Biological Conservation 169 (2014) 100–108

Contents lists available at ScienceDirect

Biological Conservation
journal homepage: www.elsevier.com/locate/biocon

Conservation genetics of Australasian sailfin lizards: Flagship species

threatened by coastal development and insufficient protected area
coverage
Cameron D. Siler a,⇑, Andrés Lira-Noriega b, Rafe M. Brown b
a
Sam Noble Oklahoma Museum of Natural History and Department of Biology, University of Oklahoma, Norman, OK 73072-7029, USA
b
Biodiversity Institute and Department of Ecology and Evolutionary Biology, University of Kansas, Lawrence, KS 66045, USA

a r t i c l e i n f o a b s t r a c t

Article history: Despite rampant coastal development throughout Southeast Asia and the Pacific, studies of conservation
Received 31 May 2013 genetics and ecology of vulnerable, coastal species are rare. Large bodied vertebrates with highly
Received in revised form 14 October 2013 specialized habitat requirements may be at particular risk of extinction due to habitat degradation and
Accepted 23 October 2013
fragmentation, especially if these habitats are naturally patchily distributed, marginal, otherwise
geographically limited, or associated in space with high human population densities or heavy anthropo-
genic disturbance. Particularly telling examples of these conservation challenges are large Australasian
Keywords:
reptiles with obligate habitat requirements for lowland, coastal and mangrove forests. Plagued by habitat
Black markets
Coastal forests
destruction due to high human densities along coastlines, sprawling rural development, and rapidly
Conservation developing estuarine fisheries industry, coastal forest reptiles are experiencing rapid declines. And yet
Ecological niche studies of population biology, genetics, and habitat requirements of species depending on these environ-
Hydrosaurus ments are few. We undertook the present study in order to take a multifaceted approach to understand-
Mangroves ing a poignant conservation problem. We identify significant evolutionary units for conservation in
Phyloforensics large-bodied sailfin lizards (genus Hydrosaurus), model suitable habitat in the Philippines from extensive
occurrence data and evaluate the efficacy of the current protected area network, and identify the source
of hydrosaurs in the illegal pet trade. We determine that the extent of the species’ habitat coincident with
protected areas is low. Our forensic evaluation of the illegal pet trade in the Philippines determines the
existence of a natural population that is at risk of systematic exploitation by traders. Together, this
integrative study characterizes a conservation urgency of particular significance: the genetically distinct
sailfin lizards of the Bicol faunal region, with suitable habitat virtually unprotected, and clear evidence of
heavy exploitation for illegal trade. To the best of our knowledge, our study is the first conservation
genetic study to evaluate the potential effectiveness of the protected landscape coverage in the Philip-
pines, a Megadiverse nation and Biodiversity Hotspot.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction include Proboscis Monkeys (Jeffrey, 1982), Northern Spotted Owls

(Franklin et al., 2000), Dugongs (Preen and Marsh, 1995), and Ivory
Protection of species with obligate requirements of specific, Billed Woodpeckers (Schock, 2005).
patchily distributed, marginal, or heavily fragmented habitats Lowland and coastal forests, and the species that depend on
represents a special challenge for government agencies, wildlife them, represent poignant conservation challenges of particular
managers, and conservation biologists. Species unable to survive concern (Mumby et al., 2004; Millenium Ecosystem Assessment,
outside of very specific or highly restricted ecological conditions 2005; Duke et al., 2007), and nowhere is this concern more evident
are, by definition, immediately imperiled if their required habitats than in the highly degraded littoral zone mangroves and beach or
are altered, degraded, or destroyed by the activities of humans. coastal forests of Asia and islands of the Pacific (for review: Polid-
Celebrated examples of taxa threatened—or even driven to extinc- oro et al., 2010). Throughout the edge of the Asian continent and
tion—by anthropogenic destruction of their required habitats the adjacent Australasian archipelagoes, coastal and mangrove
forests have been systematically cleared for brackish water aqua-
culture ponds, firewood and timber, or residential and urban
⇑ Corresponding author. Tel.: +1 405 325 3718; fax: +1 405 325 7699.
development (e.g., Walters, 2000; Blasco et al., 2001). Globally,
E-mail addresses: camsiler@ou.edu (C.D. Siler), alira@ku.edu (A. Lira-Noriega),
rafe@ku.edu (R.M. Brown). mangrove biodiversity is highest in the Australasian region
URL: http://www.cameronsiler.com (C.D. Siler). (Polidoro et al., 2010), yet in some countries, mangrove and dry

0006-3207/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.biocon.2013.10.014
 C.D. Siler et al. / Biological Conservation 169 (2014) 100–108 101

coastal forest removal has been near complete (Primavera et al., One such group for which almost no information is known for
2004; Walters, 2004; Samson and Rollon, 2008). conservation status assessment is the spectacularly colorful and
At the same time, the critical importance of coastal forests has morphologically distinctive sailfin lizards of the genus Hydrosau-
been well documented (Millenium Ecosystem Assessment, 2005; rus; Hydrosaurus pustulatus is already considered vulnerable
Duke et al., 2007; Polidoro et al., 2010). Mangroves and beach for- (A2acd + 4acde; IUCN, 2012) as a result of habitat loss. Sailfin
ests are recognized to possess a unique flora, which acts at the crit- lizards have a Wallacean distribution, occurring from Sulawesi
ical interface between terrestrial, estuarine, and near-shore marine and the Philippines to Papua New Guinea (Fig. 1B), and inhabit
ecosystems (Polidoro et al., 2010). Coastal forests and mangroves vegetation and coastal forests immediately adjacent to beaches,
buffer against erosion, protect inland communities from damage or surrounding low-lying rivers (less than a few hundred meters
by inclement weather (Dahdouh-Guebas et al., 2005; Barbier in elevation), on the banks of estuarine areas, deltas, river mouths,
et al., 2008; Das and Vincent, 2009), and represent critical habitat and in mangroves. Tightly linked to riparian habitats in lowland
for terrestrial, estuarine and marine species (Robertson and Duke, coastal and mangrove forests, hydrosaurs occur in moderate to
1987; Luther and Greenburg, 2009). Finally, mangroves serve as high densities in suitable habitats but are conspicuously absent
source and sink for nutrients and sediments for inshore marine when their specific habitats are removed (personal observations).
habitats (i.e., seagrass beds and coral reefs; Dorenbosch et al., With their required habitat severely threatened, sailfin lizards
2004; Duke et al., 2007). represent an urgent need for conservation genetics studies.
In some archipelago nations, the near complete loss of coastal Two additional concerns accentuate the conservation urgency
forests may have already occurred. For example in the Philippines, for members of the genus Hydrosaurus. First, the taxonomy of the
in just the last 75 years, the destruction of the coastal forests and group is unresolved and has never been the subject of a systematic
mangroves has been extensive (Primavera, 2000; Walters, 2000, study using genetic data or modern phylogenetic methods. Origi-
2004; Primavera et al., 2004; Lewis, 2005; Samson and Rollon, nally described in the 1800s, the species’ original descriptions are
2008). The archipelago has lost more than 75% (>337,000 ha) of brief, based on specimens that were immature at the time of origi-
original mangrove forests (Baconguis et al., 1990; Walters, 2000; nal preservation, and are non-diagnostic by modern standards, fur-
Samson and Rollon, 2008), with >66% (or 278,657 ha) lost between ther hindering efforts to determine species boundaries and
1950 and 1990 (Primavera, 2000). The elimination of this unique taxonomic management units. Finally, with their striking dorsal
coastal ecosystem is primarily attributed to urbanization along crests, conspicuous caudal sail-like structure, and brightly colored,
the coasts and construction of brackish water fishponds (60%; aesthetically pleasing ornate coloration (Fig. 1A), hydrosaurs are
Samson and Rollon, 2008) and yet its impacts are largely unknown. heavily exploited in the local and international pet trade (Gaulke,
Despite the extensive documentation of the effects of habitat loss 1998; Welton et al., 2012, 2013a; E. Sy, personal communication).
in inland ecosystems (for review: Posa and Sodhi, 2006; Rickart This exploitation creates a potentially exacerbating situation given
et al., 2011), few, if any, studies have measured the direct effects the already tenuous prospects for continued survival of these
of habitat destruction on coastal forest obligate species. Due to species.
the paucity of such studies, there is an immediate need for atten- We undertook the present study in order to provide more infor-
tion to coastal and mangrove forest obligate species in studies of mation for conservation planning and to characterize the conserva-
species persistence, habitat availability and suitability, exploita- tion status and urgency for these largely unstudied species of
tion, and studies of genetic variation in obligate resident species suspected high conservation importance. We (1) screened multilo-
(e.g., conservation genetics). cus genetic data for divergent lineages (likely conservation targets)

Fig. 1. (A) Photograph of Hydrosaurus sp. provided by Scott Corning. (B) The known distribution of sailfin lizards. (C) Simplified phylogenetic estimate inferred from the
analysis of one mitochondrial and three nuclear genes, and summary of Bayesian species delimitation (BPP) analyses for the genus Hydrosaurus (with BPP speciation
probability values on representative nodes). No nuclear data are available for the unvouchered sample of Hydrosaurus amboinensis, and therefore, this species was not
included in BPP analyses (node highlighted with asterisk). Hydrosaurus sp. refers to the sampled population on Sulawesi.
102 C.D. Siler et al. / Biological Conservation 169 (2014) 100–108

and determine whether these correspond to currently recognized Table 1

species (Hydrosaurus amboinensis, H. pustulatus, and H. weberi), Models of evolution selected by AIC (as implemented in jModelTest) for mitochon-
drial (ND2, COI) and nuclear (L3, L52, DGL-/) data.
and we (2) assessed the distribution of suitable habitat and ecolog-
ical requirements for H. pustulatus in the Philippine archipelago Partition AIC Number of Number of variable
and determined the extent of this habitat’s coincidence with pro- model characters characters

tected areas. Finally, we (3) forensically evaluated the illegal pet ND2a GTR + C 1035 735
trade in Manila to determine which natural populations may be COIa HKY 577 60
tRNAsb HKY + C 418 65
at risk of systematic exploitation by black market traders. The L3 F81 281 1
results of this pluralistic, multi-tiered approach: (i) clarified puta- L52 HKY 516 3
tive species boundaries (clear conservation targets, including one DGL-/ HKY + C 480 5
lineage from Sulawesi not currently recognized by taxonomists a
Individual gene codons selected with identical AIC models selected by
and conservation biologists) throughout the range of the genus jModelTest.
Hydrosaurus, (ii) quantitatively defined habitat suitability mea- b
tRNAMet, tRNATrp, tRNAAla, tRNAAsn, tRNACys, and tRNATyr.
sures for Philippine hydrosaurs (H. pustulatus), (iii) quantified
status of the species’ imperiled habitat (mangroves, coastal forests,
and low-elevation riparian corridors) showing suitable environ- as implemented in jModelTest (v2.1.3; Posada, 2008), was used to
ments to be minimally coincident with existing protected areas, select the best model of nucleotide substitution for each partition
and (iv) identified one highly significant, presumably heavily (Table 1).
exploited, population (sailfin lizards of the Bicol Peninsula) as the Partitioned Maximum Likelihood (ML) analyses were conducted
sole source of trade animals sampled from Manila black markets. in RAxMLHPC v7.0 (Stamatakis, 2006). The more complex model
The integration of these combined results identifies a clear conser- (GTR + I + C) was used for all subsets (Table 1), and 100 replicate
vation priority (the Bicol region of the Philippines) where law ML inferences were performed for each analysis. Each inference
enforcement and conservation specialists could make tremendous was initiated with a random starting tree and nodal support was
strides towards stemming black market trade in Philippine reptiles assessed with 100 bootstrap pseudoreplicates (Stamatakis et al.,
while taking field-based steps to ensure future survival of these 2008) Alignments and resulting topologies are deposited in Dryad
charismatic endemic vertebrates. (doi: http://doi.org/10.5061/dryad.1fs7c).
We employed the NeighborNet algorithm (Bryant and Moulton,
2004) in the program SplitsTree (v4.10; Huson and Bryant, 2006) to
2. Methods
generate phylogenetic networks for the mitochondrial and concat-
enated nuclear datasets, respectively. To assess the support for in-
2.1. Data collection
ferred splits in the network, a bootstrap analysis was conducted
with 1000 replicates. An analysis of molecular variation (AMOVA)
Ingroup sampling of wild-caught animals included data for 78
was conducted on sequence data for H. pustulatus to investigate
vouchered and 1 non-vouchered individual representing all three
the amount of genetic variation explained at different resolutions
currently recognized species of Hydrosaurus and 20 individuals
of geography. All analyses were run with 1000 permutations in
sampled from skin and muscle biopsies (from tail tips or crests)
Arlequin (v3.5.1.3; Excoffier et al., 2005). The analysis was con-
from animals in the black market pet trade during five visits from
ducted to elucidate the amount of genetic variation that could be
2011 to 2012 (Appendices S1, S2; Supporting Information). For 81
explained by recognized faunal regions versus by islands within
ingroup samples we sequenced the mitochondrial genes NADH
faunal regions.
dehydrogenase subunit 2 (ND2) and Cytochrome Oxidase I (COI),
along with their flanking tRNAs. We also sequenced three nuclear
2.3. Species delimitation
loci (nuDNA): two anonymous loci (L3, L52), and the diacylglyceral
lipase alpha (DGL-/) gene (Alföldi et al., 2011; Welton et al.,
We tested the monophyly of H. amboinensis (as currently recog-
2013b; Appendix S2; Supporting Information). Additionally, se-
nized) within a Bayesian framework. A topology was created to
quence data were available on GenBank for one individual of H.
constrain all samples of H. amboinensis and H. sp. (Sulawesi) to
amboinensis (ND2 and COI; Okajima and Kumazawa, 2010), eight
be monophyletic. We used the proportion of 16,004 postburn-in
individuals of H. weberi (COI; Setiadi et al., 2010), and 10 species
trees consistent with this hypothesis as an estimate of the poster-
used as outgroups (ND2; Appendix S1). All novel sequences were
ior probability of the hypothesis.
deposited in GenBank (Supporting Information). Primer informa-
We explored support for hypothesized species boundaries
tion, and polymerase chain reaction (PCR) and sequencing proto-
within Hydrosaurus using the program Bayesian Phylogenetics
cols are provided in Appendix S2.
and Phylogeography (BPP v.2.0; Yang and Rannala, 2010). Due to
the single sample of H. amboinensis represented by mtDNA only,
2.2. Phylogenetic and population structure analyses species delimitation analyses were restricted to the remaining
three lineages of Hydrosaurus for which nuclear sequence data
Alignments were made in Se-Al Sequence Alignment Editor are available (H. sp. [Sulawesi], H. pustulatus, and H. weberi [Halma-
(v2.0a11; Rambaut, 2002). Preliminary analyses exploring the im- hera]). To account for phylogenetic uncertainty, we repeated
pact of missing data on inferred outgroup relationships resulted in analyses on all three possible starting topologies: (sp. [Sulawesi],
identical relationships inferred with and without the inclusion of (pustulatus, weberi)); ((sp. [Sulawesi], pustulatus), weberi); ((sp.
nuclear data. We therefore chose to include all available data for [Sulawesi], weberi), pustulatus). Samples were assigned to one of
subsequent analyses. Additionally, following the observation of the three hypothesized lineages. We phased each nuclear locus,
no strongly supported incongruence between datasets, we con- using the program PHASE v.2.1.1 (Stephens and Donnelly, 2003),
ducted subsequent phylogenetic analyses using a combined, parti- and retained haplotypes with the highest probabilities for subse-
tioned, concatenated dataset. The two datasets, ND2 and COI, were quent analyses. Using mtDNA and phased nuDNA data, we ran
partitioned by codon position, the tRNAs were treated as a single analyses for 500,000 generations, sampling every 50 generations,
subset, and each of the three nuclear loci, L3, L52, and DGL-/ were with a burn-in of 10,000. We accommodated differential rates of
analyzed as a single subset. The Akaike information criterion (AIC), mutation between mtDNA and nuDNA, and explored the impact
 C.D. Siler et al. / Biological Conservation 169 (2014) 100–108 103

of both algorithm and prior regime, following the methods of (Fig. 2). Within H. pustulatus, six major clades are supported, corre-
Setiadi et al. (2011) and Leaché and Fujita (2010) (Supporting sponding roughly to geographic regions in the Philippines with a
Information). few exceptions (Fig. 2, Clades A–F). All trade samples possess high
sequence identity with vouchered individuals from the Bicol Pen-
2.4. Habitat suitability modeling insula of southeastern Luzon Island (Fig. 2, Clade F).

We estimated habitat suitability for Philippine populations of 3.2. Network analyses and distribution of genetic diversity
H. pustulatus using ecological niche modeling and GIS queries of
minimum distances to landscape features that correspond to The results of network analyses corroborate the major results
Hydrosaurus habitat preferences (e.g., rivers and coastlines) and observed from phylogenetic analyses (Fig. 2, Supporting Informa-
known collection localities. This procedure allowed us to construct tion). Our mtDNA network (ND2 + COI) revealed eight highly diver-
a map of potential habitat suitability throughout the known range gent, well-supported groups corresponding to H. amboinensis, H.
of the species. Species occurrence points (localities for which se- weberi, and the six clades of H. pustulatus recovered in phylogenetic
quence data were collected for vouchered specimens) included analyses (Fig. 2). Although the combined nuDNA network revealed
26 unique localities across 13 islands. In order to model suitable little structure (Supporting Information), no strongly supported
habitat for the species we used Normalized Difference Vegetation conflicts with the mtDNA network were apparent. Population sam-
Index (NDVI) images derived from the Aqua sensor of MODIS at ples of H. pustulatus from the northern Philippines once again were
500 m resolution. We downloaded the NDVI tiles for the study re- recovered as members of a distinctly divergent clade among
gion for the years 2003–2011 and mosaicked them using the Philippine populations of Hydrosaurus (Fig. 2, Clade A). Analyses
MODIS Reprojection Tool 4.1 (USGS, 2012). With these images of molecular variation within H. pustulatus samples revealed signif-
we generated four sets of layers for each of the years in the follow- icant (P < 0.001) portions of genetic variation explained by both
ing way: (1) mean, (2) maximum, (3) minimum, and (4) minimum levels of the AMOVA, with more genetic variation explained among
of the driest period of the year (estimated to be from November– islands within faunal regions than among faunal regions (Fig. 2).
May).
To estimate areas of suitability, we used the maximum entropy
algorithm in MaxEnt (v.3.3.3; Phillips et al., 2006). We generated a 3.3. Species delimitation
model per layer type for unique occurrence points using the raw
output with a zero random test percentage. Other parameters re- Statistical hypothesis tests provided no support (posterior prob-
mained at default settings. Each model (corresponding to a layer ability approaching 0) for of the monophyly of H. amboinensis + H.
type) was converted to a binary prediction using the minimum sp. (Sulawesi). Subsequent Bayesian species delimitation analyses
training presence value given that all occurrence points come from aimed at exploring the support for the remaining three lineages
point localities for which we are confident about their precision. within the genus resulted in support for all three lineages as puta-
The binary predictions from the four models were then multiplied tive species (H. sp. [Sulawesi], H. pustulatus, and H. weberi) with
to obtain a map where all models suggest suitability. speciation probabilities of 1.0 (Fig. 1C), regardless of starting topol-
ogy or prior distributions.

2.5. H. pustulatus suitability map

3.4. Suitable habitat and protected areas landscapes
Because H. pustulatus is a species associated with coastal vege-
tation and mangrove forest, and low elevation fresh water rivers The estimate of total suitable habitat for H. pustulatus in the
near coasts, we generated raster layers that matched the species Philippines is 127,823 km2 of which 54,794 km2 correspond to
habitat suitability in accordance with what is known about its nat- highly suitable sites for the species after restricting it to areas
ural history (Supporting Information). This approach restricted our around mangrove forests, coastlines, and coastal river systems
best estimate of suitable habitat for H. pustulatus to data from areas where the species is most likely to be found according to its habitat
that were identified as suitable for the species from the niche preferences (Fig. 3). Of the estimated suitable habitat for the full
modeling map and that coincided with rivers, coastline, mangrove and reduced models, only 11.5% and 9.4%, respectively, are cur-
habitat, and known species’ occurrence points. Finally, we assessed rently represented in the Philippines protected areas (PA) network
the degree of protection of the modeled suitable habitats by exam- (Table 2). Within the Bicol Peninsula of Luzon Island, only 2.8% of
ining geographic overlap with the current national park system suitable, coastal habitat currently is protected (Table 2). Consider-
and protected areas network in the Philippines (IUCN and UNEP, ing the country’s mangrove forests only, based on recent estimates
2009). We also compared the suitability models to the geography of this ecosystems occurrence (Giri et al., 2011), 7.5% of this threa-
of current human population density (CIESIN, 2011). tened ecosystem is protected in the Philippines under the current
PA network (Table 2).

3. Results
4. Discussion
3.1. Phylogeny of Hydrosaurus and identification of trade samples
We define an urgent conservation challenge embodied by sailfin
We identified twenty-six unique mtDNA haplotypes and 11 lizards: heavily exploited, large-bodied reptiles with obligate
unique nuDNA haplotypes among our 89 samples with complete habitat requirements for rapidly disappearing forests, which are
data for the four genes. Analyses of the combined data insufficiently covered by protected areas. The combination of
(ND2 + COI + L3 + L52 + DGL-/) resulted in topologies with high approaches utilized here is the key to both the characterization
Maximum Likelihood (ML) bootstrap support for major clades of sailfin lizard conservation urgency and an understanding of
(Figs. 1 and 2). All analyses support the monophyly of the genus, the way in which human land use patterns along coastal and ripar-
however, the relationships among the three recognized species re- ian areas, plus an illegal pet trade’s impact on these striking and
main uncertain: H. amboinensis is recovered as the sister species to colorful vertebrates, have conspired to elevate the threat to their
a clade of H. pustulatus + (H. weberi + H. sp.) with poor support continued survival.
104 C.D. Siler et al. / Biological Conservation 169 (2014) 100–108

Fig. 2. (A) Color coded map of the Philippine archipelago (note correspondingly colored tree tips). (B) Summarized results of AMOVAs. (C) Mitochondrial haplotype network
inferred by SplitsTree (Huson and Bryant, 2006). (D) Preferred Maximum Likelihood phylogenetic tree estimate. Numbers below nodes indicate bootstrap values and clade
labels (A–F) are discussed in text. Alphanumerical labeled haplotypes correspond to identifications of vouchered and trade samples (Supporting Information). Hydrosaurus sp.
refers to the sampled population on Sulawesi.

4.1. Species boundaries and potential unrecognized diversity Sulawesi (for which the name ‘‘H. celebensis’’ [Peters, 1872] is
available). At a very minimum, these divergent lineages should
Our results partially clarify the current taxonomic arrange- now be embraced as Evolutionary Significant Units (ESUs; Ryder,
ment for sailfin lizards and, by implication, suggest four clear tar- 1986), or appropriate targets for conservation to preserve genetic
gets (probable species) for conservation efforts: H. amboinensis of diversity among and within species in an objective (Moritz,
Ambon Island, eastern Maluku, and New Guinea, H. pustulatus of 1994), and species-level targets for conservation efforts (IUCN,
the Philippines, H. weberi of Halmahera, and a fourth lineage on 2012).
 C.D. Siler et al. / Biological Conservation 169 (2014) 100–108 105

Fig. 3. Hydrosaurus pustulatus habitat suitability, with black dots representing vouchered localities sampled for this study, estimated from the full (A) and reduced (B) habitat
preference models. Federally protected areas shaded in green (C). The inset panel shows the Bicol Peninsula of Luzon in greater detail (above) with approximate human
density projected (see legend; CIESIN, 2011) and suggested priority regions for increased conservation efforts. 1Stars represent absence localities used in exploratory analyses
of presence–absence datasets (shown for reference). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Table 2
Summary of ecological estimates calculated in this study.

Measure Model Region Total (km2) Protected (km2) Protected (%)

Pertaining to Hydrosaurus pustulatus
Suitable habitat Full Philippines 127823.535 14719.414 11.5
Full Bicol Peninsula, Luzon Island 14217.669 644.135 4.5
Suitable habitat Reduced Philippines 54793.873 5152.070 9.4
Reduced Bicol Peninsula, Luzon Island 7057.757 197.999 2.8
Suitable habitat Reduced, Coastal Philippines 12937.456 732.745 5.7
Reduced, Coastal Bicol Peninsula, Luzon Island 2794.537 21.692 0.8
Pertaining to the Philippines
Mangrove forest Philippines 1705.481 128.739 7.5
Bicol Peninsula, Luzon Island 249.501 3.832 1.5

Our results reveal a unique, unrecognized lineage (currently Hydrosaurus celebensis (Peters, 1872) is the oldest available name
considered H. amboinensis) from Sulawesi, Indonesia (Figs. 1 and attributed to the Sulawesi lineage, we find it likely that future
2). Historically, sailfin lizards from Sulawesi were described as a studies will result in the resurrection of H. celebensis.
unique species, H. celebensis (Peters, 1872), but the name was Our analyses support a single Sailfin lizard species (H. pustula-
synonymized with H. amboinensis (Günther, 1873). Unfortunately, tus) distributed throughout much of the Philippines (Fig. 2); how-
the only published and available sequence of putatively true ever, haplotype networks and phylogeographic analyses (Fig. 2)
H. amboinensis comes from a genetic pet trade sample (SDNCU- reveal six distinct genetic clusters, or haplotype groups, in the Phil-
x029; Okajima and Kumazawa, 2010; Okajima and Kumazawa, ippines. The most genetically divergent/distinctive appears to be
personal communication) with a reported but unconfirmed region restricted to Aurora and Quezon Provinces of the islands of Luzon
of origin. Although there is general confidence in the samples re- and Polillo (Fig. 2, Clade A). With the exception of these distinct
ported origin from within the range of real H. amboinensis (Okajima northern populations, the remaining clades recovered in phyloge-
and Kumazawa, personal communication), there is no vouchered, netic analyses do not appear to conform to expected geographic
museum specimen available for comparison. We therefore chose patterns (Fig. 2, Clades B–F), and the observed patterns may be
to take a conservative approach in evaluating the results of this more heavily driven by the persistence of ancestral polymorphism
study, and have avoided making definitive taxonomic decisions or gene flow. Given the ecology of these coastal species, inter-pop-
at this time. Regardless of these limitations, the results of all ulation gene flow among adjacent regions is most likely a common,
analyses are conclusive in their support for the paraphyly of H. ongoing phenomenon—as would be expected with presumably
amboinensis as currently recognized. Given that the name dispersal-prone, large bodied lizards specialized in coastal forests,
106 C.D. Siler et al. / Biological Conservation 169 (2014) 100–108

lowland riparian corridors, and river-mouth habitats. Historically, density of individuals observed in the wild is located on the small
this may have been especially true during glacial periods when island of Dinagat in the southeastern Philippines. Unfortunately, so
sea levels dropped (100–140 m below current levels) resulting in little is known about the population sizes and ecology of these spe-
greater connectivity between adjacent islands previously sepa- cies, that this study represents the first approach to synthesize all
rated by shallow water (review: Voris, 2000; Siler et al., 2012). available information about this unique and threatened vertebrate
When compared to many other radiations of Southeast Asian group. As future studies attempt to better conserve the extant pop-
agamid lizards (i.e., Acanthosaura, Bronchocela, Calotes, Draco, ulations of species in this genus, it is critical that researchers focus
Gonocephalus, Japalura), the genus Hydrosaurus is a relatively spe- studies on the ecology of these riparian taxa.
cies-poor group (Manthey, 2008; Grismer, 2011). Given extensive This study adds to a growing baseline of Philippine trade foren-
geographic variation in scale features and body coloration (personal sic data. However, much remains to be done before we can have a
observations), conservation workers in the Philippines have long comprehensive understanding of how conservation efforts should
assumed that genetic analyses would reveal a considerable degree be prioritized to address the country’s illegal trade. Ideally, future
of cryptic diversity in the group. Surprisingly, the results of our studies will employ more rapidly evolving nuclear loci, possibly
study demonstrate low genetic diversity across the entire range incorporating next-generation sequencing technology as a tool
of the genus (5.7% maximum pairwise sequence divergence), and for marker discovery (SNP, SSRs; Garvin et al., 2010; Ogden,
indicate that the genus Hydrosaurus may in fact be composed of 2011), as well as more fully explore the power and utility of stan-
a small number of taxa, with relatively widespread geographical dard barcoding genes (e.g., COI) for more rapid and cost-effective
distributions. Nevertheless, these species remain a conservation identification of trade samples for other groups commonly encoun-
concern because of their dependence on coastal habitats, which tered in the black market pet trade. Finally, it is vital that phylofo-
have experienced heavy human-mediated development. rensic monitoring of the illegal Philippine pet trade should focus on
the numerous well-developed trade centers (e.g., Puerto Princesa
4.2. Phylogeographic patterns and future directions City, Palawan Island; Cebu City, Cebu Island; Davao City, Mindanao
Island).
Although the primary focus of this study was a conservation ge- Although there are clear differences between this and a previ-
netic assessment of a flagship conservation genus, and to highlight ous study (Welton et al., 2013a) in terms of collecting locations
the importance of reevaluating Philippine protected landscape targeted by poachers, both studies agree on one thing: the Bicol
strategies, phylogeographic patterns elucidated here merit discus- Peninsula of Luzon Island represents a major source of illegally
sion. With Hydrosaurus amboinensis supported as sister to all other collected individuals for the spectacular, large, and threatened rep-
lineages in the genus, our results indicate that sailfin diversity tiles illegally traded in Manila black markets. Given the moderate
originated east of Wallace’s Line, one of the most celebrated bio- to high density of human populations throughout much of the cen-
geographic barriers separating Asian and Australian biota (Wallace, tral and southern Bicol Peninsula (CIESIN, 2011; Fig. 3), increased
1860). The absence of well-supported relationships among the four conservation efforts and expansion of protected areas may have
genetically distinct lineages precludes us from inferring with con- the greatest impact in terms of preservation of critical hydrosaur
fidence the number of potential colonization events across this rec- habitat if prioritized in the northern regions of the peninsula
ognized barrier to dispersal; however, it is clear that Wallace’s Line (Camarines Norte Province, Caramoan Peninsula; Fig. 2).
was crossed historically at least once, giving rise to H. pustulatus
(Figs. 1 and 2). Timing of divergences inferred in recent higher-le- 4.4. Importance of reevaluating Philippine protected landscape
vel analyses suggest that the genus is quite old (>55 Ma; Townsend strategies
et al., 2011), begging the question of why species diversity is so
low. This work represents the first conservation genetic study aimed
Within the Philippines, phylogeographical patterns are nonsen- at integrating a species-specific, habitat suitability-driven evalua-
sical and defy expectation derived from other recent studies tion of the country’s protected landscape network with an analysis
(Brown et al., 2013). Shallow divergences and relationships of of a clear conservation threat in the form of illegal commercial
clades that defy both a south-to-north progression (expected from exploitation. The question remains: does the non-correspondence
inferred lineage diffusion from the southeast of the Philippines) between suitable habitat and protected areas represent a serious
and Pleistocene Aggregate Island Complex (PAIC) based expecta- problem for long-term survival of the H. pustulatus? And, expand-
tions of distinct lineages on former Pleistocene landmasses (Brown ing this question to other vertebrates with specific habitat require-
and Diesmos, 2009; Brown et al., 2013) both suggest the possibility ments, do the majority of land vertebrates in the archipelago
of a recent colonization and expansion through the archipelago. inhabit ranges even partially encompassed or overlapped by pro-
tected areas? We suspect that although the current protected area
4.3. Phyloforensics and origins of black market trade in Hydrosaurus system in the Philippines does overlap the geographic ranges of
many endemic species, many others possess habitat requirements
Trade sample sequences were 99.8–100% identical to a single, such that protected areas do not encompass their geographic
known-locality sequence from the Bicol Peninsula of Luzon Island. range. Until future studies continue to address this important con-
These observations stand in contrast to a recent trade forensic servation issue in a large, comparative framework, we are unable
study on monitor lizards of the Philippines (Welton et al., to predict the generality of the apparent mismatch between suit-
2013a), which recovered trade samples among seven clades and able habitat and protected area network in the Philippines that
five species. While Welton et al. (2013a) also found the majority we have documented here for H. pustulatus.
of trade samples (58–68%) to have originated from the Luzon fau- Our results clearly indicate the low extent of remaining habitat
nal region, only one-third of these samples originated from the Bi- for hydrosaurs in the Philippines (Fig. 3; Table 2). Only 5.7% of suit-
col Peninsula. Although we have observed hydrosaurs to be quite able habitat for H. pustulatus is protected under the current pro-
common in the Bicol Peninsula in comparison to Luzon proper, tected area network in the Philippines (Table 2). Moreover, in the
where they are rarely encountered, it is not clear at this time Bicol Peninsula, identified as the source region for all illegal pet trade
whether this observation necessarily is indicative of larger native samples included in this study, only 0.8% of the coastal suitable hab-
populations in this region as compared to elsewhere in the archi- itat is protected (Fig. 3; Table 2) and coastlines are some of the most
pelago where the species occurs. In fact, the site with the highest densely populated human habitations in the region. Our modeled
 C.D. Siler et al. / Biological Conservation 169 (2014) 100–108 107

estimates of suitable habitat for H. pustulatus likely are conservative References

and only partially representative of the country’s remaining coastal
habitat. Unfortunately, we expect that the percentage of protected Alföldi, J. et al., 2011. The genome of the green anole lizard and a comparative
analysis with birds and mammals. Nature 477, 587–591.
coastal forest in the archipelago is even lower because most pro- Baconguis, S., Cabahug, D., Alonzo-Pasicolan, S., 1990. Identification and inventory
tected areas are inland and do not include coastal forests (Fig. 3). of Philippine forested-wetland resource. Forest Ecol. Manage. 33 (34), 21–44.
Additionally, we have no doubt that even within protected, suitable Barbier, E. et al., 2008. Coastal ecosystem based management with non-linear
ecological functions and values. Science 319, 321–323.
habitat, poaching pressures likely exist. Unfortunately, there are no Blasco, F., Aizpuru, M., Gers, C., 2001. Depletion of the mangroves of continental
studies to date that have quantified the level of illegal poaching in Asia. Wetl. Ecol. Manage. 9, 245–256.
protected areas in the Philippines. Estimates of the volume of trade Brown, R., Diesmos, A., 2009. Philippines, Biology. In: Gillespie, R., Clague, D. (Eds.),
Encyclopedia of Islands. University of California Press, Berkeley, pp. 723–732.
would be useful in developing strategic conservation plans for the
Brown, R. et al., 2013. Evolutionary processes of diversification in a model island
development and maintenance of protected landscapes in the Phil- archipelago. Annu. Rev. Ecol. Evol. Syst. 44, 1–24.
ippines. Finally, future field studies should address the degree to Bryant, D., Moulton, V., 2004. Neighbor-net: an agglomerative method for the
which hydrosaurs exhibit tolerance for what appears to be a low construction of phylogenetic networks. Mol. Biol. Evol. 21, 255–265.
Center for International Earth Science Information Network (CIESIN), International
habitat connectivity syndrome, given that all coastal habitats Food Policy Research Institute (IFPRI), The World Bank and Centro Internacional
throughout the archipelago are experiencing degradation due to de Agricultura Tropical (CIAT). 2011. Global Rural-Urban Mapping Project
encroaching development and/or wholesale removal. version 1 (GRUMPv1): Urban Extents Grid, Columbia University, Palisades
<http://sedac.ciesin.colubmia.edu/data/set/grump-v1-urban-extents>
With only 7.5% of all mangrove forest in the country coinciding (accessed 20.08.12).
within protected areas (and a mere 1.5% Bicol region mangroves Dahdouh-Guebas, F. et al., 2005. How effective were mangroves as a defence against
protected), our study provides a first example of an integrative, the recent tsunami? Curr. Biol. 15, 443–447.
Das, S., Vincent, J., 2009. Mangroves protected villages and reduce death toll during
multidisciplinary approach to conservation and management of Indian super cyclone. Proc. Natl. Acad. Sci. B. 106, 7357–7360.
one of the Philippines’ major charismatic riparian corridor forest Dorenbosch, M., van Riel, M., Nagelkerken, I., van der Velde, G., 2004. The
and coastal forest obligate vertebrate species. Unfortunately, the relationship of reef fish densities to the proximity of mangrove and seagrass
nurseries. Estuar. Coast. Shelf Sci. 60, 37–48.
ecological estimates presented in this study likely overestimate Duke, N. et al., 2007. A world without mangroves. Science 317, 41.
the actual percent of mangrove and suitable hydrosaur habitat, Excoffier, L., Laval, G., Schneider, S., 2005. Arlequin ver. 3.0: an integrated software
as many of these protected regions likely have experienced some package for population genetics data analysis. Evol. Bioinf. 1, 47–50.
Franklin, A., Anderson, D., Gutiérrez, R., Burnham, K., 2000. Climate, habitat quality,
degree of habitat degradation, especially at the edges of parks
and fitness in Northern Spotted Owl populations in northwestern California.
and reserves (personal observations). We expect that numerous Ecol. Monogr. 70, 539–570.
additional vertebrates will be shown to exhibit similar dependence Garvin, M., Saitoh, K., Gharrett, A., 2010. Application of single nucleotide
on coastal and lowland forest habitats and are likely similarly at polymorphisms to non-model species: a technical review. Molec. Ecol. Res.
10, 915–934.
peril due to transformation of these environments by human activ- Gaulke, M., 1998. Utilization and conservation of lizards and snakes in the
ities. We have shown that by integrating multiple lines of evidence, Philippines. In: Erdelen, W. (Ed.), Conservation, Trade and Sustainable Use of
types of data, and analytical approaches, investigators may be able Lizards and Snakes in Indonesia. Mertensiella, Supplement to Salmandra, 9,
Bonn, Germany, pp. 137–142.
to take advantage of immediately applicable and emergent find- Giri, C., Ochieng, E., Tieszen, L., Zhu, Z., Singh, A., Loveland, T., Masek, J., Duke, N.,
ings for addressing previously understudied, but at-risk species 2011. Status and distribution of mangrove forests of the world using earth
of high conservation urgency. Great strides towards imperiled observation satellite data. Global Ecol. Biogeogr. 20, 154–159.
Grismer, L., 2011. Lizards of Peninsular Malaysia, Singapore and their adjacent
mangrove, coastal, and riparian lowland forest obligate species archipelagos. Edition Chimaira, Frankfurt, 728 pp.
conservation could be made by undertaking similar comparative Günther, A., 1873. Notes on some reptiles and batrachians obtained by Dr. Bernhard
methods across taxa of varying life history strategies, demographic Meyer in Celebes and the Philippine Islands. Proc. Zool. Soc. London 1873, 165–
172.
patterns, and microhabitat requirements. Huson, D., Bryant, D., 2006. Application of phylogenetic networks in evolutionary
studies. Mol. Biol. Evol. 23, 245–267.
IUCN, 2012. IUCN Red List of Threatened Species version 2011.3 <http://
Acknowledgements www.iucnredlist.org/> (accessed 01.11.12).
IUCN and UNEP, 2009. The World Database on Protected Areas (WDPA). UNEP-
WCMC. Cambridge, UK. (Update from 2012).
We thank the Protected Areas and Wildlife Bureau (PAWB) of
Jeffrey, S., 1982. Threats to the proboscis monkey. Oryx 16, 337–339.
the Philippine Department of Environment and Natural Resources Leaché, A., Fujita, M., 2010. Bayesian species delimitation in West African forest
(in particular T.M. Lim, C. Custodio, J. de Leon, A. Manila, and A. geckos (Hemidactylus fasciatus). Proc. Roy. Soc. B 277, 3071–3077.
Lewis, R., 2005. Ecological engineering for successful management and
Tagtag) for facilitating this study. Thanks are due to A. Diesmos,
rehabilitation of mangrove forests. Ecol. Eng. 24, 403–418.
E. Sy, and J. Barns for logistical support. Sampling efforts followed Luther, D., Greenburg, R., 2009. Mangroves: a global perspective on the evolution
protocols approved by the University of Kansas (KU) IACUC and and conservation of their terrestrial vertebrates. Bioscience 59, 602–612.
MOA between KU and PAWB, as outlined in a GP administered Manthey, U., 2008. Agamid Lizards of Southern Asia, Draconinae 2 (Terralog 7b). –
Frankfurt (Chimaira), 168 pp.
by PAWB. We thank B. Evans and J. McGuire for access to Halma- Millenium Ecosystem Assessment, 2005. Ecosystems and human well-being:
hera and Sulawesi material, respectively. Fieldwork was supported Synthesis. Island Press, Washington DC, 137 p.
by a National Science Foundation Doctoral Dissertation Improve- Moritz, C., 1994. Defining ‘evolutionary significant units’ for conservation. Trend.
Ecol. Evol. 9, 373–375.
ment grant (DEB 0804115 to CDS), Fulbright and Fulbright-Hayes Mumby, P. et al., 2004. Mangroves enhance the biomass of coral reef fish
Fellowships (to CDS); and an NSF Biotic Surveys and Inventories communities in the Caribbean. Nature 427, 533–536.
grant (DEB 0743491 to RMB). Jake Esselstyn and three anonymous Ogden, R., 2011. Unlocking the potential of genomic technologies for wildlife
forensics. Mol. Ecol. Res. 11, 109–116.
reviewers provided constructive comments on earlier drafts of the Okajima, Y., Kumazawa, Y., 2010. Mitochondrial genomes of acrodont lizards:
manuscript. A special thanks to L. Campbell and Q. Huijie for help timing of gene rearrangements and phylogenetic and biogeographic
in processing MODIS datasets, and to Scott Corning for providing implications. BMC Evol. Biol. 10, 141.
Peters, W. 1872. Über einige von Hrn. Dr. A.B. Meyer bei Gorontalo und Auf den
photographs of adult hydrosaurs. Togian-Inseln gesammelte Amphibien. Monatsberichte der Königlich
preussischen Akademie der Wissenschaften zu Berlin, pp. 581–585.
Phillips, S., Anderson, R., Schapire, R., 2006. Maximum entropy modeling of species
Appendix A. Supplementary material geographic distributions. Ecol. Mod. 190, 231–259.
Polidoro, B. et al., 2010. The loss of species: mangrove extinction risk and
geographic areas of global concern. PLoS ONE 5, e10095.
Supplementary data associated with this article can be found, in Posa, M., Sodhi, N., 2006. Effects of anthropogenic land use on forest birds and
the online version, at http://dx.doi.org/10.1016/j.biocon.2013.10.014. butterflies in Subic Bay, Philippines. Biol. Conserv. 129, 256–270.
108 C.D. Siler et al. / Biological Conservation 169 (2014) 100–108

Posada, D., 2008. JModelTest: phylogenetic model averaging. Mol. Biol. Evol. 25, Stamatakis, A., Hoover, P., Rougemont, J., 2008. A rapid bootstrap algorithm for the
1253–1256. RAxML Web Servers. Syst. Biol. 57, 758–771.
Preen, A., Marsh, H., 1995. Response of dugongs to large-scale loss of seagrass from Stephens, M., Donnelly, P., 2003. A comparison of Bayesian methods for haplotype
Hervey Bay, Queensland, Australia. Wildlife Res. 22, 507–519. reconstruction from population genotype data. Am. J. Human Genet. 73, 1162–
Primavera, J., 2000. Development and conservation of Philippine mangroves: 1169.
institutional issues. Ecol. Econ. 35, 91–106. Townsend, T.M., Mulcahy, D.G., Noonan, B.P., Sites Jr., J.W., Kuczynski, C.A., Wiens,
Primavera, J., Sadaba, R., Lebata, M., Altamirano, J., 2004. Handbook of Mangroves in J.J., Reeder, T.W., 2011. Phylogeny of iguqnian lizards inferred from 29 nuclear
the Philippines – Panay. Philippines: SEAFDEC Aquaculture Department and loci, and a comparison of concatenated and species-tree approaches for an
UNESCO Man and the Biosphere ASPACO Project. 106 p. ancient, rapid radiation. Mol. Phylogen. Evol. 61, 363–380.
Rambaut, A., 2002. SE-AL Sequence Alignment Editor, v2.0a11. University of Oxford, USGS. 2012. MODIS reprojection tool <http://https://lpdaac.usgs.gov/tools/
Oxford, UK. modis_reprojection_tool> (accessed 20.08.12).
Rickart, E., Balete, D., Rowe, R., Heaney, L., 2011. Mammals of the northern Voris, H., 2000. Maps of the Pleistocene sea levels in South East Asia: shorelines,
Philippines: tolerance for habitat disturbance and resistance to invasive species river systems, time durations. J. Biogeogr. 27, 1153–1167.
in an endemic insular fauna. Divers. Distr. 17, 530–541. Wallace, A.R., 1860. On the zoological geography of the Malay archipelago. J. Proc.
Robertson, A., Duke, N., 1987. Mangroves as nursery sites: comparisons of the Linn. Soc. 4, 172–184.
abundance and species composition of fish and crustaceans in mangroves and Walters, B., 2000. Local mangrove planting in the Philippines: are fisherfolk and
other nearshore habitats in tropical Australia. Mar. Biol. 96, 193–205. fishpond owners effective restorationists? Restor. Ecol. 8, 237–246.
Ryder, O.A., 1986. Species conservation and systematics—the dilemma of Walters, B., 2004. Local management of mangrove forests in the Philippines:
subspecies. Trends Ecol. Evol. 1, 9–10. successful conservation or efficient resource exploitation? Human Ecol. 32,
Samson, M., Rollon, R., 2008. Growth performance of planted mangroves in the 177–193.
Philippines: revisiting forest management strategies. Ambio 37, 234–240. Welton, L.J., Siler, C.D., Diesmos, A.C., Diesmos, M., Lagat, R., Causaren, R., Brown,
Schock, D., 2005. Forest management for Ivory-billed Woodpeckers (Campephilus R.M., 2012. Genetic identity, geographic range, and major distribution records
principalis): a case study in managing an uncertainty. North Am. Birds 59, 214– for frugivorous monitor lizards of Luzon Island, Philippines. Herpetol. Rev. 43,
219. 226–230.
Setiadi, M., Hamidy, A., Abidin, Z., Susanto, D., Brown, R., Peterson, A., Li, X., Evans, Welton, L.J., Siler, C.D., Linkem, C.W., Diesmos, A.C., Diesmos, M., Sy, E., Brown, R.M.,
B., 2010. Genetic structure of herpetofauna on Halmahera Island, Indonesia: 2013a. Dragons in our midst: phyloforensics of illegally traded southeast asian
implications for Aketajawe-Lolobata National Park. Conserv. Biol. 24, 553–562. water monitor lizards. Biol. Conserv. 159, 7–15.
Setiadi, M. et al., 2011. Adaptive radiation and ecological opportunity in Sulawesi Welton, L.J., Siler, C.D., Oaks, J.R., Diesmos, A.C., Brown, R.M., 2013b. Multilocus
and Philippine fanged frog (Limnonectes) communities. Am. Nat. 178, 221–240. phylogeny and Bayesian estimates of species boundaries reveal hidden
Siler, C.D., Oaks, J.R., Welton, L.J., Linkem, C.W., Swab, J., Brown, R.M., 2012. Did evolutionary relationships and cryptic diversity in Southeast Asian monitor
geckos ride the Palawan raft to the Philippines? J. Biogeogr. 39, 1217–1234. lizards. Mol. Ecol.. http://dx.doi.org/10.1111/mec.12324.
Stamatakis, A., 2006. RAxML-VI-HPC: maximum likelihood-based phylogenetic Yang, Z., Rannala, B., 2010. Bayesian species delimitation using multilocus sequence
analyses with thousands of taxa and mixed models. Bioinformatics 22, 2688– data. Proc. Natl. Acad. Sci. 107, 9264–9269.
2690.