Genetic diversity and population structure of the threatened temperate woody bamboo Kuruna debilis ( Poaceae : Bambusoideae : Arundinarieae ) from Sri Lanka based on microsatellite analysis

Species of the temperate woody bamboo genus Kuruna Attigala, Kathriar. & L.G. Clark (Poaceae: Bambusoideae) distributed in Sri Lanka and southern India, are threatened due to deforestation and habitat fragmentation. The current study focused on the tetraploid woody bamboo Kuruna debilis (Thwaites) Attigala, Kathriar. & L.G. Clark, using twelve variable microsatellite loci to assess the genetic diversity and population structure in six known Sri Lankan populations. Due to the rarity of the species, an exhaustive sampling of accessible plants resulted in a total of only 28 individuals. Nonetheless, the allelic diversity was high at most loci and given the limited distances separating populations (< 65 km apart), they exhibited a fairly high genetic differentiation (FST = 0.113) and strong isolation by distance. Structure, neighbour-joining, and neighbour-net analyses concur in grouping the six K. debilis populations into three genetic clusters consistent with the spatial proximity of the populations: one cluster comprised populations from the Piduruthalagala Mountain and Horton Plains, the second cluster consisted of the population from Adams Peak and the last comprised the populations from the Handapan Ella Plains. Due to multiple indicators of high allelic diversity, the population from the northern Horton Plains (LA124) should be targeted for conservation. Moreover, the population found in Adams Peak (LA159) is also genetically important and critical to the conservation of these species due to its unique genetic diversity. As the first population genetics study of Bambusoideae in Sri Lanka, we anticipate that our results will provide a foundation for future comparative population genetics and conservation studies in the country.


INTRODUCTION
Bamboos (subfamily Bambusoideae, Poaceae) are an essential component of forest and tropical high altitude grassland ecosystems worldwide (Soderstrom et al., 1999;Clark et al., 2015).In Sri Lanka, bamboos occur naturally in all however, no native bamboo is found in extremely dry areas (Kariyawasam, 1998).Bamboo, in general, is an economically, culturally and ecologically important plant for Sri Lanka (De Zoysa & Vivekanandan, 1994;Gunatilleke et al., 1994).Most of the non-native bamboos are used in housing and construction due to their enduring, versatile and highly renewable nature.There are no statistics on bamboo consumption, but the forestry sector master plan (FSMP) of Sri Lanka estimated that the total annual consumption was at least 80,000 m 3 , i.e. about 700,000 culms two decades ago (FSMP, 1995).Although the native bamboos are not of economic such example is the animal biodiversity associated with the native bamboos such as the Sambar deer, many insects and fungi (Abayasinghe et al., 2014;personal observations).Bamboo studies conducted in Sri Lanka have mainly focused on its reproductive ecology (Ramanayake & Yakandawala, 1995;1998;Ramanayake & Weerawardene, 2003), vegetative propagation (Ramanayake et al., 2001;2007) and growth and development (Rajapakse, 1992;Ramanayake et al., 2001Ramanayake et al., ). et al., 2012)).Nearly one fourth of these are endemic and concentrated in the humid southwestern quarter of the country (Gunatilleke & Gunatilleke, 1990).For many years, forests in Sri Lanka have been cleared both legally and illegally, due to the rapidly increasing demand for land for settlement schemes, timber production, economic and agricultural developments and weak enforcement of land use policies in the country (Gunatilake, 1998; Government of Sri Lanka, 2000; Bandaratillake & Fernando, 2003).Several studies have reported that the studies have shown that habitat fragmentation and small diversity of populations (Ellstrand & Elam, 1993;Fischer & Matthies, 1998;Luijten et al., 2000;Paschke et al., 2002).Low levels of genetic diversity typically limit the ability of a population to adapt to adverse environmental conditions or increased competition (Fischer et al., 2000;Pluess & Stocklin, 2004).Each of the Kuruna species found in Sri Lanka has a limited distributional be under severe threat due to deforestation and habitat fragmentation.Of the six native Kuruna species, only K. debilis has several, spatially distinct populations in the understory of the upper cool mountain slopes of Based on herbarium records K. debilis populations are found at the Adams Peak, Horton Plains, Piduruthalagala Mountain, Knuckles Mountains and the Handapan Ella Plains of Sri Lanka.Muktesh Kumar (2011) reported that K. debilis has also been located recently from the Kerala part of the Western Ghats, but provided no documentation.
K. densifolia, , K. scandens, K. serrulata and K. walkeriana) are each summits and in open montane grasslands in Sri Lanka (Figure 1).Due to their restricted distribution and habitat already at risk.Population genetic studies are essential for planning conservation strategies for these Kuruna species.Such studies provide conservation managers of random genetic drift and inbreeding, and reduced these processes could potentially be used to initiate conservation planning (Ellstrand & Elam, 1993; Young et al The primary objective of this study was to assess the genetic diversity and population structure in six natural populations of the tetraploid woody bamboo K. debilis in Sri Lanka.

METHODOLOGY
Leaf samples were collected from 28 individual from 3 remote montane forests and an open montane grassland (Figure 1).Of the 5 main localities of K. debilis populations (Adams Peak, Horton Plains, Piduruthalagala Mountain, Knuckles Mountains and Handapan Ella Plains), the authors were able to collect population samples from all localities except the Knuckles Mountains.Although a collection trip was made to Knuckles Mountains we were unable to locate any K. debilis populations.Kuruna debilis has a pachymorph occupy a relatively large area.Due to clonal propagation, reveal the true number of genets, or genetic individuals in a population.Therefore, samples were collected from individuals that were readily distinguishable as a single plant and all propagating clones of an individual plant were considered as a single entity to allow for a number

Marker selection
Microsatellite sequences (SSRs) are highly polymorphic and readily replicable markers, evenly distributed throughout eukaryotic genomes.Based on previous studies on temperate woody bamboos (Kitamura et  ).These measures and the IS ) were estimated for each locus and each K. debilis population.The A R used in the using the rarefaction method recommended for uneven et al., 1998).Further, global F-statistics (F IT , F IS and F ST ) were estimated over all loci and populations using SPAGeDi 1.4c (Hardy & determined by permutation (10,000 replicates).The o ) was calculated manually for each population over all loci.
The spatial genetic structure was inferred using a Bayesian clustering approach, which was implemented in Structure 2.3.4 (Pritchard et al., 2000).The model parameters were set to admixture with correlated allele frequencies between populations, and 20 replicated runs were performed for each value of K (the number to 100,000 followed by 200,000 recorded Markov Chain Monte Carlo steps.Each run estimated the log probability of data, L(K).Following Evanno et al. (2005), the differences in log L(K) for successive values of clusters, a process implemented using the Structure Harvester (Earl & vonHoldt, 2012).The preferred value of K using this method is the one associated with the 2007) and results were graphically represented using the Population differentiation was assessed by analysis of molecular variance (AMOVA) using the software et al., 2005).The K optimal genetic clusters detected with Structure determined the hierarchical levels in the AMOVA analysis and resulted in three estimates of genetic differentiation: ST , genetic differentiation among subpopulations SC , genetic differentiation among subpopulations within a genetic CT , genetic differentiation among genetic replicates).Isolation by distance (IBD) was evaluated by assessing the correlation between the log transformed genetic distance measure F ST / (1-F ST ) (Rousset, 1997) by the Mantel test in the programme IBDWS 3.22 (Jensen et al., 2005).For the IBD analyses, geographic was based on 10,000 permutation replicates.For both AMOVA and IBD analyses, the tetraploids were treated as diploids (Saltonstall, 2003) as there was no evidence of inbreeding within populations of K. debilis and also the analysis programmes allowed only diploid or haploid data.The diploid data matrix was generated by randomising the genotypes within each population.
A rooted majority rule consensus tree was constructed using the neighbour-joining (NJ) method with Cavalliwere generated using a combination of SEQBOOT, woody bamboos Oldeania alpina (K.Schum.)Stapleton and Chimonocalamus montanus J.R. Xue & T.P. Yi were used as outgroups to root the NJ tree.The resulting NJ tree was visualised with FigTree software v1.4 (http:// ).
further means of visualising the genetic clustering of sample populations, a network-building distancebased algorithm (Neighbour-Net) was performed with

Allelic variation at microsatellite loci
All twelve microsatellite loci assayed were polymorphic, and the number of alleles detected for each locus varied between 3 (FAN30) to 20 (Sasa500) leading to 94 alleles selected loci on average generated 8 alleles per locus and the genetic diversity of the 12 microsatellite loci.

Genetic variation within populations
Despite the small number of plants available within genetic diversity were relatively high (Table 3 The proportion of the observed genetic variation between clusters ranged from F ST = -0.053for locus FAN27 to 0.394 for locus FAN30 with an average value of 0.113 Table 4).Individual locus estimates of the inbreeding IS Structure analyses using the Evanno method in Structure Harvester grouped the six populations into K = 3 clusters (Figures 2 and 3A).The estimated shown in Figure 3A.The three clusters correspond well to the geographic distribution of the populations (Figure 1), with populations 1, 2 and 3 (LA120, LA124 and LA130) forming an Eastern cluster, populations 4 and 5 (LA148 and LA154) a Southern cluster, and in the Eastern cluster were sampled from the Piduruthalagala Mountain (LA120) and the Horton Plains (LA124 and LA130).These two localities are in relatively close proximity and are separated by ca. 15 km.The two populations in the Southern cluster are from Handapan Ella Plains and Western cluster is from Adams Peak.The Mantel test of the correlation between Rousset's genetic distance and geographic distance indicated the six K. debilis populations (r 2 (Figure 3D).

Genetic relationships within the geographic groups
The rooted NJ tree for the six K. debilis populations (Figure 3B) resulted in three clades corresponding to the three genetic clusters indicated by the Structure analysis (Figure 3A).The neighbour-net network derived from the SplitsTree analysis also revealed the same three population genetic clusters (Figure 3C).

Locus and population level genetic diversity
Of the six K. debilis populations sampled for this study, population 1 (LA120), which was collected from mount Piduruthalagala was unusual in displaying lower allelic and genetic diversity compared to the or more missing loci, as these were similar to the other is due to failure to amplify certain loci.
Of the various dimensions of genetic diversity, allelic richness is often considered to be of key relevance in conservation programmes (Petit et al., 1998; Simianer, and genetic drift, and may be an important indicator of a population's adaptive potential, as the limit of selection response is mainly determined by the initial number of alleles regardless of the allelic frequencies ( 4).Population 2 (LA124) from the northern part of the Horton Plains showed the highest levels of N A , N Ae and A R , and is a particularly good candidate for conservation.The single formed its own genetically distinct cluster and is also a potential candidate for conservation due to its unique genetic diversity.Although the three genetic clusters do not provide evidence for morphologically distinct ecotypes, their unique genetic diversity must be taken into consideration by conservation managers when identifying conservation management units.Many studies show that, while ecotypes do matter when conservation are commonly made based on the genetic uniqueness of the populations (Soule & Simberloff, K. debilis recommend that populations separated by more than ca.35 km (as is the case for these clusters) should be treated as distinct units for management and conservation, while those within ca. 15 km should be managed jointly.These recommendations serve as an initial step towards identifying management units for threatened bamboos in smaller one must use these results cautiously.variation were found to be relatively high within populations.Further, tests of genetic differentiation among populations were found to be important insight into the genetic diversity and connectivity of K. debilis towards the conservation management of this threatened temperate woody bamboo species.For these rare and threatened species with limited distribution areas, a single stochastic event, such as a serious insect attack or pathogen infection, could cause catastrophic reductions extinction (Ma et al., 2013).Taking proper measures to protect their current populations is required.
Genetic diversity and population structure of the threatened temperate woody bamboo Kuruna debilis Arundinarieae) from Sri Lanka based on microsatellite analysis † Lakshmi Attigala * , Timothy Gallaher, John Nason and Lynn G. Clark Department of Ecology, Evolution and Organismal Biology, College of Liberal Arts and Sciences, Iowa State University, 251 Bessey Hall, Ames, IA, 50011-1020, USA.
al., 2009; Zhan et al., 2009), 25 primer pairs were tested with Of the 25 microsatellite primers, 12 were successfully DNA extraction and genotyping Total genomic DNA extractions were performed from silica gel-dried specimens using the Iowa State extraction robot.Genotyping was performed on an ABI 3730 DNA analyser (Perkin-Elmer, Applied Biosystems Division, Norwalk, Connecticut) by the DNA Sequencing sequencing reactions were performed in an MJ Research remove unincorporated primers and dNTPs from the PCR products.Genotyping of microsatellite loci were subjected to standard error checking procedures, as described in DeWoody et al Individual tetraploid genotypes were scored from the electropherograms following the microsatellite DNA allele counting-peak ratios (MAC-PR) method of Esselink et al. (2004) using the GeneMapper ® v4.1 (Applied Biosystems) software.March 2017 Journal of the National Science Foundation of Sri Lanka 45(1) The software SPAGeDi 1.4c (Hardy & Vekemans, 2002) was used to compute the percentage of missing genotypes and the following estimates of genetic diversity: the number of alleles per locus (N A ); the effective number of alleles (N Ae ) estimated following Nielsen et al. (2003); R ) expressed as the expected number of alleles among k gene copies (12 gene copies e

Structure
Harvester results of structure analyses for K = 2 -5 putative genetic clusters of K. debilis individuals.A: mean L(K) L(K-1); C: absolute values of the second order rate of change of the likelihood distribution (mean ± SD) calculated according to is the most probable number of clusters or the uppermost level of structure, here three clusters.

Figure 3 .
Figure 3. Analyses of genetic structure among the six K. debilis populations; A: Bayesian clustering Analyses of genetic structure among the six K. debilis genetic clusters.Each individual is represented by a vertical column and the populations are separated by a vertical black line.Different colours in the same column for each individual indicate the percentage of estimated membership in a cluster; B: rooted