Reconsideration of the Taxonomic Status of Mason's Lilaeopsis – A State-Protected Rare Species in California

2011·10 MB·English

by  Peggy L Fiedler

APK 12?ni?
Madrono, Vol. 58, No. 3, pp. 131-144, 2011
RECONSIDERATION OF THE TAXONOMIC STATUS OF MASON'S
LILAEOPSIS - A STATE-PROTECTED RARE SPECIES IN CALIFORNIA
Peggy L. Fiedler
Natural Reserve System, UC Office of the President, Office of Research and
CA
Graduate Studies, 1111 Franklin Street, 6^*^ Floor, Oakland, 94607
[email protected]
EsA K. Crumb
Wetlands and Water Resources, 818 Fifth Ave, STE 208, San Rafael, CA 94901
A. Kate Knox
WA
7730 19^^^ Avenue NE, Seattle, 98115
Abstract
Lilaeopsis masonii is a California state-listed rare species with a wide range of morphologies
observed in the field throughout its range, and in herbaria collections. This extensive variation
confounds reliable taxonomic identification, particularly forthose specimens intermediate between L.
masonii and its sister taxon, L. occidentalis. To investigate the genetic basis of this morphological
variation, weexamined two portions ofthe Lilaeopsisgenome in seven species. Specifically we sought
todeterminewhetherL. masoniiissufficientlydistinctfromitsclosely related, widespreadcongenerto
continuetowarrant specificstatus. DNAsequenceanalysisofITS1, 5.8S, and ITS2nuclearribosomal
DNA revealed no differences between L. occidentalis and L. masonii California collections, and
minimaldifferencesbetweenthese samples and L. occidentaliscollected fromthe stateofWashington,
suggesting strongly that these two species form a single clade. A combination offragment data from
three AFLP primers yielded 274 fragments from 29 samples. Genetic Manhattan distance values
mcailncourlatdeifdfefrreonmcetsheamAoFnLgPalmlatsramipxlweist.hiUnPsGpeMciAescrlaunsgteerdpfhreonmoagrlaomwsosfu1p.p4ortotathheigrhesuolfts6.6o,frtehfelePctCinAg
analysis, illustrating a cluster of L. occidentalis + masonii samples distinct from other Lilaeopsis
species. Because conservation dollars should protect unique evolutionary entities, we suggest that L.
masonii be subsumed under L. occidentalis and therefore no longer receive formal state protection.
Key Words: AFLP, Apiaceae, California endangered species Act, goldilocks conundrum, ITS,
UPGMA.
lilaeopsis masonii, lilaeopsis occidentalis,
Lilaeopsis masonii Mathias & Constance (Ma- can genus Neogoezia as suggested by Petersen
son's lilaeopsis) is one of 15 wetland or aquatic et al. (2002). The New World endemics clade of
species ofthewidespread genus Lilaeopsis Greene tribe Oenantheae is native to North America and
within the Apiaceae. The genus Lilaeopsis is comprises a monophyletic group that appears to
comprised of perennial herbs characterized by a be evolving much faster than any other major
horizontal stem with leaves commonly in clusters clade recognized in the tribe (Hardway et al.
("ramets") bornedirectlyfromthestem, although 2004).
rarely leaves occur individually. Lilae—opsis is Early taxonomic work on the genus in
notable in its morphologic simplicity entire, California by Hill (1927) and Mason (1957)
generally linear leaves; simple umbels; absence included mention of comparatively smaller and
of a carpophore; and, a strongly reduced habit narrower leaves in Lilaeopsis specimens occurring
(Petersen et al. 2002; Downie et al. 2000; Downie away from the coast, in contrast to a relatively
et al. 2008). Such simple morphology has led to more robust coastal form. Professor Herbert
a long history of taxonomic uncertainties and Mason, an early expert on the wetland flora of
difficulty in the reconstruction ofits phylogeny. California, first collected a relatively smaller
Evidence for monophyly ofLilaeopsis is strong Lilaeopsis from Brannan Island of the San
(Petersen et al. 2002). However, recent research Francisco Bay/Sacramento-San Joaquin Delta
based on molecular evidence from nuclear and (Bay-Delta). He referred to the smaller form as
chloroplast genes suggests that the genus is best the "San Francisco Bay and river-mouth" form
placed in the Oenantheae tribe within the (Mason 1957: 631). This specimen, according to
Apioideae (Downie et al. 2008) and that Lilaeop- Mason (unpublished) was "definitely distinct
sis is sister to the clade comprising Ptilimnium, from the coastal L. occidentalis.''' Western
Limnosciadium, Daucosma, Cynosciadium and lilaeopsis {Lilaeopsis occidentalis J. M. Coulter
rachis-leaved species of Oxypolis, not the Mexi- & Rose) is a widespread, common species.
MADRONO
132 [Vol. 58
j
ranging from the Queen Charlotte Islands of masonii, but noted that it was a geographical
i
British Columbia, Canada to Marin County, outlier for the rare, Bay-Delta endemic species.
California (Affolter 1985). Considered to be a Today, CNPS (Tibor 2001) acknowledges that \
coastal species confined to salt water or brackish this specimen is likely to be L. occidentalis, not |
water intertidal habitats, collections of L. occi- the rare L. masonii, but provides no explanation.
dentalis from inland fresh water lentic and lotic Several attempts to relocate this Lilaeopsis
habitats are known, but considered "uncharac- material at Chicken Ranch Beach by the authors
teristic'^ (Affolter 1985). have failed as the population appears to be
Lilaeopsis masonii was not described as a extirpated, thereby making an independent spe-
distinct taxon for two decades after the smaller cies corroboration impossible.
form in the Bay-Delta was first observed. In It is important to note that (1) numerous
1977, Mathias and Constance formally recog- collections ofL. occidentalis from the beaches of
nized the diminutive nature of a specimen Marin and Sonoma counties exist, (2) leaflengths
obtained from Twitchell Island in the Bay-Delta range by an order of magnitude or more within
as L. masonii (Mathias and Constance 1977). and between adjacent populations of L. occiden-
Mathias and Constance described L. masonii as talis, (3) the number and clarity of internal
distinct from L. occidentalis based upon the crosswalls considered important diagnostic char-
former (rare) taxon bearing narrower, typically acters are more likely a function ofrelative plant
shorter, and more or less terete leaves, and an size, exposure, or both, and (4) inland collections
inland distribution. They honored Herbert Ma- ofthe common species are known from the state
son's expertise in the wetland flora of the State of Washington (e.g., UC 1594452; 4 September
with the specific epithet. 1962). Also noteworthy, L. masonii has never
Mason's lilaeopsis was one ofthe first vascular again been collected on the Pacificcoast ofNorth
plant species to be protected as "rare" under the America beyond Schreiber's Marin Co. collection
California Endangered Species Act (CESA) in 1936.
(California Fish & Game Code §§2050, et seq.). Further, Affolter (1985) remarked that leaves
At the time ofits listing in November 1979, only from a collection of L. masonii (derived from
seven population occurrences were known Sherman Island immediately down river of
(CNDDB 2009). Since formal protecfion, the Twitchell and Brannan islands) cultivated for
documented extent of geographic distribution his greenhouse comparisons were "remarkably
and population abundance of L. masonii has longer than any of the herbarium material"
increased nearly three-fold, primarily as a result (Affolter 1985:70). He suggested the observed
ofconcentrated field survey efforts conducted in overall larger and more robust greenhouse
the early 1990's by Golden, Fiedler, and Zebell material was evidence of how difficult it is to
(Golden and Fiedler 1991; Golden 1992; Fiedler understand vegetative plasticity from herbarium
and Zebell 1993; Zebell and Fiedler 1996). Today, material alone. However, the relatively robust
Mason's lilaeopsis is known to occur within 24 response of Mason's lilaeopsis to the mild
USGS quadrangles and seven counties (CNPS conditions ofa greenhouse suggests strongly that
2008), spanning across roughly 690 square miles. strict morphological distinctions between the two
One hundred eighty-six documented occurrences taxa are problematic.
are on record with the state (CNDDB 2009), Additional morphological characters further
although most, but not all are extant. support the assertion that L. masonii is not
distinctly different from L. occidentalis. Affolter
A History ofTaxonomic Uncertainty (1985:70) noted that the "two taxa are similar in
several respects," including similar (1) leafshapes
Confusion over the taxonomic limits of this (linear), (2) rhizome branching architecture, (3)
rare species existed from the beginning of its fruit shapes, (4) fruit cell types, (5) fruit venation
description. Two examples are relevant. First, a patterns, (6) habitats, and they have (7) overlap-
long-controversial Lilaeopsis specimen collected ping geographic distributions. Despite all these
by Schreiber (#2266 UC, 28 June 1936) from similarities, Affolter (1985:71) supported their
Chicken Ranch Beach in Marin County derives separate specific status, primarily because "when
from outside the circumscribed geographic range grown under a common-garden environment in
of the endemic inland taxon. Leaf lengths from the greenhouse, the two species retained the
this specimen range between 15 to 42 mm, a vegetative characteristics that distinguish them
morphological range characteristic of compara- in the field."
tively larger leaflengths for L. masonii. However, Subsequent laboratory studies conducted by
it is possible to key these larger leaved Chicken the principal author and her students (Golden
Ranch Beach specimens to L. occidentalis in and Fiedler 1991; Golden 1992; Fiedler and
every relevant flora (e.g., Hickman 1993). Zebell 1993; Zebell and Fiedler 1996) have
Affolter (1985) examined this specimen in his provided little clarity. Most importantly, no
monograph of Lilaeopsis, and accepted it as L. nucleotide variation was found among nine
2011] FIEDLER ET AL.: RECONSIDERATION OF MASON'S LILAEOPSIS 133
populations ofL. masonii or between L. occiden- be subsumed within L. occidentalis, and contin-
talis and L. masonii when the 204 nucleotides of ued protection under the California Endanger-
the ITS2 nuclear genome were ascertained ed Species Act for L. masonii should be
(Fiedler and Zebell 1993). Fiedler and her reconsidered.
colleagues thus concluded tentatively that the Fallon (2007) noted that genetic information is
rare species wasmost likely an inland ecotype not being used increasingly to resolve taxonomic
clearly distinct from its widespread congener. issues for protection at the federal level under
the U.S. Endangered Species Act of 1973 (ESA).
Field Observations She conducted a rev&iew oflisting decisions made
by the U.S. Fish Wildlife Service and the
Decades of field observations of Lilaeopsis National Marine Fisheries ofspecies, subspecies,
throughout the Bay-Delta, Suisun Marsh, and or distinct population segments (DPSs) proposed
Napa River ecosystems do not reinforce many of for protection under the ESA. Fallon determined
the conclusions offered by Hill (1927), Mason that the listing fate of a DPS based upon data
(1957, unpubhshed), Mathias and Constance from more than one genetic marker resulted in a
(1977), and Affolter (1985) supporting the higher probability of protection than candidate
recognition oftwo distinct taxa. Rather, the few taxon or population segment whose discreteness
vegetative characteristics that typify this genus was determined by a single genetic marker. With
are highly variable both within and between the cautionary tale of Fallon's findings in mind,
populations throughout this region. Occurrences we examined the ITS region of the nuclear
of L. masonii in the lower Napa River, studied genome and, to corroborate our ITS findings,
since 2001 (WSP 2007, unpubhshed; Blasland, conducted an amplified fragment length poly-
Bouck, & L, Inc. unpublished; Entrix unpub- morphism (AFLP) analysis on a similar suite of
lished; L.C. Lee & Associates unpublished; Still- taxa.
water Sciences and Fiedler unpublished) include We chose the ITS region in large part because
a full spectrum of individual ramet sizes. Often, Hardway et al. (2004) found evidence for
both large and small forms ofLilaeopsis species, particularly rapid evolution in the Oenantlieclade
easily identifiable to the two different species, can that includes Lilaeopsis when compared to the
be found growing in the same location. Often the rest of the taxa. Sequence divergences in this
plant stature/leaf length size gradient runs per- clade averaged 6-7 times higher (approx. 17%)
pendicular to the shoreline, where the small than between species in Oenanthe (approx. 2.8%)
''masonic form (approx.1.5-^.5 cm in height) or Cicuta (approx. 2.4%) (Hardway et al. 2004).
grows relatively close to the water's edge, while AFLP analysis was selected as a secondary
increasing larger and more robust ''occidentalis'' marker system basedDoNnAthe increasing popularity
(approx. >11 cm in height) can be found further of this form of fingerprinting as a
from the water. "Intermediate" or medium-sized complementary system in phylogenetic studies
Lilaeopsis material (approx. >6.25 and <11 cm (Holland et al. 2008). Additionally, AFLP
in height) is common throughout this shoreline/ fingerprinting offers a reliable, robust, and
river bank habitat and geographic range, and genomically comprehensive method of genetic
keys to either (or both) the rare o—r the common analysis for taxa lacking complex nuclear and
species. We call this phenomenon i.e., range in organellar markers (Vos et al. 1995).
size of a critical morphological c—haracter, with
significant overlap between taxa the "Goldi- Materials and Methods
locks Conundrum" to highUght the problem that
theintermediate-sizedmaterial is not "just right," Field Collection
but rather, highly problematic.
To resolve our conundrum and determine Lilaeopsis masonii specimens were collected in
whether L. masonii is a discrete species distinct the spring of2007 from locations along the Napa
from L. occidentalism we initiated a genetic River and in the Sacramento/San Joaquin Delta.
analysis of seven species of this genus. We Lilaeopsis occidentalis was collected from Bodega
hypothesized that there were no significant Head, Cahfornia, and Masonand Lawrencelakes
differences between diagnostic portions of the in Washington State (Fig. 1). Leafmaterial to be
genome selected for this study ofthe two species, usedinDNAextractionwaspreserved in silicagel
L. masonii and L. occidentalis. Based on these at the time ofcollection. Voucherswere deposited
analyses, we then explored whether L. masonii at the herbarium at San Francisco State Univer-
warrants continued recognition as a distinct sity (SFSU) (Table 1). Material for L. hrasiliensis
species or rather, should be subsumed under the (Glaz.)AffolterandL. mauritianaG. Petersen&J.
widespread and common L. occidentalis. If no Affolterwasobtainedfromacommercialaquarium
significant differences were shown to exist be- supplier (freshwateraquariumplants.com). The
tween diagnostic portions of the L. occidentalis dataset is composed of 35 nrDNA ITS sequences
and L. masonii genomes, then L. masonii should representing seven taxa, including three sequences
MADRONO
134 [Vol. 58
T:^|Washington Samples
S3W-A06S
(2E-B07
I2E-A07
12W-B02
A^2E-E07
Maxwell Brg
I
I1B-S07
I1A-H07
riA-B04
California^
Novada
Bodega Head ^
Ij-y I
^ptIIHillSlough/GrizzlyIsland IN
I
rH Collinsvllle
3 Mile SloughK--^(y^^LibertyIsland
y^|Twitchell Island
M
Brannan Is. Boat Ramp 0 37.5 75 150 Mites
1 1 I I I
Fig. 1. Map of the geographic locations of Lilaeopsis masonii and L. occidentalis specimens collected for
this study.
from Genbank (L. carolinensis J. M. Coulter & study for Lilaeopsis is in preparation (S. Downie,
Rose, L. novae-zelandiae (Gand.) A. W. Hill, and Univ. of Illinois, Urbana-Champaign, personal
L. occidentalis), twospecimensfromtheaquarium communication).
trade labeled as L. brasiliensis and L. mauritiana,
and two specimens ofL. schaffneriana(Schltdl.)J. ITS Methods
M. Coulter & Rose subsp. recurva (A. W. Hill)
DNA
Affolter courtesy ofthe Desert Botanical Garden from leaf tissue of five of the seven
staff. Comprehensivesamplingwasconducted for species was extracted using the Qiagen DNeasy
L. masonii and L. occidentalis as the purpose of Plant Mini Kit (QIAGEN, Inc., Valencia, CA),
this study was to resolve the taxonomic classifi- following the manufacturer's protocol with slight
cation forthese two species. A detailed systematic modifications. Lilaeopsis carolinensis and L.
—
'
2011] FIEDLER ET AL.: RECONSIDERATION OF MASON'S LILAEOPSIS 135
3^
03
B > OX)
D D 0D0 00 0p0 D00 00 D00 000 :0d0 :0:0) 00 000 D 00 00 000 D00 00 ^ 0D0 ^^Oioo;o:>o::o)oD;:^Qco/D5
tin tin tin tin Ph Ph Ph Uh Uh Uh UhUhU.^ Ph Uh
00 00 00 00 00 00 00 00 00 00 00 oo 00 00 00 00 00 00 oo oo oo hJ
or- o o o o O O O o o ro- o O o o
o o o o ro- ro- o o o o o o O O o o o o o o o o
o o o o o(N o(N (oN JiNn ifNn ^ ^ o (rNn fN fN fN fN fN '2007 <z '2007 'i20n08 'i20n08 <z '1991 1992 <zz<
00 00 00 (N fN fN fN fN fN fN o fN fN
in in in in in in In ON 00 00 00 00 00 00 00
o
fOfm^NNN OfmNN vrOiOon (moN o(m^No 0f(rm-0NN) fO(mr-NN~ m OfO(ONNNOO 0fmfoiNN0on m(r1-N^- Omm(rr--NN-- mO(NN 0rm(f0NN- O0Oim(nN0Nn (0oN0 (o(0iNoN0n f(OONNN 0fm\0ND OrOfOr<NNn-i <mOmN 0—mm01 \—^frD^ or^0ro0> <^^: >< c^ ON <z <z
in < <
ON ON 00 mo ON ON o OmN fN 00 m imn oo mm •n in ON ON OmN NO ON
fioNn \\omficNno> f'iosNnO mfioNn f\ioNDn mf\oiNDn f\oiiNDnn 0f\oiN0nD o^ioinno oof\oiNDoon fNioNOn \(ioNon NoinO NfNioNOOn N(iioNOnn (iNn OioNn mONONO ONoNO NoiinOn mNoiOn OOoiiiOOnnn OooNoNO f0^O^0Nl irmo«n-) immon •2<. I- O \D <ZZ<
< <
O0O00n0OC/3C/51/5 O0GOGO0OO0GOGOO0C/:iC/3O0C/:iGOiy5O0O0< H H < < < < <
z z z z
u<<<
<<<<<<< <<<<<<<<< ^cd 2 2^ 2 ^ ^ <
u u u u u u u uuuuuuuuu Pj c c G U U U
2 £ £ S o" d d
zccdxzazCXzQhzcCldizccQd3hzcQdh ZZZZZZZZZUooooooc/5ooooZ <zz<
f^i
o NaO
oo
X X
ON
t
^ ^ o
PQ NO X NO Oa*s ^ ^ NmrjO NQN1-v^OO'f ONh^OH o o
_0 O<fNN<OfNN<OOQ<f1of^l<oK10^3<^o<03 00 < roO-Q- oP0J3 o-rW-D- oO0Q3 -oOOQ oO ^o po J3j:^ 03 ?)^^C NXO NO ^ a£ £a
W T3
o OQ PJ PJ pj — 03 03
f^4 (N f^J r^j fN c iS O o
"Sa 'c>S '>c5>'f5>5>"cS>'c>5 S>o^'>^>"(5 Cdi^ aa:dPC p:!^ cc > ^ c ^ i^s =i -<rU ^NO HH ^03 0>3 kN.O Uc03 Uc03 'cdd cdd
00 z0ac3d zcc&dd zcacdd zac0d3zccaddzca0d3 zccdd P^ zcacddz00a33 Zca3Zoa3Zoa3 zoac3dzaoc3dzacd Zad"SO^ H HJ K u •• u 2^ d
ro OQ p: OQ
•2 ^
00
MADRONO
136 [Vol. 58
Table 2. AFLP Primer and Adapter Sequences.
Primer Sequence
AdlEcoRI 5' -CTCGTAGACTGCGTACC- 3'
AdlMsel 5' -GACGATGAGTCCTGAG- 3'
prampEcoRI 5' -GACTGCGTACCAATTCA- 3'
prampMsel 5' -GATGAGTCCTGAGTAAC- 3'
FAM-EcoRI 5' -GACTGCGTACCAATTCAAC- 3'
HEX-EcoRI 5' -GACTGCGTACCAATTCACG- 3'
Msel + CAA 5' -GATGAGTCCTGAGTAACAA- 3'
Msel + CAT 5' -GATGAGTCCTGAGTAACAT- 3'
Msel + CAG 5' -GATGAGTCCTGAGTAACAG- 3'
novae-zealandiae were excluded from this analysis pre-selective reaction condition consisted of 30
DNA
due to technical difficulties with the cycles of94° for 30 sec, 56° for 1 min, and 72° for
extraction from the leaf material. Dilutions of 1 min.
DNA
the genomic extract of 1:10 in ultrapure Three combinations of selective primer sets
water were used in PCR reactions. The contigu- were used to produce a final AFLP fingerprint
ous ITSl, 5.8S, and ITS2 regions of nuclear for each sample (Table 2). Each set of selective
DNA
ribosomal were PCR-amplified using the primersconsisted ofa primerregionmatchingthe
primers ITSLEU (Baum et al. 1998) and ITS4 knownadaptersequence, aswell as three selective
(White et al. 1990) in final reaction volumes of nucleotides on the 3' end ofthe Msel primer and
t2h5ejiMl. OPosiBtIivOe UamlptlriafCilceaatnionPsCwRerCelpeuarni-fuiepdDusNinAg othnretehese3l'ecetnivdetnhuecloefoEticdoeKslplpursimaerf.loTreemspcelnattelafboerl
Purification Kit (MO BIO Laboratories, Inc., the pre-selective PCR was diluted 6-fold and then
Solana Beach, CA). Internal primers ITS2 and combinedwith a mastermixcontainingone set of
ITS3 (White et al. 1990) were used in addition to selective PCR primers. A step-down PCR was
ITSLEU and ITS4 in cycle-sequencing reactions used to amplify the selective fragments in a
in order to extend fragments and clarify ambigu- program consisting of 13 cycles of94° for 30 sec,
ities. Fragments were sequenced with the BigDye 65° for 30sec(—0.7° percycle), and 72° for 1 min,
3.1 kit (Applied Biosystems, Foster City, CA) followed by 24 cycles of 94° for 30 sec, 56° for
following the manufacturer's protocols, and 30 sec, and 72° for 10 sec.
visualized using the ABI PRISM 3100 Genetic The final selective PCR product fragments
Analyzer (Applied Biosystems). Sequences were containing a fluorescently labeled EcoKl end and
manually aligned using Sequencher 3.1.1 (Gene- unlabeled Msel end were analyzed undiluted
Codes Corp., Ann Arbor, MI) and MacClade using an ABI 3100 genetic analyzer (Applied
4.04 (Maddison and Maddison 2001). Biosystems). Initial fragments were sized first
using the analysis software GeneScan (Applied
AFLP Methods Biosystems) and using the program by GeneMar-
ker© (SoftGenetics, State College, PA). After a
AFLP fingerprinting was conducted follow- comparisonoffragmentcallingusingbothprograms,
ing a modified protocol based on the methods all samples were analyzed using GeneMarker®.
described by Vos et al. (1995). DNA extracts Fragments were recorded for each sample in a
prepared for ITS analysis also were used for data matrix based on a binary system (1 for
this study undiluted. Approximate DNA con- presence, 0 for absence); a data matrix was
centrations for all samples were estimated to developed for each primer combination and then
contain a range ofconcentrations from 10 ng/|j,l all data was collated into a single data matrix. To
to 50 ng/|il using an ethidium dot test. DNA test reproducibility of results, twenty percent of
template of each sample was digested using the all samples selected at random for each primer
infrequent endonuclease cutter EcoKX and the pair were re-analyzed, starting with the initial
DNA
frequent endonuclease cutter Msel. Immediately extracts. Fragment peaks that were deter-
following digestion, the entire digestion reaction mined to be consistently low (below 300 peak
was combined with an equal volume of ligation intensity) or unpredictable were dropped from
mix. the matrix table.
DNA
The resulting fragmented template con-
taining "sticky ends" was diluted five-fold and Data Analysis
subsequently amplified by PCR using a pre-
selective primer mix. This step effectively reduces Phylogenetic analyses of ITS sequences were
the number of possible fragments by approxi- conducted using Phylip version 3.68 (Felsentein
mately 1/16*'^ (Meudt and Clarke 2007). The 2004). All characters were weighted equally.
2011] FIEDLER ET AL.: RECONSIDERATION OF MASON S LILAEOPSIS 137
character state transformations were treated as data were then analyzed using a principal
unordered, and gapswere treated asmissingdata. coordinates analysis (PCA) using the Microsoft
Most-parsimonious trees were obtained in Phylip Excel® add-in program GenAlEx 6.2 (Peakall
using the "branch-and-bound" method of exact and Smouse 2006). To determine whether a
search implemented by the analysis unit DNA- measurable degree of genetic dissimilarity
PENNY. Bootstrap re-sampling (1000 replicates) among L. occidentalis (WA and CA samples)
was used to assess nodal support (Felsenstein and L. masonii (CA samples) could be attributed
1985). Most parsimonious trees were generated to geographic distance, an additional test of
from a search of 100,000 trees and a final tree was molecular variance based on geographic origin as
derived using a strict consensus tree method measured by Global Position System (GPS) also
(Felsenstein 2004). Additional tree searches were was tested. A two-way analysis of variance was
conducted in greater volumes, up to 1,000,000. assessed for collections of L. occidentalis and L.
However, larger searches produced the same final masonii using an analysis of molecular variance
tree, thus a smaller tree search was selected to (AMOVA) with GenAlEx 6.2 (Peakall and
reduce run-time during bootstrapping. Several Smouse 2006). Significance was assessed using
combinations of Lilaeopsis species outgroups 99 permutations.
were explored before selecting L. novae-zelandiae
as the outgroup. This selection was based on Results
indications ofa potential sistergroup relationship
between L. novae-zelandiae and L. occidentalism DNA sequence analysis of the ITS regions
which was supported by ITS phylogenetic anal- ITSl, 5.8S, and ITS2 ofnuclear ribosomal DNA,
ysis of this genus within the Apiaceae tribe based on a most parsimonious search of 64,631
described in Downie et al. (2008). Genetic trees using a branch-and-bound method, revealed
distances to determine branch lengths were no differences between California samples of L.
calculated in Phylip using the Jukes-Cantor occidentalis and L. masonii samples, including a
method implemented in DNADIST and a GenBank accession forL. occidentalis (100 of 100
Fitch-Margoliash (FITCH) search. trees) (Fig. 2). Within the L. occidentalisXL.
AFLP phylogenetic analysiswas performed us- masonii clade, samples collected from Washing-
ing the program Phylip version 3.68 (Felsenstein ton clustered separately. Distance values between
2004). A genetic distance matrix was created Washington and California samples were low for
using the techniques described by Nei and Lei single collections from Lawrence Lake and
(1979) as implemented by RestDist in Phylip Mason's Lake (0.1 and 0.3%, respectively).
(Felsenstein 2004). The output matrix was then However, a second sample from Lawrence Lake
NEIGHBOR UPGMA
input into using the exhibited higher distance values (1.5%), which
method of cluster analysis (Felsenstein 2004). may be due to missing data. Distance values for
Using this approach, an output tree was con- California samples ofL. masonii and L. occiden-
structed by successive clustering using an aver- talis were 0% across all samples. Comparatively,
age-linkage method ofclustering. The output file distance values between the additional species
was then plotted as both a rooted and unrooted used for this study ranged from 1.2-8%. Distance
tree. A search for the most parsimonious trees based analysis of ITS sequences found an
was implemented first using the branch-and- identical tree structure as the strict consensus
bound algorithm of DOLPENNY (100,000 trees tree inferred from most parsimonious results
searched) in Phylip following bootstrap analysis implemented by DNAPENNY. Bootstrap esti-
using 100 replicates. A 50% majority-rule con- mates from 1000 replicate analyses yielded 100%
sensus tree was then generated to condense the nodal support for all branches. Branch placement
results into a final tree, which is presented here. and relationship ofLilaeopsis species used in this
Previous studies of Lilaeopsis using AFLP studyareconsistent with results ofa previous ITS
analysis have not previously been reported. As phylogenetic analyses by Downie et al. (2008),
such, outgroup selection for AFLP parsimony though that study excluded L. masonii.
analysis was determined following variable, The three AFLP primer combinations gener-
preliminary analysis replicates. Lilaeopsis schaff- ated 274 unique fragments among 29 samples,
neriana was selected as this species is the closest with only 21 fragments shared or monomorphic
geographically to both L. masonii and L. between the five species used in this study.
occidentalis. Further, L. schaffneriana also dem- Although a small sample size was used, specifi-
onstrated sufficient genetic differences to be used cally for L. brasiliensis, L. schajfneriana, and L.
as an appropriate outgroup. mauritiana, the large number ofshared fragments
To further visualize potential multi-dimen- is potentially indicative of low genetic diversity
sional correlation of AFLP data based on within this genus, which is consistent with species
genetic similarities, an additional genetic dis- exhibiting high morphologic plasticity (Linhart
tance matrix was derived using Manhattan and Grant 1996). The number oftotal diagnostic
distance (StatisiXL; www.statistixl.com). These bands from the three markers combined data set
MADRONO
138 [Vol. 58
0.003 substitutions/base Loccidentalis(BodegaBay,CA)
1A-B04a,Napa
—L.occidentalis(MasonLake,WA)
L.occidentalis(LawrenceLake2,WA)
L.occidentalis(LawrenceLake,lWA
L.occidentalis(GenBank)
3MileSlough,CA
GrizzlyIsland,CA
BrannanIsland,CA
lA-H07b,Napa
lA-29a,Napa
Collinsville,CA
2W-B02a,Napa
2w-B02b,Napa
2E-A04,Napa
McAvoyHarboy,CA
Reach3,Napa
2E-B07a,Napa
2E-A07,Napa
1A-B04b,Napa
LibertyIsland,CA
1A-H07a,Napa
2W-B02a,Napa
1A-001,Napa
lA-29b,Napa
MaxwellBridge,Napa
TwitchellIsland,CA
1B-S07,Napa
L.brasiliensis2
c
brasiliensisi
L.mauntiana
-Lcarollnensis(GenBank)
1^L-schaffnerianal
L.schaffneriana2
I
L.novae-zelandiae(GenBank)
Fig. 2. Strict consensus tree derived from ITS sequence data using the branch and bound method implemented
by DOLPENNY. Branch values are Bootstraps. Upper left tree illustrates distance values, branch lengths are
proportional to the number ofnucleotide substitutions per base.
varied between species, ranging from a low of98 masonii (Napa River #2W-B02), to high of 11.9
bands for one sample of L. schaffneriana, to a for L. schaffneriana and L. occidentalis (Lake
WA WA
high of 135 bands for the Mason Lake, Lawrence, collection).
sample ofL. occidentalis, with a mean number of Principal coordinate analysis (PCA) illustrated
AFLP bands equaling 123 (SD = 9.5) (Table 3). an overlapping association between samples ofL.
Within species genetic distance values (Man- occidentalis and L. masonii (Fig. 3), but a clear
hattan distance calculated from the AFLP differentiation between the L. occidentalislL.
matrix) ranged from a low of 1.4 for L. masonii masonii cluster and all other Lilaeopsis species
(Napa River site #1A-H07, large and small examined in this study, i.e., L. schaffneriana, L.
forms), to a high of 6.6 for L. masonii (Napa mauritiana, and L. brasiliensis. The small separa-
site 1A-B04 and Twitchell Island collections). tion observed between the L. occidentalis and L.
Between species values ranged from a low of4.1 masonii data may be attributed to geographic
for L. occidentalis (Mason Lake, WA) and L. distance. Results of AMOVA analysis derived
1
2011] FIEDLER ET AL.: RECONSIDERATION OF MASON'S LILAEOPSIS 139
Table 3. AFLP Fragment Numbers eor Each Species.
Mean or Total No. ofFragments
Species CAA CAG CAT Total
L. masonii 53.5 46.1 21.9 125.1
L. occidentalis 52.7 50 22.9 128.8
L. schaffneriana subsp. recurva 47.5 34.5 16 99.5
L. hrasiliensis 52 39 21 1 14
L. niauritiana 43 38 16 99
All Fragments 123
from grouping samples based on geographic lution within this taxon is less certain of the
location indicated that approximately 73% of specific placement of Lilaeopsis samples, based
genetic variation was distributed between groups on geographic location (Fig. 5).
and thus 23% among groups, supporting the
conclusion that most observed geneticvariation is
due to geographic distance (P = 0.01). Discussion
UPGMA
The cluster phenograms provides
additional support for the results of the PCA Taxonomic implications. Within the last two
analysis, illustrating a combined grouping of L. decades, the use ofgenetic techniques to distinguish
occidentalis (CA and WA) and L. masonii discrete evolutionary units has become common
samples (Fig. 4). Within the L. occidentalis and place in systematic biology. Use of genetic data in
L. masonii clade, samples collected from Wash- theprotection ofendangered specieswhen morpho-
ington clustered separately from samples col- logical (or other character) information is either
lected within Cahfornia, corroborative of results unreliable or impossible isjust one reason why this
AMOVA
of indicating that variation within this approach to species identification and delimitation
clade is due in large part to geographic distance. is so important (Avise 2003). Thus, sole reliance on
Samples of L. hrasiliensis and L. niauritiana morphological, geographic, reproductive behavior
cluster separately but are sister to the L. or some combination of non-genetic characters to
occidentalislL. masonii clade; L. schaffnericma delimit taxa is no longerdefensiblewhen diagnostic
samples also cluster separately but are sister to genetic information is available and can be readily
all other specimens/species used for this study. assessed. In the case of Lilaeopsis masonii and L.
The most parsimonious tree from the maximum occidentalis, neither ITS sequence nor AFLP
parsimony analysis supports a single L. occiden- fragment length data support the recognition of
talislL. masonii clade; however, additional reso- the L. masonii as a distinct evolutionary entity.
L.schaffneriana
Lmauritiana
L.brasilensis
Coord.
Fig. 3. Principal coordinates analysis (PCA) ofAFLP fragment data matrix. Codes for Napa collections, e.g.,
1A-H07, indicate different collection locations and dates along the Napa River specific to the Napa River Flood
Protection Project.
MADRONO
140 [Vol. 58
±1. .LawrenceLake1
•LawrenceLake2 Li'aeopsis
occidentalis
12.8 •LawrenceLake3 (Washington)
9.5
MasonLake1
Mason Lake2
1A-29a
1A-B04a
1A-29b L.masonii
(Napa,CA)
p1A-H07a
n-1A-H07b
l^MaxwellBridge
39.8
J—2W-B02a
L-2W-B02b
McAvoyHarbor
3MileSlough
BrannanIsland
L.masonii
LibertyIsland (California)
—
Collinsville
—
GrizzlyIsland
—
^IA-001
I—
* Reach3— L(.Nampaas,onCiAi)
8.3
2E-B07
—
9.2 •BodegaHead L(.Caolcicfoirdneinat)alis
13.7 -1A-B04b- L.masonii
(Napa,CA)
14 TwitchellIsland L.masonii
(California)
49.5
Lilaeopsisbrasiliensis
Lilaeopsismauritiana
5.5
59 Lilaeopsisschaffnerianal
Lilaeopsisschaffneriana2
Fig. 4. UPGMA Cluster Phenogram (rooted and unrooted trees) from AFLP data matrix of five species of
Lilaeopsis. Numbers are distance values. Codes for Napa collections, e.g., 1A-H07, indicate different collection
locations and dates along the Napa River specific to the Napa River Flood Protection Project.
The morphological and geographic information use of best available science such as existing or
to support two distinct taxa is weak, ambiguous, generating new genetic information is equally
and unreHable at best. valid for the periodic reviews of listed species
Fallon's (2007) arguments regarding the im- required of both the federal and state agencies.
portance of using genetic information to resolve Further and relevant to L. masonii, use ofgenetic
taxonomic issues for species protection is borne data is likely to be essential during a de-Hsting
out in our study. While her review focused solely review process.
on vertebrates, and on only those species, Based upon several lines ofevidence, including
infraspecific or population segments proposed decades offieldwork throughout the range of L.
for listing, not those already listed, our results masonii, observations from the most recent
add further emphasis for use of molecular monograph (Affolter 1985), and our molecular
techniques in conservation efforts. We concur genetic analyses, we urge that this rare taxon no
that multiple genetic markers are essential for a longer be recognized as a separate taxonomic
thorough assessment oftaxonomic or population entity. Rather, L. masonii should be subsumed
unit (or at any appropriate level) when consider- within the larger, much more widespread, com-
ing offormal protection. We further suggest that mon, and equally variable species, L. occidentalis.