation is unresolved among 11 species of
Malacothamnus currently recognized in California and Baja California. To evaluate variation among
characters used in previous taxonomic treatments and to identify characters most usefulfordiscriminating
among four species closely associated with M. fasciculatus, we analyzed 26 morphological characters
from 46 exemplarspecimens using cluster, ordination, and statistical analyses. Although many vegetative
and reproductive characters exhibited high levels ofvariation among and within OTU's, bractletandcalyx
characters demonstrated maximum utility for discriminating among species in principal components and
discriminant functions analyses. OTU's representing previously segregated "varieties" ofM.fasciculatus
(Torrey & A. Gray) E. Greene did not group together nor did conspecific OTU's, except forthree OTU's
representing the San Clemente Island endemic, M. clementinus (Munz & I. M. Johnston) Kearney. The
absence ofwell-defined gaps in many characters and intergradation resulting from gene flow and/orrecent
divergence from ancestral species impede delimitation at the species and infraspecific levels whether a
phenetic or phylogenetic approach is employed. Malacothamnusfasciculatus and M. fremontii A. Gray
are particularly variable and difficult to define unambiguously using morphological data.
Introduction logically distinct within Malvaceae, species and va-
rietal delimitations are problematic because ofchar-
Malacothamnus Greene is a genus of woody or acter intergradation and morphological variation
suffrutescent perennials with 11 species currently both within and among populations (Kearney
recognized in California and Baja California (Bates 1951b; Bates 1963, 1993). The degree oftaxonom-
h19a6m3a, C1o9u93n)t.ieDsisitnrinbourttehderfnroCmalMiefonrdnoicaitnooRaonsdarTieo-, siciduenrcienrgtatihnetywiadnedlyinfstlaubcitluiattyinigs envuimdbeentr wohfesnpeccoine-s
Baja California Norte, species of Malacothamnus
recognized in different treatments. Whereas Greene
are primarily associated with chaparral and coastal
sage scrub communities. The common names of (1906) recognized nine species, subsequent treat-
Malacothamnus, chaparral mallow and bush mal- ments recognized from one (Bates 1963) to twenty-
seven (Eastwood 1936) species and a variable num-
low, reflect its habit and habitat preferences.
Established by E. L. Greene (1906), Malacoth- ber of infraspecific taxa. In the most recent treat-
agemdn,uscoismpclheatreacltyeridzeehdisbcyenutniocvaurlpaetles,; unaaspcpeennddian-g rmeecnotgsniozfedthe19Caslpiefcoirensiaansdpesciixesv,arKieetairenseaynd(1M95u1nbz)
ovules; and a shrubby habit. Later treatments place and Keck (1968) delimited 18 species, one subspe-
Malacothamnus in Tribe Malveae, Subtribe Abutil- cies, and six varieties. In the Jepson Manual (Hick-
inae (Kearney 1951a, b). Bates (1968) did not rec- man 1993), Bates recognized ten species without
ognize subtribes, but regrouped the Malveae into infraspecific taxa; taxa previously ranked as sub-
"alliances" and suggested a Malacothamnus alli- species (Bates 1963) were elevated to species level
ance consisting of Malacothamnus (n = 17), Ili- without changes in circumscriptions (Bates, person-
amna Greene (n = 33), and Phymosia Desvaux (n al communication). Skinner et al. (1995) noted that
= 17). Based on cytological evidence coupled with several Malacothamnus taxa are of "uncertain dis-
inference of Tertiary floristic evolution, Bates tinctiveness" and calculated that 75% of the taxa
(1967, 1968) suggested that the three genera are in the genus are taxonomically uncertain.
derived from a Phymosia-like ancestor. Eastwood Few characters unique to any species have been
(1939) proposed that the California Malvastrum identified in previous taxonomic treatments. Gen-
{Malacothamnus) was an old genus of previously erally, combinations ofoverlapping characters have
greater abundance, but now in decline. Kearney been used to delimit species (Kearney 1951b; Bates
(1951b) noted that the lack ofpronounced morpho- 1963, 1993). These recent treatments have em-
logical differences among species of Malacotham- ployed type and density of calyx pubescence,
nus suggested a recent origin for the genus. length and shape of the calyx lobes, length and
Although the genus Malacothamnus is morpho- shape of the bractlets, relative length of the invol-
1999] BENESH AND ELISENS: MALACOTHAMNUS 143
ucel bractlets and calyx lobes, type of inflores- and the high level of variability found within spe-
cence, and shape and texture ofthe leaves (Kearney cies, there have been no explicit numerical studies
1951b; Bates 1963, 1993). For example, M. fre- ofthe amount and pattern ofvariation nor statistical
montii A. Gray and M. davidsonii (Robinson) E. analyses of characters that differentiate species.
Greene can be identified by thick, often densely Malacothamnus fasciculatus is morphologically
pubescent leafblades, but other species are not eas- and geographically diverse, has been examined at
ily distinguished based on leaf characters. Bates the molecular level (Swensen et al. 1995), and in-
(1963) provided an example where polymorphism tergrades commonly with other sympatric species,
in leaf shape in a single population of about 50 particularly in the southern portion ofits range. The
individuals of M. fasciculatus (Torrey & A. Gray) present study focused on variation in morphological
E. Greene encompassed almost all the variation characters among five intergrading species: M. cle-
found throughout the genus. Differences in habit mentinus (Munz & I. M. Johnston) Kearney, M.
among species sometimes may be observed, which davidsonii, M. densiflorus, M. fasciculatus, and M.
may result from environmental factors such as wa- fremontii. The primary objectives ofthe study were
ter stress, fire, and competition, affecting growth to assess variation among characters used in pre-
form (Bates 1963). vious taxonomic treatments, to identify characters
The basic floral arrangement within Malacoth- that discriminate among species, and to test hy-
amnus is a cyme arranged into terminal heads, potheses of taxonomic delimitation of species and
spikes, or open panicles (Bates 1963); variation can varieties.
be found in the length, numberofnodes, branching, The five species examined in this study are wide-
internode length, and leafiness of the inflorescence spread in southern California and Baja California
(Bates 1963). Examples include M. densiflorus (S. Norte, and one species is distributed in northern as
Watson) E. Greene, which is distinguished by a well as southern California. Malacothamnusfascic-
densely flowered, spikeM-like inflorescence with ulatus as delimited by Kearney (1951b) ranges
elongate internodes; and palmeri (S. Watson) E. from Santa Barbara County to Baja California Nor-
Greene, which has a densely flowered, head-like or te. Kearney (1951b) recognized five varieties: vars.
short, spike-like inflorescence subtended by con- fasciculatus Greene, laxiflorus (Gray) Kearney, and
spicuous bracts (Bates 1993). Involucel bractlet and nuttallii (Abrams) Kearney are restricted to main-
calyx characters also have been used to delimit spe- land California, whereas var. catalinensis (Eastw.)
cies, but exhibit polymorphism within and among Kearney (found on Santa Catalina Island and coast-
species (Kearney 1951b; Bates 1963, 1993). Within al Ventura to San Diego Counties) and var. nesi-
Malacothamnus, bractlet length ranges from 2.5 to oticus (B. L. Rob.) Kearney (restricted to Santa
mm mm
21 and calyx length varies from 5 to 19 Cruz Island) have populations on the California Is-
(Bates 1993). A single species, M. densiflorus, en- lands. Also widely distributed are M. fremontii,
compasses most of the range for these characters; ranging from Tehama to Riverside Counties and M.
mm
its bractlets vary from 5 to 15 in length and densiflorus, which is distributed from Riverside
its calyces from 6 to 14 mm. Unlike leaf, inflores- County to northern Baja California Norte. The fed-
cence, bractlet, and calyx characters, morphology erally and state listed endangered species M. cle-
ofthe corolla, pistil, stamen, carpels, and seeds pro- mentinus (Smith & Berg 1988) is endemic to San
vide few characters used to segregate species Clemente Island. Populations of M. davidsonii are
(Kearney 1951b; Bates 1963). separated by the Transverse Ranges between Los
Similar to problems in species delimitation and Angeles County and San Luis Obispo and Monte-
identification, infraspecific taxa in M. fasciculatus rey Counties.
are notoriously difficult to identify. Although ex-
treme morphological forms are evident, intergra- Materials and Methods
dation among habit, leaf shape and pubescence, in-
florescence structure, and carpel characters is wide- We examined variation among 26 morphological
spread. Five varieties of M. fasciculatus were de- characters among five species of Malacothamnus
scribed previously by Kearney (1951b) and eight using clustering, ordination, and statistical analyses.
subspecies by Bates (1963) in an unpublished dis- Taxonomic identity was determined using keys and
sertation. Later, Bates (1993) recognized no infra- descriptions from the Jepson Manual (Hickman
specific taxa in M. fasciculatus. In a study of mo- 1993) and, for "varieties" ofM.fasciculatus, Kear-
lecular variation in M. fasciculatus, Swensen et al. ney's treatment (1951b) and Munz and Keck (1968)
(1995) provided data that indicated var. nesioticus were used. One representative individual from each
is genetically distinct from othervarieties ofM.fas- of 46 collections representing natural populations
ciculatus. (Table 1, Fig. 1) was selected as an exemplar op-
We conducted this study to initiate analyses of erational taxonomic unit (OTU). The OTU's were
character variation within and among species of chosen from herbarium specimens or from collec-
Malacothamnus. Although previous taxonomic and tions of the first author; they were selected to span
floristic treatments (Kearney 1951b; Bates 1963) the geographic and morphological range of each
emphasized the lack of constant, unique characters species analyzed (Fig. 1). The widespread species
MADRONO
144 [Vol. 46
Table 1. Specimens representing numbered OTU's of Table 1. (Continued).
five species of Malacothamnus used in investigations
of morphological variation. All locations are in the "var. nuttallii (Abrams) Kearney"
state ofCalifornia unless otherwise noted. 3. Santa Barbara Co., upper Cachuma Dam area, Smith
M. clementinus (Munz & Johnston) Kearney 4ta9i4n0s,, SHBoBwGe.241.64L,osSADn.ge2l5e.sSCaon.t,aSaBnatrabaMroaniCco.a,MSoaunnt-a
13. San BernardinoCo.,Victorville,NorthVerdeRanch, Barbara, Benesh 63, OKL. 26. Ventura Co., Lake Cas-
Raven 16625, CAS. 14. Los Angeles Co., SanClemente itas, near East Casitas Pass, Benesh 64, OKL. 43. Santa
Island, Lemon Tank, Beauchamp & Douglas 3236, SD. Barbara Co., Gaviota Pass, Abrams 5030, UC.
y1o5.n,LMoosrAanng,elBeesauCcoh.,amSpa,n&CleOmbeenrtbeauIesrlan2d2,69C7h,inUaC.Can- M.fremontii (Torr. ex Gray) Torr. ex Greene
1. Santa Barbara Co., Vandenburg AFB, Nichols 664,
M. davidsonii (B. L. Rob.) Greene JER 2. Sutter Co., North Butte, Ewan & Ewan 9597,
11. San Luis Obispo Co., Bee Rock, Hardham s.n., JER 8. Madera Co., Oakhurst, Wolf2433, SBBG. 28.
CAS. 12. Los Angeles Co., San Fernando Wash, David- Los Angeles Co., Pine Canyon, Hoffman s.n., SBBG.
son s.n., CAS. 16. Los Angeles Co., Big Tujunga, Lyon 29. Ventura Co., Lockwood Valley, unknown s.n.,
s.n., CAS. 17. Monterey Co., Lake San Antonio,Jensen SBBG. 38. San Bernardino Co., Baldwin Lake, Tilforth
45, UC. & Wisura 2107, RSA. 42. Riverside Co., San Jacinto
Mtn., Purer 6472, CAS.
M. densiflorus (S. Watson) Greene
9. Mexico, Baja, Valle Redondo, Fosberg 8368, UC. 10.
RiversideCo., SantaAnaMtns., aboveGlen Ivy,Howell
1056, UC. 18. San Diego Co., Barrett Lake, Benesh & M. fasciculatus and M. fremontii were dispropor-
Anderson 35, OKL. 39. San Diego Co., Ramona, Bran- tionately represented in their southern range where
degee s.n., UC. 46. Mexico, Baja, San Pedro Martir they and other species are sympatric.
Mtn., Robertson 55, UC. Eleven quantitative and 15 qualitative characters
M.fasciculatus (Nutt. ex Torr. & Gray) Greene were chosen to represent vegetative and reproduc-
"var. catalinensis (Eastwood) Kearney" tive structures (Table 2). Characters were selected
5. Los Angeles Co., Santa Catalina Island, Middle based on their inclusion as diagnostic features in
Ranch Dam, Fosberg 54770, UC. 6. Orange Co., La- previous taxonomic treatments (Kearney 1951b;
guna, Peirson 4662, SD. 19. Los Angeles Co., Santa Munz and Keck 1968; Bates 1963, 1993). For each
Catalina Island; Black Jack Mtn., Thome 36518, OTU, we obtained a minimum of 10 measurements
SBBG. 21. Ventura Co., Santa Monica Mtns. beyond for each character whenever possible. Several mea-
Hidden Valley on Sherwood Rd., Holbrook s.n., surements ofbractlet, calyx, and leaf length and/or
SBBG.
width were converted to a ratio (characters 2, 7, 17,
"var.fasciculatus (Nutt. ex Torr. & Gray) Greene" 18; Table 2) to describe shape and relative dimen-
20. San Diego Co., 2 mi EofSan Ysidro on Otay Mesa, sions rather than absolute size. Lack of sufficient
Gander 219.13, SD. 31. Riverside Co., El Toro Peak, suitable material on herbarium specimensprecluded
San Jacinto Mtns., SantaRosaRange, Hall 765, UC. 32. the use of leaf size, carpel characters, and seed
San Diego Co., Fallbrook; ca. 6 mi NE, Benesh & An- characters in our analysis. Preliminary analyses
dArebroslointo3s9,,TOhKoLm.e &33.ChMaerxlitcoon,58B6aj8a5,, PRuSnAt.a3B4a.nMdeaxinceoa,r also revealed that several characters not included in
Baja, Rancho Estela, Moran 24380, SD. 35. Mexico, our analysis were either variable on a single indi-
Baja, El Condor, Sierra Juarez, Moran 13578, SD. 37. vidual, too subjective to be coded accurately, or
San Mateo Co., Spring Valley, Cougdon s.n., JEPS. 40. difficult to detect (e.g., leaflobing, leafmargin, leaf
Mexico, Baja, Upper Arroyo Copal, Sierra San Pedro apex, leaf thickness, pubescence color, inflores-
Martir, Moran 15499, SD. 41. Mexico, Baja, Santa cence type, and bractlet shape).
Rosa, Moran & Thome 14427, RSA. 44. San Bernar- Analyses were performed using NTSYS-pc
dino Co., San Bernardino Mtns., State Hwy 18 ca 0.5 (Rohlf 1993) and SYSTAT (SPSS Inc. 1997). A
mi above Waterman Canyon Road, Raven & Beeks standardized data matrix was used to compute a
H1o67n5d9o,),UCT.hom45e.,MWeixsiucroa,, BPaejtae,rsoSno,cor&roAnWnaasbhle(5A8r0r1o1y,o dissimilarity matrix using the average taxonomic
RSA. distance coefficient and acophenetic correlation co-
efficient. A clustering phenogram was generated
"var. laxiflorus (Gray) Kearney" using the UPGMA (unweighted pair-group method
22. Los Angeles Co., Altadena, Ramsey & Ramsey using arithmetic averages) method. A Principal
3084, RSA. 23. Orange Co., Crystal Cove State Park, Components Analysis (PCA) was performed by
El Moro Canyon, Benesh 57, OKL. 24. San Bernardino computing a correlation matrix, calculating eigen-
CRPoio.vi,netrsCiMadjueognuCo,.P,aVssGa,anrdBneeerrnweiVseahrlle&5y3,,MTuOirKlfpLoh.rethy27&.78,WVieSsnuBtruBarGa.19C1o36.0,., OveTcUt'orssofnotrotthheefcirosmtptohnreenetsa.xes, and projecting the
RSA. 36. San Diego Co., Camp Pendleton, Horno Can- Using the SYSTAT software package (SPSS Inc.
yon, Beauchamp & Swickard 1518, SD. 1997), discriminant functions analysis was used to
determine which combination ofcharacters provid-
"var. nesioticus (B. L. Rob.) Greene" ed maximum discrimination amongrecognized spe-
71.888S,antUaC.Barbara Co., Santa Cruz Island, Brandegee cies and "varieties" of M. fasciculatus. Because
this method does not allow missing data, seven of
1999] 145
Fig. 1. Collection localities for46 specimens offive species ofMalacothamnus analyzed formorphological variation.
Solid symbols represent species described in The Jepson Manual (Hickman 1993; Bates 1993). Shaded symbols rep-
resent "varieties" ofM.fasciculatus as delimited by Kearney (1951b; Munz & Keck 1968).
1288 entries (0.54%) from the data matrix were es- Results
tsiammaetedspbecyieuss.inSgtathnedaarvderaangde ojfacoktkhneirfeOdTUc'lasssoiffitchae- The UPGMA phenogram (Fig. 2) had four major
tions were produced to test group relationships. In clusters and a cophenetic correlation value of
othfeliskteantdaaxradwparsocceodmupruet,edth,eacnednttrhoeindeoafcheaOcThUgrwoausp f0r.e8m4o0n.tiCliuastnedraInwOasTUcormepporseesedntoifngtwMo.OfTasUc'iscuolfatMu.s
"var.fasciculatus" ofunusual appearance from the
esixmailmairn.edWhtoendettehremiOnTeUtowwahsichmogsrtouspimiitlwarastomotshte southern limits of its range in Baja California Nor-
group representing that taxon, it was scored as a tpeh.eCnlougsrtaerm,IIawnadsctoheulmdosbtecfoumrpthleerxsgurboduipviindgedinitnhteo
correct identification. When it was most similar to three groups, excluding the outlying "var.fascicu-
a group representing another taxon, it was scored latus" (OTU 37) from San Mateo County. With the
calsasasinfiicantcioornr,ecatncOlaTssUifiwcaatsionr.emFoovredt,hethjeacckeknntirfoeidd rtewporesaefnotrienmgenMt.ionfeadsciecxucleapttuisonsw,eraell footuhnedr OwTitUh'isn
recalculated, and the OTU identified as to which cluster II along with OTU's of three other species.
group it was most similar. This was repeated for Cluster Ha consisted of 10 OTU's representing M.
each OTU, producing a less biased number of cor- fremontii, M. davidsonii, and three "varieties" of
rect classifications. Three discriminant functions M. fasciculatus. The closest phenetic similarity
analyses were performed: species level only, spe- among all OTU's was in this cluster between M.
cies and all "varieties" ofM.fasciculatus, and spe- davidsonii and M. fasciculatus "var. laxiflorus"
cies and "varieties" ofM.fasciculatus except "var. (OTU's 17 and 24). Cluster lib comprised OTU's
nesioticus". The latter analysis was performed be- representing all five "varieties" of M. fasciculatus
cause "var. nesioticus" was represented by only a as well as one OTU of M. davidsonii from Los
single OTU, which confounded thejackknifed clas- Angeles County. The OTU representing the Santa
sification. Cruz Island endemic M. fasciculatus "var. nesioti-
MADRONO
146 [Vol. 46
Table 2. Morphological characters examined among five species of Malacothamnus and purported varieties
of M. fasciculatus. Characters useful for differentiating among species in principal components analysis (PCA) are
indicated by an asterisk (*). Characters useful fordifferentiating species in discriminantfunctions analysis are indicated
by a plus sign (+).
1. stem pubescence type: loose (0), appressed (1)
2. leafblade length/width
3. leafbase: cordate (0), truncate (1)
* + 4. bractlet length
* 5. bractlets concolorous throughout: absent (0), present (1)
* 6. bractlet concolorous with calyx: absent (0), present (1)
* + 7. bractlet length/calyx length
8. calyx length from base to tip oflobe
* 9. calyx outer pubescence type: rigid (0), soft or pliant (1)
* + 10. sepal lobes in bud recurved at margins: absent (0), present (1)
* 11. sepal lobes in bud: divergent (0), connivent (1)
12. sepal lobe shape: triangular (0), ovate (1)
13. sepal lobe apex: acute (0), acuminate (1)
* 14. sepal lobe constricted at base: absent (0), present (1)
* 15. sepal lobe length
16. sepal lobe width at base
17. sepal length/width
18. sepal lobe length/calyx tube length
19. sepal lobes obscured by pubescence: absent (0), present (1)
+20. sepal pubescence: short-stalked (0), long-stalked (1)
21. sepal pubescence: number of arms
* 22. sepal pubescence: length of arms
23. sepal lobe pubescence type within: glabrous (0), tomentose (1), pilose (2), hispid (3)
24. inflorescence: unbranched (0), branched (1)
25. flower clusters: dense (0), open (1)
26. length of staminal column
cms" clustered most closely with "var. nuttallii" fasciculatus. As in the UPGMA phenogram, the
from Los Angeles County, whereas OTU's of "var. PCA model placed M. davidsonii within theM.fas-
catalinensis"\ including two from Santa Catalina ciculatus complex.
Island, did not group together. Discriminant functions analysis identified four
Only M. densiflorus and M. fasciculatus OTU's characters as most useful for classifying species: 4,
were contained in Cluster He, which included spec- 7, 10, and 20 (Table 2). The OTU's were correctly
imens from south-central Riverside County to the classified to species an average of 67% (Table 4)
southern limits of the North American range in in the standard classification. Whereas M. densiflo-
Baja California. Cluster III was composed of two rus was classified correctly 100% of the time, the
subclusters ofOTU's representing M.fremontii and lowest value, 56%, resulted when 12 M. fascicu-
M. clementinus. The M. clementinus OTU's includ- latus OTU's were misclassified as M. davidsonii
ed a (cultivated?) specimen from San Bernardino (10) or M. densiflorus (2). In the jackknifed clas-
County. Cluster IV comprised two M. densiflorus sification, a 52% average correct classification was
OTU's that are less similar to each other than are obtained. The lowest value, 44%, resulted when 15
most of the other clusters. OTU's of M. fasciculatus were incorrectly classi-
The three-dimensional PCA model (Fig. 3) re- fied.
vealed groupings similar to those in the UPGMA When "varieties" of M. fasciculatus were in-
phenogram. The first three components constituted cluded as taxa, characters 7 and 10 (Table 2) were
49 percent of the total variation: 28%, 11%, and the most useful for discriminating among species
10%, respectively. Characters with high loadings and "varieties" using discriminant functions anal-
(Table 3) included characters 4, 7, 11, 15, and 22 ysis. The correct classification percentages were
in the first component; 9, 10, and 14 in the second relatively unchanged by the exclusion of the one
component; and 5 and 6 in the third component. OTU of "var. nesioticus" The standard classifica-
.
The PCA model also did not reveal discrete group- tion was correct 43% of the time when "var. nesi-
ings corresponding to species boundaries, but in- oticus" was included (Table 4) and 47% when it
stead depicted loose groups with considerable over- was removed compared to jackknifed percentages
lap among taxa. OTU's representing M. clementin- of 35% when "var. nesioticus" was included and
us, M. densiflorus, and M.fremontii were generally 36% when it was excluded. Malacothamnus dav-
separated from M. fasciculatus and M. davidsonii idsonii and M. fasciculatus "var. laxiflorus^ were
along the first axis. There were no discrete group- consistently classified incorrectly in analyses in-
ings among OTU's corresponding to varieties ofM. cluding varieties. For example, OTU's of M. dav-
.
BENESH AND ELISENS: MALACOTHAMNUS 147
1999]
1 frem
2 frem
41 fasc fasc _~
3 fasc nutt
25 fasc nutt
44 fasc fasc
26 fasc nutt
11 dav
17 dav
ila
24 fasc lax
30 fasc lax
42 frem
12 dav _~
4 fasc nutt
7 fasc nes
5 fasc cat
36 fasc lax
6 fasc cat
22 fasc lax
21 fasc cat
20 fasc fasc lib
23 fasc lax
27 fasc lax
45 fasc fasc
34 fasc fasc
16 dav
32 fasc fasc
19 fasc cat
43 fasc nutt _
9 dens
40 fasc fasc
31 fasc fasc
35 fasc fasc lie
39 dens
46 dens
33 fasc fasc _
37 fasc fasc
8 frem
-d 29 frem
38 frem
13 clem
14 clem
15 clem
28 frem
10 dens
IV
18 dens
2.0 1.5 1.0 0.5 0.0
AverageTaxonomic (Morphological) Distance
Fig. 2. UPGMA phenogram derived from average taxonomic distance coefficients using 26 morphological characters
from 46 OTU's of five species ofMalacothamnus. OTU numbers and locations are listed in Table 1. The cophenetic
correlation is 0.840.
idsonii were classified as "var. nesioticus", "var. levels of variability within species, populations,
nuttalliV or "var. laxiflorus" and, even, single individuals. This point often was
commented upon to varying degrees by previous
Discussion workers (Kearney 1951b; Bates 1963, 1993), but
Character variation and life history traits. The there has been no quantification or documentatiWoen
diverse taxonomic interpretations within Malacoth- of character variation until the present study.
amnus reflect complex patterns of morphological found that most vegetative characters used previ-
variation. Our observations and analyses confirm ously to delimit and characterize species, particu-
that many characters used previously exhibitedhigh larly those describing the vestiture, lobing, apex,
3
MADRONO
148 [Vol. 46
O M. clementinus A M. densiflorus V M. fremontii
M. davidsonii q M. fasciculatus
Fig. 3. Three-dimensional model produced in principal components analysis using 26 morphological characters in
five species ofMalacothamnus. OTU numbers and locations are listed in Table 1.
and margins of leaves, were too variable on spec- whose "coarse" habit, "stout" branches, and
imens to code or to describe accurately the phe- "densely" pubescent leaves are readily observable
notype ofa collection orOTU. Our study is at least in field populations, but difficult to observe and
partially supportive of Bates' (1963) comment that quantify on herbarium specimens. Consequently,
some characters are vague or difficult to quantify, only one stem and two leafcharacters were includ-
but are apparent to those familiar with the genus. ed in the analyses out of 26 total characters. Ofthe
This was particularly noteworthy forM. davidsonii, three vegetative characters identified, none were in-
dicated as "useful" for discriminating among spe-
cies in our analyses.
Table 3. Eigenvalues produced in PCA for 26 mor- Reproductive characters, especially those de-
phological CHARACTERS AMONG FIVE SPECIES OF MALA-
COTHAMNUS. Character numbers refer to Table 2.
Character Component 1 Component 2 Component 3 Table 4. Percentage of taxa correctly classified in
discriminant functions analysis of 26 morphological
1 -0.185 -0.233 0.361 characters among five species of malacothamnus. re-
2 -0.070 0.235 0.059 sults are presented for comparisons among species only
3 0.019 0.248 -0.458 and among species and all "varieties" ofM.fasciculatus.
4 0.899 0.168 0.117 See text for discussion of results when "var. nesioticus"
5 0.490 -0.068 -0.632 was removed from the analysis.
6 0.256 0.129 -0.750
7 0.802 0.343 0.244 Percent correctly classified
8 0.678 -0.320 -0.223
9 0.426 -0.496 -0.020 Species only Including
10 -0.630 0.483 -0.422 "varieties"
-0.780 0.231 -0.370 ofM.
1121 -0.040 0.303 0.177 fasciculatus
13 0.012 0.416 0.088 Stand- Jack- Stand- Jack-
14 0.433 0.563 -0.239 Taxa ard knifed ard knifed
15 0.788 -0.308 -0.327
16 0.152 -0.215 -0.210 M. clementinus 67 67 100 67
1187 00..76811 --00..228278 --00..333224 MM.. ddeanvsiidfsloonriuis 17050 5600 800 800
19 -0.563 -0.395 -0.146 M.fremontii 86 71 29 29
20 0.480 -0.418 0.388 M.fasciculatus 56 44
21 -0.524 -0.365 -0.365 "var. catalinensis" 75 75
22 0.760 0.308 0.217 "var.fasciculatus" 45 36
23 0.092 0.305 -0.273 "var. laxiflorus" 0 0
24 -0.489 -0.436 -0.191 "var. nesioticus"'' 100 0
25 -0.500 -0.385 -0.060 "var. nuttallii" 40 20
26 0.125 -0.211 0.086 Total 67 52 43 35
1999] BENESH AND ELISENS: MALACOTHAMNUS 149
scribing the bractlets and calyx, were more consis- intergradation in Malacothamnus (Loveless and
tent within OTU's, were easier to code, made up Hamrick 1984; Briggs and Walters 1985). Addi-
the majority of characters subjected to numerical tionally, replenishment of alleles from the soil seed
analyses, and comprised all 11 characters that dis- bank may contribute to heterogeneity within and
criminated most effectively among species in the among populations (Huenneke 1991). Germination
PCA and/or discriminant functions analyses (Table experiments indicate that Malacothamnus seeds re-
2). Other reproductive characters describing the in- main viable for more than 50 years (Bates 1963;
florescence and staminal column exhibited little Benesh unpublished data) and germinate rapidly
utility for discriminating among species. Three following a fire (Bates 1963).
characters exhibited high levels of discrimination
among species in both PCA and discriminant func- Taxonomic implicationsandspeciesdelimitation.
tions analyses, and represent characters of maxi- Among 46 OTU's included in clustering and ordi-
mum taxonomic utility: bractlet length, bractlet nation analyses, only those representing M. clemen-
length/calyx length, and the presence or absence of tinus formed a distinct cluster and grouped with
rqueacluirtvaitnigveseapnadl tlhorbeeemqauragnitnitsatiinvebucdh.arAancottehrserwfeirvee OspTeUc'iessowfeMr.efirnetmeornmtiiix.edSianmtphleeUdPspGeMciAmepnhseonfoogtrhaemr
significant in either PCA or discriminant functions and PCA model. Malacothamnusfasciculatus was
analyses: bractlets concolorous with calyx, bractlets the most widely sampled species and encompassed
concolorous throughout, outer calyx pubescence the most morphological variation. Similar to the in-
texture, divergence ofsepal lobes in bud, sepal lobe terspecific pattern, OTU's representing segregated
constriction at base, sepal lobe length, and length "varieties" of M. fasciculatus were completely in-
of the arms and stalks of the sepal pubescence. termixed and clustered with those of three other
Variation patterns observed among species re- species: M. fremontii, M. davidsonii, and M. den-
flect historical factors such as biogeography and sifiorus. The 16 specimens in cluster lib of the
phylogeny, and genetic factors such as population phenogram (Fig. 2) formed the most internally con-
structure, mating system, reproductive isolation, sistent taxonomic grouping, with one M. davidsonii
and seed dispersal (Avise 1994). Populations ofM. in a group of M. fasciculatus OTU's.
fasciculatus and related species vary in size from The general lack ofresolution delimiting species
few to numerous individuals, consist often of of Malacothamnus using numerous characters and
clones because ofreproduction via rhizomes (Bates a phenetic approach indicate polymorphism among
1963; Swensen et al. 1995), and are characterized species was extensive. In our study, multivariate
by lack of fertility barriers (Bates 1963). Artificial techniques did not identify any well-defined mor-
hybridization experiments revealed widespread phological and geographic clusterwithin Malacoth-
cross-compatibilities and interfertilities among spe- amnus. The pattern of morphological and geo-
cies (Bates 1963). Natural hybridization between graphical variation was widespread among the five
species also has been postulated; Bates (1963) iden- species sampled, and character intergradation was
tified four regions of sympatry with interspecific not restricted to particular species pairs or to a few
hybrid swarms. Bates (1963) suggested that "ran- areas of sympatry. Conspecific OTU's were gen-
dom but recurrent hybridization and isolation of erally spatially separated on the phenogram (e.g.,
populations has probably been a major factor in the OTU's 16+32+19 and OTU's 17+24+30), except
evolution of morphological diversity in Malacoth- forclusterlie where three M. densiflorus specimens
amnus.'''' grouped with four OTU's of M. fasciculatus in an
Variable compatibility systems may be present area of sympatry and purported hybridization
among species of Malacothamnus; M. fasciculatus (Bates 1963). Although found at elevations less
and M. palmeri are self-incompatible, whereas self- than 650 meters throughout much of its range
compatibility has been observed in M. fremontii (Bates 1963), inland populations ofM.fasciculatus
and M. jonesii (Bates 1963). Additionally, Bates in Riverside and San Diego Counties and northern
(1963) observed that although the flowers of Ma- Baja California Norte can be found at higher ele-
lacothamnus are bisexual and pollen is released vations, resulting in sympatry with M. densiflorus
while stigmas are receptive, insects commonly visit (Fig. 4). Bates (1963) recognized this area as a zone
the flowers. He concluded that species of Mala- of introgression and suggested that delimitation of
cothamnus have a mixed mating system but arepre- M. densiflorus and M. fasciculatus is arbitrary in
dominantly outcrossing. However, in an examina- this region.
tion of genetic variation within M. fasciculatus, Because comparatively few characters exhibited
Swensen et al. (1995) reported a value of isozyme significant taxonomic utility (Table 2), it is useful
differentiation among ten populations (G = 0.584) to consider whether a phylogenetic approach to
st
which is most concordant with that ofa selfing spe- species delimitation (e.g., Nixon and Wheeler 1990,
cies (Loveless and Hamrick 1984; Hamrick and Davis and Nixon 1992), based on uniform and
Godt 1990). Because of the lack of interspecific unique characters or combinations of characters, is
reproductive barriers, gene flow among populations more useful than the phenetic methods used in this
and species may be partly responsible for character study. Population (OTU)-based datahave been used
MADRONO
150 [Vol. 46
Fig. 4. Collection locations for specimens of five species of Malacothamnus analyzed for morphological variation.
Cluster designations correspond to the clusters produced in the UPGMA phenogram (Fig. 2). Only clusters II and III
are shown.
successfully to identify lineage-defining characters encompass the largest geographical range and mor-
in several groups (e.g., Davis and Manos 1991), phological variation among the species investigated
and several species of Malacothamnus can be de- and in the genus (Bates 1963). Illustrating the dis-
limited using morphological data with this proce- cordant views of species boundaries and the extent
dure. For example, M. densiflorus can be defined ofphenotypic variation among these species, Bates
by a unique character combination of an unbran- (1993) submerged five and three specific epithets
ched, elongated inflorescence, dense flower clus- to synonymy within M. fasciculatus and M. fre-
ters, and large bractlet and calyx lengths. Mala- montii, respectively. Other than geographic locality
cothamnus clementinus is identified by the associ- (e.g., "var. catalinensis", "var. nesioticus"), vari-
ation of isodiametric leaves (similar to M. fascicu- eties of M. fasciculatus were equally difficult to
latus) and recurved sepal lobes in bud (similar to characterize and were not recognized in the Jepson
M. fremontii). The combination of "stout" branch- Manual (Hickman 1993). Because of complex pat-
es and "dense" leaf pubescence of M. davidsonii terns of intergradation among morphological char-
also are unique in the genus and are used to define acters, we were not able to identify any unique
this species by several workers (Kearney 1951b; character combinations for M. fasciculatus or M.
Bates 1963, 1993). However, in M. davidsonii as in fremontii.
M. fasciculatus, M. fremontii, and other species in Using molecular data, Swensen et al. (1995) ex-
Malacothamnus, lineage-defining characters are of- amined variation among varieties ofM.fasciculatus
ten difficult to detect in herbarium specimens, vary to assess the genetic distinctness ofthe two known
greatly among and within populations, and are dif- populations of the Santa Cruz Island endemic, M.
ficult to divide into discrete character states. The fasciculatus "var. nesioticus". Sampling within M.
absence of well-defined gaps in many characters fasciculatus was from six counties and was based
and intergradation resulting from gene flow and/or on ten populations in the isozyme studies and seven
recent divergence from ancestral species pose se- populations in the RAPD and rDNA analyses.
rious problems to species delimitation within Ma- Swensen et al. (1995) concluded that the pattern of
lacothamnus whether a phenetic or phylogenetic molecular variation supported Kearney's (1951b)
approach is employed. varietal delimitations within M. fasciculatus, and
Malacothamnus fasciculatus and M. fremontii that "var. nesioticus" was the most divergent ge-
.
1999] BENESH AND ELISENS: MALACOTHAMNUS 151
netically among sampled populations with unique Transverse Ranges (Valentine and Lipps 1967, Sor-
isozyme and RAPD alleles and rDNA restriction lien 1994).
sites. Our morphological analyses indicate a con- The rare species M. davidsonii is disjunct be-
clusion similar to Bates (1963, 1993) is warranted. tween the South Coast Ranges and Los Angeles
There is extensive morphological intergradation County. The four OTU's representing these disjunct
among "varieties" of M. fasciculatus and we are regions did not group together, but exhibited the
unable to characterize any variety unambiguously, closest morphological similarity to M. fasciculatus
including "var. nesioticus." and, to a lesser extent, M. fremontii. Our data sup-
port Bates' (1963, 1993) observation of the inter-
Biogeographic implications. Populations of two gradation between M. davidsonii with M. fascicu-
species examined in this study, M.fasciculatus and latus. As in this and other examples of character
M. clementinus, are considered California Island intergradation between species, this study cannot
endemics or island-mainland disjuncts. Whereas M. differentiate between hypotheses of interspecific
clementinus is restricted to San Clemente Island, hybridization or primary (and incomplete) diver-
previously segregated "varieties" of M. fascicula- gence from ancestral species. Future studies em-
tus included a Santa Cruz Island endemic ("var. ploying molecular markers may represent the best
nesioticus") and a Santa Catalina Island and main- approach to investigate specific instances of gene
land disjunct ("var. catalinensis""1). Our results sup- flow, species delimitations, and interspecific rela-
ported the status of the endangered species M. cle- tionships.
mentinus as a distinct morphotype in Malacoth-
amnus that may be found occurring naturally only Acknowledgments
on San Clemente Island. Malacothamnus clemen-
tinus forms the most morphologically coherent We thank K. Anderson, A. Fone, andJ. Takara forfield
assistance, S. Junak forpermission tocollectat SantaBar-
group for any species examined, and its three bara Botanic Garden, and the California Native Plant So-
OTU's are most similar morphologically to four ciety and University ofOklahoma Graduate Student Sen-
OTU's ofM.fremontii collected at elevations rang- ate for monetary assistance with field work. The Okla-
ing from 1160 to 1500 meters in Los Angeles, Ma- homa Biological Survey is acknowledged forgranting use
dera, San Bernardino, and Ventura Counties. Our of computing facilities. Dan Hough and Ian Butler pro-
analyses included an OTU (P. H. Raven #16625, vided assistance in computer use and computations. Scott
annotated by Bates) from San Bernardino County, cRuosmsmeleln,tGsoorndoann eUanrol,ierandrdafJtohofntLhiasDpuapkeer.oAffesrpeedcivaalltuhaabnlke
but it was unclear if it was a human-facilitated in- you to David Bates for his clarifications, comments, and
troduction. Bates (1963, 1993) previously noted the support.
similarities between M. clementinus and M. fre-
montii in calyx characters, and suggested that M.
clementinus represented a "generalized" form of Literature Cited
Malacothamnus that persisted in refugial habitats in Avise, J. C. 1994. Molecular markers, natural history and
San Clemente Island (Raven and Axelrod 1978). evolutioMn. Chapman and Hall, New York.
Mainland and island OTU's representing M.fas- Bates, D. 1963. The Genus Malacothamnus. Ph.D.
ciculatus "var. catalinensis""1 did not group together dissertation. University ofCalifornia, Los Angeles.
in our morphological analyses, indicating polymor- . 1967. Chromosome numbers in the Malvales.
phism among Santa Catalina Island and mainland Gentes Herbarium 10:39-46.
populations formerly placed in "var. catalinensis" . 1968. Generic relationships in the Malvaceae,
We present no morphological evidence to charac- trib.e1M9a9l3v.eaPpe.. 7G5en1t-e7s54HeirnbJa.riCu.mHi10c(k2)m:a1n17(-ed1.)3,5.The
terize "var. catalinensis" or to support its status as Jepson manual: higher plants of California. Univer-
a California Island endemic or island-mainland dis- sity ofCalifornia Press, Berkeley, CA.
junct. For M.fasciculatus "var. nesioticus", its co- Briggs, D. and S. M. Walters. 1985. Plant variation and
herence as a distinctive morphotype within M.fas- evolution, 2nd ed. Cambridge University Press,Cam-
ciculatus and as a Santa Cruz Island endemic could bridge.
not be tested because only one OTU was examined. Davis, J. I. and P. S. Manos. 1991. Isozyme variation and
Swensen et al. (1993) found that the two known cspoemcpileesxde(lPiomaicteaatei)o:n ainn atphpelicPautcicoinnelolfiathenutpthayllloigaen-a
populations of "var. nesioticus" were more dissim- netic species concept. Systematic Botany 16(3):431-
ilar genetically to each other than either population 445.
was to other "varieties". Our PCA indicated that and K. C. Nixon. 1992. Populations, genetic vari-
the Santa Cruz Island OTU was most similar to a ation, and the delimitation of phylogenetic species.
collection of M. fasciculatus "var. nuttallii" from Systematic Biology 41(4):421-435.
the Santa Monica Mountains in Los Angeles coun- Eastwood, A. 1936. The shrubby Malvastrums of Cali-
fornia with descriptions of new species and a key to
ty. This morphological grouping is geographically the known species. Leaflets ofWestern Botany 1(18):
proximate and also is consistent with the hypothe- 213-220.
sized physiographic and geological correspondence . 1939. New California plants. Leaflets ofWestern
of the northern California Islands with the western Botany II(10):186-188.