On the origin of species assemblages of Bornean microsnails
Hendriks, Kasper
DOI:
10.33612/diss.124819761
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.
Research goal
During the past five years I studied the community assembly of Bornean microsnails,
and tried to answer the main question: Have Bornean microsnail communities randomly
assembled, and if not, what factors were of influence? Because natural ecological
communities (especially in the tropics; see below) are such tremendously complex
systems (Vellend et al. 2014), I set out on a multi-disciplinary approach to understand
better the various community aspects (Figure 5.1). It is one thing to test for community
(non-)neutrality to learn how much of an observed pattern can be explained by
simple, stochastic processes, but it is additionally interesting to test for (more)
deterministic processes (Chase and Myers 2011), such as the influence of community
interactions and environmental aspects. I decided to study (i) community composition
on habitat islands and to test for neutrality, (ii) phylogeography and demographics to
understand possible linkage among island populations and communities, and (iii)
individual diets to test for their variation among species and communities. Halfway
the project, I decided to add data on (iv) the communities’ environs, and (v) individual
microbiomes to test for possible influences of these on diet choice and the overall
community. The latter dataset could be added with little additional effort, after I
learned that laboratory techniques to obtain these data were very similar to the ones
I used to obtain diet data (Box 5.1).
Figure 5.1 The study of ecological communities can take a stochastic approach, such as a direct tests of neutrality, following e.g. the Unified Neutral Theory of Biodiversity (Hubbell, 2001), or a deterministic approach, such as traditionally studied by niche theory (Chase, 2014). In many cases, it is not at all clear if, and how much, can be explained by either deterministic or stochastic processes (such as described by neutral theory), and most likely both are important to some degree.
Ne
ut
ra
l t
he
or
y
Nich
e th
eo
ry
Stochastic Deterministic direct neutrality tests (SAD) diet µbiome geographyphylo- environment
? ? ?
Main research findings
Schilthuizen (2011) reported that the species abundance distributions (SADs) of
Bornean land snail communities follow a logseries-like distribution (Figure 1.3).
Using census data from the snail shells we collected from ca. 100 plots, distributed
over 15 limestone outcrops in the Lower Kinabatangan Floodplain, we studied the
SADs more closely (results presented in Box 5.2). We found that they could generally
be fitted well by a neutral model from the Unified Neutral Theory of Biodiversity
(UNTB; Hubbell 2001). However, especially at the local scale the most dominant species
were more abundant than the model predicts. Pooling community data into a regional
pool somewhat ‘evened out’ these extremes. Our phylogeographic and demographic
analyses highlighted the highly segregated gene pools in three target snail species-
complexes within the region (Chapter 2). Local populations were usually old
(> 500,000 years) and often originated from long-distance dispersal (LDD). These
findings thus support the application of the UNTB, which assumes neutral community
assembly to result, among others, from dispersal limitation. With regard to the
influence of traits (niche theory), we did not find support for a direct correlation
between plant diet diversity and snail consumer community diversity, although both
these communities were positively (though not strongly) correlated to a third
community: that of the snail microbiome (Chapter 3). We show that correlations
between these three communities may be the result of similar responses to
environmental variables. Most notably, consumer and diet diversities correlated
negatively with distance to anthropogenic activities (such as plantations and villages),
and positively with distance to the nearest cave entrance, which might have an
influence as a potential source of nutrients from guano runoff. Consumer community
and microbiome diversities both correlated positively with habitat island size,
which is in general agreement with the Island Biogeography Theory (MacArthur and
Wilson 1967). Snail plant diet richness, a trait that is potentially strongly associated
with non-neutral species interactions, differed significantly among the species we
studied, but we also found a significant positive correlation with snail size (Chapter
4). From these results, we cannot conclude that the snail diet is dictated by niche
partitioning, as larger snails are expected to show a richer diet simply from random
feeding with larger mouthparts. While many individual snails showed a sign of a
phylogenetic selection of plant diet items, suggesting individual snails have (strong)
dietary preferences, we cannot be certain that these results are not the by-product of
other, outside-community, selection pressures, such as snails hiding underneath
specific plants from predators, and only eating from these locations.
Box 5.1 Sampling the diet and microbiome from Bornean microsnail guts.
This box’ contents are an edited, summarized version of our research report as published in
The Malacologist, newsletter of The Malacological Society of London (Hendriks et al. 2019b).
Introduction
In the light of our interest in seemingly neutral communities of land snails on limestone outcrops in the Lower Kinabatangan Floodplain in Sabah (Schilthuizen 2011), Malaysian Borneo, we set up a project to study possible influences of traits on the community assembly. More specifically, we applied modern metabarcoding techniques that allow the reconstruction of snail diet and gut microbiome based on genetic data (Pompanon et al. 2012, Taberlet et al. 2012). With these data we tried to answer the question of whether the snail diet and/or microbiome influence the assembly of the species into communities (see Chapters 3 and 4).
Methods
We visited the Lower Kinabatangan Floodplain in November 2017. Our goal was to resample plots from which we had gathered community data (shells) during visits in 2015 and 2016, so that we could correlate newly gathered diet and microbiome data to previously gathered community data. We visited seven different limestone outcrops within the Lower Kinabatangan Floodplain. From each outcrop, samples were taken from three plots (four in a single outcrop, Keruak) along the base of the outcrop, with a between-plot distance of 50 m (sometimes more if not possible otherwise due to too dense vegetation). Plots measured two by two metres (Figure 5.2). We focussed on three target species of unrelated gastropod:
Plectostoma concinnum (Fulton, 1901), Georissa similis E. A. Smith, 1893 s.l.1, and Alycaeus
jagori Von Martens, 1859. Studies using standardised plots along a transect that spans both
limestone and non-limestone substrate have shown that the prosobranch microsnail genera
Plectostoma and Georissa tend to occur nearly strictly on limestone (Schilthuizen et al.
2003a), while Alycaeus was found also away from limestone, but in very low numbers (personal observations). We aimed to collect 40 individuals per target species per plot, with
1 Georissa similis E. A. Smith, 1893 was, until recently, considered a single species, endemic to the
Kinabatangan Floodplain. Hendriks et al. (2019a) suggested, based on phylogeographic studies, that the taxon could best be treated as a species-complex, characterized by high levels of endemism due to many long-distance colonization events. Recent taxonomic research by Khalik et al. (2019), based on combined phylogenetic and conchological studies, also suggests that the taxon is in fact best treated as a complex of closely related species: G. flavescens (found along the Kinabatangan River at limestone outcrops Pangi, Keruak, Tomanggong Besar, and Tomanggong 2), G. bangueyensis (widely distributed over northern and eastern Sabah), G. nephrostoma (along the Kinabatangan River on outcrops Keruak and Tandu Batu), G. xesta (widely distributed over Sabah), and G. similis (widely distributed over eastern Sabah). The samples used in the current study were identified as general ‘G. similis s.l.’ only. Because G. similis s.l. is composed of closely related, genetically nested species, evolutionarily (and likely ecologically) widely different from all other species in the region (Khalik et al. 2019), we treat the taxon as a species-complex here and refer to it simply as ‘G. similis s.l.’.
a minimum of 20, evenly distributed over the plot. To this end, we subdivided each plot into four quadrants of one by one metre (Figure 5.2), and sampled each quadrant for 30 minutes. In addition, we aimed to collect each individual snail from at least 10 cm distance from previous ones of the same species (not always possible for G. similis s.l. due to a combination of low numbers and our target number). All other snail species encountered within the plot were collected as well, with a maximum of 20 individuals per species per plot. We conserved samples in plastic tubes containing 96% ethanol and froze them directly in a styrofoam box filled with ice after the 30-minute search session had ended.
After registration of samples and deposition into the BORNEENSIS collection of the Universiti Malaysia Sabah (UMS), Kota Kinabalu, Malaysia, samples were exported to the Netherlands as a long-term loan. We performed our laboratory work at Naturalis Biodiversity Center, Leiden, the Netherlands, in January and February 2018. See Chapter 3 for a detailed description of the applied laboratory methods and bioinformatics. In short, we (1) double- checked identifications, (2) performed genomic DNA-extractions on the snail gut contents (see Figure 5.3 for an example of how faecal pellets usually are visible when shells are translucent), (3) amplified and sequenced both plant and microbial DNA from the gut using metabarcoding, and (4) identified genetic read data by comparison to benchmark databases. We used general Figure 5.2 Sampling microsnails from limestone. Plots were defined by a grid (made of rope) of two by two metres, subdivided into four quadrants of one by one metre (highlighted in yellow). Each quadrant was sampled for 30 minutes. Researchers (from left to right): Karen Bisschop, James Kavanagh, Anaïs Larue, and Hylke Kortenbosch. Photo by Kasper P. Hendriks.
genetic markers that proved to work effectively in metabarcoding studies before: rbcL for the plant diet (Hofreiter et al. 2000) and 16S V3-4 region for the microbiome (Liu et al. 2007, Andersson et al. 2008). Sequencing was performed on an Illumina MiSeq at BaseClear, Leiden, the Netherlands.
Results
We collected a total of 1,712 individual snails (excluding empty shells) belonging to 31 different gastropod species (Table 5.1). Plectostoma concinnum was particularly abundant in all but two plots. Alycaeus jagori, a rather conspicuous species due to its relatively large size, was usually common, too, but sometimes absent, such as from outcrop Batangan. Georissa
similis s.l., at two millimetres the smallest of the three target species, was often difficult to
locate, and in 11 out of 22 plots we found less than 10 individuals. It is clear that the three target species, as anticipated, were most abundant.
Sequence data for the diet, based on rbcL reads and after filtering, were obtained for 822 samples, covering 29 species. Sequence data for the gut microbiome, based on 16S V3-4 region reads and after filtering, were obtained for 823 samples, covering the same species as Figure 5.3 Specimen of the euconulid snail Kaliella calculosa (Gould, 1852) with faecal pellets in the gut clearly visible through the translucent shell. Such pellets were extracted from the snail in a sterile environment and used for genomic DNA-extraction. The white scale bar equals 1 mm. This is specimen BORMOL13455.01. Photo by Kasper P. Hendriks.
Table 5.1 Overview of collected species during our fieldwork in the Lower Kinabatangan Floodplain, November 2017, with totals of non-empty shells per species (n). A total of 1,712 specimens were collected, with highest numbers for the three target species (in bold).
NON-PULMONATE SNAILS n
Assimineidae
Acmella cyrtoglyphe Vermeulen, Liew & Schilthuizen, 2015 5 Acmella striata Vermeulen, Liew & Schilthuizen, 2015 14 Cyclophoridae
Alycaeus jagori Von Martens, 1859 359
Chamalycaeus sp. 4
Japonia kinabaluensis (E.A. Smith, 1895) 3
Japonia sp. 5
Leptopoma pellucidum (Grateloup, 1840) 1 Leptopoma sericatum (Pfeiffer, 1851) 15
Pterocyclos / Opisthoporus sp. 8
Diplommatinidae
Diplommatina asynaimos Vermeulen, 1993 1 Diplommatina calvula Vermeulen, 1993 3 Diplommatina gomantongensis (E. A. Smith, 1894) 5 Diplommatina rubicunda (Von Martens, 1864) 6 Plectostoma concinnum (Fulton, 1901) 893
Plectostoma simplex (Fulton, 1901) 31 Helicinidae
Sulfurina sp. 15
Hydrocenidae
Georissa kinabatanganensis Khalik, Hendriks, Vermeulen & Schilthuizen, 2018 33 Georissa similis E. A. Smith, 1894 s.l. 267 Georissa nephrostoma Vermeulen, Liew & Schilthuizen, 2015 5
PULMONATE SNAILS n
Ariophantidae
Everettia sp. 4
Macrochlamys tersa (Issel, 1874) 6
Microcystina appendiculata (Von Moellendorff, 1893) 2 Euconulidae
Kaliella accepta (Smith, 1895) 7
Kaliella barrakporensis (Pfeiffer, 1852) 2
Kaliella calculosa (Gould, 1852) 4
Kaliella scandens (Cox, 1872) 5
Rathousiidae
Atopos sp. 1
Trochomorphidae
Videna froggatti (Iredale, 1941) 1
Videna metcalfei (Pfeiffer, 1845) 4
Videna sp. 2
Valloniidae
Ptychopatula orcula (Benson, 1850) 1
for the diet. Most metabarcoding data could be classified down to at least the taxonomic level of the family (for plants) or phylum (for microbiome). With these data, it was possible to create a distribution of each individual’s plant diet and microbiome (examples for six specimens are shown in Figure 5.4). We found that the diet consisted of multiple plant families for each individual, with a diet richness of up to six plant families for A. jagori (excluding the category ‘unassigned’, which is a bin category for all rbcL reads that could be identified as seed plant, but not specifically to a plant family; causes may vary). Not all plant families were retrieved from all specimens, although families like Brassicaceae and Fabaceae were identified in most samples (see Chapter 4 for a detailed analysis). The microbiome for each specimen usually contained a large proportion of Proteobacteria, while 13 more bacterial phyla were found from the six example specimens (Figure 5.4).
Discussion
Our study demonstrates the potential of DNA metabarcoding techniques for the study of land snail community ecology and trait evolution. Direct observations of foraging snails, or a dissection of the gut and visual identification of its contents, are often hardly possible due to the small size of the many snail species in the communities we study. Visual identification of gut content is usually only possible down to broad categories (see e.g. Schamp et al. 2010). Instead, we reconstructed the diet using modern next generation sequencing techniques in combination with the latest genetic barcode reference databases for seed plants and bacteria. This allowed us to collect plant dietary and gut microbial overviews (both from one and the same individual) for many hundreds of snails in just several months of research (including preparations, fieldwork, laboratory work, and computational analyses). A study of similar size based solely on field observations would have taken many years and would likely be far less detailed.
We have created an overview of plant diet for 29 species of land snail, but it is important to be aware of the diet components that our technique cannot pick up, viz. all materials without chloroplast genes. Although the literature on the subject is sparse, Barker and Efford (2004) give a diet overview for various pulmonate families. At least for several families also encountered in our study (Ariophantidae, Euconulidae, Rathouisiidae, and Trochomorphidae), it is known that the diet includes also fungi and algae. Hence, to obtain a more complete diet overview, it is suggested for future studies to include genetic markers for these groups, too.
One sample worth mentioning specifically is that of a single individual of the rathouisiid
Atopos sp. we collected from Pangi (Figure 5.4). This is a genus of carnivorous slugs, in the
region mainly preying on P. concinnum (Schilthuizen and Liew 2008, Liew and Schilthuizen 2014c). We found its diet to include at least some different plants, which conforms to the description by Van Benthem Jutting (1953), who mentions fungi and plant materials in the diet of Atopos. Based on our dataset, however, we cannot know if the plants we encountered
A
B
C
1 2 3 4 5 6 0 25 50 75 10 0re
la
tiv
e
ab
un
da
nce
0 25 50 75 10 0re
la
tiv
e
ab
un
da
nce
Fa m ily Aquifo lia ce ae Ast er ace ae Br assi ca ce ae Fa ba ce ae M or ace ae O rch id ace ae Ru bi ace ae So la na ce ae un assi gn ed Ur tica ce ae Vi ta ce ae Ph ylu m Aci do ba ct er ia Act in ob act er ia Ar m at im on ad et es Ba ct er oi de te s Ch la m yd ia e Ch lo ro bi Ch lo ro fle xi Fi rm icu te s G em m at im on ad et es Ni tro sp ira e Pr ot eo ba ct er ia Sp iro ch ae te s TM 6 un assi gn ed Ve rru co m icr ob ia(A)
(B)
(C
)
The influence of scale
Over thirty years ago Wiens (1989) made a case for the explicit consideration of
spatial scaling in ecology, writing: “Plant ecologists long ago recognized the importance
of sampling scale in their descriptions of the dispersion or distribution of species. However,
many ecologists have behaved as if patterns and the processes that produce them are
insensitive to differences in scale and have designed their studies with little explicit
attention to scale.” Our results from comparisons of SADs from various regional and
local gastropod communities again highlight the importance of the spatial scale
(Box 5.2). The skew in abundance between dominant and middle-rank species is
stronger in local communities, and ‘evened-out’ at the regional scale, the result of
different species being the dominant species at different locations. Fits of the UNTB
point-mutation model to our empirical land snail data from the Lower Kinabatangan
Floodplain, by application of the recently published sampling strategy by Haegeman
in the gut derive directly from an herbivorous or omnivorous diet, or indirectly from eating other, herbivorous animals.
Acknowledgements
This research was funded by The Malacological Society of London (Early Career Research Grant, awarded to K.P.H.), as well as NWO (VICI grant, number 865.13.003, awarded to R.S.E.), KNAW (Fonds Ecologie Beurs, reference Eco/1711, awarded to K.P.H.), and the Leopold III-Fonds (awarded to K.B.). All samples were collected (permit numbers JKM/MBS.1000-2/2 JLD.6 (107, 112, 114, 116, and 118)) and exported (JKM/MBS.1000-2/3 JLD.3 (51)) under license of Sabah Biodiversity Council and have been deposited into the BORNEENSIS collection of UMS.
Figure 5.4 (Opposite page) Preliminary results of diet and gut microbiome compositions for randomly chosen individuals from six different species. (A) Drawings of species’ representatives of the six species: (1) Atopos sp. (data are for specimen BORMOL13662.01), (2) Alycaeus jagori Von Martens, 1859 (BORMOL13479.02), (3) Kaliella accepta (Smith, 1895) (BORMOL13455.01), (4) Diplommatina calvula Vermeulen, 1993 (BORMOL13425.01), (5) Georissa similis E. A. Smith, 1893 s.l. (BORMOL13410.01), and (6) Plectostoma concinnum (Fulton, 1901) (BORMOL13403.01). Drawings by Bas Blankevoort, Naturalis Biodiversity Center. Black scale bars equal 1 mm. (B) Relative distribution of seed plant families found from the diet, based on rbcL read numbers and normalized to 100%. (C) Relative distribution of microbial phyla found from the gut micro- biome, based on 16S V3-4 region and normalized to 100%.
and Etienne (2017), showed that, especially at the local scale, the few dominant species
in communities are more abundant than expected under neutral theory (Box 5.2).
That is, the UNTB appeared to fit empirical data at the larger spatial scale better than
at smaller spatial scales, and our future analyses should include different datasets
(such as presented in Box 5.2) and find out whether this is a general pattern.
The influence of stochasticity is considered to be stronger at smaller spatial
scales, resulting in larger between-grain variations when grain size is smaller (grain
being the spatial scale of sampling, such as the plot, or the location; Leibold et al.
2004). The explanation suggested by Leibold et al. (2004) is that at larger spatial scales,
there is a stronger coupling between physical (climate, vegetation) and biological
processes, whereas at smaller spatial scales, the effects of interactions between
individuals are stronger. This could explain why, in our study system, local dominance
is higher, and different species are dominant in different locations. Overall, all snail
species in the communities we studied are adapted to the general tropical, limestone
environment, but within localities, they compete. In his ambitiously titled 2014
commentary, Spatial scale resolves the niche verses neutral theory debate, Chase even
argues that any neutrality we find at small spatial scales is really only caused by
individuals from a niche-structured community at a larger scale that were pushed
towards peripheral, sub-optimal locations (Chase 2014). Hence, neutrality would be
what we perceive because we study communities at the wrong (too small) spatial
scale. While this idea seems plausible, and was backed up by a large study of tropical
tree distributions at different scales (Garzon‐Lopez et al. 2014), it seems at odds with
our preliminary results in Box 5.2 that suggest a better fit of the neutral model to the
regional Kinabatangan snail dataset. More work in this direction is needed, preferably
including multiple community datasets, the basis of which is presented in Box 5.2.
The influence of complexity
As stated in Chapter 1 (Introduction), the main objective of science is to find patterns
and to infer generalisations of the world around us. But to start off such a daunting
task, assumptions need to be made, and study systems defined, simply to be able to
handle the enormous complexity encountered in nature. Community ecologists use
different approaches to reduce the apparent complexity: they focus on well-defined
food webs, specific interactions between species (e.g. predator and prey), mutualistic
networks, or, like in this thesis, horizontal communities (Vellend 2016 p. 12).
But complexity is not constant throughout ecosystems around the world
(Figure 5.8). It is generally assumed that ecological complexity increases towards
lower latitudes and altitudes, and reaches a maximum in tropical rainforests, as was
shown for tree communities (Holdridge 1967, Lugo and Brown 1991). Furthermore,
Box 5.2 Community assembly using SADISA.
This box contains preliminary, unpublished research results that are interesting or important in the light of this thesis’ Synthesis.
Introduction
Species abundance distributions (SADs) are very useful in describing ecological communities: their interpretation is straightforward, and they capture much more information than single-number metrics like richness, diversity, and evenness (McGill et al. 2007).
One of the major benefits of the Unified Neutral Theory of Biodiversity (UNTB) is that it predicts empirical SADs remarkably well. It provides sampling formulas, which give the likelihood of the observed abundance data given a set of model parameters (Haegeman and Etienne 2017). These formulas can be used to fit the models to empirical data and these models can subsequently be compared using likelihood-based information criteria, such as Akaike Information Criterion (AIC). An exact sampling formula was published by Etienne (2005), but this was computationally demanding, especially for larger sample sizes, and therefore not easily accessible to the general ecologist.
A leap forward was recently made by the publication by Haegeman and Etienne (2017) of a new method to compute sampling formulas, which was implemented in the R package ‘sadisa’ (Species Abundance Distributions under the Independent Species Assumption). The package allows very fast fitting of the original point-mutation model to observed abundance data, but also of a whole suite of more recently proposed models, such as a multiple- sample model (Etienne 2007, 2009), a multiple-guilds model (Janzen et al. 2015), a random-fission model (Etienne and Haegeman 2011), a protracted speciation model (Rosindell et al. 2010), and a species-level density dependence model (Haegeman and Etienne 2017). The key innovation that made this fast model fitting possible, was to replace the analytical assumption of constant community size (‘zero-sum assumption’) with the assumption of independent species abundances, after which new, computationally less demanding, sampling formulas could be derived. Based on model fits to empirical data using both the original and newly developed sampling formulas, Haegeman and Etienne (2017) showed almost identical results for the models with and without zero-sum assumption, proving that their new approach is useful.
We used SADISA to study the fit of the original point-mutation neutral model to our empirical Bornean land snail community data, plus a collection of seven other worldwide gastropod abundance datasets from the literature. We fitted the point-mutation model to all regional datasets (i.e. pools of the local datasets), using both a single-sample and multiple- sample approach. For all local communities from all the datasets, we compared three different models (point-mutation, protracted-speciation, and density-dependence models), selected the best model, and tested for correspondence between best models and community richness. Because we noted a difference in abundance of the commonest species (‘dominance’) between regional and local datasets, we studied model predictions for our Bornean dataset.
The following results are a preliminary overview from analyses done so far, and we intend to publish an elaborate analysis later.
Methods
For our land snails on limestone outcrops from the Lower Kinabatangan Floodplain, Borneo, we used community data based on shells collected from the soil, for which sampling details are given in the Methods section of Chapter 3. Additional gastropod community data were collected from literature and unpublished data from peers (Table 5.2): Australian land snails (Cameron and Pokryszko 2005), European fen snails (Horsák et al. unpublished), Lake Tanganyika freshwater snails (Meyer et al. 2011), Afromontane land snails (Wronski et al. 2016), and Atlantic Ocean Pteropoda and Heteropoda (Burridge et al. 2017).
We used the function ‘SADISA_ML’ from the R package ‘sadisa’ to fit two neutral point-mutation models (both single-sample and multiple-samples) to the regional datasets, where regional data were simply the sum of the local species abundances. To overcome model optimizations to get stuck in local optima, we used all combinations of different starting values for θ and I (namely 10 and 10,000). Default settings were used for other function arguments. This resulted in four fitted models for each dataset, the best of which was selected based on lowest AIC value.
We then fitted three different models to all local datasets (with a sample size of at least 250, because fitting models to too small datasets can result in errors): the point-mutation (‘pm’) model, the protracted-speciation (‘pr’) model, and the density-dependence (‘dd’) model. Again, each model was started with all combinations of different starting values for θ and I (namely 10 and 10,000), with additionally in the pr-model φ (again 10 and 10,000), and in the dd-model α (-10.0 and 0.9). Best results were selected, based on AIC, for each of the three models, and subsequently the three models were compared mutually for each location, again based on AIC. Because for the different local communities, different models performed best, we tested correspondence between best models and an important community characteristic: richness (based both on observed data and Chao1-estimates, the latter taking into account data not sampled and thereby correcting for different sample sizes).
Finally, we used θ and I-values from the optimal point-mutation models for the Borneo datasets (both regional and local) and function ‘SADISA_sim’ to simulate 999 neutral communities, and compared their SADs to those of the observed community. Because we noted a difference in accuracy of fit between regional and local datasets, we briefly studied differences in dominance between all regional and local datasets.
T abl e 5 .2 O ve rv ie w o f g as tr op od c om m u ni ty d at as et s i nc lud ed i n t hi s s tud y, w it h c om m u ni ty d es cr ip ti on s a nd r es u lt s f ro m S A D IS A po in t-m ut at io n m od el s f ro m b ot h s in gl e-sa m pl e a nd m u lt ip le -s am pl es a pp ro ac he s. A bb re via ti on s a re a s f ol lo w s: n – n u m be r o f l oc al co m m u ni ti es w it h t he r eg io na l d at as et , n>2 50 – n u m be r o f l oc al c om m u ni ti es w it h a s am pl e s iz e o f a t l ea st 2 50 , S S – r eg io na l s am pl e s iz e, RO – r eg io na l o bs er ve d r ic hn es s, R C ha o1 – r eg io na l C ha o1 e st im at e o f r ic hn es s, D R – r eg io na l d om in an ce , DL – m ea n l oc al d om in an ce f ro m lo ca l c om m u ni ti es w it h a s am pl e s iz e o f a t l ea st 2 50 , w it h l ow er a nd u pp er 2 .5 % q u an ti le s b et w ee n b ra ck et s. θ a nd I a re t he m od el pa ra m et er e st im at es . N ot e t ha t l og -li ke li ho od v al ue s f ro m s in gl e-sa m pl e a nd m u lt ip le -s am pl e a pp ro ac he s c an no t b e c om pa re d. D at as et ha bitat sour ce n n>250 SS RO RChao1 DR DL appr oach loglik A IC θ I Borneo terr estrial this study 15 14 26063 69 72.0 0.23 0.39 (0.21-0.61) single -211.93 427 .86 1.13E+1 1.85E+3 m ultip le -2255. 72 4515.45 1.62E+1 1.85E+1 U ganda, A frica terr estrial W rons ki et al. (2016) 40 4 4823 91 97 .4 0.29 0.31 (0.15-0.56) single -140.6 7 285.34 1.80E+1 2.85E+3 m ultip le -2220. 78 4445.56 2.90E+1 8.07E+0 Lak e T angan yika, A frica fr es hwater Me yer et al. (2011) 20 5 3722 11 11.0 0.47 0.91 (0. 78-0.99) single -53.64 111.28 7.42E+0 3.65E+0 m ultip le -461.66 927 .32 2.66E+0 1.84E+0 A tlantic O cean (Heter opods ) marine Burridge et al. (2017) 25 1 1812 19 19.0 0.15 0.54 (0.54-0.54) single -69.46 142.91 7.21E+0 2.09E+1 m ultip le -700.24 1404.49 4. 75E+0 4.68E+0 A tlantic O cean (Pter opods ) marine Burridge et al. (2017) 30 10 7238 42 44.5 0.32 0.6 7 (0.45-0.91) single -109.86 223. 73 6.11E+0 1.48E+4 m ultip le -1393. 72 2791.44 1.11E+1 5.07E+0 N ew S out h W ales, A ustr alia terr estrial Camer on and Po kr ys zk o (2005) 2 2 8731 38 40.0 0.22 0.40 (0.30-0.50) single -115.84 235.6 7 5. 77E+0 3.15E+3 m ultip le -228.59 461.17 1.91E+1 1.01E+1 Centr al Eur ope (liv e data ) terr estrial Hor sák et al. (unpu blis hed) 15 12 9323 52 59.0 0.30 0.50 (0.20-0. 71) single -135.08 27 4.16 4.39E+8 7.26E+0 m ultip le -1305.42 2614.83 1.47E+1 7.83E+0 Centr al Eur ope (s he ll data ) terr estrial Hor sák et al. (unpu blis hed) 15 11 8720 58 59.5 0.20 0.24 (0.16-0.33) single -156 .90 317 .80 3.96E+1 2.20E+1 m ultip le -1438.33 2880.65 1.60E+1 1.01E+1
Results
We compared point-mutation neutral models for each of the regional datasets (Table 5.2). Because log-likelihoods from single-sample and multiple-sample approaches cannot be directly compared, we need further analyses to study which approach results in a better model fit. We used results on θ and I from best models from single-sample approaches to simulate neutral communities (see below).
We found that for different local abundance datasets, different models fit best. In 44 out of 59 locations the point-mutation model fits best, whereas in 15 out of 59 locations the density-dependence model fits best (Figure 5.5, Table S5.1). Differences were significantly correlated with local community richness (both observed richness and richness from Chao1 estimation), with the dd-model performing better when local community richness was higher. Simulations from optimal point-mutation model parameter-values (from Table 5.2 and Table S5.1) for the Borneo datasets (regional and local) were characterized by a wide band of variation (Figure 5.6). Upon visual inspection of SADs, mean values from neutral model simulations indicated an underestimation of the observed abundances of the most dominant one (locally) to three (regionally) species, and an overestimation of the abundances of the middle-rank species. This discrepancy was strongest for the local communities.
Comparisons of regional and local SADs from all datasets showed a consistent difference between local and regional patterns, with dominant species being more abundant in the local (at least the mean values) than in the regional communities (Figure 5.7, Table 5.2).
Discussion
Our results show the potential of fitting neutral models to observed community abundance data. We found that the density-dependence model, a non-neutral model in which the per capita birth rate varies with species abundance (Haegeman and Etienne 2017), outperforms the point-mutation model in many speciose local communities, suggesting that indeed the abundance of each species has an influence on the assembly of the community when richness increases.
We also found that dominance in local communities is on average higher than dominance regionally, in all of the datasets we studied, resulting in stronger-skewed abundance distributions locally. Thus, spatial scale is very important in community assembly research. The point-mutation model apparently has difficulty in fitting strongly skewed data. As explained by Rosindell et al. (2011), it is these deviations from neutrality that help us search for important biological details that are important in explaining community assembly.
We suggest to expand this study by further comparing single-sample and multi-samples approaches for regional abundance data.
Figure 5.5 Local community species richness based on (A) observed richness (RO) and (B) the
Chao1 estimate (RChao1), versus best metacommunity model. Community data from studies
listed in Table 5.2, but local communities with a sample size smaller than 250 were excluded. Three different metacommunity models were fitted to the data from each location: the point- mutation model (‘pm’), the protracted-speciation model (‘pr’), and the density-dependence model (‘dd’). The best-fitting model for each location was chosen based on lowest AIC-value. The pr-model was never selected as the best model, and therefore not shown in the plots. A Wilcoxon signed-rank test (for non-normal data) was used to calculate significance of difference in means between the two models and their associated community richness.
p < 0.001 10 20 30 40 model RO • p < 0.001 20 40 60 pm (44) (15)dd model RCha o1 pm (44) (15)dd
(A)
(B)
Figure 5.6 Species abundance distributions (SADs) from observed data (thick black lines) and 999 simulated neutral communities (red violins), for (A) our regional dataset from the Lower Kinabatangan Floodplain, and locations (B) Keruak, (C) Tomanggong 2, and (D) Tomanggong Kecil. Neutral community data were generated using R package ‘sadisa’, function ‘SADISA_ sim’, based on values for θ and I from model optimization using function ‘SADISA_ML’ with a point-mutation model. Note how observed SAD lines show a steep drop at the lowest ranks, i.e. the dominant species, then a dip, and then a pattern well-fitted by the neutral data. In general, the neutral models underestimated the abundances of the most dominant one or two species, and overestimated the abundances of the middle-rank species.
6 1 6 6 1 0.01 0.1 1 10 0.1 1 10 0.1 1 10 1 species rank pe rce nt re la tiv e sp eci es ab un da nce 0.01 0.1 1 10 1 species rank 11 21 31 41 51 61 71 81 91 11 16 21 26 31 36 41 46 11 16 21 26 31 36 41 46 51 56 11 16 21 26 31 36 41 46 51
(A)
(C)
(B)
(D)
Figure 5.7 Species abundance distributions (SADs) from observed data for eight different gastropod community datasets. In each panel, the red lines show the SADs for the local communities and the thick black line the SAD for the regional community (i.e. the collection of the local communities). Note that in the local communities the dominant species is more dominant than in the regional community (see insets). Therefore, we need to keep in mind that the scale at which we study a community influences our findings, and possibly alters the fit to a neutral assembly process. Datasets are as follows: (A) Lower Kinabatangan Floodplain land snails, from this study; (B) Australian land snails, from Cameron and Pokryszko (2005); (C) European fen land snails, collected alive, from Horsák et al. (unpublished); (D) European fen land snails, collected as shells, from Horsák et al. (unpublished); (E) African rift lake freshwater snails, from Meyer et al. (2011); (F) African land snails, from Wronski et al. (2016); (G) Atlantic Ocean Pteropoda, from Burridge et al. (2017); and (H) Atlantic Ocean Heteropoda, from Burridge et al. (2017).
0.01 0.1 1 10 100 0 20 40 60 0.01 0.1 1 10 0 10 20 30 0.01 0.1 1 10 100 0 10 20 30 40 50 0.01 0.1 1 10 0 20 40 60 0.1 1 10 100 3 6 9 pe rce nt re la tiv e sp eci es ab un da nce 0.1 1 10 100 0 25 50 75 0.01 0.1 1 10 100 0 10 20 30 40 species rank 0.1 1 10 100 5 10 15 species rank
0
10
20
30
40
50
0.01
0.1
1
0
3
6
9
0.1
1
10
100
0
(F)
1
10
100
(H)
0
20
40
60
0
0
10
20
30
40
50
0.01
0.1
1
10
0
(D)
3
6
9
0.1
1
10
100
0
(F)
100
(H)
0.01
0.1
1
10
100
0
20
40
60
(A)
0.01
0.1
1
10
100
0
10
20
30
40
(C)
10
100
pe
rce
nt
re
la
tiv
e
sp
eci
es
ab
un
da
nce
(E)
0.01
0.1
1
0
20
40
60
0.01
0.1
1
10
100
0
10
20
30
40
(C)
0.1
1
10
100
pe
rce
nt
re
la
tiv
e
sp
eci
es
ab
un
da
nce
(E)
0
20
40
60
0.01
0.1
1
10
100
0
10
20
30
40
(C)
0.1
1
10
100
3
6
9
pe
rce
nt
re
la
tiv
e
sp
eci
es
ab
un
da
nce
(E)
100
(G)
0.01
0.1
1
0
10
20
30
40
0.1
1
10
100
3
6
9
pe
rce
nt
re
la
tiv
e
sp
eci
es
ab
un
da
nce
(E)
1
10
100
(G)
0
20
40
60
0.01
0.1
1
10
0
(B)
0
10
20
30
40
50
0.01
0.1
1
10
0
(D)
10
100
(F)
0
20
40
60
0.01
0.1
1
0
0
10
20
30
40
50
0.01
0.1
1
10
0
(D)
0.1
1
10
100
(F)
(
E
)
(F)
(G)
(H)
(C)
(D)
(A)
(B)
there is a general belief that species diversity increases with habitat complexity
(MacArthur and MacArthur 1961, August 1983). Thus, we need to keep in mind that
the assembly of tropical snail communities in the Lower Kinabatangan Floodplain is
more complex than we might have expected at the start. In this light, it is useful to
consider how we defined communities of snails living on the limestone outcrops, and
how habitat complexity in the form of different microhabitats may be important for
the survival of different snail species (Box 5.3). Our assumption was that all land
snails together form a single horizontal community, i.e. belong to the same guild.
Some of our preliminary results in Box 5.3 suggest that about half of the snail species we
studied are actually found in just a single microhabitat within the overall limestone
habitat. This might be interesting in combination with our finding that diet richness
and snail size are positively correlated in prosobranchs, but apparently not in
pulmonates (in the case of higher rank classification of the diet; Chapter 4), although
such higher taxonomy differences were not found for microhabitat use (Box 5.3).
Another level of complexity is not formed by the definition of the community,
or its members habitat uses, but by the taxonomy of the communities’ members. In the
species-rich communities of snails we studied, the taxonomic detail in knowledge of
the different taxonomic groups varies (Box 5.4). While some groups have been
studied rather thoroughly, such as the family Diplommatinidae (Vermeulen 1991,
1994, 1996a, Webster et al. 2012, Liew and Schilthuizen 2014c, Liew 2019b) and, very
recently, the genus Georissa (Khalik et al. 2018, 2019), knowledge on others is still very
limited, as is the case in the genera Japonia and Microcystina. Many species are rarely
encountered alive, and species descriptions are based mainly on shells. In Chapter 2
we showed for three species complexes (Plectostoma concinnum (Fulton, 1901) s.l.,
Georissa similis E. A. Smith, 1893 s.l., and Alycaeus jagori Von Martens, 1859) that
genetic variation within the region is surprisingly large, with same-species
populations on neighbouring outcrops in some cases having different haplotypes
with several mutations in the mitochondrial COI gene. This suggests these populations
have been separated for a long time, and it is possible that their ecologies have
diverged over time, too, adapting to (slightly) different circumstances (different
predators, competitors, environs, communities). It may thus be questioned whether
such different populations are not best considered different species, but such
systematic and taxonomic considerations were outside the scope of current study.
Acknowledgements
We thank Robert Cameron for sharing community abundance data on Australian land snails, and Michal Horsák and colleagues for sharing unpublished abundance data on European fen snail communities.
Concluding remarks and future directions
In this thesis, I presented our studies aimed at answering the one main question:
Have Bornean microsnail communities randomly assembled, and if not, what factors
were of influence? I tried to show deviations from neutral assembly theory, not to
reject neutral theory, but with the goal to increase our understanding of how these
communities came to be, compared with the null model of full neutrality. I elucidated
how the neighbouring populations of several prominent community members have
been separated for hundreds of thousands of years, and how their evolution has been
influenced by long-distance dispersal, which means that local communities really
are local entities. I found that, contrary to findings in many other herbivore-plant
systems, direct correlations between the snail communities and their plant diet
are absent, or at least masked by their microbiomes and the joint responses to
environmental variables. I also showed how plant diet richness can differ strongly
among community members, but is correlated with snail size, and thus likely still the
result of random feeding behaviour.
Figure 5.8 Holdridge life zones system with values within hexagons indicating ‘forest complexity index’, showing how complexity increases towards tropical lowlands, i.e. bottom right (Holdridge, 1967). The complexity index is the product of basal area, tree height, tree density, and the number
Based on my results, I was not able to unequivocally reject the idea that stochasticity
plays an important role in the assembly of Bornean microsnail communities. Still, I
found several indications that species differ in their ecological traits (community
correlations with the microbiome and environment; individual plant dietary
preferences; the dominant species are more abundant than expected from neutrality;
and differences in microhabitat preference). But these differences may be the result
of various other selection mechanisms, such as community members trying to avoid
predation or desiccation.
To better understand the assembly of Bornean microsnail communities, I suggest
that the following issues receive more attention in the future. First, community
composition is clearly influenced by spatial scale. At the very least, it should be
studied how, and to what extent, spatial scaling influences community assembly, and
whether neutral theory better fits the local or regional spatial scale. The outcome of
such studies should be helpful in understanding community assembly if differences
between scales can be explained by (environmental) variables (think of the
micro-habitat!). Second, the communities of Bornean microsnails are of a complex composition,
with still partly unexplored taxonomic variation. Future investigations should include
an analysis of the influence of this unknown complexity. This could be done by a
hypothetical fission-by-location of one or more species into multiple species, thereby
simulating the raise of a species-complex to a collection of full species.
Even though the communities of Bornean microsnails offer many advantages to
the community ecologist, the past five years have shown that understanding their
assembly is a hard nut to crack. I made progress, but propose to continue these
studies, because a proper understanding of these communities would benefit both
community ecologists and the conservation of a unique, rich ecosystem under threat
from an industrialised world.
Box 5.3 Microsnail microhabitat study.
This box contains preliminary, unpublished results.
Introduction
So far in this thesis, when we described the communities of microsnails on limestone outcrops in the Lower Kinabatangan Floodplain, we considered these communities to be composed of the totality of same-guild species that co-exist in a given location, so-called ‘horizontal communities’ (Vellend 2016 p. 11). This implicitly assumed that the community members are very similar ecologically, and are likely to share resources, in turn suggesting possible competition. In Chapter 4 we describe how the plant diet of the community members largely overlaps, with different species eating from the same plant families. But there are other ways in which competing community members can release competition (Hutchinson 1961). One of the most obvious ways is for the different species to harvest resources from very specific ‘microhabitats’, which is not uncommon in gastropods (Harris and Charleston 1977, Cook et al. 1985, Cowie 1985, Chapman 1994, Otero-Schmitt et al. 1997, Crowe and Underwood 1998, Książkiewicz et al. 2013).
Methods
During our fieldwork in the Lower Kinabatangan Floodplain in 2015 and 2016 we collected microsnails for phylogenetic and population genetic purposes (see Chapter 2 for detailed Method). At one or two plots from each of 12 limestone outcrops (Batangan plot 3, Batu Payung plot 1, Kampung plots 4 and 6, Keruak plot 7, Materis plots 4 and 5, Mawas plot 1, Pangi plot 2, Tandu Batu plot 1, Tomanggong Besar plot 1, Tomanggong 2 plot 5, Tomanggong Kecil plot 5, and Ulu Sungai Resang plot 3), we additionally performed a 30-minute census for live snails from each of five different microhabitats: bare rock, mossy rock, bark, fresh leaves, and dead leaves (Figure 5.9). We only selected plots that had all five microhabitats. We recorded, for each plot and per microhabitat, the different snail species and their numbers. Because we performed this fieldwork with different researchers and needed to exclude systematic errors resulting from individual differences in experience, search strategy, and
Figure 5.9 Definition of five different microhabitats in which microsnails can be found on limestone outcrops in the Lower Kinabatangan Floodplain. Photos by the Kasper P. Hendriks.
technique, the person to sample a specific microhabitat was randomized by drawing straws just before the start of each search. To prevent cross-contamination in the form of snails having unintentionally been moved from one microhabitat to another (e.g. by heavy rains), we scored only snails found alive, and ignored empty shells (which are known to be flushed across short distances).
We created a ‘microhabitat community matrix’ from totals of collected snail species. We counted the number of species found from each microhabitat community, and the number of species found only from each microhabitat (and thus not from any other). For each species, the number of microhabitats from which it was collected was counted. We calculated distances among microhabitat communities based on occurrence and abundance data and used principal component analysis on community occurence and abundance data to study microhabitat differences.
Results
We found different numbers of specimens and species from different microhabitats (Table 5.3). The mossy rock microhabitat was most species rich (16 species), while abundance was highest on bare rock (786 specimens, of which 78.1% of individuals belonged to Plectostoma concinnum). Each microhabitat had one or more species unique to it, i.e. species not found from any other microhabitat we studied, although this is of course subject to search intensity. Many species were found in a subset of microhabitats only, with 17 species (50.0%) found in just a single microhabitat (Figure 5.10). Georissa similis s.l. was the only omnipresent species, found in all five microhabitats. At higher snail taxonomic levels (Neritimorpha, Prosobranchia, and Pulmonata) the number of microhabitats used did not differ.
Figure 5.10 Count of species found from one or more microhabitats. Most species (17) were found in a single microhabitat only. Georissa similis s.l. was the only species found in all five microhabitats. Higher order taxonomic groups (indicated by different colours) were found throughout.
0 5 10 15 1 2 3 4 5 number of microhabitats sp eci es co un t Georissa similis s.l. Taxonomy Pulmonata Neritimorpha Prosobranchia
Table 5.3 (A) Community matrix and (B) summary of results from a microsnail microhabitat study on limestone outcrops in the Lower Kinabatangan Floodplain. Data are totals from 14 plots from 12 different limestone outcrops.
Microhabitat bare
rock mossy rock bark fresh leaves leavesdry (A) Community matrix
Acmella cyrtoglyphe Vermeulen, Liew & Schilthuizen, 2015 11 Allopeas clavulinum (Potiez & Michaud, 1838) 1 1 Allopeas gracile (Hutton, 1834) 1
Alycaeus jagori Von Martens, 1859 33 42 7 Arinia borneensis E. A. Smith, 1894 1 12
Chamalycaeus sp. 3
Charopa jugalis sp. n. 1
Diplommatina asynaimos Vermeulen, 1993 2 1 1 Diplommatina gomantongensis (E. A. Smith, 1894) 1
Everettia sp. 1
Georissa gomantongensis E. A. Smith, 1894 7 Georissa nephrostoma Vermeulen, Liew & Schilthuizen, 2015 10 4 9 Georissa kinabatanganensis Khalik, Hendriks, Vermeulen &
Schilthuizen, 2018
3
Georissa similis E. A. Smith, 1894 s.l. 89 56 132 2 41
Japonia sp. 1
Kaliella calculosa (Gould, 1852) 4 Kaliella dendrophila (Van Benthem Jutting, 1950) 4 Kaliella microconus (Mousson, 1865) 2 4 3 Kaliella punctata Vermeulen, Liew & Schilthuizen, 2015 1
Kaliella scandens (Cox, 1872) 12 28 1 Leptopoma pellucidum (Grateloup, 1840) 1 9 Leptopoma sericatum (Pfeiffer, 1851) 2 4
Leptopoma sp. 2 1
Leptopoma undatum (Metcalfe, 1851) 1 1 Microcystina appendiculata (Von Moellendorff, 1893) 5
Microcystina microrhynchus Vermeulen, Liew & Schilthuizen, 2015 1 Philalanka kusana (Aldrich, 1889) 1 1 Plectostoma concinnum (Fulton, 1901) 614 363 3
Plectostoma simplex (Fulton, 1901) 20
Ptychopatula orcula (Benson, 1850) 4 1 Pupina hosei Godwin Austen, 1889 2
Sulfurina sp. 1 7 47
Videna froggatti (Iredale, 1941) 1 1
Videna sp. 1
(B) Summary
Specimen count 786 496 165 118 61
Species count 12 16 8 14 11
Species count, unique to microhabitat 5 7 1 3 1
Shannon diversity 0.85 1.02 0.83 1.90 1.24
Based on occurrence data, communities on bark and dry and fresh leaves were most similar (Figure 5.11). Based on abundance data, communities on bare and mossy rock were most similar, likely the result of the dominance of P. concinnum and Alycaeus jagori in both these microhabitats.
Figure 5.11 Redundancy analysis (RDA) of community matrix presented in Table 5.3A, based on (A) occurrence data (i.e. presence vs. absence), and (B) abundance data. For readability, species loadings (arrows) are shown for the five most abundant species only. Note Georissa
similis s.l. had a loading of zero in plot (A) and was not drawn
mossy rock bare rock
barkdry leaves
fresh leaves
Alycaeus jagori Kaliella scandens
Plectostoma concinnum Sulfurina martensi -1.0 0.0 1.0 −1.5 −1.0 −0.5 0.0 0.5 1.0 PC1 PC 2 PC 2 bare rock mossy rock bark dry leaves fresh leaves Alycaeus jagori Georissa similis s.l. Kaliella scandens Sulfurina martensi -1.0 0.0 −1.0 −0.5 0.0 0.5 1.0 1.5 PC1 Plectostoma concinnum 0.5 1.5 -0.5 -1.5 -0.5 0.5 1.0 1.5
(A)
(B)
Discussion
Our results show that the communities of snails on limestone outcrops in the Lower Kinabatangan Floodplain show a rather strong (spatial) substructure, with different species having different preferences for microhabitats. We did not study the causes of these differences, but expect these to be associated with differences in food preferences or strategies to evade predation. Future studies on community assembly in snails should profit from these results by including microhabitat occupancy in the analyses, for example by definition of different guilds, or by including microhabitat preference as a covariate in statistical tests.
Our results are based on a rather small dataset and it would be interesting to expand this study with more data from different locations. Furthermore, our definition of the different microhabitats is rather arbitrary, and may not at all be how snails perceive their environment. It is also possible that snails occupy different microhabitats at different times of the day or year, or change their preference during their lifetime or reproduction cycle. We thus suggest future studies to be repeated at different times and to score explicitly the age of each individual, when possible.
Box 5.4
Microsnail species definition: there’s still much
to be discovered!
This box contains preliminary, unpublished results.
We identified snails based mainly on the taxonomic descriptions by Vermeulen (1991, 1993, 1994, 1996a, 1999), several online taxonomic databases (Liew 2019a, 2019b), and descriptions from nearby regions (Vermeulen and Whitten 1998, Liew 2019c), all of which have a conchological basis. Whereas some genera and species are easily told apart simply based on overall structure and size of the shell, other groups are composed of species that are very much alike and need some experience to be identified (Figure 5.12). Most taxa have taxonomically been described in much detail. This is particularly true for the species that belong to the family Diplommatinidae (Vermeulen 1991, 1993, 1994, 1996a). Not surprisingly, many of this family’s members have very attractive shells, with beautiful ornamentation and some have a weird, sinistrally-orientated aperture combined with a dextrally coiled shell, and many are abundant and easily found. Other taxa have not been studied in detail yet, and have been only roughly described, or only very recently (e.g. Vermeulen et al. 2015). This is mainly true for representatives of the genera Japonia and Leptopoma. Especially the first genus shows much conchological variation, without clearly delineated species, within our study region. Being interested in community ecology and not in taxonomy per se, we have taken the liberal, practical approach of lumping all together as morphospecies Japonia sp. (cf. Krell 2004).
On top of such undescribed taxonomic variation, we expect further variation to be hidden in the details. In Chapter 2 we showed that there can be strong genetic differentiation among local populations of the same species, or ‘species-complex’. These cases are maybe best interpreted as examples of radiations of morphologically cryptic species. One of the species-complexes we studied in Chapter 2, which we referred to as Georissa similis s.l.,
Figure 5.12 An overview of the conchological land snail diversity on the limestone outcrops of the Lower Kinabatangan Floodplain. While (A-E) show five different species that can easily be distinguished from each other based on general shell form, (F-I) show four species from the genus Kaliella that look very similar, are probably closely related, and are often found together. It takes some experience and the study of characteristics like relative height of whorls, depth of suture, shape of aperture, and microsculpture to distinguish such closely related species. Photos not to scale, but all species with a shell height in the order of several millimetres, except (E), which is ca. 15 mm tall. All samples were collected by Kasper P. Hendriks and colleagues and subsequently deposited to the BORNEENSIS collection of ITBC in Kota Kinabalu, Sabah, Malaysian Borneo. Sample details: (A) Allopeas gracile (Hutton, 1834), BORMOL7701; (B) Diplommatina rubicunda (Von Martens, 1864), BORMOL7352; (C) Diplommatina calvula Vermeulen, 1993, BORMOL13425; (D) Microcystina appendiculata (syn. lissa) (Von Moellendorff, 1893), BORMOL13664; (E) Leptopoma sericatum (Pfeiffer, 1851), BORMOL7854; (F) Kaliella
dendrophila (Van Benthem Jutting, 1950), BORMOL7489; (G) Kaliella punctata Vermeulen, Liew
and Schilthuizen, 2015, BORMOL7470; (H) Kaliella accepta (Smith, 1895), BORMOL7509; and (I)
Kaliella calculosa (Gould, 1852), BORMOL7533. Photos by Kasper P. Hendriks
(A)
(B)
(C)
(D)
(E)
was recently taxonomically investigated by Zac bin Khalik and colleagues, who included several of our specimens from the Lower Kinabatangan Floodplain (Khalik et al. 2019). They combined the studies of detailed conchology, such as shell microsculpture, made visible using the latest microCT-scanning tools, and phylogenetics. They concluded that species of ‘non-scaly Georissa’ on Borneo are indeed best treated as several different species (see also Box 5.1).
In conclusion, the studies presented in this thesis are likely to suffer from an unknown level of underestimation of snail diversity. We expect our results, therefore, to be overall conservative, but suggest future studies to further focus on the inclusion of the non-obvious variation present. This could be done, for example, by genotyping all individuals studied, such that ecological variation can be tested directly against genetic variation.
Supplementary material
Ta bl e S 5.1 S A D IS A fi ts t o l oc al s pe ci es a bu nd an ce d is tr ib ut io ns f ro m c en su s d at a o n d iff er en t g as tr op od c om m u ni ti es , c ol le ct ed f ro m t hi s st ud y, l it er at u re , a nd u np ub li sh ed d at a f ro m H or sá k e t a l. T hr ee d iff er en t m et ac om m u ni ty m od el s w er e fi tt ed t o t he d at a f ro m e ac h l oc at io n: th e p oi nt -m ut at io n m od el ( ‘p m ’), t he p ro tr ac te d-sp ec ia ti on m od el ( ‘p r’ ), a nd t he d en si ty -d ep en de nc e m od el ( ‘d d’ ). B es t m od el s ( ba se d o n l ow es t A IC -v al ue ) a re p ri nt ed i n b ol d. A bb re via ti on s a re a s f ol lo w s: R O – o bs er ve d r ic hn es s, R C ha o1 – C ha o1 e st im at e o f r ic hn es s, D – d om in an ce , i.e . r el at iv e a bu nd an ce o f t he m os t c om m on s pe ci es , A IC – A ka ik e I nf or m at io n C ri te ri on .θ
,φ
,α
, a nd I a re t he m od el p ar am et er e st im at es . D at as et Location n RO RChoa1 D mode l loglik A IC θ φ α IRegion: Borneo, Asia Habitat: terr
estrial Sour ce: t his study Batangan 774 39 41.5 0.360 pm -58.36 120. 73 1.06E+1 3.56E+2 pr -58.04 122.08 1.11E+1 6.69E+2 7.86E+4 dd -56 .87 119. 74 2.54E+0 0.51 7.38E+1 Batu P ay ung 3562 28 28.6 0.315 pm -80.58 165.15 4.28E+0 8.65E+3 pr -80.55 16 7.10 4.30E+0 1.55E+4 4.05E+8 dd -80.66 16 7.31 5.59E+3 -10.00 4.28E+0 Batu T ai 451 36 39.5 0.470 pm -42.41 88.82 9.45E+0 2. 74E+3 pr -42.40 90. 79 9.51E+0 4.40E+3 2.39E+8 dd -39.88 85. 77 1.03E+0 0.64 7.12E+1 K ampung 1046 34 34.1 0.214 pm -63.96 131.91 7.6 7E+0 6.95E+2 pr -63.61 133.23 7.89E+0 1.29E+3 2. 74E+7 dd -63.88 133. 75 6.03E+0 0.11 3.32E+2 K eruak 1571 38 60.0 0.459 pm -65.07 134.13 4.39E+8 7.02E+0 pr -65.07 136 .13 1.60E+8 6.22E+7 7.04E+0 dd -62.34 130.6 7 2.54E+0 0.23 4. 79E+8 Materis 1808 31 34.0 0.284 pm -70. 78 145.56 1.35E+8 5.32E+0 pr -70. 78 147 .56 1.36E+7 1.09E+8 5.31E+0 dd -70. 75 147 .50 4.80E+0 0.03 4. 79E+8 Ma was 1522 31 34.3 0.362 pm -64.29 132.58 5. 75E+0 3.36E+3 pr -64.29 134.58 5. 75E+0 4.85E+8 3.36E+3 dd -63.06 132.12 1. 71E+0 0.38 1.69E+2 N ew Location 1 381 29 40.3 0.339 pm -38. 78 81.56 1.97E+8 7.30E+0 pr -38. 78 83.56 1.53E+7 6.60E+7 7.29E+0 dd -37 .65 81.31 3.39E+0 0.23 4.65E+8 Pangi 518 34 41.2 0.247 pm -46 .58 97 .15 2.60E+8 8.16E+0 pr -46 .58 99.15 1.01E+7 2.49E+7 8.18E+0 dd -45.52 97 .04 2.40E+0 0.43 1.44E+2 Tandu Batu 857 5 40 43.3 0.639 pm -101.25 206 .50 5.53E+0 3.55E+4 pr -101.25 208.50 5.53E+0 2.17E+5 5.16E+4 dd -98.53 203.06 1.60E+0 0.27 1. 74E+3 Tomanggong 2 828 29 32.8 0.557 pm -51.66 107 .32 1.97E+8 5.85E+0 pr -51.66 109.32 2.65E+7 2.96E+7 5.92E+0 dd -49.84 105.68 7.46E-1 0.58 6.44E+1 Tomanggong Bes ar 543 36 39.0 0.204 pm -50.68 105.36 9.10E+0 1.57E+3 pr -50.65 107 .31 9.18E+0 2. 77E+3 4.52E+6 dd -50. 78 107 .55 4. 78E+3 -10.00 9.17E+0 Tomanggong Kecil 3120 35 37 .5 0.465 pm -87 .15 178.30 5.64E+0 1.27E+4 pr -87 .15 180.30 5.64E+0 2.82E+4 2.94E+5 dd -86 .32 178.63 1.50E+0 0.55 4.17E+1
U lu Sungai Resang
1117 35 36. 7 0.490 pm -62.37 128. 75 7.12E+7 6.87E+0 pr -62.15 130.31 7.32E+0 3.87E+3 1.39E+8 dd -60.90 127 .81 1.25E+0 0.61 4.42E+1 Region: U ganda, A frica Ha bitat: terr estrial Sour ce: W rons ki et al. (2016) Plot 14 469 25 26 .4 0.316 pm -44.32 92.64 5. 71E+0 5.33E+3 pr -44.31 94.63 5. 74E+0 7.77E+3 6.00E+6 dd -44.30 94.60 5.14E+0 0.04 2.82E+3 Plot 15 577 31 31.3 0.206 pm -53.33 110.6 7 1.06E+1 9. 76E+1 pr -52.64 111.27 1.11E+1 2.43E+2 1.89E+8 dd -53.21 112.42 1.35E+1 -0.21 2.46E+2 Plot 16 859 33 36 .3 0.141 pm -64.83 133.66 3.05E+1 1.45E+1 pr -64.83 135.66 3.05E+1 4.85E+8 1.45E+1 dd -64.83 135.65 2.65E+1 0.08 1.48E+1 Plot 21 1130 23 32.3 0.577 pm -49.12 102.23 3.23E+8 4.09E+0 pr -49.12 104.23 7.27E+6 3.51E+7 4.14E+0 dd -46 .56 99.11 1.09E+0 0.28 4.81E+8 Region: Lak e T angan yika, A frica Habitat: f res hwater Sour ce: Me yer et al. (2011) Hilltop N ort h 451 4 5.0 0.989 pm -14.51 33.02 1.18E+8 6.05E-1 pr -14.51 35.02 1.59E+8 1.24E+8 6.05E-1 dd -14.51 35.02 3.61E+8 0.98 6.05E-1 Hilltop S out h 274 4 4.0 0.898 pm -16 .08 36 .17 5.09E+7 6.64E-1 pr -16 .08 38.17 1.33E+8 2.40E+7 6.64E-1 dd -16 .08 38.17 4.82E+8 0.99 6.64E-1 K ata be Inner 309 5 6. 0 0.945 pm -16 .96 37 .91 3.21E+7 8.47E-1 pr -16 .96 39.91 4.08E+7 9.94E+7 8.47E-1 dd -16 .96 39.91 4.83E+8 0.99 8.47E-1 K ata be S out h 408 3 3.0 0.971 pm -14.29 32.58 8.20E+7 4.39E-1 pr -14.29 34.58 3. 75E+7 1.34E+7 4.39E-1 dd -14.29 34.58 4. 76E+8 0.99 4.39E-1 N ondwa 379 5 5.0 0. 765 pm -23.09 50.18 1.25E+0 2.40E+1 pr -22.63 51.25 1.55E+0 4.13E+1 3.31E+8 dd -22.96 51.92 2.24E+0 -0.34 7.91E+1 Region: A tlantic O cean (Heter opods ) Ha bitat: marine Sour
ce: Burridge et al. (2017)
Station 20 259 10 13.0 0.541 pm -24.92 53.83 1.97E+8 2.07E+0 pr -24.92 55.83 4.45E+7 4.21E+7 2.07E+0 dd -24.92 55.83 4.85E+8 0.99 2.07E+0 Region: A tlantic O cean (Pter opods ) Ha bitat: marine Sour
ce: Burridge et al. (2017)
Station 10 326 10 16 .0 0.552 pm -26 .44 56 .88 2. 72E+8 1.95E+0 pr -26 .44 58.88 6.02E+6 2.44E+8 1.95E+0 dd -26 .44 58.88 4. 77E+8 0.99 1.95E+0 Station 13 292 12 40.0 0.918 pm -24.55 53.11 4.14E+8 2.52E+0 pr -24.55 55.11 1.21E+7 4.51E+7 2.53E+0 dd -24.55 55.11 4.85E+8 0.99 2.52E+0 Station 15 271 14 16 .0 0. 790 pm -25.80 55.59 1.97E+8 3.13E+0 pr -25.80 57 .59 1.04E+7 2.19E+7 3.14E+0 dd -23. 77 53.53 1.36E-1 0.63 6.27E+1 Station 16 613 11 12.0 0.881 pm -27 .65 59.30 1.18E+8 1.90E+0 pr -27 .65 61.30 9. 71E+7 1.03E+8 1.90E+0 dd -27 .65 61.30 4.32E+8 0.98 1.90E+0 Station 27 816 3 3.0 0.647 pm -18.33 40.66 1.22E+0 2.19E+0 pr -18.29 42.57 6.24E+5 3.81E+0 2.43E+0 dd -18.26 42.53 2.43E+0 -0.53 2. 76E+0
Ta bl e S 5.1 C on ti nu ed . D at as et Location n RO RChoa1 D mode l loglik A IC θ φ α I Region: A tlantic O cean (Pter opods ) Ha bitat: marine Sour
ce: Burridge et al. (2017)
Station 28 1779 6 6. 5 0.811 pm -24.94 53.88 1.97E+8 7.7 5E-1 pr -24.94 55.88 2.94E+7 2.28E+7 7.7 6E-1 dd -24.94 55.88 4. 79E+8 0.99 7.7 5E-1 Station 29 351 6 7.0 0.692 pm -22.28 48.55 8.20E+7 1.03E+0 pr -22.28 50.55 1.90E+7 5.63E+7 1.03E+0 dd -22.28 50.55 4.85E+8 -0. 78 1.03E+0 Station 7 271 15 15.2 0.454 pm -30.55 65.10 3.62E+0 6.24E+2 pr -30.50 67 .00 3.69E+0 9.46E+2 1.68E+7 dd -30.66 67 .32 4.84E+8 0.99 3.42E+0 Station 8 337 20 23.3 0.501 pm -36 .6 7 77 .34 4.44E+7 4.66E+0 pr -36 .6 7 79.34 2.45E+7 3. 72E+8 4.66E+0 dd -36 .14 78.28 1.52E+0 0.53 3.06E+1 Station 9 313 16 19.3 0.457 pm -30.52 65.04 2. 72E+8 3.57E+0 pr -30.52 67 .04 2.24E+6 9.61E+6 3.56E+0 dd -30.52 67 .04 4. 72E+8 0.99 3.57E+0 Region: N ew S out h W ales, A ustr alia Ha bitat: terr estrial Sour ce: Camer on and Po kr ys zk o (2005) Site 12 6141 27 27 .0 0.290 pm -95.35 194. 71 4.30E+0 1.33E+3 pr -94.81 195.62 4.52E+0 2.48E+3 5.19E+7 dd -96 .13 198.25 4.87E+2 -10.00 5.90E+0 Site 16 2590 27 30.0 0.509 pm -7 4.63 153.26 5.04E+0 6. 79E+2 pr -7 4.63 155.26 5.04E+0 4.85E+8 6. 79E+2 dd -7 4.08 154.15 1.87E+0 0.41 5.14E+1 Region: Centr al Eur ope (liv e data ) Ha bitat: terr estrial Sour ce: Hor sák et al. (unpu blis hed) Bie la v oda 913 22 27 .6 0. 700 pm -44.53 93.06 3.00E+8 4.06E+0 pr -44.53 95.06 2. 77E+7 1.72E+8 4.04E+0 dd -41.13 88.26 7.43E-1 0.34 1.00E+4 Blatna do lina 1250 18 20.0 0. 710 pm -43.36 90. 71 4.39E+8 2.98E+0 pr -43.36 92. 71 3.93E+8 1.57E+8 2.98E+0 dd -43.36 92. 71 4.84E+8 0.99 2.98E+0 Blatnica 510 17 20.0 0.402 pm -38. 71 81.42 3.91E+7 3.39E+0 pr -38. 71 83.42 3.58E+8 2.30E+7 3.39E+0 dd -38. 71 83.42 4.35E+8 0.97 3.39E+0 Bor oto va 567 21 26 .0 0.510 pm -41.23 86 .47 1.97E+8 4.29E+0 pr -41.23 88.47 3.09E+7 7.46E+7 4.30E+0 dd -40. 73 87 .46 2.44E+0 0.16 4. 75E+8 Malejo v 413 18 28.0 0.513 pm -34.09 72.17 1.44E+8 3.84E+0 pr -34.09 74.17 7.24E+6 7.91E+6 3.84E+0 dd -32.60 71.20 5.24E-1 0.52 8.65E+1 Omsenie 274 19 22.8 0.496 pm -30.84 65.6 7 1.89E+8 4.64E+0 pr -30.84 67 .67 2.45E+8 4.41E+8 4.64E+0 dd -29.93 65.87 1.62E+0 0.31 5.20E+2 Pr ecin 867 19 29.5 0.6 71 pm -39.82 83.65 1.97E+8 3.43E+0 pr -39.82 85.65 1.78E+7 4. 75E+8 3.44E+0 dd -39.82 85.65 4.64E+8 0.99 3.43E+0 Pribo vce 624 18 18.0 0.232 pm -47 .73 99.45 5.34E+0 6.27E+1 pr -47 .28 100.55 5. 72E+0 1.60E+2 7.11E+6 dd -47 .52 101.03 7.81E+0 -0.27 1.95E+2
