Ecogeographic Variation in the Morphology of Two Asian Wild Rice Species, Oryza nivara and Oryza rufipogon Author(s): Maria Celeste N. Banaticla-Hilario, Marc S. M. Sosef, Kenneth L. McNally, Nigel Ruaraidh Sackville Hamilton, and Ronald G. van den Berg Source: International Journal of Plant Sciences, Vol. 174, No. 6 (July/August 2013), pp. 896- 909 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/10.1086/670370 . Accessed: 16/07/2013 08:49 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. . The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to International Journal of Plant Sciences. http://www.jstor.org This content downloaded from 137.224.252.10 on Tue, 16 Jul 2013 08:49:58 AM All use subject to JSTOR Terms and Conditions
Transcript
Ecogeographic Variation in the Morphology of Two Asian Wild Rice Species, Oryza nivara andOryza rufipogonAuthor(s): Maria Celeste N. Banaticla-Hilario, Marc S. M. Sosef, Kenneth L. McNally, NigelRuaraidh Sackville Hamilton, and Ronald G. van den BergSource: International Journal of Plant Sciences, Vol. 174, No. 6 (July/August 2013), pp. 896-909Published by: The University of Chicago PressStable URL: http://www.jstor.org/stable/10.1086/670370 .
Accessed: 16/07/2013 08:49
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp
.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].
.
The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access toInternational Journal of Plant Sciences.
http://www.jstor.org
This content downloaded from 137.224.252.10 on Tue, 16 Jul 2013 08:49:58 AMAll use subject to JSTOR Terms and Conditions
Int. J. Plant Sci. 174(6):896–909. 2013.� 2013 by The University of Chicago. All rights reserved.1058-5893/2013/17406-0005$15.00 DOI: 10.1086/670370
ECOGEOGRAPHIC VARIATION IN THE MORPHOLOGY OF TWO ASIAN WILD RICESPECIES, ORYZA NIVARA AND ORYZA RUFIPOGON
Maria Celeste N. Banaticla-Hilario,1,*,† Marc S. M. Sosef,†,‡ Kenneth L. McNally,*Nigel Ruaraidh Sackville Hamilton,* and Ronald G. van den Berg†
*T. T. Chang Genetic Resources Center, International Rice Research Institute, Los Banos, Laguna, Philippines; †Biosystematics Group,Wageningen University, Wageningen, The Netherlands; and ‡Naturalis Biodiversity Center (Section NHN),
Wageningen University, Wageningen, The Netherlands
To search for variation patterns and diagnostic features between Asian wild rice species, several numericalmethods were applied to phenotypic data obtained from 116 accessions representing sympatric populationsof Oryza nivara and Oryza rufipogon from tropical continental Asia and O. rufipogon populations frominsular Southeast Asia and Australasia. Ordination and cluster analyses separate O. rufipogon from O. nivara,indicating the presence of two sympatric morphological species occupying different ecological niches. Oryzanivara and O. rufipogon are morphologically more differentiated in South Asia than in mainland SoutheastAsia, implying more recent divergence and/or more interspecific gene flow among sympatric populations inthe latter region. Oryza nivara exhibits South and Southeast Asian phenotypes while the Australasian pop-ulations of O. rufipogon appear as distinct from the rest of the species. Seedling height, culm number, anddiameter; leaf length and width; and anther length were significantly correlated to certain geoclimatic factorsand displayed contrasting correlation directions for O. nivara and O. rufipogon, implying that the two speciesrespond differently to geographic and climatic gradients. Diagnostic characters are provided to delineate thespecies morphologically. The results suggest the strong influence of ecology on species morphology, existenceof geographic races within species and morphological divergence between O. nivara and O. rufipogon.
In the International Rice Genebank (IRG) at the Interna-tional Rice Research Institute (IRRI), Oryza nivara Sharma &Shastry and Oryza rufipogon Griff. make up the majority ofthe wild rice collection. Oryza nivara is an annual self-polli-nator found in seasonally dry areas of tropical continentalAsia, while O. rufipogon is a perennial cross-pollinator foundin permanently wet areas (e.g., ponds and swamps) of the Asia-Pacific region (South China, south and Southeast Asia to north-ern Australia). Under Philippine conditions, O. rufipogonflowers mainly from September to December, whereas O. ni-vara produces inflorescences throughout the year. Morpho-logically distinct accessions are common, but mislabeling stilloccurs due to the close physical resemblance of the seeds ofthe two species. Plant descriptions are often incomplete, andthe morphological distinction between these species is oftennot clear from the literature. A number of accessions also ap-pear to represent intermediate forms of O. nivara and O.rufipogon.
Although these two taxa are generally recognized as distinct
Manuscript received November 2012; revised manuscript received February2013; electronically published June 7, 2013.
species (Sharma and Shastry 1965; Ng et al. 1981; Aggarwalet al. 1999; Lu 1999; Lu et al. 2001; Duan et al. 2007; Kurodaet al. 2007; Xu et al. 2012), some rice scientists consider O.nivara to be the annual ecotype or subspecies of O. rufipogon(with O. rufipogon sensu stricto as the perennial ecotype) dueto their continuous (morphological and genetic) variation andinterfertility (Tateoka 1963; Second 1985; Oka 1988; Hiroi etal. 1990; Morishima et al. 1992; Iwamoto et al. 1999; Vaughanet al. 2003; Park et al. 2003; Zhu and Ge 2005; Kwon et al.2006; Zhou et al. 2008; Zheng and Ge 2010). In this study,O. nivara and O. rufipogon are provisionally treated as sep-arate species. A clear perception of the morphological variationbetween and within O. nivara and O. rufipogon and well-defined phenotype-based species delineation will help genebank managers in dealing with misidentified accessions and inmanaging intermediate forms.
This study reexamines the morphology of the two speciesacross their entire geographic range. Sympatric populationpairs (defined here as pairs of accessions that belong to dif-ferent species collected from the same geographic locality butnot necessarily growing next to each other) of O. nivara andO. rufipogon were used to assess species divergence at the localscale. Geographical variation patterns within species were alsoexamined.
The objectives were to (1) evaluate the morphological dif-
This content downloaded from 137.224.252.10 on Tue, 16 Jul 2013 08:49:58 AMAll use subject to JSTOR Terms and Conditions
BANATICLA-HILARIO ET AL.—MORPHOLOGY OF TWO ASIAN WILD RICE SPECIES 897
Fig. 1 Geographic distribution of the studied populations.
ferences between O. nivara and O. rufipogon to determine theappropriate taxonomic level of their distinction, (2) determinegeographical patterns of morphological variation within andbetween the two taxa, and (3) provide a set of distinguishingcharacters that can clearly delineate these taxa morpho-logically.
Material and Methods
Selection and Georeferencing of Plant Material
The IRG-IRRI provided the accessions used in this study(app. A). The collection locality data of the accessions wereobtained from IRG collection reports and from the Interna-tional Rice Genebank Collection Information System(IRGCIS; http://www.irgcis.irri.org:81/grc/Irgcishome.html),and these were georeferenced using the following softwareand databases: Biogeomancer workbench (http://www.biogeomancer.org/workbench.html), Biogeomancer spatialattribute lookup (http://bg.berkeley.edu/sal/), World Gazet-teer (http://world-gazetteer.com/), and the US National Im-agery and Mapping Agency database (http://earth-info.nga.mil/gns/html/index.html). Location data were also validatedusing DIVA-GIS (Hijmans 2001) and Google Earth (http://earth.google.com/).
The selected accessions (fig. 1) represent (a) 52 sympatricpopulation pairs of Oryza nivara and Oryza rufipogon from
tropical continental/mainland Asia (South China, Vietnam,Laos, Cambodia, Thailand, Myanmar, Bangladesh, Nepal, In-dia, and Sri Lanka; 104 accessions), (b) nine populations ofO. rufipogon from insular Southeast Asia (Philippines and In-donesia), and (c) five populations of O. rufipogon from Aus-tralasia (Irian Jaya and northern Australia). Altitude, meanannual temperature, and mean annual precipitation data oneach geographic location were extracted from the 5-arc-minutegrids downloaded from the WorldClim website (http://worldclim.org/) and obtained using DIVA-GIS (Hijmans2001).
Phenotyping
Five individuals from each accession were planted in theGenetic Resources Center screenhouse at IRRI, Philippines,following a randomized complete block design. Unfortunately,the seeds of two O. nivara accessions from China (N17) andMyanmar (N30) failed to germinate. Their corresponding sym-patric O. rufipogon accessions (R17 and R30) were still in-cluded in the analysis (app. A).
Phenotyping was conducted using 8 qualitative and 22 quan-titative characters (table B1, available online) as described inthe list of descriptors for wild and cultivated rice (Oryza spp.)published by Bioversity International, IRRI, and the AfricaRice Center (2007). Four additional quantitative characters(stigma length, style length, ratio of anther length to spikelet
This content downloaded from 137.224.252.10 on Tue, 16 Jul 2013 08:49:58 AMAll use subject to JSTOR Terms and Conditions
length, and ratio of spikelet width to spikelet length) were alsoobtained (table B1).
Data Validation
Phenotype data and herbarium specimens were cross-refer-enced to detect labeling errors. In the few cases of apparentmixtures within accessions, individuals that did not belong tothe population (e.g., an Oryza sativa or O. nivara plant in anO. rufipogon accession) were removed from the subsequentanalyses. Intermediate phenotypes were retained. PopulationsN26, R5, and R29 (app. A) were recoded as subpopulationsN26A and N26B, R5A and R5B, and R29A and R29B, re-spectively, as they exhibited two distinct plant types of whichone appeared to be an intermediate form.
Data Analyses
All analyses were conducted using the free software R, ver-sion 2.14 (R Development Core Team 2011) with additionalpackages. A pairwise character correlation analysis was per-formed by the cor() function using the Pearson’s product-moment coefficient on interval- and ratio-type data and theSpearman correlation coefficient on ordinal data. Rescaling ofquantitative characters and principal component analysis(PCA) were performed by the prcomp() function.
A hierarchical cluster analysis (HCA) of qualitative andquantitative characters was conducted with the cluster package(Maechler et al. 2011). The daisy() function was applied inconstructing a dissimilarity matrix based on Gower’s coeffi-cient and the hclust() function in performing agglomerativehierarchical clustering using the unweighted pair groupmethod with arithmetic mean (UPGMA). Bootstrapping with10,000 replicates was conducted by the consensus() functionof the agricolae package (Mendiburu 2010).
K-means clustering was performed by the kmeans() functionof the base package. To determine the appropriate number ofclusters, the within-group sum of squared error (SSE) of eachcluster (from to ) was plotted and screened fork p 1 k p 15the cluster solution, marked by a bend in the plot of SSE againstk. Principal components (PCs) are said to be the continuoussolution of the k-means clustering membership indicators(Ding and He 2004). Cluster solutions were plotted againstthe first two PCs using the clusplot() function of the clusterpackage (Maechler et al. 2011).
The standardized Euclidean distance between O. nivara andO. rufipogon in each sympatric accession pair was calculatedbased on quantitative traits using the daisy() function of thecluster package (Maechler et al. 2011). The cor.test() functionwas used to detect the correlation of latitude, longitude, alti-tude, annual mean temperature, and annual precipitation withthe Euclidean distances between sympatric O. nivara and O.rufipogon accessions and also with quantitative phenotypicdata within species.
To establish whether the Euclidean distances between sym-patric O. nivara and O. rufipogon differ significantly betweengeographic regions, an ANOVA and Tukey’s HSD test wereperformed using the agricolae package (Mendiburu 2010). Thegeographic regions compared are South Asia (Bangladesh, In-dia, Nepal and Sri Lanka) and continental Southeast Asia(Cambodia, Laos, Myanmar, Thailand, and Vietnam).
A MANOVA was applied to the quantitative characters totest the overall effect of the characters on inter- and intra-specific differences, followed by a univariate ANOVA and Tu-key’s HSD test using the agricolae package (Mendiburu 2010)to identify specific characters that are significantly differentbetween species and between populations of each species. Thegroups compared were the geographical populations of O.nivara in South Asia and continental Southeast Asia, and ofO. rufipogon in south Asia, continental Southeast Asia, insularSoutheast Asia and Australasia.
Results
Character Correlation
To reduce data redundancy, one of the characters was elim-inated from each strongly correlated (dcorrelation coef-ficientd ≥ 0.6) and functionally or structurally linked pair. Eightquantitative characters (flag leaf length and width, ligulelength, panicle number, spikelet fertility, awn thickness, ratioof anther length to spikelet length, and ratio of spikelet widthto spikelet length) were excluded from the PCA, HCA, and k-means analysis, and one qualitative character (rhizome andstolon formation) was removed from the HCA (table 1).
Principal Component Analysis
The first two principal components (PCs) accounted for44.8% of the total variance (table 2). PC1 explains 25.4% ofthe total variance and separates the relatively broad-spikeleted,short-anthered, short-culmed, and early-flowering Oryza ni-vara from the relatively narrow-spikeleted, long-anthered,long-culmed, and late-flowering Oryza rufipogon (fig. 2). Thetwo O. rufipogon accessions (R27 and R45) included in theO. nivara cluster were confirmed to represent misidentified O.nivara populations. PC1 also divides O. nivara in two sub-groups, with a predominantly South Asian cluster showinggreater divergence from O. rufipogon than the second, pre-dominantly Southeast Asian cluster (fig. 2).
Eighteen accessions (including R5A and R29A) detected asintermediate forms based on field and herbarium observationsfell between the two main clusters. Seven of these (N11, N31,R10, R15, R37, R43, and R50) are typical weedy forms orhybrid swarms between cultivated/escaped Oryza sativa andeither O. nivara or O. rufipogon). The remaining 11 accessionsexhibit a more ambiguous morphology and could either beweedy forms or hybrids/introgressed populations of the twowild species. These accessions were not included in the suc-ceeding analyses since hybrid populations as well as those in-fluenced by cultivation are beyond the scope of this study.
PC2 explains 19.4% of the variation and does not showany clear separation between O. nivara and O. rufipogon.However, it produces geographical clustering patterns withinspecies (fig. 2). The majority of South Asian O. nivara acces-sions (characterized by few, broad culms are separated fromtheir Southeast Asian counterparts (with numerous, narrowculms). Within O. rufipogon, most of the South Asian acces-sions as well as the Australasian populations cluster togetherwhile the (insular and continental) Southeast Asian popula-tions do not exhibit a distinct clustering pattern.
Sympatric population pairs from different geographic
This content downloaded from 137.224.252.10 on Tue, 16 Jul 2013 08:49:58 AMAll use subject to JSTOR Terms and Conditions
BANATICLA-HILARIO ET AL.—MORPHOLOGY OF TWO ASIAN WILD RICE SPECIES 899
Table 1
Highly Correlated Characters and Their CorrelationCoefficient (Denoted by r)
Characters r
Anther length to spikelet length ratio:a
Anther length .98Spikelet width �.85Spikelet width to spikelet length ratioa .81Spikelet fertilitya �.73Flag leaf lengtha �.69Awn thicknessa �.64Number of days from seeding to first heading .61
first heading .3503 �.1086 �.0679Standard deviation 2.138 1.8701 1.3554Proportion of variance .254 .1943 .1021Cumulative proportion .254 .4483 .5504
a Components providing factor loadings above 0.3 areunderlined.
regions display different clustering patterns along PC2. Oryzarufipogon and O. nivara from South Asia occupy separatepositions while those from continental Southeast Asia do notshow a distinct pattern as O. rufipogon populations are scat-tered over the PC2 axis. Oryza rufipogon from Australasiaseems separated from Southeast Asian O. nivara (fig. 2). PC3accounts for 10.2% of the total variance and does not produceany distinct clustering pattern between or within O. nivaraand O. rufipogon.
Hierarchical Cluster Analysis
The UPGMA dendrogram branches into two main clustersthat correspond to species groups (fig. 3). Both the O. nivara
and O. rufipogon clusters are supported by 60% bootstrapvalues.
Within the O. nivara cluster, accession N45 is the last tomerge with the rest of the group. South Asian accessions (ex-cept N19, N21, N23, N24, and N37) are separated from theSoutheast Asian populations (albeit with a bootstrap supportof less than 50%; fig. 3). Oryza rufipogon accessions R27 andR45 both join the Southeast Asian O. nivara group. In the O.rufipogon cluster, accessions from the same geographic regionare not consistently grouped together. Grouping of O. rufi-pogon according to sympatry and nonsympatry with O. nivarais not observed.
K-Means Analysis
K-means converged after three iterations. A four-cluster so-lution best fitted the data as the decrease in SSE started tobecome gradual at (fig. B1, available online). Table B2,k p 4available online shows how the accessions were partitioned inthe analyses run with to . At , O. nivara isk p 2 k p 4 k p 4separated into South Asian and Southeast Asian clusters (tableB2; fig. 4A) with the exception of two O. rufipogon (R27 andR45) and seven O. nivara accessions that do not cluster intheir supposed geographic group. The geographic division ofO. nivara is similarly depicted in the PCA and HCA results.Oryza rufipogon is also divided into two groups. Althoughthe geographical grouping is not as well defined as in O. nivara,the majority of the continental (South and Southeast) Asian
This content downloaded from 137.224.252.10 on Tue, 16 Jul 2013 08:49:58 AMAll use subject to JSTOR Terms and Conditions
Fig. 2 Principal component analysis results obtained from 18 quantitative characters. The encircled populations were tentatively identifiedas intermediate forms (i.e., intermediate between Oryza nivara and Oryza rufipogon, O. nivara and Oryza sativa, or O. rufipogon and O. sativa)based on screenhouse and herbarium observations.
populations are separated from most of the insular SoutheastAsian and all of the Australasian populations (table B2; fig.4A). At , the South Asian–Southeast Asian split in O.k p 3nivara is retained while O. rufipogon is recognized as a group(table B2; fig. 4B). At , k-means produces a clear-cutk p 2separation of the two species (table B2; fig. 4C).
Correlation AnalysesCorrelation of geographic position and environmental fac-
tors with Euclidean distances between sympatric populationsof O. nivara and O. rufipogon. The Euclidean distances be-
tween sympatric accessions of O. nivara and O. rufipogonexhibit no correlation with annual precipitation, moderate cor-relation with longitude ( , ) and alti-r p �0.4224 P p 0.0092tude ( , ), and a relatively strong cor-r p 0.3551 P p 0.0310relation with latitude ( , ) and meanr p 0.5804 P p 0.0002annual temperature ( , ).r p �0.6084 P p 0.0001
Correlation of geographic position and environmental factorswith intraspecific character variations. Ten characters in O.nivara and 11 in O. rufipogon show a geographical trend (table3). A particularly strong correlation is displayed by latitude withleaf width, spikelet width, sterile lemma width, and anther length
This content downloaded from 137.224.252.10 on Tue, 16 Jul 2013 08:49:58 AMAll use subject to JSTOR Terms and Conditions
Fig. 3 UPGMA dendrogram based on Gower’s distance of 101 populations of Oryza nivara and Oryza rufipogon. Bootstrap values over50% are displayed above the branches.
This content downloaded from 137.224.252.10 on Tue, 16 Jul 2013 08:49:58 AMAll use subject to JSTOR Terms and Conditions
Fig. 4 K-means cluster solutions plotted against the principal component analysis scatter plot (component 1 on the X-axis and component2 on the Y-axis). A, ; B, ; C, .k p 4 k p 3 k p 2
This content downloaded from 137.224.252.10 on Tue, 16 Jul 2013 08:49:58 AMAll use subject to JSTOR Terms and Conditions
Within Oryza rufipogon:Number of days from seeding to first heading �.2928 .0317Culm diameter �.4038 .0025Spikelet length �.2734 .0455Culm number .3547 .0085 �.3649 .0067Culm length �.3542 .0086 .4799 .0002Leaf length �.4426 .0008 .4670 .0004Panicle length �.3167 .0197 .3324 .0141Distance from panicle base to lowest spikelet insertion �.2684 .0498 .3289 .0152Stigma length �.3340 .0136 .3685 .0061Awn length .3279 .0055Seedling height .3205 .0182
Note. Correlation coefficient is denoted by r. Values with relatively strong correlations are underlined.
and also by longitude with leaf width and spikelet width withinO. nivara. Eight characters in O. nivara and six in O. rufipogonare significantly correlated (at level) to at least one ofP ! 0.05the examined environmental factors (e.g., altitude, annual meantemperature, annual precipitation; table 4).
Eight characters that are significantly correlated to one ormore of the geoclimatic factors are common to O. nivara andO. rufipogon (tables 3, 4). In both species, awn length tendsto increase with longitude while the number of days fromseeding to first heading tends to decrease with increasing al-titude. The remaining common characters exhibit contrastingcorrelation directions. As latitude increases, culm diameter andleaf length increase in O. nivara and decrease in O. rufipogon,while culm number decreases in O. nivara and increases in O.rufipogon. As longitude increases, seedling height and leaflength decrease in O. nivara and increase in O. rufipogon,while culm number increases in O. nivara and decreases in O.rufipogon. With increasing annual mean temperature, antherstend to lengthen in O. nivara and shorten in O. rufipogon,while with increasing annual precipitation, leaves tend to nar-row in O. nivara and broaden in O. rufipogon.
Analysis of Variance
Euclidean distances of sympatric species in geographicregions. The ANOVA reveals that the Euclidean distancesbetween sympatric O. nivara and O. rufipogon populationsare significantly greater (at level) in South AsiaP ! 0.05( ) than in Southeast Asia ( ).mean p 10.29 mean p 9.07
Interspecific character differences. Highly significant dif-ferences (at level) between O. nivara and O. ru-P ! 0.0001fipogon have been observed in 25 out of the 34 characters(table 5). The combined effect of all characters on the differ-ences between species is highly significant at level.P ! 0.0001
Intraspecific character differences. The combined effectsof the 34 characters on the differences between populationsoccurring in different geographic regions within O. nivara andwithin O. rufipogon were also highly significant at P !
level. In O. nivara, 17 characters were highly signifi-0.0001cantly different (at level) between the South AsianP ! 0.0001and Southeast Asian populations. South Asian populationstend to have taller seedlings; fewer, more erect and broaderculms; larger leaves and flag leaves; more erect flag leaf angle;less exserted and more compact panicles; broader spikelets;and sterile lemmas and shorter awns and anthers comparedto the Southeast Asian populations.
Within O. rufipogon seven characters differentiated the Aus-tralasian populations from the other geographic populations(at significance level). The ANOVA revealed that O.P ! 0.001rufipogon from Australasia possess significantly fewer culms,taller seedlings, and longer leaves, flag leaves, ligules, panicles,spikelets, awns, anthers and stigmas than the rest of the species.
Discussion
Two Morphological Species
Hierarchical, ordination, and partitional (k-means) cluster-ing methods all discriminate Oryza nivara from Oryza rufi-
This content downloaded from 137.224.252.10 on Tue, 16 Jul 2013 08:49:58 AMAll use subject to JSTOR Terms and Conditions
Correlation of Altitude, Annual Mean Temperature, and Annual Precipitation to Characterswithin Species at P ! 0.05 Significance Level
Character
AltitudeAnnual meantemperature
Annualprecipitation
r P value r P value r P value
Within Oryza nivara:Culm length �.3122 .0368Spikelet length .3110 .0376Number of days from seeding to first heading �.3136 .0359Leaf width .3466 .0197 �.3770 .0107 �.3414 .0217Spikelet width �.4359 .0028 �.3316 .0261Sterile lemma width �.3883 .0084Anther length .4419 .0024Sterile lemma length �.3746 .0112
Within Oryza rufipogon:Anther length �.4004 .0027Number of days from seeding to first heading �.3486 .0098 .4821 .0002Culm diameter �.2698 .0485 .3335 .0137
Stigma length .2821 .0388Leaf width .3636 .0069Number of basal primary branches of the panicle .4591 .0005
Note. Correlation coefficient is denoted by r.
pogon. This corroborates Sharma and Shastry’s (1965) tax-onomic decision to identify the annual element as a distinctmorphological species. In Laos, natural populations of O. ni-vara and O. rufipogon can be easily differentiated by flag leaflength and width, panicle branching type and distance fromthe panicle base to the lowest spikelet insertion (Banaticla-Hilario and Almazan 2010). Ng et al. (1981) and Barbier(1989) also found phenotypic discontinuities between the twotaxa, disputing the Asian wild rice perennial-annual continuumpostulated by Morishima et al. (1980, 1984).
Some of the phenotypic differences between the two wildOryza species can be seen as responses to the differences intheir habitat, breeding system, and life cycle. The presence ofstolons and long, strong, and spreading culms make O. rufi-pogon suitable for permanently inundated habitats while theshorter, less decumbent culms of O. nivara are probably moresuited for seasonally dry habitats. Stigmas and anthers arelonger and panicles are more exserted and open in the out-crossing O. rufipogon than in the inbreeding O. nivara. Leafsenescence tends to be earlier, and flowering is not as photo-period sensitive in the annual species. Adaptation to differenthabitat conditions (particularly to different hydrological re-gimes) led to differentiation of morphology, reproductive sys-tem, and life cycle (Morishima et al. 1984; Vaughan et al.2003). Oryza nivara and O. rufipogon have undergone andare still undergoing ecological speciation. However, the firstPC, which discriminated between the two species, accountedfor only 25% of the variation among the populations of thetwo species, indicating a comparatively large degree of intra-specific variation and/or introgression between the species.
The 18 accessions that held an intermediate position in thePCA plot (fig. 2) are probably of hybrid origin and requirefurther examination. In continental Asia, gene flow betweenpopulations of O. rufipogon (sensu lato) and Oryza sativa is
highly probable due to their overlapping distribution in thisregion (Vaughan et al. 2003). There are numerous accountsof hybridization between cultivated and wild rice (Oka andChang 1961; Oka 1988; Vaughan et al. 2008). However, hy-brid populations between O. nivara and O. rufipogon arerarely reported as they are hardly found growing side by sidein the same area (Morishima et al. 1984). Banaticla-Hilarioand Almazan (2010) observed considerable ecogeographic iso-lation between O. nivara and O. rufipogon in Vientiane, Laos,where the nearest populations were as far as 7 km apart. Twostudies identified genetically (Sano et al. 1980) and morpho-logically (Morishima et al. 1984) intermediate populations ofthe two wild species in Thailand. Therefore, it is highly possiblethat most of the intermediate accessions detected in this studyare actually hybrids between cultivated/escaped O. sativa andO. rufipogon or O. nivara rather than between the two wildspecies. Molecular techniques such as single nucleotide poly-morphism (SNP) genotyping could determine the genetic iden-tity of hybrid populations. Hybrids and nonhybrid individualscan coexist in mixed accessions as exemplified by the two O.rufipogon accessions R5 and R29 (app. A).
Geographic Races
As indicated in the PCA (fig. 2), HCA (fig. 3), k-means (fig.4), and ANOVA results, O. nivara can be partitioned intoSouth Asian and Southeast Asian populations that differ inculm, leaf, panicle and spikelet characters. Bayesian clusteringusing simple sequence repeat (SSR) data recognizes the geneticseparation of these regional groups at higher levels of popu-lations structure (Banaticla-Hilario 2012). Oryza nivara is con-fined to areas with a pronounced dry season and its occurrencehas not been reported in the more humid, western part ofMyanmar (Vaughan et al. 2008). This area defines the regional
This content downloaded from 137.224.252.10 on Tue, 16 Jul 2013 08:49:58 AMAll use subject to JSTOR Terms and Conditions
Note. Mean standard error as well as F and P values of each character are shown.
boundary of tropical continental Asia and this geoclimatic fac-tor probably restricts gene flow between the South and South-east Asian populations of O. nivara.
The PCA (fig. 2) and HCA (fig. 3) did not divide O. rufi-pogon into geographic groups, but at , the k-means anal-k p 4ysis separated all the Australasian and most (six out of nine)of the insular Southeast Asian populations from the majority(70.7%) of the continental Southeast Asian ones (fig. 4A). Thiscontinental-insular variation reflects the typically high geneticdifferentiation observed in island populations resulting fromtheir geographic isolation (Fedorenko et al. 2009). A similardivergence of mainland and island populations has been ob-served in species of Castilleja (Helernum et al. 2005), Ptero-carpus (Rivera-Ocasio et al. 2006), and Festuca (Fedorenko etal. 2009). Divergence of the Australasian populations of O.rufipogon from the rest of the species is also well supportedby molecular (Banaticla-Hilario 2012; Waters et al. 2012) andhybridization data (Banaticla-Hilario et al. 2013). This geo-graphic population is reportedly more similar to O. meridi-onalis than to Asian O. rufipogon based on chloroplast ge-nome sequences (Waters et al. 2012). Yet SSR analyses revealedthat at higher population structure levels, the Australasianpopulations are genetically nested within O. rufipogon (Ban-aticla-Hilario 2012). Similarly, the results of the current study(figs. 2–4) indicate that the Australasian O. rufipogon still fallwithin the morphological variation of the perennial species.
More Distinct in South Asia, More Alikein Southeast Asia
The South and Southeast Asia populations of O. nivara arephenotypically separated from the perennial species at differentdegrees. The ANOVA and PCA shows that O. nivara and O.rufipogon are more differentiated in South Asia than in South-east Asia (fig. 2). The correlation analysis likewise reveals thatthe Euclidean distance between sympatric O. nivara and O.rufipogon is correlated positively with latitude and negativelywith longitude. The strong negative correlation of Euclideandistance and annual mean temperature is largely influenced bythe inversely proportional relationship between the latter andlatitude. The results indicate that O. nivara and O. rufipogonin Southeast Asia are more recently diverged and/or had moreinterspecific gene flow compared to their South Asian coun-terparts. Accordingly, Mantel correlations of geographic andmorphological distances as well as genetic analysis based onSSRs suggest a similar geographic pattern of differentiationbetween the said species (Banaticla-Hilario 2012). Molecularanalysis using SNPs data and hybridization studies could beapplied to validate this conclusion and provide more detailedinformation on the migration of genes between O. nivara andO. rufipogon.
This content downloaded from 137.224.252.10 on Tue, 16 Jul 2013 08:49:58 AMAll use subject to JSTOR Terms and Conditions
Contrasting Variation Patterns and Responses toGeoclimatic Gradients
In continental Asia, regional differentiation is evident in O.nivara but lacking in O. rufipogon. This contrasting variationpattern agrees with the genetic diversity patterns observed inthe two species (Kuroda et al. 2007; Zhou et al. 2008; Zhengand Ge 2010) in which O. nivara showed higher interpopu-lation genetic differentiation than O. rufipogon. Again, dif-ferences in habitat and breeding system may have influencedthe different behavior of the two species. Inbreeding plants likeO. nivara generally exhibit low intrapopulation diversity butin situations where local adaptation is important, the differentpopulations that are isolated from each other would maintaindifferent alleles at certain loci, hence resulting in greater in-terpopulation than intrapopulation diversity (Silvertown andCharlesworth 2001). Furthermore, O. rufipogon is found incomparatively stable habitats while O. nivara occupies dis-turbed habitats (Kariali et al. 2008). Fluctuating environmentalconditions could have induced O. nivara to adapt to differ-ences in geoclimatic factors, subsequently leading to regionaldivergence of the annual species. In fact, four characters in O.nivara and none in O. rufipogon exhibited strong correlationwith latitude and annual mean temperature (tables 3, 4).
Several characters that are significantly correlated to certaingeoclimatic factors (tables 3, 4) exhibit different correlationdirections for O. nivara and O. rufipogon. Such opposing cor-relation directions suggest that the two species behave differ-ently towards the existing geographic and climatic gradientsand indicate phenotypic differentiation beyond the scope ofadaptive divergence. The contrasting morphological responsesprobably contribute to the phenotypic dissimilarities observedbetween sympatric O. nivara and O. rufipogon populations.This further supports our conclusion that O. rufipogon shouldbe recognized as being taxonomically different from O. nivaraat the species level.
Diagnostic Characters
The ANOVA detected 25 species-discriminating characters(table 5). Culms, stigmas, anthers, and distance from paniclebase to lowest spikelet insertion are longer in O. rufipogonthan in O. nivara. Anther length to spikelet length ratio is alsohigher in the perennial species. Leaves, flag leaves, and ligulesare longer; flag leaves, spikelets, and sterile lemmas are broaderand spikelet fertility and spikelet width to spikelet length ratioare higher in O. nivara. Anther length and spikelet width canreadily distinguish the two species from each other. Even so,it is not advisable to rely mainly on these characters. Thefollowing descriptions provide a clear delineation of each spe-cies as observed at IRRI.
Oryza rufipogon is characterized by a spreading habit, sto-lon formation, long culms ([74] 136 to more than 170 cm)that are moderately resistant to lodging, short leaves (15–37[70] cm), flag leaves (7–20 [40] cm), and ligules ([2.4] 18–30[48] mm), narrow flag leaves (0.4–0.95 [1.2] cm) that aremostly horizontal, well-exserted panicles with spreadingbranches, long distances from the panicle base to lowest spike-let insertion ([7] 16–26 [54] mm), slender spikelets (1.8–2.3[2.8] mm) and sterile lemmas (0.5–0.76 [0.90] mm), short ([30]
40–67 [100] mm) and thin (0.2 mm) awns, long anthers (13.7mm) and stigmas ([1.2] 1.8–2.8 mm), low spikelet width tospikelet length ratios (0.4–0.8), high anther length to spikeletlength ratios (0.4–0.8), low spikelet fertilities (2–32 [88]%),late leaf senescence, photoperiod sensitivity, and a perenniallife cycle. Culm number was detected by ANOVA as signifi-cantly lower in O. rufipogon. However, this character shouldonly be considered when dealing with potted plants or plantsin a contained environment. It was reported that in their nat-ural habitat, O. nivara produces fewer tillers while O. rufi-pogon generates abundant tillers (Kariali et al. 2008).
Oryza nivara can be distinguished by semierect to open hab-its, absence of stolons, short plant height (50–150 [!170] cm),weaker culms (compared to O. rufipogon), long leaves (23–50 [79] cm), flag leaves ([16] 30–49 cm) and ligules ([9.4] 21–44 mm), broad flag leaves ([0.40] 1.1–1.8 cm) that are mostlysemierect, moderately well-exserted panicles with compact tosemicompact branches, short distances from the panicle baseto the lowest spikelet insertion (9–18 [31] mm), broad spikelets([2.2] 2.8–3.4 mm) and sterile lemmas ([0.6] 0.83–1.2 mm),long (50–88 [133] mm)] and thick (0.3 mm) awns, short an-thers (!3.7 mm) and stigmas ([0.9] 1.4–1.8 [2.3] mm), lowanther length to spikelet length ratios (usually !0.4), highspikelet width to spikelet length ratios (0.28 to 0.4), high spike-let fertilities ([24] 30–64 [96]%), early leaf senescence, lack ofphotoperiod sensitivity, and an annual life cycle.
These descriptions delineate the two species morphologicallyand can be used to identify Oryza series Sativae plants (fromthe Asia-Pacific region) or reclassify those that have been mis-identified at the species level. We hope this will help gene banksand herbaria in providing correct taxonomic labels for theirOryza accessions and specimens.
It should be noted that this research used gene bank acces-sions that may represent only a portion of the diversity foundin natural populations. Furthermore, sampling five individualsper accession may not be enough to capture the entire variationin populations. This poses a challenge for gene banks to ensurethat sufficient diversity (to represent natural populations) iscontained in the accessions and preserved throughout differentgene bank procedures (such as regeneration and characteri-zation). Additionally, more in situ studies are needed to un-derstand the diversity of O. nivara and O. rufipogon. Nev-ertheless, the observed variation patterns in this study areremarkable and reflect how ecology and geography steer theevolution of Asian wild rices.
Acknowledgments
This study was supported by the T. T. Chang Genetic Re-sources Center (TTC-GRC) in the International Rice ResearchInstitute and the Biosystematics Group of Wageningen Uni-versity. The TTC-GRC wild rice nursery staff (Ma. SocorroAlmazan, Remegio Aguilar, Vicente Arcillas, Edwin Jarabejo,Nora Kuroda, Yolanda Malatag, Mae Merluza, and Liza Yon-zon) provided assistance in phenotyping. The PBGB-CRILteam (Violeta Bartolome, Leilani Nora, and Rose Imee ZhellaMorantte) and Nadine Singson were consulted about statisticalmethods.
This content downloaded from 137.224.252.10 on Tue, 16 Jul 2013 08:49:58 AMAll use subject to JSTOR Terms and Conditions
BANATICLA-HILARIO ET AL.—MORPHOLOGY OF TWO ASIAN WILD RICE SPECIES 907
Appendix A
Codes, Accession Numbers, and Geographic Origin of the Studied Accessions
Shown are code, species, International Rice Genebank (IRG) accession number, country of origin, and latitude and longitude.Code was assigned by the authors. Oryza nivara and Oryza rufipogon accessions were georeferenced on the basis of availablelocation data in the IRG Information System (IRGCIS); latitude and longitude are in decimal degrees. Please note: N3 is currentlylabeled as O. rufipogon in IRGCIS and tentatively reclassified by the author as O. nivara on the basis of seed morphology. ForN17 and N30, seeds did not germinate. R65 is currently labeled as O. meridionalis in IRGCIS and tentatively reclassified bythe author as O. rufipogon on the basis of seed morphology.
N1, O. nivara, IRGC 105882, Bangladesh, 23.0166, 89.4666; N2, O. nivara, IRGC 103830, Bangladesh, 24.3333, 88.9000;N3, O. nivara, IRGC 93129, Cambodia, 10.6333, 103.7833; N4, O. nivara, IRGC 105719, Cambodia, 11.4167, 104.9667;N5, O. nivara, IRGC 89216, Cambodia, 11.4467, 104.4889; N6, O. nivara, IRGC 89187, Cambodia, 11.4969, 104.8183; N7,O. nivara, IRGC 89172, Cambodia, 11.5094, 104.8189; N8, O. nivara, IRGC 89185, Cambodia, 11.5297, 104.8342; N9, O.nivara, IRGC 89212, Cambodia, 11.5317, 104.8283; N10, O. nivara, IRGC 106320, Cambodia, 11.8333, 104.7933; N11, O.nivara, IRGC 88939, Cambodia, 12.5333, 106.6166; N12, O. nivara, IRGC 92886, Cambodia, 12.6166, 103.9833; N13, O.nivara, IRGC 92677, Cambodia, 12.6500, 104.9166; N14, O. nivara, IRGC 92823, Cambodia, 13.0500, 104.5166; N15, O.nivara, IRGC 106339, Cambodia, 13.3833, 103.8333; N16, O. nivara, IRGC 92943, Cambodia, 14.1667, 103.2667; N17e,O. nivara, IRGC 103821, China, 23.1833, 114.4500; N18, O. nivara, IRGC 80549, India, 19.6667, 81.6833; N19, O. nivara,IRGC 80560, India, 19.7500, 81.3333; N20, O. nivara, IRGC 80601, India, 20.0000, 81.5000; N21, O. nivara, IRGC 80677,India, 21.1333, 81.3333; N22, O. nivara, IRGC 106101, India, 21.7500, 87.5000; N23, O. nivara, IRGC 80645, India, 22.0333,81.9167; N24, O. nivara, IRGC 106051, India, 22.0833, 86.4167; N25, O. nivara, IRGC 106054, India, 22.1667, 86.0000;N26, O. nivara, IRGC 81837, India, 25.5000, 84.3333; N27, O. nivara, IRGC 86699, Laos, 15.4100, 106.7000; N28, O.nivara, IRGC 106148, Laos, 17.9167, 102.7500; N29, O. nivara, IRGC 106151, Laos, 18.2000, 102.7500; N30e, O. nivara,IRGC 106358, Myanmar, 16.6667, 96.4667; N31, O. nivara, IRGC 106347, Myanmar, 17.2500, 96.6667; N32, O. nivara,IRGC 93192, Nepal, 27.5119, 83.2428; N33, O. nivara, IRGC 93185, Nepal, 28.0200, 81.6033; N34, O. nivara, IRGC 93191,Nepal, 28.0211, 81.6011; N35, O. nivara, IRGC 93184, Nepal, 28.0283, 81.6322; N36, O. nivara, IRGC 93195, Nepal,28.1008, 82.2683; N37, O. nivara, IRGC 103422, Sri Lanka, 8.9167, 81.0000; N38, O. nivara, IRGC 105803, Thailand,17.0833, 104.0833; N39, O. nivara, IRGC 104724, Thailand, 14.4667, 100.1166; N40, O. nivara, IRGC 105765, Thailand,14.3333, 102.9167; N41, O. nivara, IRGC 105755, Thailand, 14.5000, 102.0833; N42, O. nivara, IRGC 104743, Thailand,14.9666, 102.1166; N43, O. nivara, IRGC 104756, Thailand, 15.1000, 104.3333; N44, O. nivara, IRGC 105859, Thailand,14.6667, 102.3333; N45, O. nivara, IRGC 104473, Thailand, 16.8333, 100.2500; N46, O. nivara, IRGC 105801, Thailand,17.0000, 104.0833; N47, O. nivara, IRGC 105809, Thailand, 17.2500, 103.7500; N48, O. nivara, IRGC 105825, Thailand,17.6667, 102.9167; N49, O. nivara, IRGC 105828, Thailand, 17.8450, 102.5841; N50, O. nivara, IRGC 104736, Thailand,18.1500, 100.1333; N51, O. nivara, IRGC 86496, Vietnam, 12.5681, 107.7556; N52, O. nivara, IRGC 86493, Vietnam,13.6178, 108.1171; R1, O. rufipogon, IRGC 105881, Bangladesh, 23.0666, 89.3500; R2, O. rufipogon, IRGC 103827,Bangladesh, 24.1500, 89.0500; R3, O. rufipogon, IRGC 93085, Cambodia, 10.6333, 103.7833; R4, O. rufipogon, IRGC 105720,Cambodia, 11.4167, 104.9833; R5, O. rufipogon, IRGC 89228, Cambodia, 11.4467, 104.4889; R6, O. rufipogon, IRGC 89230,Cambodia, 11.4969, 104.8183; R7, O. rufipogon, IRGC 89223, Cambodia, 11.5094, 104.8189; R8, O. rufipogon, IRGC 89227,Cambodia, 11.5297, 104.8342; R9, O. rufipogon, IRGC 89232, Cambodia, 11.5317, 104.8283; R10, O. rufipogon, IRGC106321, Cambodia, 11.8333, 104.7833; R11, O. rufipogon, IRGC 89007, Cambodia, 12.5333, 106.6166; R12, O. rufipogon,IRGC 110408, Cambodia, 12.6166, 103.9833; R13, O. rufipogon, IRGC 99538, Cambodia, 12.6500, 104.9166; R14, O.rufipogon, IRGC 93063, Cambodia, 13.0500, 104.5166; R15, O. rufipogon, IRGC 106335, Cambodia, 13.3833, 103.8500;R16, O. rufipogon, IRGC 110409, Cambodia, 14.1667, 103.2667; R17, O. rufipogon, IRGC 103823, China, 23.2333, 114.0333;R18, O. rufipogon, IRGC 80550, India, 19.6667, 81.7667; R19, O. rufipogon, IRGC 80562, India, 19.7500, 81.3167; R20,O. rufipogon, IRGC 80600, India, 20.0000, 81.5000; R21, O. rufipogon, IRGC 80680, India, 21.1333, 81.5667; R22, O.rufipogon, IRGC 82983, India, 21.7500, 87.5000; R23, O. rufipogon, IRGC 80643, India, 22.0333, 82.8167; R24, O. rufipogon,IRGC 82982, India, 22.0833, 86.3667; R25, O. rufipogon, IRGC 106055, India, 22.1667, 85.9167; R26, O. rufipogon, IRGC81881, India, 25.5000, 84.1333; R27, O. rufipogon, IRGC 86697, Laos, 15.4100, 106.7000; R28, O. rufipogon, IRGC 106149,Laos, 17.9167, 102.7500; R29, O. rufipogon, IRGC 106152, Laos, 18.2000, 102.7500; R30, O. rufipogon, IRGC 106357,Myanmar, 16.6667, 96.5000; R31, O. rufipogon, IRGC 106346, Myanmar, 17.2500, 96.6333; R32, O. rufipogon, IRGC 93221,Nepal, 27.5119, 83.2439; R33, O. rufipogon, IRGC 93218, Nepal, 28.0200, 81.6033; R34, O. rufipogon, IRGC 93220, Nepal,28.0211, 81.6011; R35, O. rufipogon, IRGC 93210, Nepal, 28.0283, 81.6322; R36, O. rufipogon, IRGC 93216, Nepal, 28.1008,82.2683; R37, O. rufipogon, IRGC 103423, Sri Lanka, 8.5000, 81.1667; R38, O. rufipogon, IRGC 105804, Thailand, 14.4667,100.1166; R39, O. rufipogon, IRGC 104713, Thailand, 14.5000, 102.0833; R40, O. rufipogon, IRGC 105766, Thailand,14.5000, 102.9167; R41, O. rufipogon, IRGC 105758, Thailand, 14.6667, 102.3333; R42, O. rufipogon, IRGC 104742,Thailand, 14.9666, 102.1166; R43, O. rufipogon, IRGC 104757, Thailand, 15.1000, 104.3333; R44, O. rufipogon, IRGC105860, Thailand, 16.8333, 100.2500; R45, O. rufipogon, IRGC 104474, Thailand, 17.0000, 104.0833; R46, O. rufipogon,IRGC 105800, Thailand, 17.0833, 104.0833; R47, O. rufipogon, IRGC 82979, Thailand, 17.1667, 103.7500; R48, O. rufipogon,IRGC 105823, Thailand, 17.6667, 102.9167; R49, O. rufipogon, IRGC 105829, Thailand, 17.8450, 102.5841; R50, O.
This content downloaded from 137.224.252.10 on Tue, 16 Jul 2013 08:49:58 AMAll use subject to JSTOR Terms and Conditions
rufipogon, IRGC 104737, Thailand, 18.1500, 100.1333; R51, O. rufipogon, IRGC 86512, Vietnam, 12.5681, 107.7556; R52,O. rufipogon, IRGC 86506, Vietnam, 13.8333, 107.8667; R53, O. rufipogon, IRGC 80774, Philippines, 7.8803, 125.0061;R54, O. rufipogon, IRGC 81976, Indonesia, �6.5569, 106.7617; R55, O. rufipogon, IRGC 81977, Indonesia, �6.1081,106.7097; R56, O. rufipogon, IRGC 81978, Indonesia, �6.3333, 106.3000; R58, O. rufipogon, IRGC 105567, Indonesia,�0.7500, 117.2667; R59, O. rufipogon, IRGC 105952, Indonesia, �6.4167, 107.0000; R60, O. rufipogon, IRGC 105958,Indonesia, �6.0631, 106.1189; R61, O. rufipogon, IRGC 106452, Indonesia, �3.2500, 104.6667; R62, O. rufipogon, IRGC106453, Indonesia, �3.0833, 104.5000; R63, O. rufipogon, IRGC 86542, Australia, �12.5333, 131.7167; R64, O. rufipogon,IRGC 93274, Indonesia, �8.2930, 140.4089; R65, O. rufipogon, IRGC 105283, Australia, �12.5833, 131.3333; R66, O.rufipogon, IRGC 105293, Australia, �15.6667, 145.2500; R67, O. rufipogon, IRGC 105303, Australia, �13.5000, 141.7500.
Literature Cited
Aggarwal, RK, DS Brar, S Nandi, N Huang, GS Khush 1999Phylogenetic relationships among Oryza species revealed by AFLPmarkers. Theor Appl Genet 98:1320–1328.
Banaticla-Hilario, MCN 2012 An ecogeographic analysis of Oryzaseries Sativae in Asia and the Pacific. PhD thesis. WageningenUniversity.
Banaticla-Hilario, MCN, MSR Almazan 2010 Field observation ofwild rice populations in Vientiane, Laos. International Rice ResearchInstitute, Los Banos, Laguna, Philippines.
Banaticla-Hilario MCN, KL McNally, RG van den Berg, HR SackvilleHamilton 2013 Crossability patterns within and among Oryza se-ries Sativae species from Asia and Australia. Genetic Resources andCrop Evolution, doi:10.1007/s10722-013-9965-4.
Barbier P 1989 Genetic variation and ecotypic variation in the wildrice species Oryza rufipogon. ii. Influence of the mating system andlife-history traits on the genetic structure of populations. Jpn J Genet64:273–285.
Bioversity International, International Rice Research Institute, andWARDA 2007 Descriptors for wild and cultivated rice (Oryzaspp.). Bioversity International, Rome, Italy; International Rice Re-search Institute, Los Banos, Philippines; WARDA, Africa Rice Cen-ter, Cotonou, Benin.
Ding C, X He. 2004. K-means clustering via principal componentanalysis. In R Greiner, D Schuurmans, 2004 eds. Proceedings of the21st International Machine Learning Conference. Association forComputing Machinery, New York.
Duan S, B Lu, Z Li, J Tong, J Kong, W Yao, S Li, Y Zhu 2007Phylogenetic analysis of AA-genome Oryza species (Poaceae) basedon chloroplast, mitochondrial, and nuclear DNA sequences.Biochem Genet 45:113–129.
Fedorenko OM, MV Gritskikh, IE Malysheva, TS Nikolaev-skaya 2009 Genetic diversity of insular natural populations of Fes-tuca pratensis huds.: RAPD analysis. Russ J Genet 45:1134–1138.
Helenurm K, R West, SJ Burckhalter 2005 Allozyme variation in theendangered insular endemic Castilleja grisea. Ann Bot 95:1221–1227.
Hijmans RJ, L Guarino, M Cruz, E Rojas 2001 Computer tools forspatial analysis of plant genetic resources data: 1. Diva-gis. PlantGenet Res Newsl 127:15–19.
Hiroi K, A Al Mamun, T Wada, Y Takeoka 1990 A study on theinterspecific variation of panicle structure in the genus Oryza. JpnJ Crop Sci 59:815–823.
Iwamoto M, H Nagashima, T Nagamine, H Higo, K Higo 1999 Atourist element in the flanking region of the catalase gene cata revealsevolutionary relationships among Oryza species with various ge-nome types. Mol Genet Genom 262:493–500.
Kariali E, S Rani Kuanar, P Kumar Mohapatra 2008 Individual tillerdynamics of two wild Oryza species in contrasting habitats. PlantProd Sci 11:355–360.
Kuroda Y, Y-I Sato, C Bounphanousay, Y Kono, K Tanaka 2007Genetic structure of three Oryza AA genome species (O. rufipogon,
O. nivara and O. sativa) as assessed by SSR analysis on the VientianePlain of Laos. Conserv Genet 8:149–158.
Kwon S-J, JK Lee, S-W Hong, Y-J Park, KL McNally, N-S Kim 2006Genetic diversity and phylogenetic relationship in AA Oryza speciesas revealed by rim2/hipa cacta transposon display. Genes Genet Syst81:93–101.
Lu B-R 1999 Taxonomy of the genus Oryza (Poaceae): historical per-spective and current status. Int Rice Res Notes 24:4–8.
Lu B-R, S Ge, T Sang, J-K Chen, D-Y Hong 2001 The current tax-onomy and perplexity of the genus Oryza (Poaceae). Acta Phyto-taxon Sin 39:373–388.
Maechler M P Rousseeuw, A Struyf, M Hubert, K Hornik 2011Cluster analysis basics and extensions. R package version 1.14.1
Mendiburu de F. 2010. Agricolae: statistical procedures for agriculturalresearch. R package version 1.0–9. http://CRAN.R-project.org/packagepagricolae.
Morishima H, Y Sano, HI Oka 1980 Observations on wild and cul-tivated rices and companion weeds in the hilly areas of Nepal, Indiaand Thailand. National Institute of Genetics, Misima, Japan.
——— 1984 Differentiation of perennial and annual types due tohabitat conditions in the wild rice Oryza perennis. Plant Syst Evol144:119–135.
——— 1992 Evolutionary studies in cultivated rice and its wild rel-atives. Oxford Surv Evol Biol 8:138–184.
Ng NQ, TT Chang, JT Williams, JG Hawkes 1981 Morphologicalstudies of Asian rice and its related wild species and the recognitionof a new Australian taxon. Biol J Linn Soc 16:303–313.
Oka HI 1988 Origin of cultivated rice. Japan Science Society, Tokyo.254 pp.
Oka HI, WT Chang 1961 Hybrid swarms between wild and culti-vated rice species, Oryza perennis and O. sativa. Evolution 15:418–430.
Park KC, NH Kim, YS Cho, KH Kang, JK Lee, NS Kim 2003 Geneticvariations of AA genome Oryza species measured by MITE-AFLP.Theor Appl Genet 107:203–209.
R Development Core Team. 2011. R: a language and environment forstatistical computing. R Foundation for Statistical Computing,Vienna.
Rivera-Ocasio E, TM Aide, WO McMillan 2006 The influence ofspatial scale on the genetic structure of a widespread tropical wet-land tree Pterocarpus officinalis (Fabaceae). Conserv Genet 7:251–266.
Sano Y, H Morishima, H-I Oka 1980 Intermediate perennial-annualpopulations of Oryza perennis found in Thailand and their evolu-tionary significance. Bot Mag Tokyo 93:291–305.
Second G 1985 Evolutionary relationships in the sativa group ofOryza based on isozyme data. Genet Sel Evol 17:89–114.
Sharma SD, SVS Shastry 1965 Taxonomic studies in the genus OryzaL. III. O. rufipogon Griff. sensu stricto and O. nivara Sharma etShastry nom. Nov. Indian J Genet Plant Breed 25:157–167.
Silvertown J, D Charlesworth 2001 Introduction to plant populationbiology. Blackwell, Oxford.
This content downloaded from 137.224.252.10 on Tue, 16 Jul 2013 08:49:58 AMAll use subject to JSTOR Terms and Conditions
BANATICLA-HILARIO ET AL.—MORPHOLOGY OF TWO ASIAN WILD RICE SPECIES 909
Tateoka T 1963 Taxonomic studies of Oryza. III. Key to the speciesand their enumeration. Bot Mag Tokyo 76:165–173.
Vaughan DA, B-R Lu, N Tomooka 2008 The evolving story of riceevolution. Plant Sci 174:394–408.
Vaughan DA, H Morishima, K Kadowaki 2003 Diversity in theOryza genus. Curr Opin Plant Biol 6:139–146.
Waters DLE, CJ Nock, R Ishikawa, N Rice, RJ Henry 2012Chloroplast genome sequence confirms distinctness of Australianand Asian wild rice. Ecol Evol 2:211–217.
Xu X, X Liu, S Ge, JD Jensen, F Hu, X Li, Y Dong, et al 2012Resequencing 50 accessions of cultivated and wild rice yields mark-
ers for identifying agronomically important genes. Nat Biotechnol30:105–111.
Zheng X, S Ge 2010 Ecological divergence in the presence of geneflow in two closely related Oryza species (Oryza rufipogon and O.nivara). Mol Ecol 19:2439–2454.
Zhou H-F, X-M Zheng, R-X Wei, G Second, DA Vaughan, SGe 2008 Contrasting population genetic structure and gene flowbetween Oryza rufipogon and Oryza nivara. Theor Appl Genet 117:1181–1189.
Zhu Q, S Ge 2005 Phylogenetic relationships among a genome speciesof the genus Oryza revealed by intron sequences of four nucleargenes. New Phytol 167:249–265.
This content downloaded from 137.224.252.10 on Tue, 16 Jul 2013 08:49:58 AMAll use subject to JSTOR Terms and Conditions
