Content uploaded by Marten - Sørensen
Author content
All content in this area was uploaded by Marten - Sørensen
Content may be subject to copyright.
ORIGINAL ARTICLE
A morphometric study of the Abies religiosa–hickelii–
guatemalensis complex (Pinaceae) in Guatemala and Mexico
Uffe Strandby Æ Knud Ib Christensen Æ
Marten Sørensen
Received: 28 May 2008 / Accepted: 15 February 2009 / Published online: 8 April 2009
Ó Springer-Verlag 2009
Abstract This morphometric study of the geographic
variation in the Abies religiosa–hickelii–guatemalensis
complex is based on samples from 15 Guatemalan and 12
Mexican populations, two populations of A. religiosa s.str.
and A. hickelii s.str., and herbarium specimens of A.
hickelii, A. vejarii and varieties of A. guatemalensis. The
multivariate methods employed were principal components
analysis, and UPGMA clustering. The multivariate and
univariate analyses based on 231 operational taxonomic
units imply that although morphological differences exist
distinct morphospecies cannot be recognized within the
A. religiosa–hickelii–guatemalensis complex. A Mantel’s
test reports that taxonomic dissimilarities are significantly
related to geographic distance. We suggest, therefore, that
A. religiosa, A. hickelii and A. guatemalensis are merged
so that A. hickelii is referred to as A. religiosa subsp.
hickelii (Flous & Gaussen) U. Strandby, K.I. Chr. & M.
Sørensen, comb. et stat. nov. and A. guatemalensis as
A. religiosa subsp. mexicana (Martı
´
nez) U. Strandby, K.I.
Chr. & M. Sørensen, comb. nov. According to our analyses
A. vejarii cannot retain its status as a separate taxon as the
material studied is nested within A. religiosa subsp.
mexicana.
Keywords Morphometric study PCA UPGMA
Abies guatemalensis A. hickelii A. religiosa
A. vejarii
Introduction
Protecting species and habitats involves numerous actors
(governments, NGOs, communities, etc.) and disciplines
(forestry, ecology, economy, etc.) to reach successful
results (Carney 1998). However, applying well-recog-
nized conservation measures entails that the species in
question are well-defined. An unresolved or even incor-
rect taxonomy means that action at the species level
cannot efficiently address those taxa most in need of
conservation (Williams and Humphries 1994; Heywood
and Iriondo 2003; Hollingsworth 2003; Richard and
Evans 2006). The survival of certain species can thereby
become threatened and resources will be inefficiently
used (Hollingsworth 2003; McGough 2006; Rich 2006;
Richard and Evans 2006). Mountain ranges separated by
cultivated valleys result in forest population fragmenta-
tion and can produce differentiation among populations
both in terms of genetic composition and morphological
characters (Parker et al. 1981; Furnier and Eguiarte
1997; Aguirre-Planter et al. 2000; Ledig et al. 2000;
Jaramillo-Correa et al. 2008). Furthermore, the likely
increase in human population pressure will lead to rising
demand for land and water, and a higher incidence of
fragmented montane populations (Briggs and Walters
1997). Therefore, in terms of conservation, it is essential
to record the extant population variation in order to
correctly address conservations measures (Williams and
Humphries 1994; Hollingsworth 2003; Richard and
Evans 2006).
U. Strandby (&) M. Sørensen
Department of Agriculture and Ecology,
University of Copenhagen, Rolighedsvej 21,
1958 Frederiksberg C, Denmark
e-mail: uffes@post.cybercity.dk; usa@life.ku.dk
K. I. Christensen
The Botanical Garden and Museum, The Natural History
Museum of Denmark, University of Copenhagen,
Ø. Farimagsgade 2B, 1353 Copenhagen K, Denmark
123
Plant Syst Evol (2009) 280:59–76
DOI 10.1007/s00606-009-0164-x
Of the roughly 17 species of Abies Mill. occurring in
the New World, eight occur in Central America including
Mexico, and one (A. guatemalensis Rehder) or possibly
two [A. religiosa (Kunth) Schltdl. & Cham.] in Guatemala
(Martı
´
nez 1948; Liu 1971; Farjon 1990; Edwards 2008;
Furnier and Eguiarte 1997). The taxonomy of the genus
Abies has always been problematic: the interspecific
morphological differences are quite small and hybridiza-
tion may occur between even distantly related species
(Liu 1971; Edwards 2008; Furnier and Eguiarte 1997;
Scaltsoyiannes et al. 1999). Several authors have ques-
tioned the taxonomic status of and differentiation within
and between A. guatemalensis, A. hickelii Flous &
Gaussen, A. religiosa and A. vejarii Martı
´
nez (Farjon
1990; Debreczy and Ra
´
cz 1995; Furnier and Eguiarte
1997; Aguirre-Planter et al. 2000; Strandby Andersen
et al. 2006; Jaramillo-Correa et al. 2008).
Until 1932, A. religiosa, the Mexican fir, was the only
known Central American fir, when Flous and Gaussen
(1932) proposed the new species A. hickelii from Oaxaca,
Mexico. In 1939, A. guatemalensis Rehder (1939) was
described from Huehuetenango, Guatemala. In 1942,
Martı
´
nez described A. vejarii from Tamaulipas, and
A. mexicana from Nuevo Leo
´
n both in Mexico. A. religiosa
is the most widely distributed fir in Mexico (Farjon 1990).
A. hickelii is a species with a very restricted distribution
in the states of Chiapas, Oaxaca and Veracruz in Mexico.
A. guatemalensis occurs from northern Honduras to Jalisco
in West Mexico and Tamaulipas in East Mexico.
The main distribution area of A. guatemalensis
is,
however, in Guatemala, where montane conifer forests
with A. guatemalensis are estimated to cover 25,812 ha
(INAB 1999). A. vejarii has a limited range within the
Sierra Madre Oriental of Northern Mexico (Farjon 1990).
Furthermore, within each of these species a number of
infraspecific taxa have been proposed (Table 1).
Information on the geographic range of the Mexican firs
is inadequate; according to estimates from 1978 the genus
Abies covered approximately 32,000 ha with A. religiosa
constituting by far the major part (Rzedowski 1978).
Abies guatemalensis and A. hickelii are both listed as
vulnerable in the IUCN Red List of Threatened Species,
whereas both A. religiosa and A. vejarii are listed within
the category ‘lower risk–least concern’ (IUCN 2008).
Furthermore, in Mexico at the national level, both
A. hickelii and A. guatemalensis are regarded as facing high
risk of extinction, whereas A. vejarii is categorized as
threatened and A. religiosa does not appear on the list
(CONABIO 2001). In addition, A. guatemalensis is inclu-
ded on CITES Appendix 1 (CITES 2008), which signifies
Table 1 Meso-American Abies used in this morphometric study
Species Origin of type Year described No. of
infraspecific taxa
Abbreviation used in the
numerical analyses
A. religiosa Masatlan et Chilpantzingo, Mexico 1830 7 rel (16)
rem
b
(2)
A. hickelii Oaxaca, Mexico 1932 3 hic (11)
A. guatemalensis
a
Huehuetenango, Guatemalan 1939 7 gua (2)
jal
c
(8)
ixp
d
(1)
lon
e
(4)
taa
f
(2)
A. vejarii Tamaulipas, Mexico 1942 3 Vej (6)
A. flinckii Nevada de Colima, Jalisco, Mexico 1989 – Not included
g
A. zapotekensis Sierra de Jua
´
rez, Oaxaca, Mexico 1995 – Zap (1)
Number in parentheses refers to the number of sheets analysed
a
Two varieties of A. guatemalensis were excluded due to unavailability of herbarium material (A. guatemalensis var. tamaulipasensis Silba) or
important character states could not be scored (A. guatemalensis var. rushforthii Silba). However, the authors’ collections from Tamaulipas,
TAM, were collected at the type locality of A. guatemalensis var. tamaulipasensis
b
Refers to A. religiosa var. emarginata Loock & Martı
´
nez
c
Refers to A. guatemalensis var. jaliscana Martı
´
nez
d
Refers to A. guatemalensis var. ixtepejiensis Silba
e
Refers to A. guatemalensis var. longibracteata Debreczy & Ra
´
cz
f
Refers to A. guatemalensis var. tacanensis (Lundell) Martı
´
nez
g
The varieties A. guatemalensis var. jaliscana (jal) and A. religiosa var. emarginata (rem) are regarded as synomynous with A. flinckii.
Furthermore, our own collection from Jalisco (LCU) is within the distribution of A. flinckii Rushforth (1989)
60 U. Strandby et al.
123
that all international trade is prohibited, also in Guatemala
all felling and branch cutting is prohibited by national law
(INAB 1996).
The aims of this study were to: (1) analyse the geo-
graphic variation within the A. religiosa–hickelii–guate-
malensis complex, thereby making it possible to evaluate
the amount of morphologic differentiation, i.e. morpho-
species or infraspecific taxa of one morphospecies, and to
produce a robust identification/determination key usable
for both the specialist and the non-specialist, i.e. essential
to ensure efficient conservation procedures, (2) clarify
the distribution of A. guatemalensis in Guatemala and
Mexico and (3) test the validity of the infraspecific taxa of
A. guatemalensis. One of the primary goals of any sys-
tematic study is to produce well-defined, identifiable taxo-
nomic entities, thereby, ensuring the best possible basis for
ecological studies, decision making concerning conserva-
tion measures, etc. Furthermore, it should be noted that in
practice the conservationist is conserving populations not
taxa per se, and conservation politics should not influence
which rank is applied to a certain taxon. It may be just as
important to conserve an infraspecific taxon as a species.
Materials and methods
Materials
This morphometric study of the geographic variation in the
A. religiosa–hickelii–guatemalensis complex in Guatemala
and Mexico is based on material collected in natural forests
in Guatemala in December and January 2004/2005, in
December 2006 and in Mexico in January 2007. Sites were
chosen to represent the range of the natural distribution and
to cover the varieties described (Table 2). The closely
related A. religiosa, A. hickelii and A. vejarii were also
included by means of new material collected in the field
(A. religiosa: ABR and ARE; A. hickelii: IXT and IXJ) and
herbarium specimens. The intention was to sample
approximately 10 individuals per site. However, at some
locations this number had to be compromised due to a low
frequency of cone-bearing trees. The study sites include 15
forests in five Guatemalan departments and 12 forests in
six Mexican states (Fig. 1 and Table 2).
Individual trees were randomly selected. In this study,
individual trees or specimens were regarded as opera-
tional taxonomic units (OTUs). A point within the forest
was identified; a compass direction from this point and a
distance of 25 or 50 m were randomly chosen; the nearest
tree within a radius of 10 m was selected and labelled
with a semi-permanent label. If no tree met the agreed
terms (a healthy tree with a substantial cone production)
within the circle, a new compass direction and distance
was randomly chosen. The material was collected by
climbing the trees. Collections for this study were taken
from cone-bearing trees and only cone-bearing branches
from the upper part of the crown facing south were
analysed. One cone and one cone-bearing branch were
collected from each tree. Material was collected from the
upper part of the crown of the tree as it minimizes the
possibility of character differences due to environmental
differences and also eliminates the within-tree leaf vari-
ation found to be high in other firs (Parker et al. 1981;
Panetsos 1992).
Herbarium material was also included in the analysis in
order to test the validity of the A. guatemalensis varieties
described and to improve the foundation for comparisons
with the closely related species A. religiosa, A. hickelii,
A. vejarii and A. zapotekensis Debreczy, I.Ra
´
cz &
G.Ramı
´
rez (abbreviations used for herbarium specimens,
Table 1).
Vouchers are kept at C, GH, K and MICH (herbarium
acronyms according to Holmgren et al. 1990
).
Characters
All leaf and cone characters used were quantitative except
one multistate character, i.e. shape of leaf apex (Table 3).
All leaf measurements were scored on leaves from a
2-year-old annual shoot. The leaves analysed (±10 per
individual) were randomly chosen from the central part of
the previous year’s growth. Transverse leaf sections (20–
30 lm thick) were cut (from the mid section of the leaf)
with a razor blade and affixed to slides unstained. Mea-
surements of resin duct characters were done on the two
marginal ducts of three leaves per individual. The mea-
surements were made with a Nikon photomicroscope. For
all leaf characters, except NAP and ASL, the leaves were
preserved in 50% ethanol for at least 24 h before measur-
ing. Cone length and width characters were scored on one
representative cone per tree. The cone width was measured
at the mid section. All other cone characters represent
average values from ten ovuliferous scales, ten bracts and
ten seeds. The above mentioned leaf and cone characters
were chosen, because they are generally used in studies of
Abies (Lester 1968; Liu 1971; Parker et al. 1979, 1981;
Farjon 1990; Wu and Hu 1997) and could be measured
unequivocally.
Certain features previously used in studies of Abies
taxonomy were omitted because they proved invariant
among populations, e.g. number of stomatiferous rows on
the lower surface of the leaf and number of stomata per
2 mm of a stomatiferous row. A few features varied too
much within a population, e.g. female cones: sessile, sub-
sessile or pedunculate, or were difficult to score, e.g. male
strobili.
Morphometric study of the geographic variation 61
123
Table 2 The Abies religiosa–hickelii–guatemalensis complex—study sites and collection data in Guatemala and Mexico
Location site (department) Abbreviation Elevation
(m)
Latitude and
longitude
Number of
individuals sampled
Guatemala
Totonicapa
´
n (Totonicapa
´
n) TOT 3,100 14° 54
0
41
00
N6
91° 18
0
41
00
W
El Eden Palestina de los Altos (Quetzaltenango) EPA 2,865 14° 57
0
34
00
N11
91° 39
0
33
00
W
La Laguna Sibilia (Quetzaltenango) SLA 3,102 14° 57
0
25
00
N13
91° 37
0
45
00
W
Mataquesquintla (Jalapa) MAT 2,600 14° 31
0
40
00
N11
90° 08
0
54
00
W
San Jose
´
Ojetenan (San Marcos) JOS 3,232 15° 13
0
25
00
N12
91° 57
0
54
00
W
Buenos Aires (San Marcos) BAI 3,068 15° 07
0
52
00
N10
91° 52
0
59
00
W
Ixchigua
´
n (San Marcos) IXC 3,381 15° 10
0
24
00
N8
91° 56
0
56
00
W
San Vicente Buenabaj (Quetzaltenango) VBU 3,122 15° 02
0
58
00
N10
91° 34
0
39
00
W
Todos Santos, (Huehuetenango) TSA 2,955 15° 29
0
18
00
N5
91° 33
0
22
00
W
Puerta del Cielo, (Huehuetenango) PCI 3,330 15° 33
0
17
00
N7
91° 36
0
00
00
W
Montan
˜
a Pen
˜
a Blanca (San Marcos) PBL 3,445 15° 29
0
51
00
N6
91° 55
0
05
00
W
San Mateo Ixtata
´
n (Huehuetenango) SMI 2,970 15° 50
0
26
00
N7
91° 32
0
05
00
W
Soloma (Huehuetenango) SOL 3,005 15° 38
0
36
00
N4
91° 31
0
09
00
W
Volca
´
n Tacana
´
(San Marcos) TAC
a
3,263 15° 08
0
29
00
N5
92° 05
0
23
00
W
Volca
´
n Tajumulco (San Marcos) TAJ 3,453 15° 04
0
01
00
N5
91° 53
0
27
00
W
Mexico
Niquivil (Chiapas) NIQ 2,755 15° 15
0
34
00
N6
92° 12
0
46
00
W
Porvenir (Chiapas) POR 2,942 15° 27
0
14
00
N5
92° 19
0
11
00
W
Zapotitlan (Oaxaca) ZAP 2,678 16° 07
0
20
00
N17
96° 29
0
08
00
W
Ixtepeji (Oaxaca) IXT
b
3,011 17° 11
0
18
00
N8
96° 38
0
31
00
W
Ixtla
´
n de Jua
´
rez (Oaxaca) IXJ
c
2,994 17° 22
0
56
00
N10
96° 26
0
53
00
W
Carrizal del Bravo (Guerrero) CAR
d
2,521 17° 36
0
12
00
N10
99° 50
0
03
00
W
Los Sauces (Michoaca
´
n) LSA
e
2,466 19° 21
0
46
00
N13
101° 20
0
10
00
W
62 U. Strandby et al.
123
Data analyses
The following numerical methods commonly employed in
numerical taxonomy were used (Dunn and Everitt 1982;
Thorpe 1983; Fielding 2007; Rohlf 2008): (1) principal
components analysis (PCA), (2) unweighted pair-group
method using arithmetic averages or UPGMA clustering,
(3) Mantel’s test, (4) univariate analyses and (5) regression
analyses.
Prior to the PCA analysis, the specimens were tentatively
referred to as either ‘religiosa’, ‘hickelii’, ‘guatemalensis’
or ‘flinckii’. The a priori classification was based on former
taxonomic classifications and geographic distribution.
Principal components analyses were used to provide a
low-dimensional summary of the morphometric data
obtained from our sample (Jeffers 1967; Everitt and Dunn
2001; Lattin et al. 2003; Fielding 2007). In order to test the
extent to which our a priori classification of OTUs was
supported by these data, as well as to evaluate any clusters
suggested by the PCAs, we used unweighted pair-group
(UPGMA) sorting to find groups algorithmically (Sneath and
Sokal 1973; Dunn and Everitt 1982; Everitt and Dunn 2001;
Gibson 2002; Fielding 2007). We also sought to determine to
what extent our data might be structured geographically by
carrying out a Mantel’s test, and by exploring their rela-
tionship (in a regression sense) with elevation, latitude and
longitude. Finally, we depict the univariate comparisons
between the taxa that our results lead us to recognize.
In the current PCA, we identified the characters of minor
importance by visual inspection of biplots illustrating the
character-loading onto the first two components combined
with ‘the broken-stick model’ (Gabriel 1971; Jolliffe 1986;
Legendre and Legendre 1998; Rohlf 2008).
The standardization method used in both PCA and
UPGMA was to subtract the minimum and then dividing
by the range according to the recommendations of Milligan
and Cooper (1987). The input dissimilarity matrix used in
the UPGMAs and the subsequent Mantel’s test was based
on the DIST or average taxonomic distance coefficient
(Rohlf 2008).
A Mantel’s test (Mantel 1967) was performed to
examine the correlation between taxonomic dissimilarities
and geographical distance matrices (Sokal 1979). The
isolation by distance, web service (IBDWS) version 3.15
(Jensen et al. 2005) was used to perform the Mantel’s
test.
Univariate analyses and regression analyses were car-
ried out with ANOVA using JMP v6.0 (SAS Institute Inc.
2006).
Taxonomic concept
The species concept used in the present work is morpho-
logical, and mostly in line with that previously employed
by the second author in studies of Pinus and Crataegus
(Christensen
1987, 1992, 2005). Species complexes often
result from a recent radiation (Givnish 2001), and since
molecular divergence tends to be limited during recent
radiation events (Givnish 2000), morphological character-
istics remain very informative and often the only practical
Table 2 continued
Location site (department) Abbreviation Elevation
(m)
Latitude and
longitude
Number of
individuals sampled
San Carlos (Tamaulipas) TAM
f
1,444 24° 31
0
42
00
N5
98° 57
0
40
00
W
La Cumbre (Jalisco) LCU
g
2,182 20° 12
0
49
00
N5
104° 43
0
49
00
W
Papagayo (San Louis Potosi) PAP 1,788 22° 26
0
59
00
N12
99° 26
0
42
00
W
Cerro Burro (Michoaca
´
n) ABR 2,538 19° 23
0
60
00
N6
101° 33
0
36
00
W
20 km from Mexico City (Mexico) ARE 2,960 19° 17
0
58
00
N11
99° 24
0
04
00
W
a
Material collected at or near the type locality of Abies guatemalensis var. tacanensis
b
Material collected at or near the type locality of Abies guatemalensis var. ixtepejiensis
c
Material collected at or near the type locality of Abies zapotekensis
d
Material collected at or near the type locality of Abies guatemalensis var. longibracteata
e
Material collected at or near the type locality of Abies religiosa var. emarginata
f
Material collected at or near the type locality of Abies guatemalensis var. tamaulipasensis
g
Material collected at or near the type locality of Abies guatemalensis var. jaliscana
Morphometric study of the geographic variation 63
123
means available to diagnose or describe the subunits of
species complexes (Thorpe 1983; Wendt et al. 2000;
Gengler-Nowak 2002; Otieno et al. 2006). For reviews of
the assumptions, methods and applications of various
species concepts in taxonomy, see e.g. Baum and Donoghue
(1995), Davis (1995), Doyle (1995), Luckow (1995),
McDade (1995) and Olmstead (1995). The taxonomic ranks
used are defined as follows:
– Species of a genus differ from each other in several,
distinct characters and have a characteristic distribution
area of their own. Where closely related species meet
occasional hybridization and introgression may occur.
– Subspecies of a species are both regionally and locally
allopatric or at least parapatric. They differ from each
other in a few, distinct characters, but intergrade in
areas where their distributions overlap.
Results
PCA
In a PCA, characters of minor importance should be
identified and excluded from the final analysis. In the
current PCA, this was done on the basis of visual
inspection of biplots showing the character-loading onto
the first two principal components in combination with
the broken-stick model. Those dimensions with observed
proportions greater than those of the broken-stick model
explain more of the variance than expected due to chance
alone (Jolliffe 1986), i.e. the first THREE components of
the initial PCA (21.9508 ? 14.4992 ? 9.8003%—broken-
stick model expected 13.3166 ? 9.9833 ? 8.3166%), and
the first TWO components of the final PCA (34.8508 ?
Fig. 1 The Abies religiosa–hickelii–guatemalensis complex. a Over-
view of Mexico and Guatemala with the Abies populations indicated
with an asterix. b Locations of populations sampled in Mexico.
c Locations of populations sampled in Guatemala. For explanation of
abbreviations, see Table 1
64 U. Strandby et al.
123
18.7317%—broken-stick model expected 21.1296 ?
14.8796%). All ratio characters, as well as ND, NAP,
RDI and BW were excluded from the final PCA
(Table 3 and Fig. 2). Most ratio characters turned out
to be poor discriminators as compared to the characters
on which they were based. The first axis of the final PCA
is primarily described by various cone characters, the
second axis mostly by leaf characters.
In the PCA, three overlapping, poorly defined groups are
recognizable (Fig. 2): (1) the ‘religiosa’ group, (2) the
‘hickelii’ group and (3) the ‘guatemalensis’ group. The
specimens tentatively referred to ‘flinckii’ plot either
among the ‘religiosa’ specimens (rem) or where the
‘hickelii’ and ‘guatemalensis’ groups overlap (jal and the
LCU material).
UPGMA
The input dissimilarity matrix used in the UPGMA was
based on the DIST or average taxonomic distance coefficient
calculated on the 16 characters used in the final PCA.
The UPGMA dendrogram based on the 231 individual
OTUs shows that it is possible to recognize three large
clusters within A. religiosa–hickelii–guatemalensis com-
plex (Figs. 3, 4, 5): (1) the ‘guatemalensis’ group, (2) the
‘religiosa’ group and (3) the ‘hickelii’ group. It should be
noted that the clustering of several OTUs deviates from the
original, tentative identification. This is, however, hardly
surprising, when considering the amount of overlap between
the ‘guatemalensis’, ‘religiosa’ and ‘hickelii’ groups found
in the PCA (Fig. 2). On the other hand, the three groups are
Table 3 The Abies religiosa–
hickelii–guatemalensis
complex—list of characters
used in numerical analyses
Characters used in the final PCA
and UPGMA are given in bold.
All characters are quantitative
except for character 6 (shape of
needle apex) which is
qualitative. All quantitative
characters were measured in
millimetre
Plant
part
Character
Leaf 1. Length (NL)
2. Width (NW)
3. Thickness (NT)
4. Thickness/width (NT/NW)
5. Density (ND)
6. Shape of needle apex (NAP) [emarginate (0), obtuse (1), acute (2)]
7. Length of adaxial needle surface covered by stomata (ASL)
8. Length of adaxial needle surface covered by stomata relative to total needle length (ASL/NL)
9. Depth of the notch on the adaxial needle surface (NA)
10. Depth of the notch on the adaxial needle surface relative to needle thickness (NA/NT)
11. Number of resin ducts (RD)
12. Diameter of resin duct (RDI)
13. Position of the resin duct relative to the upper and lower needle surface (Ab/Ad)
14. Distance from the centre of the resin duct to the needle margin (LD)
Cone 15. Cone length (CL)
16. Cone width (CW)
17. Cone width relative to cone length (CW/CL)
18. Seed scale length (SSL)
19. Seed scale width (SSW)
20. Seed scale length relative to seed scale width (SSL/SSW)
21. Bract length (BL)
22. Bract width (BW)
23. Bract width relative to bract length (BW/BL)
24. Bract length relative to seed scale length (BL/SSL)
25. Cusp length (CUL)
26. Seed length (SL)
27. Seed wing width (SWW)
28. Seed wing length (SWL)
29. Seed length relative to seed wing length (SL/SWL)
30. Seed wing width relative to seed wing length (SWW/SWL)
Morphometric study of the geographic variation 65
123
rather well correlated with geography (Figs. 6, 7, 8). Gen-
erally, OTUs belonging to the ‘guatemalensis’ group may
occur in most of the populations studied, but the group is
centred in the Mexican–Guatemalan border-zone, as well as
in the Mexican states of Tamaulipas, San Luis Potosı
´
and
Guerrero (Fig. 6). Although OTUs of the ‘religiosa’ group
are found from the Mexican state of Jalisco in the Northwest
to Guatemala in the Southeast, the group is only common in
the Mexican states of Michoacan and Mexico (Fig. 7).
Finally, the ‘hickelii’ group is restricted to the Mexican state
of Oaxaca with outliers in Guatemala (Fig. 8).
In the UPGMA dendrogram, original material of A.
religiosa var. emarginata (=A. flinckii, FL-rem2) is basal to
the entire A. religiosa–hickelii–guatemalensis cluster,
while the authors’ material collected near the type locality
of var. emarginata (RE-LSA) is nested within the ‘religi-
osa’ cluster (Fig. 3). Furthermore, original material and/or
material collected near the type locality of A. guatemal-
ensis var. tamaulipasensis (GU-TAM), A. guatemalensis
var. longibracteata (GU-lon, GU-CAR), A. guatemalensis
var. tacanensis (GU-TAC), A. vejari (GU-vej), A. guate-
malensis var. jaliscana (=A. flinckii, GU-jal, GU-LCU) and
A. zapotekensis (GU-zap) are nested within the ‘guate-
malensis’ cluster, while material collected near the type
locality of the latter taxon (HI-IXJ) is nested within the
‘hickelii’ cluster (Figs. 3, 4, 5). Finally, the type material of
A. guatemalensis (GU-gua) is deeply nested within the
‘guatemalensis’ cluster (Fig. 5).
Mantel’s test
To test whether the differences identified among the pop-
ulations actually are related to the geographic distances
between them, we used a simple Mantel’s test on data from
the 27 populations studied by the authors in agreement with
the theory and use of Mantel’s test according to Sokal
(1979
), Bonnet and Van de Peer (2002) and Fielding (2007).
The standardization method employed here was to subtract
the minimum and then dividing by the range (Milligan and
Cooper 1987). The input dissimilarity matrices used in the
IBDWS were based on the DIST or average taxonomic
distance coefficient. The test based on the raw distances
gave a correlation coefficient r = 0.4562 with an associated
P-value \0.001 from 1,000 randomizations, while the test
based on log-transformed distances gave a correlation
coefficient r = 0.5865 with an associated P-value \0.001
from 1,000 randomizations. In both tests the correlation
coefficient is significantly different from 0 with P \ 0.001,
and, therefore, the morphological differences are correlated
to the geographic distances (Fig. 9).
Univariate analyses
In order to further test the grouping suggested by the UP-
GMA analysis, univariate analyses based on the 16 previ-
ously identified characters (Table 3) were conducted. From
the univariate analysis (Fig. 10) it is seen that the three
subspecies differ. Especially the characters NL (Fig. 10a),
ASL (Fig. 10d), RD (Fig. 10f), CL (Fig. 10h), SSL
(Fig. 10f), BL (Fig. 10l) and SWL (Fig. 10p) appear most
dissimilar. At the significance level of P \ 0.05, A. reli-
giosa subsp. religiosa differs from the other two subspecies
in 10 of the 16 characters. A. religiosa subsp. hickelii
differs in six characters from the other two subspecies and
A. religiosa subsp. mexicana differs in ten characters from
the other two.
Fig. 2 The Abies religiosa–
hickelii–guatemalensis
complex. The final PCA based
on 16 quantitative characters
and 231 OTUs
66 U. Strandby et al.
123
Regression analyses
Simple linear regressions were performed to examine the
relationship between taxonomic characters and geographic
attributes (altitude, latitude and longitude). Latitude and
altitude were significantly correlated with a Pearson prod-
uct-moment correlation coefficient of 0.72 (P \ 0.0001).
Quite a few characters (Table 4) are correlated with the
geographic attributes and 7 of the 16 characters used in the
final PCA and UPGMA analyses are correlated with the
geographic attributes.
Discussion and conclusion
Morphometric analysis provides a very useful tool for
elucidating phenetic relationships among morphologically
closely related taxa or species complexes. In this study,
multivariate analyses of quantitative morphological char-
acters show that it is possible to recognize three rather
poorly defined groups within the A. religiosa–hickelii–
guatemalensis complex (Figs. 2, 3, 4, 5): (1) the ‘religiosa’
group, (2) the ‘hickelii’ group and (3) the ‘guatemalensis’
group. Furthermore, the Mantel’s test performed indicates
that the taxonomic dissimilarities are correlated to geo-
graphic distances (Fig. 9). Each group is predominant
within a certain geographic area, but individuals belonging
to one group often occur in sites where another group is
common (Figs. 6, 7, 8). The overlapping distributions and
the lack of distinct morphological dissimilarities between
the three groups of the A. religiosa–hickelii–guatemalensis
complex indicate that there is little evidence for supporting
the continued recognition of A. religiosa, A. hickelii and
A. guatemalensis as separate morphospecies. Consequently,
the three taxa are merged as follows: (1) A. religiosa subsp.
religiosa, (2) A. religiosa subsp. hickelii (=A. hickelii s.str.)
Fig. 3 The Abies religiosa–hickelii–guatemalensis complex. Lower part of the UPGMA dendrogram based on 16 characters and 231 OTUs.
Arrows in the left margin indicate herbarium specimens studied
Morphometric study of the geographic variation 67
123
and (3) A. religiosa subsp. mexicana (including A. guate-
malensis and A. vejarii). For details, see below.
The studies of Central American Abies by Furnier and
Eguiarte (1997) and Aguirre-Planter et al. (2000) demon-
strated that Abies species occurring in the south of Mexico
and in Guatemala are very similar and a merger of cur-
rently accepted species may prove necessary. Furnier and
Eguiarte (1997) argued that the reason for the slight dif-
ferences among these Abies species can be ascribed to the
facts that: (1) previously when the climate was cooler and
the human impact lower the Abies populations were more
coherent, and (2) the fragmentation of the Abies popula-
tions occurred fairly recently, therefore, the populations
have not had sufficient time to diverge genetically and
morphologically. On the other hand, Ledig et al. (2000)
demonstrated that among Mexican conifers (including
Abies) the levels of genetic differentiation are two to three
times higher than the levels among boreal and north
temperate conifers. The extreme topography and associated
climate in Mexico mean that plant species often occur as
small, insular populations situated far apart, the migration
among these populations is restricted, and that genetic drift
and selection in isolation may speed up the speciation
processes (Lienert et al. 2002; Oostermeijer et al. 2002).
In a study of the molecular biogeography of the Mexican–
Guatemalan A. religiosa complex Jaramillo-Correa et al.
(2008) found that mitochondrial genetic differentiation
within the complex was not related to the taxonomy, while
the chloroplast genetic differentiation was significantly
correlated with the taxonomy: (1) A. flinckii was the
most deviant species, and (2) A. religiosa, A. hickelii and
A. guatemalensis formed a more or less continuous group of
taxa. Several of the populations studied by Jaramillo-Correa
et al. (2008) are also included in the current mophometric
study of the A. religiosa complex, and although there are
disagreements concerning the identity of a couple of
Fig. 4 The Abies religiosa–hickelii–guatemalensis complex. Middle part of the UPGMA dendrogram based on 16 characters and 231 OTUs.
Arrows in the left margin indicate herbarium specimens studied
68 U. Strandby et al.
123
populations, most populations are referred to the same taxon
in the two studies (Table 5). A single population studied by
Jaramillo-Correa et al. (2008) but not by us requires a
comment. Their A. religiosa population AR22 from Nevado
de Colima is de facto collected at the type locality of
A. flinckii. In spite of this, the AR22 population is situated
on the wrong side of the genetic boundaries of A. flinckii as
defined by Jaramillo-Correa et al. (2008: Figure 3).
Rasmussen et al. (2008) used microsatellite markers
to investigate genetic diversity in 18 populations of A.
religiosa subsp. mexicana in Guatemala (including all
Guatemalan populations examined in this study) and found
that the majority of these populations could be considered
as one metapopulation with two geographically marginal
populations as possible outliers (MAT of this study and
Sierra de las Minas). The genetic boundary E found in
Guatemala by Jaramillo-Correa et al. (2008: Figure 3) may
well explain this result.
Could the differences in morphology be explained by
phenotypic plasticity? We were able to demonstrate that 15
of the 30 measured characters were correlated with altitude
(9 characters), latitude (8) or longitude (11). Myers and
Bormann (1963) also document the idea of an altitudinal
cline in their study on A. balsamea (L.) Mill. with respect
to scale/bract ratios and leaf length. Parker et al. (1981)
suggest that A. balsamea and A. lasiocarpa (Hook.) Nutt.
consist of only one widely distributed gene pool, and that
different local selection pressures have led to the devel-
opment of the differences in morphology. We interpret our
findings as an indication of clinal variation within the Abies
species in southern Mexico and Guatemala. The inclusion
of additional abiotic data in the current study (soil types,
precipitation, etc.) could improve the possibility of further
analysis of the phenotypic plasticity effect.
Another reason for the slight dissimilarity among Abies
species in Mexico and Guatemala could be hybridisation
Fig. 5 The Abies religiosa–hickelii–guatemalensis complex. Upper part of the UPGMA dendrogram based on 16 characters and 231 OTUs.
Arrows in the left margin indicate herbarium specimens studied
Morphometric study of the geographic variation 69
123
and introgression. Hybridization between sympatric spe-
cies of Abies is well documented (Hawley and DeHayes
1985; St. Clair and Critchfield 1988; Edwards 2008),
although there are also examples of sympatric species, e.g.
A. concolor (Gord. & Glend.) Hildebr. and A. magnifica
Andr. Murray, which are unable to hybridize (St. Clair and
Critchfield 1988). However, there is little or no evidence
supporting the hypothesis that hybridization events should
be the reason for the variation patterns found in Abies in
Mexico and Guatemala.
During the last 60–70 years several new Abies taxa have
been described within the study area and from neighbour-
ing regions. The results of the current morphometric study
(Figs. 2, 3, 4, 5) show that original material or topotypes of
the majority of these taxa are referrable to either: (1)
A. religiosa subsp. hickelii, i.e. A. zapotekensis (Aguirre-
Planter et al. 2000), and A. guatemalensis var. ixtepejiensis,
or (2) A. religiosa subsp. mexicana, i.e. A. guatemalensis,
A. vejari, A. guatemalensis var. jaliscana (Liu 1971),
Fig. 6 Frequency of Abies guatemalensis-like OTUs according to the
UPGMA (Figs. 3, 4, 5). DMAP for Windows (Morton 2001)
Fig. 8 Frequency of Abies hickelii-like OTUs according to the
UPGMA (see Figs. 3, 4, 5). DMAP for Windows (Morton 2001)
Fig. 7 Frequency of Abies religiosa-like OTUs according to the
UPGMA (Figs. 3, 4, 5). DMAP for Windows (Morton 2001)
Fig. 9 The Abies religiosa–hickelii–guatemalensis complex. Mantel’s
test. Biplot of log-transformed geographic and taxonomic distances of
27 populations
70 U. Strandby et al.
123
A. guatemalensis var. longibracteata, A. guatemalensis var.
tacanensis (Liu 1971), A. guatemalensis var. tamaulipas-
ensis, etc.
Rushforth (1989) treated A. religiosa var. emarginata
and A. guatemalensis var. jaliscana as belonging to A.
flinckii Rushforth. In this study it has, primarily due to
insufficient sampling, not been possible to clarify the exact
relationships of A. flinckii to the rest of the A. religiosa
complex. In the PCA, the few OTUs initially referred to
this taxon generally plot where the ‘hickelii’ and ‘guate-
malensis’ groups overlap (Fig. 2). In the UPGMA, how-
ever, the material of A. religiosa var. emarginata (Fl-rem2
in Fig. 3) is basal to the entire A. religiosa–hickelii–gua-
temalensis cluster which indicates that it might represent a
separate taxon within the A. religiosa complex, and studies
of the genetic variation of Abies in southern Mexico and
Guatemala support Rushforth’s opinion (Aguirre-Planter
et al. 2000; Jaramillo-Correa et al. 2008).
A consensus on the taxonomy of Abies occurring in
southern Mexico and Guatemala is undoubtedly still
remote, but in our opinion a merger of currently accepted
morphospecies is necessary. Consequently, we suggest the
following taxonomic treatment of Abies within this
region.
Abies sect. Oiamel Franco subsect. Religiosae
(Matzenko) Farjon & Rushforth
Abies sect. Oiamel Franco subsect. Hickelianae Farjon
& Rushforth.
Abies religiosa (Kunth) Schltdl. & Cham.
Schltdl. & Cham., Linnaea 5:77 (1830) : Pinus religiosa
Kunth in Humboldt et al., Nov. Gen. Sp. Pl. 2,5: 6 (1817)—
Described from Mexico, Mazatla
´
n to Chilpancingo.
Leaves (11.4–)24.0–26.2(–71.5) mm long; upper surface
without or with a groove (0–)10–13(–53)% of leaf thick-
ness; resin ducts 2–10. Female cones (4.8–)9.2–9.7(–15.2)
cm long; bracts (9.2–)18.9–20.2(–35.2) mm long, each
bract longer or shorter than the ovuliferous scale it sub-
tends, cusp (0.2–)1.6–1.9(–6.3) mm long.
Fig. 10 Character variation among the three Abies religiosa subspecies. Box plots of 16 quantitative characters in the 266 OTU sample (subsp.
religiosa = 52; subsp. mexicana = 173 and subsp. hickelii = 41). For abbreviations, see Table 3
Morphometric study of the geographic variation 71
123
Distribution
Occurring in Mexico Northwest (Sinaloa), Mexico North-
east (Chihuahua, Coahuila, Nueva Leo
´
n, Tamaulipas,
Hidalgo), Mexico Gulf (Veracruz), Mexico Central (Mex-
ico State, Distrito Federal, Tlaxcala, Puebla, Morelos),
Mexico Southwest (Jalisco, Michoacan, Guerrero, Oaxaca)
and Mexico Southeast (Chiapas), as well as in Guatemala
(Huehuetenango, San Marcos, Quetzaltenango, Tot-
onicapa
´
n, Solola
´
, Jalapa, Quiche
´
) (Figs. 6, 7, 8). Also
occurring in Honduras and possibly in El Salvador.
Abies religiosa subsp. religiosa
Leaves (18.0–)29.2–36.7(–71.5) mm long; upper surface
grooved (4–)16–21(–40)% of leaf thickness; resin ducts
2(–5). Female cones (8.0–)10.9–12.0(–15.2) cm long; bracts
(17.8–)25.7–28.4(–35.2) mm long, each bract generally
longer than the ovuliferous scale it subtends, cusp (0.9–)2.2–
2.7(–5.7) mm long.
Distribution
Occurring in Mexico Northwest (Sinaloa), Mexico North-
east (Chihuahua, Nueva Leo
´
n, Tamaulipas, Hidalgo),
Mexico Gulf (Veracruz), Mexico Central (Mexico State,
Distrito Federal, Tlaxcala, Puebla, Morelos) and Mexico
Southwest (Jalisco, Michoacan, Guerrero, Oaxaca). The
occurrence of A. religiosa subsp. religiosa in Guatemala
reported by Farjon (1990) is confirmed (Fig. 7).
Abies religiosa subsp. hickelii (Flous & Gaussen) U.
Strandby, K.I. Chr. & M. Sørensen comb. et stat. nov.
Abies hickelii Flous & Gaussen, Bull. Soc. Hist. Nat.
Toulouse 64:24–30 (1932)—Holotype: Mexico, Oaxaca,
1,650 m, s.d., Conzatti s.n. (Herb. Bonaparte LY).
A. oaxacana Martı
´
nez, Anales Inst. Biol. Univ. Nac.
Me
´
xico 19:39 (1948) A. hickelii var. oaxacana (Martı
´
nez)
Farjon & Silba, Phytologia 68: 20 (1990)—Holotype:
Mexico, Oaxaca, Ixtla
´
n de Jua
´
rez, Oct. 1942, Martı
´
nez
27900 (MEXU).
Illustration: Fig. 11a–c. Leaves (11.4–)21.7–24.5(–33.6)
mm long; upper surface with a deep groove, (6–)18–
24(–37)% of leaf thickness; resin ducts (3–)6–7(–10).
Female cones (6.2–)8.3–9.2(–11.2) cm long; bracts (13.6–)
18.4–20.3(–24.0) mm long, each bract usually longer than
the ovuliferous scale it subtends, cusp (0.6–)2.0–2.9(–6.3)
mm long.
Distribution
Occurring in Mexico Gulf (Veracruz), Mexico Southwest
(Oaxaca) and possibly Mexico Southeast (Chiapas). Also
occurring in Guatemala (Huehuetenango and San Marcos)
(Fig. 8).
Abies religiosa subsp. mexicana (Martı
´
nez) U.
Strandby, K.I. Chr. & M. Sørensen comb. nov.
A. vejarii subsp. mexicana
(Martı
´
nez) Farjon, Regnum
Veg. 121: 103 (1990) A. mexicana Martı
´
nez, Anales Inst.
Biol. Univ. Nac. Me
´
xico 13:626 (1942)—Described from
Mexico, Nuevo Leo
´
n, Sierra de Sta. Catarina, Can
˜
on de
Vivanco. Original material not located, but probably kept
at MEXU.
A. guatemalensis Rehder, J. Arnold. Arbor. 20:285–287
(1939)—Holotype: Guatemala, Huehuetenango, Cumbre
del Aire, 10,000 ft, 14 Dec. 1937, J.H. Faull 13104 (A!;
iso- NY!).
A. tacanensis Lundell, Amer. Midl. Naturalist 23: 175
(1940) A. guatemalensis var. tacanensis (Lundell) Martı
´
nez,
Anales Inst. Biol. Univ. Nac. Mexico 19:70 (1948)—
Holotype: Mexico, Chiapas, Mt. Tacana
´
, 2,000–4,038 m,
Aug. 1938, E. Matuda 2367 (MICH!, iso- MO!, NY).
A. vejarii Martı
´
nez, Anales Inst. Biol. Univ. Nac.
Me
´
xico 13:629 (1942)—Type kept at MEXU.
A. zapotekensis Debreczy, Ra
´
cz & Ramı
´
rez, Phytologia
78:225 (1995)—Holotype: Mexico, Oaxaca, Sierra de
Jua
´
rez, near Portillo, 2,700 m, 30 Jun. 1994, Debreczy
et al. 40675a (BP, iso- A, CHAP, E, MEXU, NA).
Table 4 The Abies religiosa–hickelii–guatemalensis complex—
regression analyses testing the dependence of morphological char-
acters on altitude, latitude and longitude
Characters Altitude Latitude Longitude
R
2
P value R
2
P value R
2
P value
NT 0.32 0.0019 0.48 \0.0001
NT/NW 0.16 0.0384
RDI 0.42 0.0002 0.52 \0.0001 0.24 0.0096
LD 0.17 0.031
ND 0.46 \0.0001 0.4 0.0004 0.44 0.0002
NL 0.32 0.002
NW 0.16 0.04 0.29 0.004
CW 0.16 0.046
SSL/SSW 0.31 0.0025 0.44 0.0001 0.54 \0.0001
BW 0.46 0.0001 0.31 0.0025
BW/BL 0.25 0.0072
BL/SSL 0.17 0.0344
SL 0.23 0.0143
SL/SWL 0.59 \0.0001 0.46 0.0002 0.28 0.0054
SWW/SWL 0.17 0.036 0.26 0.0073 0.22 0.0149
Characters in bold are among the 16 identified characters used in the
multivariate analyses
72 U. Strandby et al.
123
A. guatemalensis var. longibracteata Debreczy & Ra
´
cz,
Phytologia 78:227 (1995)—Holotype: Mexico, Guerrero,
Sierra Madre del Sur, near Yextla, 2,400 m, 10 Jan. 1994,
Debreczy et al. 34763 (BP, iso- A!, CHAP, E, K!, MEXU,
NA).
A. guatemalensis var. tamaulipasensis Silba, J. Int.
Conifer Preserv. Soc. 5:45 (1997)—Holotype: Mexico,
Tamaulipas, La Buffa el Diente, 1,219–1,280 m, s.d., J.
Silba M-00357 (type not located, no material at K or NY).
A. guatemalensis var. ixtepejiensis Silba in J. Int.
Conifer Preserv. Soc. 7(1):19 (2000)—Holotype: Mexico,
Oaxaca, 18 June 1942, M. Martı
´
nez s.n. (A!).
A. guatemalensis var. rushforthii Silba, J. Int. Conifer
Preserv. Soc. 7(1):19 (2000)—Holotype: Honduras, Santa
Fig. 11 Line drawing of Abies
religiosa subsp. hickelii and
subsp. mexicana. a–c A.
religiosa subsp. hickelii (a–c:
IXJ6). d–k A. religiosa subsp.
mexicana (d: VB1. e, f: Faull
13104, isotype of A.
guatemalensis, g: CAR9. h, i:
CAR3, j: CAR6. K: CAR6).a,
d, g Simplified transverse
sections of leaves. b, e, h Bracts
and ovuliferous scales. c, f, i
Seeds. J. Twig. K. Partly
disintegrated female cone.
Scales: 1 = 10 mm (b, c, e, f, h,
i), 2 = 10 mm (j), 3 = 10 mm
(k), 4 = 0.2 mm (a, d, g). Del.:
Knud Ib Christensen
Morphometric study of the geographic variation 73
123
Barbara, summit of Cerro Santa Barbara, Apr. 1951,
Armour & Chable 6098 (holotype: GH!).
Illustration: Fig. 11d–k. Leaves (11.5–)22.3–24.3(–52.4)
mm long; upper surface without or with a shallow or deep
groove, (0–)6–10(–53) % of leaf thickness; resin ducts 2–
3(–6). Female cones (4.8–)8.8–9.3(–13.6) cm long; bracts
variable (9.2–)16.9–18.0(–27.0) mm long, shorter or longer
than ovuliferous scales; cusp (0.2–)1.3–1.5(–3.3) mm.
Distribution
Occurring in Mexico Northeast (Coahuila, Nuevo Leo
´
n,
Tamaulipas), Mexico Central (Mexico State), Mexico
Southwest (Jalisco, Guerrero, Oaxaca) and Mexico
Southeast (Chiapas), as well as in Guatemala (Huehueten-
ango, San Marcos, Quetzaltenango, Totonicapa
´
n, Solola
´
,
Jalapa, Quiche
´
) (Fig. 7). Also occurring in Honduras and
possibly El Salvador.
Abies flinckii Rushforth
Rushforth, Notes Roy Bot Gard Edinburgh 46:101–105
(1989). Holotype: Mexico, Jalisco, Nevada de Colima, in
and above the Pueblo of El Izote, 2,350 m, 9. Nov. 1984,
Rushforth 621 (XAL, iso- E).
A. religiosa var. emarginata Martı
´
nez, Anales Inst. Biol.
Univ. Nac. Me
´
xico 19:60 (1948)—Holotype: Mexico,
Michoaca
´
n, Mil Cumbres, 27 Jul. 1947, Loock 128
(holotype: MEXU).
A. guatemalensis var. jaliscana Martı
´
nez, Anales Inst.
Biol. Univ. Nac. Mexico. 19:73 (1948)—Holotype: Mexico,
Jalisco, Municipio Talpa, Nov. 1944, Martı
´
nez 28500
(MEXU!).
Distribution
Occurring in Mexico Southwest (Jalisco, Michoacan)
(Rushforth 1989; Jaramillo-Correa et al. 2008).
In the current paper it has, primarily due to insufficient
sampling, not been possible to clarify the exact relation-
ships of Abies flinckii to the rest of the A. religiosa complex.
However, it is likely that this taxon should be treated as yet
another subspecies of A. religiosa as circumscribed here.
Diagnostic key to the Abies religiosa complex
(A. flinckii excluded)
Notice, in the key and the descriptions given above, leaf
characters refer to the leaves of fertile shoots bearing
female cones, and bracts and ovuliferous scales are those of
the central portion of the female cone.
1. Female cones usually at least 10 cm long; bracts
usually at least 25 mm long and each bract generally
longer than the ovuliferous scale it subtends—subsp.
religiosa.
– Female cones usually \10 cm long; bracts usually
\25 mm long and each bract often shorter than the
ovuliferous scale it subtends—2.
2. Resin ducts 2 or if 2–3(–6), then the upper surface of
leaf without or with a shallow groove, 0–14(–31) % of
leaf thickness—subsp. mexicana.
– Resin ducts (3–)6–7(–10). Upper surface of leaf
with a deep groove, (6–)18–24(–37) % of leaf
thickness—subsp. hickelii.
Acknowledgments This project was embarked upon 4 years ago
when the project proposal was approved by the Danish Research
Council for Development Research (grant No. 91160) with an addi-
tional PhD grant by the University of Copenhagen. We gratefully
acknowledge the advice and assistance provided by Jose
´
Pablo Prado
Co
´
rdova, Juan Jose
´
Castillo Mont, Gamaliel Alexander Martı
´
nez
Marroquin, Martin Schiøtz, Karen Munk Rysbjerg, Anne Marie
Thonning Skov, Gabriela Garcı
´
a Besne
´
, Tulio Lot del Angel, Erika
Aguirre-Planter, Eduardo Estrada and the Guatemalan National Seed
Centre (BANSEFOR). The curators of A, GH, K, MEXU, MICH, MO
Table 5 The Abies religiosa–hickelii–guatemalensis complex—overview of populations studied by both Jaramillo-Correa et al. (2008) and the
present authors
Bayesian group of
Jaramillo-Correa
et al. (2008) based on
cpDNA haplotype
composition
Abbreviation of
population
sample used by
Jaramillo-
Correa et al. (2008)
Abbreviation of
population sample used
in the current study
Taxon according to
Jaramillo-Correa
et al. (2008)
Taxon according
to the current study
2 AF16 LSA flinckii A. religiosa subsp. religiosa (=the
‘religiosa’ group)
6 AH3 IXJ hickelii A. religiosa subsp. hickelii (=the
‘hickelii’ group)
8 AG2 IXT guatemalensis A. religiosa subsp. hickelii (=the
‘hickelii’ group)
8 AG10, AG11, AG42, AG43,
AG44, AG51
TAC, POR, MAT, TOT,
CAR, PAP
guatemalensis A. religiosa subsp. mexicana (=the
‘guatemalensis’ group)
74 U. Strandby et al.
123
and NY kindly provided material for study. We are grateful to the
reviewers for their comments which helped to improve the
manuscript.
References
Aguirre-Planter E, Furnier GR, Eguiarte LE (2000) Low levels of
genetic variation within and high levels of genetic differentiation
among populations of species of Abies from Southern Mexico
and Guatemala. Am J Bot 87:362–371
Baum DA, Donoghue MJ (1995) Choosing among alternative
‘phylogenetic’ species concepts. Syst Bot 20:560–573
Bonnet E, Van de Peer Y (2002) zt: A software tool for simple and
partial Mantel tests. J Stat Softw 7:1–12
Briggs D, Walters SM (1997) Plant variation and evolution.
Cambridge University Press, Cambridge, pp 399–433
Carney D (1998) Implementing the sustainable rural livelihood
approach. In: Carney D (ed) Sustainable rural livelihoods—what
contribution can we make?. Department for International
Development, London
Christensen KI (1987) Taxonomic revision of the Pinus mugo
complex and P. 9 rhaetica (P. mugo 9 sylvestris) (Pinaceae).
Nord J Bot 7:383–408
Christensen KI (1992) Revision of Crataegus sect. Crataegus and
nothosect. Crataeguineae (Rosaceae–Maloideae) in the Old
World. Syst Bot Monogr 35:1–199
Christensen KI (2005) A morphometric study of the geographic
variation in Pinus contorta (Pinaceae). Nord J Bot 23:563–575
CITES (2008) Convention on International Trade in Endangered
Species of Wild Fauna and Flora, Geneva. Available at:
http://www.cites.org/eng/app/appendices.shtml (accessed April
2008)
CONABIO (2001) Comisio
´
n nacional para el conocimiento y uso
de la biodiversidad. Mexico City. http://www.conabio.gob.mx/
conocimiento/ise/fichas/doctos/plantas.html (accessed April
2008)
Davis JI (1995) Species concepts and phylogenetic analysis. Syst Bot
20:555–559
Debreczy Z, Ra
´
cz I (1995) New species and varieties of conifers from
Me
´
xico. Phytologia 78:217–243
Doyle JJ (1995) The irrelevance of allele tree topographies for species
delimitation, and a non-topological alternative. Syst Bot 20:574–
588
Dunn G, Everitt BS (1982) An introduction to mathematical
taxonomy. Cambridge University Press, London
Edwards DGW (2008) Abies. In: USDA FS agriculture handbook 727.
Dry Branch, USDA Forest Service, Washington, DC, pp 149–
198. Available at: http://www.nsl.fs.fed.us/wpsm/ Abies.pdf
(accessed May 2008)
Everitt BS, Dunn G (2001) Applied multivariate data analysis. Oxford
University Press, New York
Farjon A (1990) Pinaceae—drawings and descriptions of the genera
Abies, Cedrus, Keteleeria, Nothotsuga, Tsuga, Cathaya, Pseud-
otsuga, Larix and Picea. Koeltz Scientific Books, Ko
¨
nigstein
Fielding AH (2007) Cluster and classification techniques for the
biosciences. Cambridge University Press, Cambridge
Flous F, Gaussen H (1932) Une nouvelle espe
´
ce de sapin Mexicain
Abies hickelii. Bull Soc Hist Nat Toulouse 64:24–30
Furnier GR, Eguiarte LE (1997) Niveles y patrones de variacio
´
n
gene
´
tica del ge
´
nero Abies en Me
´
xico. Informe final del proyecto
B138. Universidad Nacional Auto
´
noma de Me
´
xico, Mexico City
Gabriel KR (1971) The biplot graphic display of matrices with
application to principal component analysis. Biometrika
58:453–467
Gengler-Nowak K (2002) Phenetic analyses of morphological traits in
the Malesherbia humilis complex (Malesherbiaceae). Taxon
51:281–293
Gibson DJ (2002) Methods in comparative plant population ecology.
Oxford University Press, Oxford
Givnish TJ (2000) Adaptive radiation and molecular systematics:
issues and approaches. In: Schluter D (ed) The ecology of
adaptive radiation. Oxford University Press, Oxford, pp 1–54
Givnish TJ (2001) The rise and fall of plant species: a population
biologist’s perspective. Am J Bot 88:1928–1934
Hawley GJ, DeHayes DH (1985) Hybridization among several North
American firs. I. Crossability. Can J For Res 15:42–49
Heywood VH, Iriondo JM (2003) Plant conservation: old problems,
new perspectives. Biol Conserv 111:321–335
Hollingsworth PM (2003) Taxonomic complexity, population genet-
ics and plant conservation in Scotland. Bot J Scotl 55:55–63
Holmgren PK, Holmgren NH, Barnett LC (1990) Index herbariorum.
Part I: the herbaria of the World, 8th edn (Regnum Veg. 120).
International Association for Plant Taxonomy, New York
Botanical Garden, New York
INAB (1996) Ley Forestal: Decreto Legislativo Nu
´
mero 101–96.
Instituto Nacional de Bosques, Guatemala City
INAB (1999) Diagno
´
stico de las poblaciones naturales de Pinabete
(Abies guatemalensis R.) en Guatemala y estrategia para su
conservacio
´
n. Co-ediciones te
´
cnicas, documento No. 11. Insti-
tuto Nacional de Bosques, Guatemala City
IUCN (2008) Abies guatemalensis—vulnerable. IUCN, Gland. Avail-
able at: http://www.iucnredlist.org/search/details.php/42285/all
(accessed March 2008)
Jaramillo-Correa JP, Aguirre-Planter E, Khasa DP, Eguiarte LE,
Pin
˜
ero D, Furnier GR, Bousquet J (2008) Ancestry and
divergence of subtropical montane forest isolates: molecular
biogeography of the genus Abies (Pinaceae) in southern Me
´
xico
and Guatemala. Mol Ecol 17:2476–2490
Jeffers JNR (1967) Two case studies in the application of principal
component analysis. Appl Stat 16:225–236
Jensen JL, Bohonak AJ, Kelley ST (2005) Isolation by distance, web
service (v.3.15). BMC Genet 6:13. Available at: http://ibdws.
sdsu.edu/
Jolliffe IT (1986) Principal components analysis. Springer, New York
Lattin J, Carrol JD, Green P (2003) Analyzing multivariate data.
Thomson Learning, Toronto
Ledig FT, Bermejo-Vela
´
zquez B, Vargas-Herna
´
ndez J (2000) Genetic
differentiation in Mexican conifers. Available at: www.fs.fed.
us/global/nafc/silvi/activities/2000/ conference/posters/Ledig
PosterNAFBW.doc
Legendre P, Legendre L (1998) Numerical ecology, 2nd edn.
Elsevier, Amsterdam
Lester DT (1968) Variation in cone morphology of balsam fir, Abies
balsamea. Rhodora 70:83–94
Lienert J, Fischer M, Schneller J, Diemer M (2002) Isozyme
variability of the wetland specialist
Swertia perennis (Gentian-
aceae) in relation to habitat size, isolation, and plant fitness. Am
J Bot 89:801–811
Liu TS (1971) A monograph of the genus Abies. Department of
Forestry, College of Agriculture, National Taiwan University,
Taipei
Luckow M (1995) Species concepts: assumptions, methods and
applications. Syst Bot 20(4):589–605
Mantel N (1967) The detection of disease clustering and a generalized
regression approach. Cancer Res 27:209–220
Martı
´
nez M (1942) Genero Abies. Anales Inst Biol Univ Nac Mexico
13:629–634
Martı
´
nez M (1948) Los Abies Mexicanos. Anales Inst Biol Univ Nac
Mexico 19:11–104
Morphometric study of the geographic variation 75
123
McDade LA (1995) Species concepts and problems in practice:
insight from botanical monographs. Syst Bot 20:606–622
McGough HN (2006) Legislation: a key user of taxonomy for plant
conservation and sustainable use. In: Leadlay E, Jury S (eds)
Taxonomy and plant conservation. Cambridge University Press,
New York, pp 255–265
Milligan GW, Cooper MC (1987) A study of variable standardization.
College of Administrative Science Working Paper Series 87–63.
Ohio State University, Columbus
Morton A (2001) DMAP for Windows. Version 7.1. Winkfield.
http://www.dmap.co.uk/
Myers O, Bormann FH (1963) Phenotypic variation in Abies
balsamea in response to altitudinal and geographic gradients.
Ecology 44:429–436
Olmstead RG (1995) Species concepts and plesiomorphic species.
Syst Bot 20:623–630
Oostermeijer JGB, Luijten SH, Ellis-Adam AC, den Nijs JCM (2002)
Future prospects for the rare, late-flowering Gentianella germa-
nica and Gentianopsis ciliata in Dutch nutrient-poor calcareous
grasslands. Biol Conserv 104:339–350
Otieno DF, Balkwill K, Paton AJ (2006) A multivariate analysis of
morphological variation in the Hemizygia bracteosa complex
(Lamiaceae, Ocimeae). Plant Syst Evol 261:19–38
Panetsos KP (1992) Variation in the position of resin canals in the
leaves of Abies species and provenances. Ann Sci For 49:253–
260
Parker WH, Bradfield GE, Maze J, Lin SC (1979) Analysis of
variation in leaf and twig characters of Abies lasiocarpa and A.
amabilis from north-coastal British Columbia. Can J Bot
57:1354–1366
Parker WH, Maze J, Bradfield GE (1981) Implications of morpho-
logical and anatomical variation in Abies balsamea and
A. lasiocarpa (Pinaceae) from Western Canada. Am J Bot
68:843–854
Rasmussen KK, Strandby Andersen U, Frauenfelder N, Kollmann J
(2008) Microsatellite markers for the endangered fir Abies
guatemalensis (Pinaceae). Mo Ecol Resour 8(6):1307–1309
Rehder A (1939) The firs of Mexico and Guatemala. J Arnold Arbor
20:281–287
Rich TCG (2006) The role of the taxonomist in conservation of
critical vascular plants. In: Leadlay E, Jury S (eds) Taxonomy
and plant conservation. Cambridge University Press, New York,
pp 212–220
Richard E, Evans D (2006) The need for plant taxonomy in setting
priorities for designated areas and conservation management
plans: a European perspective. In: Leadlay E, Jury S (eds)
Taxonomy and plant conservation. Cambridge University Press,
New York, pp 162–176
Rohlf FJ (2008) NTSYSpc—numerical taxonomy and multivariate
analysis system, Version 2.20q. Exeter Software, New York
Rushforth KD (1989) Two new species of Abies (Pinaceae) from
western Mexico. Notes R Bot Gard Edinb 46:101–109
Rzedowski RJ (1978) Vegetacio
´
ndeMe
´
xico. Limusa, Mexico City
SAS Institute Inc. (2006) JMP v6.0. SAS Institute Inc., Cary
Scaltsoyiannes A, Tsaktsira M, Drouzas D (1999) Allozyme differ-
entiation in the Mediterranean firs (Abies, Pinaceae). A first
comparative study with phylogenetic implications. Plant Syst
Evol 216:289–307
Sneath PHA, Sokal RR (1973) Numerical taxonomy. Freeman, San
Francisco
Sokal RR (1979) Testing statistical significance of geographical
variation patterns. Syst Zool 28:227–232
St. Clair JB, Critchfield WB (1988) Hybridization of a Rocky
Mountain fir (Abies concolor) and a Mexican fir (Abies
religiosa). Paper No. 2255, Forest Research Laboratory, Oregon
State university, Corvallis
Strandby Andersen U, Prado Co
´
rdova JP, Sørensen M, Kollmann J
(2006) Conservation and utilisation of Abies guatemalensis
Rehder (Pinaceae)—an endangered endemic conifer in Central
America. Biodivers Conserv 15:3131–3151
Thorpe RS (1983) A review of the mumerical methods for recogniz-
ing and analyzing racial differentiation. In: Felsenstein J (ed)
Numerical taxonomy. Springer, Berlin, pp 404–423
Wendt T, Canela MBF, Morrey-Jones JE, Henriques AB, Rios RI
(2000) Recognition of Pitcairnia corcovadensis (Bromeliaceae)
at the species level. Syst Bot 25:389–398
Williams PH, Humphries CJ (1994) Biodiversity, taxonomic related-
ness and endemism in conservation. In: Forey PL, Humphries
CJ, Vane-Wright RI (eds) Systematics and conservation evalu-
ation. Clarendon Press, London, pp 269–287
Wu H, Hu Z (1997) Comparative anatomy of resin ducts of the
Pinaceae. Trees 11:135–143
76 U. Strandby et al.
123
