IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from
it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date:
2020
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Hendriks, K. (2020). On the origin of species assemblages of Bornean microsnails. University of
Groningen. https://doi.org/10.33612/diss.124819761
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
Kasper Pjotr Hendriks
PhD thesis
to obtain the degree of PhD at theUniversity of Groningen on the authority of the Rector Magnificus Prof. C. Wijmenga
and in accordance with the decision by the College of Deans. This thesis will be defended in public on
Friday 15 May 2020 at 16.15 hours
by
Kasper Pjotr Hendriks born on 3 September 1982 in Leiden, the Netherlands
Chapter 1 General introduction 7 Chapter 2 Phylogeography of Bornean land snails suggests
long-distance dispersal as a cause of endemism
29 Chapter 3 The microbiome masks correlations between consumers
and their diet in Bornean snail-plant-interactions
101 Chapter 4 Herbivore community members share their plant diet,
but diets differ in richness
187 Chapter 5 Synthesis 243 References Summary Samenvatting Acknowledgements Curriculum vitae
Author affiliations and contact information
277 293 301 309 317 323
Niche theory - neutral theory
The main objective of science is to find patterns and to infer generalisations of the world around us. This is especially clear in some subdisciplines of physics, in which the ultimate generalisations are formalized in universal laws. Classic examples are the laws of gravitation, relativity, and thermodynamics. Biologists, including ecologists, aim to find such laws too. However, the living world seems just too complex, and biologists have to make due with theories and concepts instead (Mayr 1982).
Arguably the most influential concept in the field of community ecology is that of ‘the niche’ (Grinnell 1917). The concept aids in explaining how it is possible that many species can co-exist. Or: why does the one, better-adapted species, not drive the other, less-adapted species, to extinction (Gause 1934)? Niche theory’s explanation is surprisingly simple: no two co-existing species can have exactly the same niche (Hutchinson 1953, Whittaker 1972), and seemingly very similar co-occurring species partition resources in one way or the other (Schoener 1968, 1974). Thus, co-existing species always differ in at least some aspect (‘niche dimension’), even if this is sometimes difficult to detect, as in the famous ‘paradox of the plankton’ (Hutchinson 1961).
But there is more to this story. Over time, thousands of communities of trophically similar species (i.e. from the same guild, or with very similar resource use; sometimes called ‘horizontal communities’, Vellend 2016 p. 11) from widely different taxonomic groups have been studied empirically, and there are fascinating generalisations to be made. Most importantly, the species abundance distributions (SADs) of nearly all communities in nature are strikingly similar: few species are highly abundant, and many species are rare or very rare, a pattern described well by a logseries distribution (Figure 1.1). Enter the Unified Neutral Theory of Biodiversity (UNTB), Stephen Hubbell’s theoretical solution to this ‘problem’ of unexplained generalisation (Hubbell 2001), a theory with parallels to neutrality in population genetics (Kimura 1968, Leigh 2007). The main idea is that community assembly follows very general rules, and species- specific characteristics, taken to be of utmost importance in niche theory (and by Darwin; Darwin 1859, Alonso et al. 2006), are not at all that important in explaining community structure (Rosindell et al. 2011). Hubbell was also inspired by MacArthur and Wilson’s Island Biogeography Theory (IBT; MacArthur and Wilson 1967, Warren et al. 2015), but added species abundances to the IBT model. Apart from the neutrality assumption (i.e. functional equivalence of the species in the community), the theory emphasizes the importance of dispersal limitation, speciation, and ecological drift, processes considered of paramount importance in community assembly (Vellend 2010).
The basic UNTB model consists of a local community with a finite number of individuals, where dying individuals can be either replaced by offspring from a local community member (with probability 1-m, m being a measure of dispersal limitation), or by immigrants from the regional metacommunity (with probability m). The SAD of
the metacommunity, in turn, is determined by either replacement from offspring of randomly chosen individuals, or a speciation event (with probability v). Communities can be quantified based on the dispersal limitation number m, plus a so-called ‘biodiversity number’ θ (a function of metacommunity size JM and speciation rate v). Ecological drift follows from the chance events related to the survival and death of one species or another. This simple model allows many predictions, including that of the local SAD based on fits to empirical data (Volkov et al. 2003, Etienne and Olff 2005, McGill et al. 2007, Rosindell et al. 2011). As explained above, the model is general, and does not predict exactly which species are -or are not- abundant (Harpole 2010).
Of course, the simplifications made by the UNTB have resulted in critique and scepticism and led to discussion (Alonso et al. 2006, Clark 2009, Harpole 2010, Rosindell et al. 2012). Especially field ecologists, having ‘observed’ the many differences between species themselves, started looking for ways to refute neutral theory, or how it contradicts their observations (e.g. Dornelas et al. 2006, Chase 2007, Purves and Turnbull 2011, Chase and Myers 2011). But Hubbell and his followers never claimed that the world of ecological communities is really neutral, nor that niches are not important (Rosindell et al. 2012). Instead, the UNTB actually is informative because it
Figure 1.1 Patterns of relative species abundance in a diverse array of ecological communities, copied from Hubbell (2001). It is the striking similarity in shape among these curves, even though they are from widely different taxa, that drove Hubbell to study a general theory that could explain this. Curves are for the following taxa: (1) A tropical wet forest in Amazonia, (2) a tropical dry deciduous forest in Costa Rica, (3) a marine planktonic copepod community from the North Pacific gyre, (4) terrestrial breeding birds of Britain, and (5) a tropical bat community from Panama.
is simple and because it fails to explain certain details; these are apparently the biological details that are important (Rosindell et al. 2011). Furthermore, it allows for tests against other, competing models, such as those related to density-dependence (Etienne and Haegeman 2012), and extensions of neutral models, such as the inclusion of protracted speciation (Rosindell et al. 2010), metacommunity composition (Haegeman and Etienne 2017), and the definition of different guilds (Janzen et al. 2015).
Bornean microsnails: an ideal empirical system
in community assembly
To study the origins of community assembly empirically in the light of the UNTB, I wanted to work with a (natural) community that satisfies several conditions. Community members should be from the same trophic guild, i.e. use broadly the same resources (Root 1967, Simberloff and Dayan 1991), thus without obvious niche partitioning. Speciation, dispersal limitation, and stochasticity should be important. Preferably, the system needed to be simple, without complex trophic interactions. And, as is generally true in the studies of ecology and evolution, clear community boundaries (such as on islands) would be most helpful in community description and analysis (Darwin 1859, MacArthur and Wilson 1963, 1967, Hubbell 2001, Losos and Ricklefs 2009, Warren et al. 2015).
The communities of land snails (with most species microscopically small, hence the term ‘microsnails’ in this thesis’ title) on limestone outcrops in the Lower Kinabatangan Floodplain in Sabah, Malaysian Borneo, fulfil most of these conditions. These outcrops are isolated (see also the next paragraph), and dispersal and colonization between outcrops are considered to be rare. Due to high pH and the availability of calcium carbonate, used by the snails to build and maintain shells, snail abundance is much higher on limestone compared to that around limestone (Schilthuizen et al. 2003a, Schilthuizen 2011). The snail communities thus live on so-called ‘habitat islands’. The multiple island communities can be interpreted as natural replicates, with similar processes leading to (dis)similar outcomes (Gillespie 2004, Gillespie et al. 2018). When results agree among islands, this can increase our trust in trends found. When they do not, this allows us to learn about important factors in the assembly of communities. Furthermore, the regional collection of island communities can be considered to represent a single metacommunity.
Levels of endemism within these snails are high, especially in the classification group of the Prosobranchia, reaching 50% (depending on the spatial scale; Vermeulen and Whitten 1999, Clements et al. 2008a, 2008b), from which it can be inferred that local speciation is important. Ecologically, these snails appear very similar, with no obvious resource partitioning (though research is limited).
species are several orders of magnitude less abundant, collecting hundreds of individuals is not likely to have any negative effects on the species’ populations. The small size of most snail species also makes collecting, transportation, and frozen storage of thousands of snails in a pack with the size of a cigar box very manageable. While some species are difficult to find alive, probably due to a combination of their small size and a hidden lifestyle, others are found rather easily on exposed limestone rocks. More importantly, all snails that die leave a fantastic, readily collected
Figure 1.2 Plectostoma concinnum (Fulton, 1901) is the most studied and therefore best-known microsnail species from Borneo. At ca. 2 mm shell height -here tens are placed on a five-cent coin- this species is considered a classic example of a microsnail. Due to the enormous densities of the species on limestone outcrops, these snails are readily found in the field. With a stereo microscope at 40× magnification they are easily studied in the laboratory. This sample was collected from Mawas, Lower Kinabatangan Floodplain, on 19 March 2015. Photo by Kasper P. Hendriks.
signature for several years after: their shells (Purchon 1977). The totality of shells, taken as a proxy for the community composition, i.e. census data, can be collected from the top soil below the rocks following standardized protocols. This only requires bringing a set volume of soil to the laboratory (usually 5 l), using flotation to take out heavy constituents (such as rocks), air dry the remaining soil, and picking out and identifying the shells (Schilthuizen et al. 2002, 2003a, Schilthuizen and Vermeulen 2003, Liew et al. 2008). Subsequently, shells can be stored for centuries in dry natural history collections for future reference, which adds to scientific testability and falsifiability (Popper 2005).
Next to high abundances, these snail communities are characterized by a high species diversity. Interestingly, tropical snail communities were hardly quantified until the 1980s, with the few researchers interested focussing on taxonomy instead (Box 1.1). And, diversity was claimed to be very low compared to that in temperate regions (Solem 1984), contrasting the classic latitudinal diversity gradient (i.e. highest diversity near the equator; Rohde 1992, Hillebrand 2004). Subsequent field studies, including those on Borneo, proved this claim wrong and showed local diversity (at ca. 100 species average) to be somewhat higher, and point diversity (10-30 species from a plot) comparable to that observed in temperate regions (Schilthuizen 2011). A peculiarity of the Bornean land snail communities is formed by the almost equal proportion of pulmonate and non-pulmonate (mainly Prosobranchia) species (Schilthuizen et al. 2002). In virtually all other parts of the world, including tropical regions, prosobranchs make up just a few percent to one third of the land snail community (de Winter and Gittenberger 1998, Raheem et al. 2009).
Bornean snail communities seem to have most of the ingredients needed for a proper analysis based on the UNTB. Their abundance distribution is characterized, as in most natural communities, by few very common species (usually termed the ‘dominant species’), with additionally a rather long tail of species that are rare or very rare. And indeed, these snail communities fit the classic, lognormal abundance distribution, as was previously shown by Schilthuizen (2011), and confirmed during our recent studies (Figure 1.3).
The Lower Kinabatangan Floodplain
All studies described in this thesis were conducted in the Lower Kinabatangan Floodplain in Sabah, Malaysian Borneo, roughly between 5.45° N to 5.60° N and 118.01° E to 118.35° E (Figure 1.5). The region is principally known for its lowland tropical rainforest, which was cut down for the most part since the 1970s to make way for agricultural enterprises, mostly in the form of vast oil palm plantations (McMorrow and Talip 2001). Now, only 4% (750 km2) of fragmented and degraded forest remains
(Ancrenaz et al. 2015). But thanks to ecotourism in the region (Goh 2017), with eager tourists boarding boats by the hundreds daily in the hope to catch a glimpse of a Bornean orang-utan, a saltwater crocodile, or one of the eight species of hornbill inhabiting the riparian forest (Figure 1.6), the remaining forest has been saved from logging. Several stretches of forest along the river are now under protection from Sabah Wildlife Department, and the forest around Gomantong Caves from Sabah Forestry Department. However, in many places the oil palm plantations have already pushed their way to the waterfront, possibly eradicating (snail) species not yet known to science.
Figure 1.3 Rank abundance distributions from Bornean microsnail communities. (A) From a representative collection of same-size microsnails from a Bornean limestone outcrop (reprinted with permission from Schilthuizen 2011). (B) From newly collected data on 26,063 snails from 69 different species during this study. These distributions show the classic community assembly pattern of few very common species (on the left), with a long tail (on the right) of many rare species (Whittaker 1972, McGill et al. 2007).
0 2000 4000 1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 species rank nu m be r o f i nd ivi du al s
Box 1.1 A brief history of Bornean land snail taxonomy.
To properly describe a community of species, it is important that its members can be identified. The species from northern Borneo are indeed rather well-described nowadays, but this is only the case since some 30 years, with Bornean land snails mostly undescribed up to 1990. Kobelt and Möllendorff were among the first authors to list land snail species from Borneo (Kobelt and Möllendorff 1899, Kobelt 1902), followed by authors like Sykes and Van Benthem Jutting describing species from nearby Java and the Malay Peninsula, with mention of samples from Borneo (Sykes 1902, van Benthem Jutting 1948, 1953). Solem (1964) described two new species based on collections made by Dr. Neville Haile in 1962 (Figure 1.4). After that, with the exception of work done by Thompson and Dance (Dance 1970, Thompson and Dance 1983), not much was published about Bornean snail taxonomy, and it wasn’t until the 1990s that orchid-specialist Jaap Jan Vermeulen collected and taxonomically described tens of species new to science from northern Borneo. The result was a series of important, beautifully illustrated, taxonomic papers in Basteria, journal of the Netherlands Malacological Society (NMV), with descriptions based mostly on shell morphology (Vermeulen 1991, 1993, 1994, 1996a, 1996b, 1999, 1999, 2007). During the following decade, Menno Schilthuizen became the main instigator of Bornean land snail taxonomic research. For the first time, modern molecular techniques were applied, not only to identify and describe morphologically cryptic species (Schilthuizen et al. 2006, Schilthuizen and Liew 2008), but also to study their phylogenetic relations (Webster et al. 2012) and population genetics (Schilthuizen et al. 1999a, 1999b, 2005a, 2012, Hoekstra and Schilthuizen 2011). Several students of Schilthuizen joined him in pursuing the proper taxonomic description of the many different (cryptic) species, among whom Liew Thor-Seng (Liew et al. 2008, 2009, 2014a, 2014b, Liew and Schilthuizen 2014a, 2014b, 2014c, Vermeulen et al. 2015, Liew 2019a, 2019b) and Mohd Zacaery bin Khalik (Khalik et al. 2018, 2019), with their students again spreading local interest in tropical malacology further (bin Marzuki and Foon 2016, Foon et al. 2017, Phung et al. 2017).
Figure 1.4 Many Bornean land snail species were described only relatively recently, often in the second half of the 20th century. Shown here is the minute species Opisthostoma hailei, newly described by Solem in 1964 based on collections made by Dr. Neville Haile in 1962 (Solem 1964).
The limestone outcrops, the regional stronghold of land snail communities, form a unique habitat within the Lower Kinabatangan Floodplain. The limestones are inhabited by a very specific suite of organisms, many of them with an affinity for the calcium-rich substrate, or completely depending on it, so-called ‘calcicoles’ (Vermeulen and Whitten 1999). Apart from the snails, there are specialised arthropods and plants (Kiew and Lim 1996, Azmi 1998). Many limestone outcrops in the region have complex cave systems (Njunjić and Schilthuizen 2019), which are home to huge numbers of bats and swiftlets (Lundberg and McFarlane 2012), attracting predators such as snakes and birds of prey. Several land snails have colonized caves as well (Schilthuizen et al. 2005a, 2012).
Probably as a result of the outcrops’ discrete, island-like occurrence, plus the abovementioned specialization of many inhabitants, levels of endemism are dis-proportionately high on limestone (Sodhi et al. 2004, Clements et al. 2008b). About two-thirds of the snail species are considered regional endemics (Vermeulen and Whitten 1999), and it is expected that morphologically cryptic endemics are still
Figure 1.5 Map of the land usage in the Lower Kinabatangan Floodplain, from Estes et al. (2012). Note that only a narrow corridor of forest along the main Kinabatangan River is now protected, while most of the region was cleared and used as oil palm plantation.
waiting to be described, as was recently the case for the snail genus Georissa (Khalik et al. 2018, 2019). Among plants, there are orchids, balsams, and Amorphophallus endemic to the limestone on Borneo (Vermeulen and Whitten 1999).
Lim and Kiew (1997) listed 59 limestone outcrops in Sabah, of which ca. 20 are to be found in the Lower Kinabatangan Floodplain (Schilthuizen et al. 2003b). Most outcrops are relatively small, with base areas ranging from 0.039 km2 for Tomanggong Kecil (‘Kecil’ being Malaysian for ‘small’) to 1.210 km2 for Gomantong (Clements et al. 2008a). Geological details on the Kinabatangan limestones are surprisingly few, but several main processes are thought to have shaped the limestone outcrops that we observe today (W. Renema, 2016, pers. comm.). The study of nannofossils and benthic Foraminifera has shown that limestone formation (originally as coral reefs) took place in the Central Sabah Basin between the Upper Oligocene and the Lower Miocene, i.e. roughly between 25 and 15 million years ago (Noad 2001, Hutchison 2005 p. 241), and possibly several million years earlier (McMonagle et al. 2011). Limestone thickness at Gomantong is probably around 300 metres. Limestone formations were pushed up
Figure 1.6 Photographic impression of the Lower Kinabatangan Floodplain and some of its characteristic vertebrate inhabitants (top, left to right: white-crowned hornbill Berenicornis
comatus, Bornean orang-utan Pongo pygmaeus, and saltwater crocodile Crocodylus porosus; photos
by Kasper P. Hendriks), with an aerial impression of a part of the Lower Kinabatangan Floodplain (bottom; photo by river-junkie.com).
can only infer that the maximum time of exposure of any limestone in the region (including parts that are possibly weathered away by now) should be ca. 20 MY. In this light, it is interesting to mention several Miocene microsnail fossil findings by Jonathan Todd, Willem Renema, and colleagues, from East Kalimantan, Indonesian Borneo (Renema et al. 2015). Upon study of several of these fossils, clearly belonging to the genus Arinia, master student Suzanne Anema and I found that most fossil
Figure 1.7 Top: fossil microsnails from the genus Arinia, collected from Miocene deposits in East Kalimantan, Indonesian Borneo, by Jonathan Todd, Willem Renema, and colleagues (Renema et al. 2015). Bottom: contemporary Bornean representatives that are very similar, and likely close relatives of the fossil taxa from the top row. These unpublished results indicate that the snails we studied, and possibly their communities, have been in existence for at least 5-10 MY. Front views from 3D images by Suzanne E. M. Anema.
Fo ssi l sp eci m en s Co nt em po ra ry re la tive s A. pertusa
specimens could surprisingly well be mapped against contemporary representatives of the genus (yet unpublished; Figure 1.7). Hence, the microsnails on limestone we studied (or more correctly: their predecessors), and possibly their communities, too, have been in existence for at least 5-10 million years, although we cannot be sure exactly where.
While some of the limestone outcrops included in present study are situated within one of the wildlife sanctuaries (e.g. Gomantong, Batangan), and hence probably well-protected, others are not, and under serious threat from human exploitation. Ulu Sungai Resang, a long and narrow outcrop on the south bank of the Kinabatangan River, was found completely surrounded by oil palm plantations during our visit on 27 April 2016, and worse, the outcrop was literally being cut in two by industrial quarrying for limestone from its centre. On the western flanks of outcrop Materis we found even more intensive quarrying during our visit on 18 April 2016 (Figure 1.8). Almost all other outcrops are either surrounded by oil palm plantations (Mawas and nearby outcrops) or bordered by them (e.g. Keruak, Tomanggong Besar, Batu Payung). Clements et al. (2008a) made suggestions for outcrop conservation based on numbers of endemic land snails, and suggested to focus conservation efforts towards the larger outcrops, which have relatively large numbers of endemic snails. Perhaps contro-versially, Vermeulen and Whitten (1999) presented a checklist for the selection of limestone sites for exploitation based on biological value of the different types of outcrop (based on size, caves, presence of nearby outcrops, type of limestone, et cetera), intended to perform damage control. The future of the limestone outcrops in the Lower Kinabatangan Floodplain is, unfortunately, far from certain, as for most of the forest ecosystems in Southeast Asia (Sodhi et al. 2004, 2010). We hope that
Figure 1.8 Two of the ca. 20 limestone outcrops in the Lower Kinabatangan Floodplain, visited for current research. Left: Batu Payung, an example of a small, relatively pristine outcrop, though bordered by oil palm plantations, on 20 April 2016. Right: Materis, an example of a limestone outcrop partly fallen prey to industrial quarrying activities on its western flanks, on 18 April 2016. Photos by Kasper P. Hendriks.
(F) (G)
increasing our knowledge and understanding of this unique limestone habitat and its inhabitants’ ecology and evolution, helps future conservation efforts.
Fieldwork to sample data
Essentially all data used during the studies described in this thesis were collected during three fieldwork trips to the Lower Kinabatangan Floodplain between 2015 and 2018. General details, not included in the following chapters, are given here in brief instead.
During all three trips, the main transportation was by 4WD car and/or fast boat with local boat driver, with subsequent travels by foot to reach the base of the limestone outcrop. The first trip was in March 2015 and was of a mostly exploratory nature, but useful samples were collected nonetheless. I was joined by Menno Schilthuizen, Iva Njunjić, and Alex Pigot. Being based at the Danau Girang Field Centre of Cardiff University (5.413° N, 118.037° E), we visited four limestone outcrops between the riverside villages of Batu Putih and Bilit.
The second trip was in April/May 2016, together with Leonel Herrera-Alsina and Giacomo Alciatore. Being based further downriver this time, at Sukau village, we focussed on sampling eight outcrops around Sukau and further downriver, towards the village of Abai. Furthermore, having a 4WD car at our disposal, we visited four -sometimes hard to find or access- outcrops along the main Sandakan- Sukau road, and checked for not-yet described outcrops following suspicious sites on
Figure 1.9 (Opposite page) Photographic impression of fieldwork in the Lower Kinabatangan Floodplain. (A) To reach limestone outcrops along the Kinabatangan River, we mostly used fast boats with local boat drivers. In the photo (left to right): Alex Pigot, Menno Schilthuizen, our local boat driver, and Iva Njunjić. (B) Some limestone outcrops were in protected forests, such as here at Pangi. Here we met with staff from the Kinabatangan Orang-utan Conservation Programme HUTAN, who kept us informed of the whereabouts of Bornean pygmy elephants, which are best avoided when by foot. In the photo on the right: Giacomo Alciatore and Leonel Herrera-Alsina. (C) Upon arrival at a limestone outcrop, seven plots of two by two metres were set out from which live snails were collected using strong headlight torches and forceps. In the photo: Alex Pigot. (D) Additionally, soil samples were taken from each plot, with a first round of sieving (to take out the largest rocks) performed directly in the field. In the photo: Kasper P. Hendriks. (E) Further heavy materials were removed from the soil samples by means of a method called ‘flotation’. Only floating debris, including all empty snail shells, were saved. (F) Remaining soil debris was air dried, here at the Danau Girang Field Centre during our 2015 fieldwork. (G) Dried soil samples were collected by plot. (H/I/J) Back in the lab, here at the Institute for Tropical Biology and Conservation (ITBC) in Kota Kinabalu, shells were picked from the soil, identified, and deposited into the BORNEENSIS collection. Photos (A-I) by Kasper P. Hendriks, photo (J) by Chee-Chean Phung.
and microbiome.
After the second and third trips, two weeks of post-fieldwork laboratory time were spent at the Institute for Tropical Biology and Conservation (ITBC) of the Universiti Malaysia Sabah (UMS) in Kota Kinabalu, Sabah, Malaysian Borneo, to order, register, and deposit samples into ITBC’s BORNEENSIS collection. Subsequent genetic analyses were performed mostly at Naturalis Biodiversity Center, Leiden, the Netherlands, based on samples made available as long-term loans from the BORNEENSIS collection.
Basically, we performed two types of sampling (Figure 1.9). In order to be able to study snail communities, we collected standardized volumes of soil and the snail shells contained therein (see above and Chapter 3). For population and phylogenetic studies (Chapter 2), as well as metabarcoding of the snail diet and microbiome (Chapters 3 and 4), we collected live snails and put these on 96% ethanol directly in the field. In the case of the metabarcoding study, samples were also put on ice already in the field and subsequently stored at -20°C, including during transportation. In general, it was possible to thoroughly sample one outcrop per day, depending on outcrop distance from our base camp, the effort needed to locate the outcrop within forest or plantation, the effort needed to be granted access to an outcrop in case it was situated within an oil palm plantation, outcrop size, number of plots, inter-plot distances, weather, density of the forest around the outcrop, and the odd charging pig-tailed macaque.
Thesis outline
This thesis focusses on the community assembly of Bornean microsnails, and tries to answer the main question: Have Bornean microsnail communities randomly assembled,
and if not, what factors were of influence? As explained above, the SADs of communities
of land snails on Borneo were before shown to be very much like the general pattern observed in nature, which in turn is well predicted by the UNTB (Schilthuizen 2011). The UNTB is founded on several simple assumptions, two of which I tested explicitly. To put my research in a wider context, I refer to Webb et al. (2002), who advocated the simultaneous and integrated study of community ecology, phylogenetics, and traits (Figure 1.10).
First, the UNTB assumes dispersal limitation, i.e. dispersal among local communities is expected to be a rare event, which explains why rare species are less numerous locally than regionally (in the metacommunity). Therefore, I studied historical colonization events using population genetic and phylogenetic analyses and present the results in Chapter 2.
Figure 1.10 The integration of the studies of community ecology and phylogenetics as advocated by Webb et al. (2002). In Chapter 2 I describe the biogeography of Bornean snails in the light of their evolutionary history and colonization (i.e. ‘phylogeography’) of the Lower Kinabatangan Floodplain. In Chapter 3 I take species assemblages (communities) of these snails, and study the community composition in the light of two traits: the plant diet and the gut microbiome. In Chapter 4 I take the same dataset, but instead focus on possible diet differences among the community members. Figure redrawn from Webb et al. (2002) and annotated.
Chapter 2
fit the first assumption of the UNTB, dispersal limitation, perfectly. Limestone outcrops in Sabah, Malaysian Borneo, are home to over 100 species of land snail, and considered habitat islands for many of their calcium carbonate-dependent inhabitants (Clements et al. 2008b). This is especially true for the many species of ‘prosobranch’ snail, which are further characterized by high levels of short-range endemism (Schilthuizen et al. 2005a). Small-scale (cryptic) speciation on limestone, such as shown for some species of Plectostoma (Schilthuizen et al. 2006) and Georissa (Khalik et al. 2018, 2019) in Sabah, further adds to the expectation that these snail species/ populations are very much isolated.
Paradoxically, land snails have also reached most oceanic islands, including the most remote archipelagos, such as the Galápagos (Parent and Crespi 2006), Hawaii (Rundell et al. 2004), Norfolk Island (Donald et al. 2015), and Madeira (Waldén 1983), a sign of extreme dispersal capabilities, or long-distance dispersal (LDD). Land snail dispersal is most probably of a passive nature, such as via the transportation by flowing water or flying birds (Purchon 1977, Dörge et al. 1999, Gillespie et al. 2012). Large numbers of snails have been observed on driftwood (Boettger 1926, Czógler and Rotarides 1938), and rafting was suggested to be the main mode of dispersal in the several littoral snail species from the family Trochidae (Donald et al. 2005).
In Chapter 2 I present the results from demographic and the phylogeographic analyses of three species of land snail from the Lower Kinabatangan Floodplain,
Plectostoma concinnum (Fulton, 1901) s.l., Georissa similis E. A. Smith, 1893 s.l., and Alycaeus jagori Von Martens, 1859. I show that spatial-genetic correlations disappear
at distances larger than 5 km, that different archipelagos of limestone habitat were colonized multiple times from other archipelagos within the region, and that in most cases (78%) this was the result of LDD or dispersal from non-adjacent outcrops (i.e. non-stepping-stone dispersal). Furthermore, colonization events upriver and downriver were about equally probable, suggesting no major effect of dispersal via flowing water.
My results show that the endemic species of land snail in the study region have colonized the Lower Kinabatangan Floodplain in multiple ways. At small spatial scales, isolation-by-distance occurs, but this does not likely result in speciation and endemism, because of maintenance of genetic connections. Instead, the relative importance of LDD and colonization of non-adjacent outcrops are expected to bring together genetically distinct populations. I expect that this mode of dispersal can form the onset of highly localised speciation, possibly resulting in a radiation of endemics.
I place the results in an evolutionary context, and the species-complexes studied are shown to be very old indeed, dating back two to ten million years (based on a molecular clock rate of 2% per million years for COI). I show LDD to be an important mode of colonization at evolutionary timescales. However, given the high genetic differentiation between different, localized populations, I believe that, from an ecological, UNTB context, these snails are dispersal-limited.
In Chapter 3 I ask the question: Are communities of consumers (directly or indirectly)
influenced by their food sources? Niche theory postulates how co-existing, ecologically
similar species can avoid competition through niche partitioning (Gause 1934, Hutchinson 1961). This has empirically been shown to be true in many taxa for one of the most fundamental niche dimensions, that of food, i.e. the diet (Whittaker 1972, Siemann et al. 1998, Knops et al. 1999, Haddad et al. 2001, 2009, Hawkins and Porter 2003, Kissling et al. 2007). In these cases, more available resources allow for more same-guild consumers, i.e. richer communities of consumers (Hutchinson 1959, Tilman and Pacala 1993).
In Chapter 3 I present results from my study on the correlation between two closely-associated trophic levels: the snail community and their plant diets, again on limestone outcrops in the Lower Kinabatangan Floodplain. I focus on the same three target species as in Chapter 2. I included the gut microbiome as a third trophic level, because empirical work by previous researchers showed how the microbiome can adapt to the host’s diet and become an important co-driver of diet choice (Muegge et al. 2011, Colman et al. 2012, Youngblut et al. 2019). In addition, I tested for the significance of environmental covariates likely of influence to dispersal and habitat suitability (Tilman and Pacala 1993, Hawkins and Porter 2003, Longmuir et al. 2007).
My study shows no direct correlation between diet and snail community diversity. However, both diet and snail community diversity are weakly, positively correlated with gut microbiome diversity. Correlations with several environmental variables are found as well, most notably with anthropogenic activity, habitat island size, and a probably important nutrient source, guano runoff from nearby cave entrances. I therefore conclude that direct community correlations may in fact be masked by a third trophic level, or a similar response to environmental variation.
of their diets to release too strong competition (Gause 1934, Hardin 1960, Hutchinson 1961)?
In Chapter 4 I present results from my study on diet differentiation among the different species of land snail on limestone outcrops, once more from the Lower Kinabatangan Floodplain. I use metabarcoding data of plant DNA retrieved from the guts of 658 individual snails from 26 snail species, including representatives from widely different taxonomic groups (Neritimorpha, Prosobranchia, and Pulmonata). I explain how I identified DNA reads down to plant family level, and that the total, regional snail community ate from at least 32 plant families (mostly Fabaceae, Asteraceae, Brassicaceae, and Moraceae).
Furthermore, I compare dietary results among species and show that different species have generally overlapping plant diets, but that species’ mean diet richness can differ up to ca. 15-fold. Importantly, diet richness is significantly positively correlated with snail size. A phylogenetic plant diet assessment shows that ca. 28% of individuals have a significant clustering of the diet, indicating active diet choice, although these patterns are weak to non-existent at the species level (i.e. diet data pooled by species). I suggest that, in the light of these findings, plant diet choice could still be random, and diet richness simply the result of the size of the snail, with larger species taking in more food, and thus more different plant taxa. Also, differences among species could be a by-product of other (selection) pressures, such as different predation levels or risk of desiccation, with different species taking different food from the places they hide. My study supports the idea that the land snail communities we study do not show any strong differentiation in a trait, the diet, that (according to niche theory) would be expected to vary among co-existing species.
In Chapter 5 I discuss results from the previous chapters together, describe important considerations regarding taxonomy, systematics, and identification of Bornean land snails, and I present preliminary results from two more studies undertaken during my PhD.
The first study deals with direct tests of neutrality using the R package ‘sadisa’ (Species Abundance Distributions under the Independent Species Assumption; Haegeman and Etienne 2017), which allows very fast fitting of neutral models (plus non-neutral models) to empirical abundance data. Here, I included not only the snail abundance data presented in Chapter 3, but also unpublished data on Central European fen snail communities (Horsák et al. unpublished), and previously published snail abundance data from Africa, Australia, and the Atlantic. The first results show that models taking density-dependence into account fit empirical data better than fully neutral models when communities are very rich (i.e. are composed of many species), suggesting species abundance is important in community assembly of snails.
The second study deals with the microhabitat (a subdivision of the general limestone habitat) and its influence on snail assemblages on limestone in the Lower Kinabatangan Floodplain. During fieldwork at 15 different plots live snails were specifically collected from five different microhabitats. I show that species richness differs between microhabitats, and that half of the species are found only in a single microhabitat; only one species is omnipresent and shows up in all five microhabitats. These results highlight the need to include the microhabitat in future attempts to explain community assembly.
suggests long-distance dispersal
as a cause of endemism
Kasper P. Hendriks, Giacomo Alciatore, Menno Schilthuizen, and Rampal S. Etienne
Published in Journal of Biogeography: 2019, 46 (5): 932-944 DOI: 10.1111/JBI.13546
land snails, using multiple genetic markers. We calculated genetic distances between populations, applied beast2 to reconstruct phylogenies for each taxon, and subsequently reconstructed ancestral ranges using BioGeoBEARS.
Results We found spatial genetic structure among nearby locations to be highly pronounced for each taxon. Genetic correlation was present at small spatial scales only, and disappeared at distances of five kilometres and above. Most archipelagos have been colonized from within the region multiple times over the past three million years, in 78% of cases as a result of long-distance dispersal or dispersal from non- adjacent limestone outcrops. The flow of the main geographical feature within the region, the Kinabatangan River, did not play a role.
Main conclusions Phylogeographic structure in these Bornean land snails has only partly been determined by small-scale dispersal, where it leads to isolation-by- distance, but mostly by long-distance dispersal. Our results demonstrate that island endemic taxa only very locally follow a simple stepping-stone model, whilst dispersal to non-adjacent islands, and especially long-distance dispersal, is most important. This leads to the formation of highly localized, isolated ‘endemic populations’ forming the onset of a complex radiation of endemic species.
Keywords
endemism, long-distance dispersal, Gastropoda, island biogeography, phylogenetics, tropical ecology, tropical land snails, Borneo
Introduction
Endemism is often associated with islands (Myers et al. 2000, Kier et al. 2009), where levels can reach impressive values, such as 89.9% in higher plants and 99.9% in land snails on Hawaii (Whittaker and Fernández-Palacios 2007). On oceanic islands, there is a clear boundary that restricts dispersal. But deserts, mountain tops, lakes, and valleys can form habitat islands with many endemics, too (Kruckeberg and Rabinowitz 1985).
The unique research opportunities offered by endemics on islands were already noted by some of the first students of biogeography (Darwin 1859, Wallace 1859), and have been exploited ever since (MacArthur and Wilson 1963, Warren et al. 2015). The probability of a migrant reaching an island from another location generally declines with distance (MacArthur and Wilson 1963). MacArthur and Wilson’s (1967) stepping-stone model detailed possible migration pathways along chains of islands. Empirical evidence supports the validity of the stepping-stone model in nature as a means of dispersal, such as in marine snails (Crandall et al. 2012), coastal fish (Maltagliati 1998, Gold et al. 2001), and plants (Harbaugh et al. 2009).
Based on the stepping-stone hypothesis we expect the order and direction of colonization of islands to be of importance in the evolution of island endemics. However, migration resulting from long-distance dispersal (LDD) could result in genetically distant populations becoming neighbours, directly facilitating local endemism. A terrestrial island system in which this idea can be tested, is the system of limestone outcrops in the tropical lowlands of Southeast Asia, where acidic soils between outcrops form impassable habitat for species dependent on calcium carbonate (Crowther 1982, Lim and Kiew 1997). These species indeed show high levels of local endemism here (Clements et al. 2008b), and migration of sedentary species between limestone outcrops is considered to be rare (Vermeulen and Whitten 1999, Sodhi et al. 2004). Many are very localized and show a differentiated population structure (Schilthuizen et al. 2006, Latinne et al. 2011, Sedlock et al. 2014). Several studies have shown regional genetic diversity between locations just tens of kilometres apart to be very high (Schilthuizen et al. 1999b, Latinne et al. 2011). More precise patterns, such as the way in which populations are connected, or the influence of archipelago layout and geology on population structure, remain unstudied.
An abundant and diverse group on these limestone outcrops are land snails (Gastropoda) (Tweedie 1961, Purchon and Solari 1968, Schilthuizen 2011). Local endemism reaches 60% in some sites (Vermeulen and Whitten 1999). A short generation time (~ 1 year) and high productivity are possible sources of high levels of genetic variation. The snails’ restricted dispersal, combined with bottlenecks and founder effects (Whittaker and Fernández-Palacios 2007 p. 168), could form a barrier to the spread of (genetic) variation.
Methods
Study system
We studied three taxa of small land snail (Gastropoda; Figure 2.1): Plectostoma
concinnum (Fulton, 1901) s.l., Georissa similis E. A. Smith, 1893 s.l., and Alycaeus jagori
Von Martens, 1859, inhabiting limestone outcrops in tropical lowland forest. On-going taxonomic studies suggest the former two taxa are in fact best considered species complexes (Methods S2.1). Both are small, with shell heights of 2 mm and 1 mm, respectively (Thompson and Dance 1983, Vermeulen 1994), while the latter reaches 10 mm (Kobelt 1902). Each taxon is locally common (tens to hundreds per square metre) in suitable habitat (Schilthuizen et al. 2003b, Liew et al. 2008). Georissa similis s.l. and
P. concinnum s.l. are restricted in range to our study region (Vermeulen 1991), while A. jagori is distributed over all of Sundaland and Sulawesi (van Benthem Jutting 1948). Plectostoma concinnum s.l. is strictly related to calcareous substrate (Schilthuizen et
al. 2002), whereas the other two taxa also occasionally occur on trees and shrubs near limestone (personal observations). Studies using standardised plots along a
Figure 2.1 Photographs of the target taxa of land snail (Gastropoda) studied. Each taxon is a common inhabitant of the limestone outcrops of the Lower Kinabatangan Floodplain, Sabah, Malaysian Borneo. (A) Georissa similis E. A. Smith, 1894, (B) Plectostoma concinnum, and (C)
Alycaeus jagori. Photos: Kasper P. Hendriks. Scale bars equal 1 mm.
transect that spans both limestone and non-limestone substrate confirm that the ‘prosobranch’ microsnail genera Plectostoma and Georissa tend to occur nearly strictly on limestone (Schilthuizen et al. 2003a).
Field procedures and sampling
Sampling took place during visits in March 2015 and April 2016. We included additional samples collected with a different purpose during visits in 2004 and 2017 (Table S2.1). We followed a hierarchical spatial structuring of the region: region > archipelago > outcrop > plot (Figure 2.2). Five archipelagos (A to E) of limestone outcrops were defined based on a between-outcrop distance of < 5 km, with archipelagos of two to seven outcrops. Based on previous studies, we considered dispersal between outcrops to be a rare event (Cowie 1984, Baur and Baur 1990, 1995, Schilthuizen et al. 2002). We defined the ‘population’ as the group of individuals from a taxon on one outcrop. We sampled 17 outcrops from at least two plots, with plots on opposite ends of the outcrop. Each plot was 10 metres wide (along the periphery of the base of the
Figure 2.2 Map of the Lower Kinabatangan Floodplain, Sabah, Malaysian Borneo. Habitat islands of limestone outcrops are indicated with different colors and named as follows: Bat (Batangan), Maw (Mawas), NL1 (New Location 1), NL2 (New Location 2), Kam (Kampung), Ker (Keruak), Pan (Pangi), TB (Tomanggong Besar), T2 (Tomanggong 2), TK (Tomanggong Kecil), USR (Ulu Sungai Resang), BP (Batu Payung), TBa (Tandu Batu), BT (Batu Tai), BTQ (Batu Tai Quarry), Gom (Gomantong), and Mat (Materis). Outcrops are grouped by geographical proximity (inter-island distance < 5 km) as indicated by dotted ellipses), resulting in archipelagos A to E (shown in bold). The main geographical feature of the region, the Kinabatangan River, is shown in grey, with direction of flow indicated by double arrows. Inset map © freevectormaps.com.
5. 45 5. 50 5. 55 longitude la tit ud e 118.0 118.1 118.2 118.3 0 5 10 km Mat Bat Maw BT Ker TK Pan USR BP TBa Kam T2 NL2 NL1 TB Gom BTQ E D C B A
SE Asia
For extractions of G. similis s.l. we used the Macherey-Nagel NucleoMag Tissue kit on a ThermoFisher KingFisher™ Flex Purification System. For P. concinnum s.l. and A.
jagori extractions were performed using Omega’s E.Z.N.A. ® Mollusc DNA Kit. We stored all DNA extraction templates at -80°C at the Naturalis Biodiversity Center, Leiden, the Netherlands (Table S2.2).
We used Sanger sequencing to study both mitochondrial and nuclear markers selected from the literature and refer to these publications for details of laboratory procedures. We amplified the mitochondrial Cytochrome Oxidase I gene (COI) for all taxa and the nuclear Histone 3 gene (H3) for G. similis s.l. and A. jagori, following (Schilthuizen et al. 1999b, 2006, Parent and Crespi 2006, Webster et al. 2012). The nuclear Internal Transcribed Spacer 1 region (ITS1) was amplified for P. concinnum s.l. and A. jagori, following Schilthuizen et al. (2006). We sent amplification products to BaseClear, Leiden, the Netherlands, for sequencing in two directions. We checked sequence reads for errors and deposited all data in the online Barcode of Life Database (BOLD, www.boldsystems.org) as dataset ‘DS-2018POP‘, and GenBank (accession numbers in Table S2.2). Due to inconsistent results in forward and reverse sequencing reads in ITS1, which are likely due to within-individual polymorphisms (Vierna et al. 2009), we based our analyses of ITS1 on reverse reads only.
Population genetic analysis
We studied 929 individual snails (362 P. concinnum s.l., 366 G. similis s.l., and 201 A.
jagori). We calculated, by both locus and taxon, nucleotide diversity π (Nei and Li
1979) and haplotype diversity Hd, at the spatial scale of the outcrop (i.e. the population) and the archipelago. We listed the encountered and normalized (Hrar) number of haplotypes. We checked for possible correlations between genetic diversity (π, Hd, and Hrar), and outcrop ‘island-area’ and archipelago size (in terms of sum of outcrop areas and archipelago outcrop number) by applying linear models. Finally, we listed the fraction of private haplotypes, Hprivate (cf. Slatkin 1985).
To determine metapopulation structure, we calculated between-population fixation indices (Weir and Cockerham 1984) as ΦST, a metric that weighs the number
of mutations (Excoffier et al. 1992, Bird et al. 2011). We used the function ‘pairwiseTest’ from R package ‘strataG’ v2.0.2 (Archer et al. 2017), with 1,000 replicates. We also calculated population differentiation as Jost’s D (Jost 2008), using function ‘pairwise_D’ from the R package ‘mmod’ v1.3.3 (Winter 2012). A value of one indicates no shared alleles between two populations (Bird et al. 2011).
We performed an Analysis of MOlecular VAriance (AMOVA; Excoffier et al. 1992) by locus, using the package Arlequin, version WinArl35 (Excoffier and Lischer 2010). After finding relatively high genetic diversity among outcrops from archipelago A (see Results), we repeated these analyses excluding data from that archipelago, and compared results. AMOVA were not performed for A. jagori due to insufficient data.
Demographic analysis
We studied the spatial component of snail dispersal by relating Jost’s D to shortest geographic distance using Mantel tests (Mantel 1967) at increasing spatial classes (i.e. geographical distances). We used the function ‘mantel.correlog’ from the R package ‘vegan’ v2.5-2 (Oksanen et al. 2017), with 15 distance classes, and logged Pearson correlations. We summarized results in so-called ‘Mantel correlograms’ (Oden and Sokal 1986, Borcard and Legendre 2012).
Phylogenetic and biogeographic analyses
We performed a Bayesian phylogenetic analysis for each taxon using beast2 (Bouckaert et al. 2014) with trees for each locus (‘gene trees’) linked to conform to the taxon tree (‘species tree’), and clock and site models unlinked. The site model for each locus followed output from jModelTest2 (Darriba et al. 2012) and analyses were repeated with a general GTR site model. We set a strict clock for each locus, which is appropriate in the study of closely related taxa (Brown and Yang 2011), with a clock rate of 2% per million years for COI (Wares and Cunningham 2001, Nekola et al. 2009). With no clock rate estimates available for the other loci, the software estimated rates for these relative to that for COI. We set a Yule tree prior. We ran analyses for 100 million generations, sampling posterior parameter values and trees every 10,000th generation, after which we discarded a 10% burn-in. We checked convergence for each run based on ESS values > 200 and proper mixing of parameters over time. We summarized trees with a posterior probability limit of 50%. We compared model results by Bayes Factor (BF; based on the harmonic mean of the log-likelihood of the posterior,) and chose the model with the highest BF (Suchard et al. 2001), or, when the BF was zero, the model with the highest posterior probabilities of tree clades. (A better, more intensive model selection method, using nested sampling, was published during time of writing (Maturana et al. 2018). We expect model selection not to be different when large absolute BFs are found.) All beast2 runs were performed on the CIPRES computing cluster (Miller et al. 2010).
our analyses (Matzke 2014). Concerns raised by Ree and Sanmartín (2018) on the DEC+J model are unlikely to have any significant effect on our results due to the strong spatial structure in our system. We scored dispersal and colonization events as follows: LDD (with a distinction between down- and upriver), within-archipelago (but not to adjacent outcrop or crossing the river), and stepping-stone (to adjacent outcrop). Based on the high genetic affinities found within the outcrop in each species complex, we did not include the possibility of ancestral ranges being smaller than the outcrop.
Finally, we repeated the demographic test of the Mantel correlogram, now using a mean pairwise phylogenetic distance between samples per population (c.f. Cadotte and Davies 2016, p. 48) as a measure of genetic differentiation.
Results
We sampled the target land snails, P. concinnum s.l., G. similis s.l., and A. jagori, from the following seventeen limestone outcrops (with numbers for each species in brackets, respectively): Batangan (21, 3, 0), Mawas (29, 20, 0), New Location 1 (30, 9, 0), New Location 2 (43, 0, 0), Kampung (25, 28, 0), Keruak (27, 28, 0), Pangi (30, 43, 45), Tomanggong Besar (30, 28, 27), Tomanggong 2 (5, 40, 28), Tomanggong Kecil (29, 46, 23), Ulu Sungai Resang (0, 28, 0), Batu Payung (28, 15, 17), Tandu Batu (39, 28, 30), Batu Tai (0, 29, 0), Batu Tai Quarry (9, 1, 16), Gomantong (16, 7, 0), and Materis (0, 13, 15).
A. jagori was not found from outcrops in archipelago A, and is likely to be absent
from this archipelago. Voucher and museum identification numbers are listed in Table S2.2.
Population genetic analysis
Nei’s π was highest in G. similis s.l. and lowest in A. jagori (Figure 2.3A, Table S2.3). In G. similis s.l., data from both markers showed a relatively high π for populations from archipelago E, while the other two taxa show low values for this archipelago.
COI data in P. concinnum s.l. showed high values of π for archipelago C (cf. Schilthuizen et al., 2006).
Hrar was highest for archipelago B (Figure 2.3B, Table S2.3). Haplotype diversity was very similar for each taxon, with highest values for archipelago B (Table S2.3). We found little correlation between genetic diversity and outcrop or archipelago area, or archipelago outcrop number (Figure S2.2). There were positive trends between archipelago outcrop number and Hd, but correlations were non-significant (possibly due to a small number of data points). We found Hprivate to be very high for all three taxa, all populations/archipelagos, and all loci (Table S2.3), indicating haplotypes rarely occurred in more than one archipelago.
Fixation indices ΦST (based on a combination of all loci studied) were generally moderate between populations within archipelagos, and high between populations from different archipelagos (Table S2.4). Mean within-archipelago values (excluding non-significant values) for P. concinnum s.l., G. similis s.l., and A. jagori were 0.046, 0.024, and 0.045, respectively; between-archipelago means were 0.113, 0.034, and 0.079. Thus, populations are, at least on average, more closely related at the spatial scale of the archipelago, and less so at a larger scale. Fixation indices between
Figure 2.3 (A) Nucleotide diversity π (Nei and Li 1979) and (B) number of haplotypes based on
rarefaction, Hrar, for Plectostoma concinnum s.l., Georissa similis s.l., and Alycaeus jagori, grouped
by genetic marker and by archipelago (A to E) in the Lower Kinabatangan Floodplain. Error bars represent standard deviations.
Plectostoma concinnum s.l. ITS1 0 5 10 15 20 Hrar (A) A B C D E 0.000 0.025 0.050 0.075 0.100 π (B)
COI Georissa similis s.l.COI H3 COI Alycaeus jagoriH3 ITS1
A B C D E A B C D E A B C D E A B C D E A B C D E A B C D E
Archipelago
(A)
The AMOVA analyses revealed that most of the genetic variation present (PV) was at the level of ‘populations within an archipelago’ (50.4% to 65.7%; Table 2.1). Remaining genetic variation was explained mostly by ‘among-archipelago’ differences (18.5% to 38.7%). The last portion of variation was ascribed to genetic differences ‘within populations’ (7.1% to 15.8%). These results show that there were large genetic differences between populations within each archipelago. When repeating the AMOVA analyses while excluding data from archipelago A (results within parentheses in Table 2.1), ‘populations within an archipelago’ now explained 69.6% for P. concinnum
Table 2.1 Results of Analyses of MOlecular VAriance (AMOVA; Excoffier et al. 1992), by genetic marker, for Plectostoma concinnum s.l. and Georissa similis s.l. Sample grouping followed the hierarchical structuring of the region (region > archipelago > outcrop (~population)). Values in parentheses are for the alternative case in which data from archipelago A were excluded. Abbreviations: d.f. (degrees of freedom), SS (sum of squares), PV (percentage of variation). Significance tests: * p < 0.05, ** p < 0.005.
COI
d.f. SS PV Fixation index
Plectostoma concinnum s.l.
Among archipelagos 4 (3) 3281 (1524) 38.7 (20.5) 0.884** (0.875**)
Populations within archipelagos 9 (6) 2970 (2390) 54.2 (69.6) 0.381** (0.205**)
Within populations 329 (216) 559 (506) 7.1 (9.9)
Total 0.929** (0.901**)
Georissa similis s.l.
Among archipelagos 4 (3) 3249 (2301) 18.5 (17.3) 0.806* (0.788**)
Populations within archipelagos 11 (9) 5854 (5029) 65.7 (65.2) 0.185** (0.173*)
Within populations 344 (315) 1946 (1835) 15.8 (17.6)
Figure 2.4 Mantel correlograms for the three taxa studied, Plectostoma concinnum s.l. (solid line),
Georissa similis s.l. (dotted line), and Alycaeus jagori (striped line). Mantel test correlations (Pearson
method) are plotted versus geographic distance. Positive values indicate positive correlations between genetic and geographic distances; black squares indicate significant values. (A) Correlations tested on genetic differentiation, using Jost’s D (Jost 2008); (B) Correlations tested on a mean pairwise phylogenetic distance between samples per population. Inset artwork: Bas Blankevoort, Naturalis Biodiversity Center.
Geographic distance (km) (A) (B) 0 5 10 15 −0 .2 0. 0 0. 2 0. 4
Mantel test correlation
0 5 10 15 −0 .4 −0 .2 0. 0 0. 2 0. 4 0. 6 ITS1 H3
d.f. SS PV Fixation index d.f. SS PV Fixation index
2 (1) 1285 (850) 38.4 (37.5) 0.820** (0.919**) 9 (6) 1173 (1007) 50.5 (51.1) 0.384** (0.375**) 198 (122) 368 (339) 11.1 (11.4) 0.889** (0.887**) 4 (3) 191 (170) 36.4 (44.7) 0.792** (0.777**) 11 (9) 243 (214) 50.4 (42.9) 0.363** (0.447*) 264 (239) 83 (79) 13.2 (12.4) 0.868* (0.876**)
correlations for P. concinnum s.l and G. similis s.l. were close to zero (Figure 2.4A).
Phylogenetic and biogeographic analyses
We chose phylogenetic results from beast2 for each species complex based on models with maximum BF (Table S2.5). Our phylogenetic studies showed that individual snails from the same outcrop are genetically closely related, with a few exceptions (Figure S2.1). At the scale of the archipelago we often found more than one genetic clade (three times in P. concinnum s.l., five times in G. similis s.l., and once in A. jagori; Figure 2.5). As a result, populations on neighbouring outcrops, just several hundred metres apart, are often not each other’s closest relatives.
We estimated most genetic clades in P. concinnum s.l. to have originated around 1 million years ago (mean clade age 1.15 ± 0.53 MYA). In G. similis s.l. (2.67 ± 1.06 MYA) and A. jagori (2.14 ± 0.67) populations were older. It should be noted that mutation rates can actually differ substantially between these three distantly related taxa, which would alter (relative) clade ages.
Calculations of most probable ancestral ranges showed different patterns for the different species complexes. (Figure 2.6; for full output see Figure S2.3). Colonization and the origin of new genetic lineages were commonly associated with dispersal to non-adjacent outcrops (LDD and within-archipelago dispersal), making up 4 out of 7, 14 out of 17, and 3 out of 3 ‘speciation events’ in P. concinnum s.l., G. similis s.l., and A. jagori, respectively (Figure 2.6, Table 2.2; based on significantly supported clades only). Stepping-stone dispersal was found to be uncommon in each of three species complexes studied (rest of the ‘speciation events’). LDD was slightly more common in an upriver than downriver direction (9 versus 7 cases, respectively).
The repeated demographic Mantel test, using mean pairwise phylogenetic distances between samples per population, pointed at spatial-genetic relationships being positive up to 3 to 5 km distance between populations, with the most pronounced result for A. jagori (Figure 2.4B).
Figure 2.5 Results of phylogenetic analyses using beast2 for (A) Plectostoma concinnum s.l., based on COI and ITS1 markers; (B) Georissa similis s.l., based on COI and H3 markers; and (C)
Alycaeus jagori, based on COI, ITS1, and H3 markers. Colours of tip nodes correspond to the
different outcrops (for which see Figure 2.2). Width of tip nodes is scaled to genetic diversity within the respective clade. Height and numbers at the tips represent sample size, letters indicate archipelagos. Posterior probability values of the clades are 1, unless indicated at the node. Previously published morpho-species are indicated as follows: * P. simplex (Fulton, 1901), ** P. mirabile (Smith, 1893), and *** G. nephrostoma Vermeulen, Liew and Schilthuizen (2015) (see Methods S2.1 for details). Full phylogenetic trees can be found in Figure S2.1. Inset artwork: Bas Blankevoort, Naturalis Biodiversity Center.
0 2.5 5 7.5 10 12.5 MYA 0.1 0 15 16 16 30 124 0.99 0.97 0.51 0.96 0.97 0.56 7 24 27 13 28 14 13 28 9 20 15 14 46 4 28 31 38 *** * ** 0.81 0.63 0.73 0.27 0.91 0.96 0.43 0.51 43 29 27 35 21 26 25 16 28 46 14 15 21 (A) (B) (C) B B A A E E B A / E B B A B B E E B / (A) (B) (C) A/D B/C
Figure 2.6 Results of ancestral range reconstructions using the R package ‘BioGeoBEARS’ for (A)
Plectostoma concinnum s.l., (B) Georissa similis s.l., and (C) Alycaeus jagori. Letters with each
(ancestral) lineage refer to the archipelago found as most likely range. Dispersal type with each dispersal and colonization event is indicated by a filled (long-distance), half filled (crossing the river, or to a non-adjacent outcrop), or open (stepping-stone, i.e. to adjacent outcrop) circle. Large circles represent ‘speciation events’ with a posterior support of ≥ 95%; small circles have a support of < 95%. With each long-distance dispersal event, line type indicates a downriver (bold line) or upriver (dashed line) dispersal event. Reconstructions follow the phylogenies from Figure 2.5 pruned to ‘species’ level for each species complex. For full BioGeoBEARS output, see Figure S2.4.
Table 2.2 Counts of dispersal and colonization events for each of the three species complexes studied, Plectostoma concinnum s.l., Georissa similis s.l., and Alycaeus jagori. We distinguished long-distance dispersal, within-archipelago dispersal (crossing the river, or to a non-adjacent outcrop), and stepping-stone dispersal (to adjacent outcrop only). Counts of downriver and upriver dispersal and colonization events are given. Only ‘speciation events’ with a posterior support of ≥ 95% are included; numbers within brackets include all ‘speciation events’.
Dispersal type / taxon Plectostoma
concinnum s.l. Georissa similis s.l. Alycaeus jagori
Long-distance 3 (6) 10 (11) 3 (3)
Within-archipelago 1 (4) 4 (4) 0 (0)
Stepping stone 3 (6) 3 (4) 0 (1)
Downriver 2 (3) 4 (4) 1 (1)
Upriver 1 (3) 6 (7) 2 (2)
Discussion
Our results show that spatial-genetic structure in land snails in the Lower Kinabatangan Floodplain is composed of two forms: local structure, as isolation-by-distance, suggesting a stepping-stone model between nearby habitat islands, and regional structure, with random connections between more distant populations. This is true for all three species complexes studied, with haplotype diversity and haplotype numbers being highest within archipelago B, which has the highest number of outcrops. The patterns found are strongest for P. concinnum s.l. and G. similis s.l., while A. jagori shows relatively higher local dispersal, which is in agreement with its more generalist character, being found more often away from limestone. We found positive, non-sig-nificant correlations between haplotype diversity and archipelago island number. Most of the genetic diversity can be explained by the spatial scale of ‘populations within an archipelago’, as supported by both AMOVA and ΦST-values. Archipelago A is genetically most isolated. Most archipelagos have been colonized multiple times from within the region. Colonization through LDD and within-archipelago dispersal (i.e. non-stepping-stone dispersal) is associated with 78% of ‘speciation events’, highlighting the importance of dispersal over long distances in the origin of endemism in our system.
We find genetic diversity to vary with both taxon and archipelago (Figure 2.3, Table S2.3). Patterns in Hrar are broadly consistent between all three taxa and markers, with highest values for archipelago B. An explanation may lie in the larger island number and island size in archipelago B (Figure S2.2). Within each outcrop, snails will encounter a matrix of suitable and unsuitable microhabitats. Larger outcrops will have a higher number of such suitable microhabitats, which likely results in more genetic diversity within the outcrop (‘islands within islands’, cf. Holland and Hadfield 2002).
An explanation for the difference in haplotype diversity between taxa and outcrops may be the difference in age of the various populations. Bottlenecks (due to a small number of colonizing individuals) and subsequent founder effects are considered important consequences of island colonization events (Whittaker and Fernández-Palacios 2007 p. 168), and results include low genetic diversity and chance effects in the sorting of alleles. Therefore, low haplotype diversity may simply indicate a relatively young local population.
A combined effect of local dispersal and LDD, as we found in our system, was also described by Crandall et al. (2012) for marine snails. In studies on the limestone- dwelling snail Gyliotrachela hungerfordiana (Von Moellendorff, 1891) of Peninsular Malaysia, a similar pattern was found (Schilthuizen et al. 1999b, Hoekstra and Schilthuizen 2011) in which dispersal acts in two different forms: “successive colonization of ever further limestone outcrops”, and “additional long-range dispersal”, where the latter is
LDD vectors in land snails are given by Purchon (1977 p. 335) and Dörge et al. (1999). Two natural passive dispersal possibilities discussed by Dörge et al. (1999) may shed some light on our system. The first is running water, which the combination of heavy tropical showers and the proximity of the regularly flooding Kinabatangan River (Estes et al. 2012) in our system amply offers. Dörge et al. (1999) make special mention of the observations made by Boettger (1926) and Czógler and Rotarides (1938) of large numbers of land snails found in driftwood. However, we found that dispersal took place in both downriver and upriver directions. The second possibility, of passive dispersal by other animals, may be more likely. Both observations (Brandes 1951) and experiments (van Leeuwen and van der Velde 2012) have shown that snails can attach to bird feathers and survive for some time inside the gut of birds after having been swallowed (Matzke 1962, van Leeuwen et al. 2012, Wada et al. 2012). We expect other animals, such as wild boar and primates, to be other likely dispersal vectors.
In this study we have shown that populations of locally common taxa, by means of LDD, can reach distant islands. When reaching such new territory, these populations are likely to be genetically distinct from their neighbouring conspecifics, which can result in local endemic species. When this happens multiple times (but not too often) in a small region, such as the Lower Kinabatangan Floodplain, the result is a radiation of highly localized endemics. Set in a geographically complex habitat island system, we see here the on-going evolution of several species complexes of endemic land snails.
Karst habitats in Southeast Asia have been dubbed ‘biodiversity hotspots’ (Hughes 2017). Due to anthropogenic activities, such as quarrying, mining, deforestation, and tourist industry, limestone outcrops in Southeast Asia are rapidly disappearing (Sodhi et al. 2010, Hughes 2017). With many inhabiting species being endemic, the disappearance of each limestone outcrop results in the extinction of species, possibly including ones that have not yet been scientifically described. This is true for our study area and our study system of small land snails (Clements et al. 2008b). The result is genetic depletion (Harrison and Hastings 1996), possibly reducing species survival chances (Simberloff 1988). It is important to understand and conserve the
genetic complexities of the uniquely high levels of endemism we find in these island systems.
Acknowledgements
We thank Liew Thor-Seng of ITBC, UMS, Kota Kinabalu, Malaysia, for help with legal issues and laboratory work. Alex Pigot, Iva Njunjić, Leonel Herrera-Alsina, and Hamidin Braim assisted during fieldwork. This research was funded by NWO (grant 865.13.003, R.S.E.), the Malacological Society of London (2015, G.A.), and Treub- Maatschappij (2015, K.P.H.). Samples were collected under license of Sabah Biodiversity Council, permits JKM/MBS.1000-2/2 JLD.3 (167), JKM/MBS.1000-2/3 (99), JKM/ MBS.1000-2/2 JLD.4 (9), and JKM/MBS.1000-2/3 JLD.2 (78).
