Download Genetic differentiation of island populations: geographical barrier or

bs_bs_banner
Biological Journal of the Linnean Society, 2013, 108, 68–78. With 2 figures
Genetic differentiation of island populations:
geographical barrier or a host switch?
MAXI POLIHRONAKIS RICHMOND1, SARAH JOHNSON1, TAMARA S. HASELKORN1†,
MICHELLE LAM1, LAURA K. REED2 and THERESE A. MARKOW1*
1
Division of Biological Sciences, University of California, 9500 Gilman Drive, La Jolla, San Diego,
CA 92093, USA
2
University of Alabama, Department of Biological Sciences, 300 Hackberry Lane, Tuscaloosa, AL
35487, USA
Received 14 May 2012; revised 26 June 2012; accepted for publication 26 June 2012
In the Sonoran desert, there exists a diverse community of cactophilic drosophilids that exploit toxic, rotting cactus
tissue as a food resource. The chemistry of the necrotic cactus tissue varies among species, and several drosphilid
species have evolved specialized detoxification mechanisms and a preference for certain cactus types. In the present
study, we compared the genetic structure of two columnar cactus species, Drosophila mettleri and Drosophila
mojavensis, and two prickly pear species, Drosophila mainlandi and Drosophila hamatofila, which have all recently
colonized Catalina Island off the coast of southern California. Because there are no columnar cactus species on
Catalina Island, the two columnar specialists underwent a host switch to prickly pear cactus, the only cactus
present on the island. Previous genetic studies of D. mettleri and D. mojavensis showed significant genetic
differentiation between mainland and island populations, which could result from restricted gene flow as a result
of the San Pedro Channel, or because of a host switch to prickly pear. To distinguish between these possibilities,
we analyzed the genetic structure of the prickly pear species aiming to isolate the effects of geography versus host
switching. The results obtained show little to no genetic differentiation for the prickly pear species, supporting the
hypothesis that the genetic differentiation of the two columnar species is a result of a host switch from columnar
cacti to prickly pear. © 2012 The Linnean Society of London, Biological Journal of the Linnean Society, 2013, 108,
68–78.
ADDITIONAL KEYWORDS: cactophilic – California – Channel Islands – Drosophila – host specialization.
INTRODUCTION
Insect-host plant interactions are often associated
with specific adaptations, such as specialized biochemical detoxification mechanisms, that allow
insects to feed on plants with secondary metabolites
that would otherwise be harmful (Krieger, Feeny &
Wilkinson, 1971; Karban & Agrawal, 2002; Li,
Schuler & Berenbaum, 2007; Matsuki et al., 2011). A
number of Drosophila species inhabiting the arid
regions of North America utilize necrotic cacti as
breeding sites, and the developing larvae subse-
*Corresponding author. E-mail: [email protected]
†Current address: University of Rochester, Department of
Biology, Rochester, NY 14627, USA.
68
quently feed on the rotting tissue and resident microfauna of yeast and bacteria, which is known to
be toxic to many insect species. The chemistry profiles
of the cactus tissue vary by species (more closelyrelated species have more similar chemistry) with
respect to a variety of toxic compounds, such as
triterpene glycosides, alkaloids, and sterols (Fogleman & Danielson, 2001). There is genetic evidence for
species-specific host detoxification mechanisms in
desert drosophilids and it has revealed evolutionary
changes in known detoxification genes (Matzkin,
2004, 2005, 2008; Matzkin & Eanes, 2003; Matzkin
et al., 2006; Bono et al., 2008).
In the Sonoran and Mojave Deserts of northern
Mexico, Arizona, and California, Drosophila mojavensis and Drosophila mettleri are two well-studied
© 2012 The Linnean Society of London, Biological Journal of the Linnean Society, 2013, 108, 68–78
GENETIC DIFFERENTIAITION OF ISLAND POPULATIONS
cactophilic drosophilid species with a broad geographic distribution. Each species is associated with a
specific cactus species in different parts of its range.
Drosophila mojavensis breeds in necrotic cactus
tissue and D. mettleri breeds in the soil soaked with
the necrotic juice surrounding the cacti (Heed, 1982,
1989). Previous studies of D. mojavensis have shown
it to be a relatively strict specialist with four geographically isolated subspecies each feeding on a
different cactus host: columnar cacti (Stenocereus
gummosus and Stenocereus thurberi), barrel cactus
(Ferocactus cylindraceus), and prickly pear (Opuntia
spp.) (Heed, 1978; Heed & Mangan, 1986; Ruiz &
Heed, 1988; Ruiz, Heed & Wasserman, 1990; Reed,
Nyboer & Markow, 2007; Pfeiler, Castrezana & Reed,
2009). On the other hand, with the exception of the
Catalina Island population, D. mettleri has little to no
genetic structure throughout its mainland range, and
has been found in association with columnar cactus
species, Carnegiea gigantean, Pachychereus pringylei,
S. thurberi, and Lophocereus schottii, as well as the
California barrel cactus (F. cylindraceus) in the most
northern part of its range (Heed, 1982; Markow, Castrezana & Pfeiler, 2002; Hurtado et al., 2004). Thus,
these two species represent two cactophilic specialists: D. mojavensis, a relatively strict specialist,
comprising geographically and genetically isolated
populations that each specialize on different cactus
species, and D. mettleri, a moderate specialist with a
contiguous population (Bono et al., 2008).
Interestingly, both D. mojavensis and D. mettleri
have been collected on Santa Catalina Island, off the
coast of southern California, where the only cacti
available as hosts are native, and non-native, species
of prickly pear (Opuntia spp.) (Millspaugh & Nuttall,
1923; Heed, 1982). This is the only portion of the
range where each of these species feed on prickly
pear, even though it is readily available throughout
their respective distributions (Heed, 1982). Both Drosophila species thus have switched host plants on this
island and both have been reported to show significant genetic differentiation from con-specific populations on the mainland (Hurtado et al., 2004; Reed
et al., 2007; Haselkorn, Markow & Moran, 2009). It is
unclear whether this genetic structure reflects single
colonization events as a result of the inherent difficulties associated with a host switch, or if the San
Pedro Channel between Catalina Island and the
mainland is sufficient to reduce or prevent gene flow.
One approach for evaluating the influences of host
shifts and founder events on the genetics of the
insular D. mojavensis and D. mettleri would be to
examine additional cactophilic Drosophila species
that breed in prickly pear, and are found on both sides
of the channel. Several other cactophilic Drosophila
species are found both on Santa Catalina Island and
69
the North American mainland. Drosophila mainlandi
and Drosophila hamatofila have been reared from a
number of different prickly pear species throughout
their ranges (Patterson, 1943; Wasserman & Wasserman, 1992; Oliveira et al., 2005). Although D. mainlandi is restricted in its distribution to southern
California and the Baja California peninsula, D. hamatofila appears to be the most widespread of the
cactophilic drosophilids (Fig. 1) (Oliveira et al., 2005).
In the present study, we use these four species to
examine patterns of genetic differentiation associated
with island colonization, and to isolate the effect of
host shifts and barriers to gene flow presented by
the 40-km San Pedro Channel off the coast of southern California. If the Catalina Island populations
of prickly pear feeders, such as D. mainlandi and
D. hamatofila, have patterns of genetic structure
similar to the columnar species D. mojavensis and
D. mettleri, this would suggest that the San Pedro
Channel is a barrier to gene flow for these drosophilid
species. On the other hand, if the Catalina Island
populations of the two prickly pear species are not
genetically differentiated, it will suggest the channel
is not a significant barrier to gene flow, and that other
factors, such as ecological adaptation required for a
host switch, are contributing to the genetic isolation
between island and mainland populations of D. mojavensis and D. mettleri.
MATERIAL AND METHODS
FOCAL SPECIES: D. MOJAVENSIS, D. METTLERI,
D. MAINLANDI, AND D. HAMATOFILA
Detailed phylogeographic treatments of D. mojavensis
and D. mettleri have previously been published
(Hurtado et al., 2004; Reed et al., 2007; Haselkorn
et al., 2009). For the present study, we added data for
two additional Drosophila repleta group species:
D. mainlandi and D. hamatofila. All four of these
cactophilic Drosophila species are found in the
Sonoran Desert (Fig. 1) and have established populations on one or more of the California Channel
Islands, including Catalina Island.
SAMPLING
AND DATA COLLECTION
Taxon sampling was focused in southern California,
northern Baja California, Mexico, and Organ Pipe
National Monument in Arizona. All newly-collected
specimens were obtained using traps baited with a
banana and yeast mixture. Genomic DNA was
extracted from single whole flies of field-collected
specimens using DNeasy Tissue extraction kits
(Qiagen), and squish preps. A portion of the cytochrome c oxidase subunit I (COI) gene was amplified
using primers LCO1490f/HCO2198r in accordance
© 2012 The Linnean Society of London, Biological Journal of the Linnean Society, 2013, 108, 68–78
70
M. P. RICHMOND ET AL.
D. mojavensis
D. mettleri
D. mainlandi
D. hamatofila
Figure 1. Distribution map for each of the four species in the south-western USA and north-western Mexico.
with standard polymerase chain reaction (PCR) conditions (Folmer et al., 1994). PCR products were purified using QIAquick PCR Purification kits (Qiagen)
and ExoSAP-IT® (USB Corp.). Forward and reverse
sequencing reactions and sequence visualization was
performed by Genewiz, Inc., and DNA sequences were
edited and aligned in SEQUENCHER, version 4.8
(GeneCodes Corp.). All gene sequences were translated to amino acids to check for stop codons, and
aligned manually (no indels were present).
STATISTICAL
ANALYSIS
The following genetic diversity indices were estimated
for each species in DNAsp, version 5.10.01 (Librado
& Rozas, 2009): haplotype diversity (Nei, 1987),
nucleotide diversity (p) (Nei, 1987), qs per site (Watterson, 1975), and the mean number of nucleotide
differences (k) (Tajima, 1996). To analyze population
structure and demography, we conducted several
analyses using ARLEQUIN, version 3.5.1.3 (Excoffier
& Lischer, 2010). For each species, all haplotypes
were grouped according to four main geographical
regions (populations): Mainland (including San Diego,
California, and Organ Pipe National Monument,
Arizona), Peninsular (northern Baja California Peninsula, Mexico), Catalina Island, and Santa Cruz Island
(D. mainlandi only). To identify limits to gene flow
and population structure between the mainland and
island populations, we computed pairwise FST values
(Excoffier, Smouse & Quattro, 1992) between all populations. To test for significance, the data were permuted in Arlequin following the null hypothesis of no
differences between populations such that the P-value
represents the proportion of permutations (N = 1000)
where the FST value is equal to or greater than the
observed value.
To determine the source of variation (within versus
among populations), and estimate the covariance
components (Va and Vb) for each species, we performed an analysis of molecular variance (Excoffier
et al., 1992). Significance of Va (variation among populations) was tested using 1000 permutations of haplotypes among populations.
Phylogenetic analysis of COI for each taxon was
performed using MrBayes, version 3.1 (Huelsenbeck
& Ronquist, 2001) to infer relationships among
individuals. Models of molecular evolution for each
© 2012 The Linnean Society of London, Biological Journal of the Linnean Society, 2013, 108, 68–78
71
GENETIC DIFFERENTIAITION OF ISLAND POPULATIONS
marker were evaluated in JMODELTEST, version
0.1.1 (Posada, 2008) using the Akaike information
criterion (AIC). Gene trees of all unique haplotypes
were inferred for each species in MrBayes, imposing
the model specified by the AIC, using default priors.
We also constructed haplotype networks for each
species using TCS, version 1.21 (Clement, Posada &
Crandall, 2000) to obtain a nonbifurcating perspective
of relationships. We used the default settings of a 95%
connection limit. Because of the close relationship of
all haplotypes within species, only the TCS networks
are presented here because they provide more
detailed information than the gene trees.
RESULTS
SAMPLING
AND DATA COLLECTION
Datasets compiled for each species included
sequences from GenBank, as well as newly-sequenced
specimens: 601 bp of COI for 46 D. mainlandi specimens from Mainland, Catalina Island, and Santa
Cruz Island (GenBank Accession numbers JX489172–
JX489217); 597 bp of COI for 32 D. hamatofila specimens from Mainland and Catalina Island (Genbank
Accession numbers JX492964–JX492995); 656 bp of
COI for 117 D. mettleri specimens from Mainland,
Baja Peninsula, and Catalina Island (Hurtado et al.,
2004) (GenBank Accession numbers AY533789–
AY533812); and 570 bp of COI for 190 D. mojavensis
specimens from Mainland, Baja Peninsula, and Catalina Island (Reed et al., 2007; Haselkorn et al., 2009)
(GenBank Accession numbers DQ383686–DQ383730
and FJ656811–FJ656997). Note that only sequences
from populations pertinent to the present study were
used: CI, ANZA, COLN, SANQ, ROSO, and SARO,
sensu Reed et al. (2007), and OPNM and CI, sensu
Haselkorn et al. (2009). Sample sizes for all populations within each species were at 15 or greater
(Table 1).
Table 1. Sample sizes and genetic diversity indices
Species
Drosophila mainalndi
(Total)
Drosophila mainlandi
(Mainland)
Drosophila mainlandi
(Catalina Island)
Drosophila mainlandi
(Santa Cruz Island)
Drosophila hamatofila
(Total)
Drosophila hamatofila
(Mainland)
Drosophila hamatofila
(Catalina Island)
Drosophila mettleri (Total)
Drosophila mettleri
(Mainland)
Drosophila mettleri
(Catalina Island)
Drosophila mettleri (Baja)
Drosophila mojavensis
(Total)
Drosophila mojavensis
(Mainland)
Drosophila mojavensis
(Catalina Island)
Drosophila mojavensis
(Baja)
Length
(bp)
Polymorphic
sites (parsinomy
informative)
Haplotypes
observed
Haplotype
diversity
p
qs
k
46
601
17 (5)
18
0.787
0.00218
0.00644
1.313
15
601
9 (1)
9
0.800
0.00219
0.00461
1.314
16
601
9 (2)
9
0.767
0.00283
0.00451
1.7
15
601
3 (0)
4
0.371
0.00067
0.00154
0.4
32
597
39 (18)
25
0.978
0.00977
0.01622
5.833
16
597
21 (12)
12
0.942
0.00872
0.0106
5.208
16
597
31 (10)
15
0.992
0.01052
0.01565
6.283
117
45
656
656
25 (10)
13 (5)
24
12
0.716
0.582
0.00209
0.00149
0.00714
0.00453
1.373
0.98
16
656
2 (1)
3
0.575
0.00095
0.00092
0.625
56
156
656
587
14 (4)
32 (23)
14
24
0.664
0.848
0.00141
0.00985
0.00465
0.00969
0.928
5.71
88
587
26 (21)
16
0.711
0.01031
0.00877
6.055
34
587
1 (1)
2
0.114
0.00019
0.00042
0.114
34
587
10 (2)
8
0.674
0.00255
0.00417
1.497
N
p, nucleotide diversity; qs, Watterson estimator per site; k, mean number of nucleotide differences.
© 2012 The Linnean Society of London, Biological Journal of the Linnean Society, 2013, 108, 68–78
72
M. P. RICHMOND ET AL.
Table 2. Pairwise FST values for Drosophila hamatofila
Catalina Island
Catalina Island
Mainland
0
0.028
Table 5. Pairwise FST values for Drosophila mainlandi
Mainland
0
Catalina Island
Mainland
Santa Cruz Island
Catalina
Island
Mainland
Santa Cruz
Island
0
0.169*
0.271*
0
-0.010
0
Table 3. Pairwise FST values for Drosophila mettleri
Catalina
Island
Catalina Island
Mainland
Baja Peninsula
0
0.678*
0.698*
Mainland
0
0.008
Baja
Peninsula
0
Significant comparisons (P < 0.05) are denoted by an
asterisk (*).
Table 4. Pairwise FST values for Drosophila mojavensis
Catalina Island
Mainland
Baja Peninsula
Catalina
Island
Mainland
Baja
Peninsula
0
0.506*
0.871*
0
0.228*
0
Significant comparisons (P < 0.05) are denoted by an
asterisk (*).
POPULATION
GENETICS AND DEMOGRAPHY
Of the four species included in the analysis, D. mojavensis and D. hamatofila had the highest genetic
diversity (Table 1). This result was particularly interesting for D. hamatofila because this species had the
smallest sample size of all four species, which means
that almost every individual had a unique haplotype.
Both the Mainland and Catalina Island populations
of D. hamatofila had comparable genetic diversity,
whereas, in D. mojavensis, the Mainland population
was much more diverse than the Catalina Island
population. Genetic diversity of D. mettleri and
D. mainlandi was relatively low, with the Mainland
and Peninsular (D. mettleri only) populations harbouring most of the genetic diversity.
For all species except D. hamatofila, the Catalina
Island population showed reduced gene flow with all
other populations (Tables 2, 3, 4, 5). Pairwise FST
values between Catalina Island and Mainland/
Peninsular populations were significant for D. mettleri, D. mojavensis, and D. mainlandi. Interestingly,
the D. mainlandi population collected from Santa
Cruz Island was isolated from the Catalina Island
population but not from the Mainland. For two species,
Significant comparisons (P < 0.05) are denoted by an
asterisk (*).
D. mettleri and D. mojavensis, which had samples from
both Mainland and Peninsular populations, there was
evidence of reduced gene flow between these populations for D. mojavensis but not for D. mettleri.
Results of the analysis of molecular variance
(Table 6) revealed that most of the genetic variance
for D. hamatofila was based on within population
variation (97.16%), with very little diversity originating from variation among populations (2.84%). A
similar result was found for D. mainlandi, which harboured 82.89% of the genetic variation within populations and only 17.11% among populations. Both
D. mettleri and D. mojavensis exhibited approximately equivalent levels of variation within and
among populations.
Haplotype networks for all species were connected
using the 95% connection limit indicating a close
relationship of haplotypes within species (Fig. 2). The
networks for D. mettleri, D. mainlandi, and D. mojavensis were structured similarly in that they were
dominated by a high frequency of a few haplotypes,
with several singleton haplotypes derived from these.
Dominant haplotypes for D. mettleri were shared
between the Baja Peninsula and Mainland. Dominant
haplotypes for D. mojavensis primarily comprised
individuals from a single locality. The two distinct
D. mojavensis Mainland groups comprise Drosophila
mojavensis mojavensis from the Anza Borrego Desert,
California, and Drosophila mojavensis sonorensis
from Organ Pipe National Monument, Arizona. The
dominant haplotype for D. mainlandi was found in all
four localities sampled (Catalina Island, Mainland,
Baja Peninsula, and Santa Cruz Island), suggesting a
relatively recent colonization of Catalina and Santa
Cruz Islands. The D. hamatofila network was very
different from the other three in that there were no
dominant haplotypes, no apparent geographical structure, and a greater number of missing haplotypes.
Analyses of relationships based on the COI haplotype networks showed relatively high levels of divergence for D. mettleri and D. mojavensis, providing
further support for strong, and almost complete
localization and isolation of Catalina Island populations. For both these species, the haplotypes sampled
© 2012 The Linnean Society of London, Biological Journal of the Linnean Society, 2013, 108, 68–78
GENETIC DIFFERENTIAITION OF ISLAND POPULATIONS
73
Table 6. Results of the analysis of molecular variance
Source of variation
Species
Among populations
within groups
Drosophila
Drosophila
Drosophila
Drosophila
Within populations
Drosophila
Drosophila
Drosophila
Drosophila
Sum of
squares
Variance
components
Percentage
of variation
hamatofila
mettleri
mainlandi
mojavensis
4.219
27.863
4.793
195.611
0.08411 Va
0.38125* Va
0.11883* Va
1.68548* Va
2.84
45.64
17.11
47.07
hamatofila
mettleri
mainlandi
mojavensis
86.188
51.761
24.750
326.984
2.87292
0.45404
0.57558
1.89526
97.16
54.36
82.89
52.93
Vb
Vb
Vb
Vb
Significant among population variance values (P < 0.05) are denoted by an asterisk (*).
from Catalina Island were unique and not shared
with any other populations. Furthermore, all D. mettleri and D. mojavensis Catalina haplotypes arose
from a single ancestor, supporting a single colonization of Catalina Island.
Both D. hamatofila and D. mainlandi had shared
haplotypes between Catalina Island and the other
populations sampled. Drosophila hamatofila had two
shared haplotypes between Catalina and the Mainland, and 13 unique haplotypes on Catalina Island.
For D. mainlandi, there were two shared Catalina
Island haplotypes: one was the dominant haplotype of
this species collected in all four populations sampled,
the other was identical to a Mainland haplotype.
There were seven unique D. mainlandi haplotypes on
Catalina Island.
Taken together, genetic analysis of the mitochondrial COI gene indicated that the two columnar
species, D. mettleri and D. mojavensis, revealed more
geographical structure, such that the major regions
were dominated by closely-related haplotypes. On the
other hand, the two prickly pear species, D. hamatofila and D. mainlandi, demonstrated less isolation
among the major geographical regions, with evidence
for gene flow between Catalina Island and Mainland
populations.
DISCUSSION
Analyzing the phylogeographical structure of cactophilic drosophilids on Catalina Island provides an
opportunity to test whether the 40-km San Pedro
Channel is a substantial barrier to gene flow, or
whether other factors determine genetic differentiation, such as evolving the necessary adaptations for
a successful host switch. Based on the mitochondrial DNA sequence data reported in the present
study, populations of the columnar cactus species
D. mojavensis and D. mettleri showed more genetic
isolation on Catalina Island than the prickly pear
cactus species D. mainlandi and D. hamatofila. This
result is interpreted as evidence for a role of ecological specialization in the genetic structuring of
D. mojavensis and D. mettleri.
ISLAND
COLONIZATION
The genetic data reported in the present study
suggest a single colonization of Catalina Island by
both D. mettleri and D. mojavensis. All of the mitochondrial haplotypes sampled from Catalina Island
for both species arose from a single ancestor, and
there was no evidence for gene flow between Catalina
Island and either of the two mainland populations.
This result parallels that of previous studies of both
species based on nuclear and mitochondrial DNA
sequence data (Markow et al., 2002; Hurtado et al.,
2004; Reed et al., 2007). Although the origin of the
D. mojavensis population that colonized Catalina
appears to be from Baja, it is difficult to infer the
origin of D. mettleri on Catalina Island because there
is no genetic structure between the Mainland and
Baja Peninsula populations (Fig. 2).
As a result of evidence of gene flow between Catalina Island and the Mainland/Baja Peninsula populations for both prickly pear species, D. hamatofila and
D. mainlandi, it does not appear the ocean channel
operates as a genetic isolating mechanism for these
drosophilids. This result is consistent with phylogeographical studies of other cactophilic drosophilid
species, including D. mettleri, which support longrange dispersal (approximately 120 km) across the
Sea of Cortez (Hurtado et al., 2004). Interestingly,
compared with the Catalina Island data, the dispersal
distance between Catalina Island and the mainland is
shorter, although the genetic differentiation is higher.
Because prickly pear is the only cactus host available
on Catalina Island, this may be a result of increased
selection pressure and decreased likelihood of a
founder event.
© 2012 The Linnean Society of London, Biological Journal of the Linnean Society, 2013, 108, 68–78
74
M. P. RICHMOND ET AL.
2
D. mettleri
D. mojavensis
61
4
14
18
45
4
3
5
3
2
12
32
7
2
9
3
2
6
3
9
Catalina Island
Mainland
Baja Peninsula
Santa Cruz Island
D. hamatofila
D. mainlandi
2
3
2
4
2
20
2
8
Figure 2. TCS Haplotype networks of cytochrome c oxidase subunit I COI for each of the four species. Colours denote
population origin; the size of circles is proportional to the haplotype frequency. The number of individuals with a
particular haplotype, if greater than one, is provided next to the haplotype circle.
Alternatively, it is possible that D. mojavensis and
D. mettleri colonized the island prior to D. hamatofila and D. mainlandi, thus increasing the time
available for genealogical sorting. However, if we
were to calibrate the number of nucleotide substitutions in each species to a molecular clock, it is
clear that D. mettleri colonized Catalina Island more
recently than D. mojavensis as a result of this population being one mutational step from a mainland
population, whereas D. mojavensis on Catalina
Island is five mutational steps from the mainland
population. We do not consider that this is an effect
of sampling as a result of the large sample size of
D. mojavensis on Catalina Island relative to the
number of haplotypes observed. Thus, it does not
appear that the results obtained in the present
study reflect the amount of time since each species
colonized Catalina Island.
© 2012 The Linnean Society of London, Biological Journal of the Linnean Society, 2013, 108, 68–78
GENETIC DIFFERENTIAITION OF ISLAND POPULATIONS
GENETIC
ISOLATION AND HOST SWITCHES
Drosophila mojavensis and D. mettleri each made
a large host switch from having their primary associations with columnar (S. gummosus, S. thurberi,
and L. schotii) and barrel (F. cylindraceus) species
to prickly pear (Opuntia spp.), when they moved to
Catalina Island. That this is a significant host jump is
supported by differences in the chemical composition and yeast communities of the different cactus
species (Heed et al., 1976; Kircher, 1982; Fogleman
& Danielson, 2001). Additional evidence comes from
previous work quantifying host-plant specificity of
cactophilic drosophilids in Sonora and Sinaloa,
Mexico, which showed a discrete difference between
the Drosophila species reared from columnar cactus
species versus those reared from prickly pear (Ruiz &
Heed, 1988). Out of 570 D. mojavensis specimens collected emerging from cactus rots of columnar and
prickly pear species collected sympatrically in Playa
Cochorit, Sonora, Mexico, only 2.8% were from the
prickly pear species. This result demonstrates that, in
this population of D. mojavensis, where both prickly
pear and columnar cactus resources are available, the
preferred host for D. mojavensis was the columnar
cactus species.
This result is particularly interesting because many
columnar specialists, such as D. mojavensis and
D. mettleri, were successfully reared in prickly pear
in laboratory experiments (Ruiz & Heed, 1988). One
suggested explanation for this success is the higher
water content and elevated levels of free sugars in
prickly pear species, as well as lower alkaloid levels
(Kircher, 1982; Ruiz & Heed, 1988). Prickly pear and
columnar species are not close relatives in the cactus
phylogeny (Gibson & Horak, 1978; Cota & Wallace,
1997), and thus it is not unusual for them to have
different levels of toxicity and nutrient resources
(Fogleman & Danielson, 2001). In a recent study of a
parallel system of cactophilic drosophilids in South
America, Soto et al. (2012) suggested that host shifts
of columnar specialists to prickly pear represent an
easier transition as a result of lower toxicity of the
prickly pear cactus tissue. Interestingly, it has been
hypothesized that prickly pear species are the ancestral host for the D. mulleri complex of cactophilic
drosophilids containing D. mettleri and D. mojavensis
(Ruiz & Heed, 1988). Thus, a host switch for D. mojavensis and D. mettleri from columnars to prickly
pear would represent a return to the ancestral condition, and may not present as many challenges as a
switch to a novel host.
However, host specificity in Drosophila is a result of
adaptive specialization of many life-history aspects,
such as olfactory cues to find hosts, oviposition preferences, and larval feeding preferences (Markow &
75
O’Grady, 2008). For example, larval feeding rates of
D. mojavensis from the Sonoran Desert were significantly higher when feeding on S. thurberi (i.e. the
preferred host in this region) versus Opuntia littoralis
(Craft, 2010). Such specialization could explain why
other columnar cactophilic specialists such as Drosophila nigrospiracula have not expanded their range
into southern California. For this species, a northward range expansion into southern California would
require a host switch to prickly pear or barrel cactus,
which has never been observed. Thus, even if this
switch is to a less toxic environment and represents a
return to the ancestral condition, our results suggest
that the evolution of derived adaptations may still
pose a significant ecological challenge (Futuyma,
Keese & Funk, 1995). This was also found at the
genetic level in a study by Matzkin (2012), which
showed differential expression of detoxification genes
in a derived D. mojavensis population when reared on
an alternate host hypothesized to be the ancestral
cactus species for the D. mojavensis species cluster.
On the other hand, one could easily speculate
that prickly pear species such as D. mainlandi and
D. hamatofila are pre-adapted to move among prickly
pear host species, and thus are not restricting the
colonization of Catalina Island to episodic founder
events. The genetic data that we present supported
this scenario and provided evidence of ongoing gene
flow between the Mainland and Catalina Island populations for D. hamatofila, and a low, albeit significant,
FST value for D. mainlandi.
Ecological specialization of D. mojavensis on Catalina Island occurred < 0.5 Mya (Reed et al., 2007;
Matzkin, 2008), and has resulted in the evolution of
an island endemic that is currently recognized as one
of four D. mojavensis subspecies based on morphological, genetic, and behavioural data (Pfeiler et al., 2009;
Richmond, Johnson & Markow, 2012). Because this
species represents a model system for studies in speciation and ecological specialization, the evolutionary
history of D. mojavensis on Catalina is better understood than that of D. mettleri. In one investigation of
D. mettleri from Catalina Island and mainland populations, there was some evidence for early stages of
post-zygotic isolation (Markow et al., 2002). Further
investigation into D. mettleri on Catalina is warranted aiming to test the hypothesis that this population is in the early stages of becoming an additional
island endemic. In addition, because the genetic
underpinnings of detoxification in D. mettleri have
been shown to rely on P450 genes (Danielson et al.,
1997, 1998), it would be particularly interesting to
examine the evolutionary history of these genes in the
D. mettleri populations on Catalina Island.
Ongoing work investigating arthropod diversity of
California’s Channel Islands has revealed that these
© 2012 The Linnean Society of London, Biological Journal of the Linnean Society, 2013, 108, 68–78
76
M. P. RICHMOND ET AL.
islands support levels of diversity comparable to
mainland southern California, and may serve as
a necessary sink for diversity in the face of developmental pressure on the mainland (Rubinoff &
Powell, 2004; Chatzimanolis, Norris & Caterino,
2010; Chatzimanolis, Caterino & Richmond, unpubl.
data). A recent study investigating the patterns of
genetic structure of four beetle species on California’s
Channel Islands revealed varying degrees of gene
flow between the islands, and between the islands
and the mainland (Chatzimanolis, Caterino & Richmond, unpubl. data). The genetic structure for all
species was suggestive of multiple founder events,
and paralleled the varying life-history traits of the
beetles investigated. For example, only one beetle
species studied was capable of flight, and this species
had the largest number of island colonization events,
in line with the results of the Drosophila species that
we report in the present study.
CONCLUSIONS
By comparing patterns of genetic differentiation
between prickly pear and columnar cactophilic drosophilid species, we were able to test how host shifts
influence patterns of genetic structure. The present
study indicates that restricted gene flow across the
San Pedro Channel between Catalina Island and
mainland southern California for two columnar
cactus specialists is a result of ecological specialization. Further work testing pre- and post-zygotic
reproductive isolation between mainland and Catalina Island populations of D. mettleri is necessary to
determine whether this population represents an offshoot from the contiguous distribution of this species,
and is in the early stages of becoming an island
endemic. Additional investigation using nuclear loci
would provide a means to supplement the COI data
with neutral markers, as well as test candidate genes
involved in the specialization process.
ACKNOWLEDGEMENTS
We would like to thank the Santa Catalina Island
Conservancy for research and collecting permits, as
well as Darcee Guttilla and Fred Starkey for their
invaluable assistance with the collection. We would
also like to thank David Holway for contributing
specimens of D. mainlandi sampled from Santa Cruz
Island. Last, we thank Giovanni Hanna, Hiroto Kameyama, Osman Tarin, Harman Singh, and three
anonymous reviewers for their helpful comments.
This work was funded by the National Science Foundation (DBI-1051420 to TAM and MPR and DEB0315815 to TAM).
REFERENCES
Bono JM, Matzkin LM, Castrezana S, Markow TA.
2008. Molecular evolution and population genetics of
two Drosophila mettleri cytochrome P450 genes involved
in host plant utilization. Molecular Ecology 17: 3211–
3221.
Chatzimanolis S, Norris LA, Caterino MS. 2010. Multiisland endemicity: phylogeography and conservation of
Coelus pacificus (Coleoptera: Tenebrionidae) darkling
beetles on the California Channel Islands. Annals of the
Entomological Society of America 103: 785–795.
Clement M, Posada D, Crandall K. 2000. TCS: a computer
program to estimate gene genealogies. Molecular Ecology 9:
1657–1659.
Cota JH, Wallace RS. 1997. Chloroplast DNA evidence for
divergence in Ferocactus and its relationships to North
American columnar cacti (Cactaceae: Cactoideae). Systematic Botany 22: 529–542.
Craft J. 2010. Feeding rates of Drosophila mojavensis
sonorensis on native and non-native hosts. Drosophila
Information Service 93: 47–49.
Danielson PB, Foster JLM, McMahill MM, Smith MK,
Fogleman JC. 1998. Induction by alkaloids and penobarbital of family 4 cytochrome P450s in Drosophila: evidence
for involvement in host plant utilization. Molecular and
General Genetics 259: 54–59.
Danielson PB, MacIntyre R, Fogleman J. 1997. Molecular
cloning of a family of xenobiotic-inducible drosophilid cytochrome p450s: evidence for involvement in host-plant allelochemical resistance. Proceedings of the National Academy
of Sciences of the United States of America 94: 10797–
10802.
Excoffier L, Lischer HEL. 2010. Arlequin suite ver 3.5: a
new series of programs to perform population genetics
analyses under Linux and Windows. Molecular Ecology
Resources 10: 564–567.
Excoffier L, Smouse PE, Quattro JM. 1992. Analysis of
molecular variance inferred from metric distances among
DNA haplotypes: application to human mitochondrial DNA
restriction data. Genetics 131: 479–491.
Fogleman J, Danielson P. 2001. Chemical interactions in
the cactus–microorganism–Drosophila model system of the
Sonoran Desert. American Zoologist 41: 877–889.
Folmer O, Black M, Hoeh W, Lutz R, Vrijenhoek R. 1994.
DNA primers for amplification of mitochondrial cytochrome
c oxidase subunit I from metazoan invertebrates. Molecular
Marine Biology and Biotechnology 3: 294–299.
Futuyma DJ, Keese MC, Funk DJ. 1995. Genetic
constraints on macroevolution: the evolution of host
affiliation in the leaf beetle genus Ophraella. Evolution 49:
797–809.
Gibson AC, Horak KE. 1978. Systematic anatomy and phylogeny of Mexican columnar cacti. Annals of the Missouri
Botanical Garden 65: 999–1057.
Haselkorn TS, Markow TA, Moran NA. 2009. Multiple
introductions of the Spiroplasma bacterial endosymbiont
into Drosophila. Molecular Ecology 18: 1294–1305.
© 2012 The Linnean Society of London, Biological Journal of the Linnean Society, 2013, 108, 68–78
GENETIC DIFFERENTIAITION OF ISLAND POPULATIONS
Heed WB. 1978. Ecology and genetics of Sonoran desert
Drosophila. In: Brussard PF, ed. Ecological genetics: the
interface. New York, NY: Springer-Verlag, 109–126.
Heed WB. 1982. The origin of Drosophila in the sonoran
desert. In: Barker JSF, Starmer WT, eds. Ecological genetics and evolution: the cactus–yeast–Drosophila model system.
New York, NY: Academic Press, 65–80.
Heed WB. 1989. Origin of Drosophila of the Sonoran Desert
revisited: in search of a founder event and the description of
a new species in the eremophila complex. In: Giddings LV,
Kaneshiro KY, Anderson WW, eds. Genetics, speciation and
the founder principle. New York, NY: Oxford University
Press.
Heed WB, Mangan RL. 1986. Community ecology of the
Sonoran Desert Drosophila. In: Ashburner M, Carson HL,
Thompson JN Jr, eds. The genetics and biology of Drosophila, Vol. 3e. 311–345. London: Academic Press.
Heed WB, Starmer WT, Miranda M, Miller MW, Phaff
HJ. 1976. Analysis of yeast flora associated with cactiphilic
Drosophila and their host plants in Sonoran Desert and its
relation to temperate and tropical associations. Ecology 57:
151–160.
Huelsenbeck JP, Ronquist F. 2001. MrBayes: Bayesian
inference of phylogenetic trees. Bioinformatics 17: 754–
755.
Hurtado LA, Erez T, Castrezana S, Markow TA. 2004.
Contrasting population genetic patterns and evolutionary
histories among sympatric Sonoran Desert cactophilic Drosophila. Molecular Ecology 13: 1365–1375.
Karban R, Agrawal AA. 2002. Herbivore offense. Annual
Review of Ecology and Systematics 33: 641–664.
Kircher HW. 1982. Chemical composition of cacti and its
relationship to Sonoran Desert Drosophila. In: Barker JSF,
Starmer WT, eds. Ecological genetics and evolution: the
cactus–yeast–Drosophila model system. New York, NY:
Academic Press, 143–158.
Krieger RI, Feeny PP, Wilkinson CF. 1971. Detoxication
enzymes in the guts of caterpillars: an evolutionary answer
to plant defenses? Science 172: 579–581.
Li XC, Schuler MA, Berenbaum MR. 2007. Molecular
mechanisms of metabolic resistance to synthetic and
natural xenobiotics. Annual Review of Entomology 52: 231–
253.
Librado P, Rozas J. 2009. DnaSP v5: a software for
comprehensive analysis of DNA polymorphism data. Bioinformatics 25: 1451–1452.
Markow TA, Castrezana S, Pfeiler E. 2002. Flies across
the water: genetic differentiation and reproductive isolation
in allopatric desert Drosophila. Evolution 56: 546–552.
Markow TA, O’Grady PM. 2008. Reproductive ecology of
Drosophila. Functional Ecology 22: 747–759.
Matsuki M, Kay N, Serin J, Scott JK. 2011. Variation in
the ability of larvae of phytophagous insects to develop on
evolutionarily unfamiliar plants: a study with gypsy moth
Lymantria dispar and Eucalyptus. Agricultural and Forest
Entomology 13: 1–13.
Matzkin LM. 2004. Population genetics and geographic
variation of alcohol dehydrogenase (Adh) paralogs and
77
glucose-6-phosphate dehydrogenase (G6pd) in Drosophila
mojavensis. Molecular Biology and Evolution 21: 276–
285.
Matzkin LM. 2005. Activity variation in alcohol dehydrogenase paralogs is associated with adaptation to cactus host
use in cactophilic Drosophila. Molecular Ecology 14: 2223–
2231.
Matzkin LM. 2008. The molecular basis of host adaptation in
cactophilic Drosophila: molecular evolution of a glutathione
s-transferase gene (GstD1) in Drosophila mojavensis.
Genetics 178: 1073–1083.
Matzkin LM. 2012. Population transcriptomics of cactus host
shifts in Drosophila mojavensis. Molecular Ecology 21:
2428–2439.
Matzkin LM, Eanes WF. 2003. Sequence variation of alcohol
dehydrogenase (Adh) paralogs in cactophilic Drosophila.
Genetics 163: 181–194.
Matzkin LM, Watts TD, Bitler BG, Machado CA, Markow
TA. 2006. Functional genomics of cactus host shifts in
Drosophila mojavensis. Molecular Ecology 15: 4635–
4643.
Millspaugh CF, Nuttall LW. 1923. Flora of Santa Catalina
Island (California). Field Museum of Natural History 212:
1–454.
Nei M. 1987. Molecular evolutionary genetics. New York, NY:
Columbia University Press.
Oliveira DCSG, O’Grady PM, Etges WJ, Heed WB,
DeSalle R. 2005. Molecular systematics and geographical
distribution of the Drosophila longicornis species complex
(Diptera : Drosophilidae). Zootaxa 1069: 1–32.
Patterson JT. 1943. The Drosophilidae of the Southwest.
University of Texas Publications 4313: 7–216.
Pfeiler E, Castrezana S, Reed LK. 2009. Genetic, ecological and morphological differences among populations
of the cactophilic Drosophila mojavensis from southwestern USA and northwestern Mexico, with descriptions of
two new subspecies. Journal of Natural History 43: 923–
938.
Posada D. 2008. jModelTest: phylogenetic model averaging.
Molecular Biology and Evolution 25: 1253–1256.
Reed LK, Nyboer M, Markow TA. 2007. Evolutionary relationships of Drosophila mojavensis geographic host races
and their sister species Drosophila arizonae. Molecular
Ecology 16: 1007–1022.
Richmond MP, Johnson S, Markow TA. 2012. Evolution of
reproductive morphology among recently diverged taxa in
the Drosophila mojavensis species cluster. Ecology and
Evolution 2: 397–408.
Rubinoff D, Powell JA. 2004. Conservation of fragmented
small populations: endemic species persistence on California’s smallest channel island. Biodiversity and Conservation
13: 2537–2550.
Ruiz A, Heed WB. 1988. Host-plant specificity in the cactophilic Drosophila mulleri species complex. Journal of
Animal Ecology 57: 237–249.
Ruiz A, Heed WB, Wasserman M. 1990. Evolution of the
Mojavensis cluster of cactophilic Drosophila with descriptions of two new species. Journal of Heredity 81: 30–42.
© 2012 The Linnean Society of London, Biological Journal of the Linnean Society, 2013, 108, 68–78
78
M. P. RICHMOND ET AL.
Soto EM, Goenaga J, Hurtado JP, Hasson E. 2012. Oviposition and performance in natural hosts in cactophilic
Drosophila. Evol. Ecol. 26: 975–990.
Tajima F. 1996. The amount of DNA polymorphism maintained in a finite population when the neutral mutation rate
varies among sites. Genetics 143: 1457–1465.
Wasserman M, Wasserman F. 1992. Inversion polymorphism in island species of Drosophila. Evolutionary Biology
26: 351–381.
Watterson GA. 1975. On the number of segregating sites in
genetical models without recombination. Theoretical Population Biology 20: 256–276.
© 2012 The Linnean Society of London, Biological Journal of the Linnean Society, 2013, 108, 68–78