Plant materials

Thirty-two Dahanala accessions were included in this study, listed in Table 1. Seeds of 26 accessions were obtained from the PGRC genebank, Gannoruwa, Sri Lanka, and six accessions from RRDI, Batalagoda, Sri Lanka. Accession 006739 is the oldest Dahanala line available in Sri Lankan genebanks. The advanced breeding line Oryza sativa Bg 360 was used as reference variety during the analysis.

Morphological seed characterization

Five replicated seeds of each Dahanala accession were evaluated with seven qualitative and quantitative characters (i.e. seed shape, colour and pubescence of lemma and palea, seed coat colour, sterile lemma colour, grain length and grain width), using a modified rice seed evaluation list based on the International Plant Genetic Resources Institute rice descriptors (Bioversity International, IRRI and WARDA, 2007). See Supplemental Data 1 for a detailed description of seed morphological characters.

Molecular characterization

Trichome density analysis

A total of 297 rice seedlings were transplanted representing all 32 accessions of Dahanala along with the reference variety (Bg 360) into pots filled with mud at the PGRC plant house. Trichome density was determined 30 days after transplanting, using the third leaves of three randomly selected plants for each accession. The trichome count was acquired by observing the first ¼ area of the adaxial surface of the leaf blade from the tip of each sample under a light microscope keeping a paper strip with a 2×3mm² square hole on it. Nine readings per individual were recorded. The average trichome density (trichome number/mm²) per accession was calculated from the mean of three counts per sample (Wickramasinghe et al., 2007).

Data scoring and analysis

Morphological characterization: Cluster analysis of morphological data was carried out through the average linkage method and Euclidean distance using Minitab version 15 (Minitab, 1991).

Molecular characterization: Gel images were scored manually by visual observation as presence (1) and absence (0) of every allele for all 31 pairs of SSR markers. DNA bands of expanded bulks were scored individually along with non-expanded bulks. Scored data were analyzed by Power Marker version 3.25 (Liu & Muse, 2005) calculating major allele frequency, polymorphic information content (PIC) value and Nei’s genetic distance (Nei, Tajima, & Tateno, 1983) across the 32 accessions to identify the prevailing genetic diversity. The phylogenetic tree was constructed based on Nei’s genetic distance (Nei et al., 1983) according to neighbour-joining method using MEGA 6.06 software (Tamura, Stecher, Peterson, Filipski, & Kumar, 2013).

Trichome density analysis: The trichome density distribution of the population was assessed through the Kruskal-Wallis test at 0.05 significance level by using Minitab version 15.

Morphological characterization of 32 Dahanala accessions

Considerable variation in seed morphology was observed among the 32 Dahanala accessions (Table 3, Figure 1). Grain shapes were either half-spindle shaped (6–7mm) or spindle shaped (7–8mm) whilst vast variation was identified in lemma and palea colour. The majority of accessions had brownish-black lemma and palea colour, only one had brownish-black furrows on straw background (Acc. No. 003971). All seeds had red pericarp and straw-coloured sterile lemma, showing no diversity in these traits. The nature of the pubescence varied between short hairs and hairs on the upper portion. Grain width ranged from 2.858mm to 3.508mm while grain length varied from 9.314mm to 6.946mm.

In a dendrogram based on the seed morphological characters, the 32 accessions clustered into eight groups at 91.18% similarity level showing their potential closer relationship (Figure 2). Accessions 003971, 002049, 002050 and 014122 clustered independently and distantly denoting morphological deviations from the rest of the accessions.

Molecular assessment of Dahanala accessions using SSR microsatellite markers

To reduce costs, the 32 DNA samples were pooled into eight bulks at 91.18% similarity level, based on the dendrogram resulting from seed morphological characterization (Figure 3). All DNA bulks were able to produce amplified fragments with all selected 31 SSR primers, of which 15 primers with uniform bands without any heterogenic banding patterns. (Figure 3 and Supplemental Figure 1).

DNA bulks (B3, B4 and B6) displaying heterogenic banding patterns or unusual stutter bands for some markers (e.g. RM515, Supplemental Figure 2) were expanded and PCR products resolved again on 8% PAGE (Figure 4; Supplemental Figure 3 and 4) along with other non-expanded bulks amplified with the same primer. DNA bulks displaying a single banding pattern were not expanded as all DNA samples contained in those bulks were assumed to be homozygous for that marker.

Allelic diversity

Three to six alleles per locus were amplified with the mean value of four by 31 SSR primers across the 32 Dahanala accessions resulting in a total of 124 alleles. Except for monomorphic primer RM255, which was discarded from further analysis, the other 30 markers produced a range of alleles among the bulks and individual accessions. RM202 primer had the highest polymorphism, yielding six alleles per locus. The lowest major allele frequency was observed for marker RM515 (0.303). Genetic diversity ranged between 0.169 (RM216; RM236) and 0.744 (RM515) with a mean value of 0.475. The PIC varied between 0.161 (RM216; RM236) and 0.697 (RM412) (Table 4).

Cluster analysis

A phylogenetic tree was constructed from Nei’s genetic distances and showed significant genetic diversity within the population by grouping all accessions into six major clusters (Figure 5). Some variation of the clustering pattern was observed compared to previously classified eight bulks based on morphology characterization. Bulks 1, 2 and 8 from morphological characterization, along with control Bg 360 clustered together, also bulks 6 and 7 formed a single cluster. On the other hand, bulk 4 is separated into two sub-clusters based on molecular characterization. Furthermore, phylogenetic results revealed accessions 003924/003327 and accessions 0010160/006165 as two sets of putative duplicates among the 32 Dahanala accessions, suggesting the possibility of having duplicated samples among conserved accessions.

Trichome analysis

The 32 Dahanala accessions were characterized for their trichome morphology and density as a proxy for potential thrips resistance. A significant variation in trichome density within the population was identified (P<0.05) (Table 5). The highest and the lowest density values were recorded from accession 003386 (30.83 trichomes/mm²) and accession 006357 (6.70 trichomes/mm²), respectively. The Suduru Samba variety, which is highly susceptible to thrips, had no trichomes (Figure 6) while moderately thrips susceptible Bg 360 showed medium trichome density (8 trichomes/mm²) (Table 5).

Besides the density variation, differences in nature and the placements of the trichomes on the adaxial surfaces of the rice leaves were observed under the light microscope (Figure 8; Figure 7).

Evaluation of the presence of RM277 and RM279 in selected Dahanala accessions

Based on previous studies of Gimhani (2010), RM277 and RM279 primers, selected for having a putative relationship with thrips resistance and trichome density, were used for PCR amplification of selected accessions. Ten accessions were selected to represent all 32 accessions of Dahanala based on the calculated trichome density values and significantly different morphology of the trichomes. These included two accessions with long trichomes (Acc. No. 003924; Acc. No. 003304), one accession with short trichomes (Acc. No. 002049), two accessions with the highest and lowest trichome density (Acc. No. 003386 and 006357, respectively), one accession with comparatively bigger gaps in between two trichomes (Acc. No. 592), and one accession which deviated in the cluster analysis (Acc. No. 003971). Moderately thrips-susceptible Bg 360 and highly thrips-susceptible Suduru Samba (two accessions, see Table 1) were used as reference samples.

No polymorphism was observed in amplified PCR products of RM277 between the Bg 360 sample and the selected Dahanala accessions with lengthy trichomes (Acc. No. 003924; Acc. No. 003304), the accession with the lowest trichome density (Acc. No. 006357), the accession with the highest trichome density (Acc. No. 003386), the accession with short trichomes (Acc. No. 002049), the accession with comparatively higher gaps between two trichomes (Acc. No. 592), and the accession that revealed to be a deviated accession in cluster analysis (Acc. No. 003971) (Figure 8 ). However, the highly susceptible Suduru Samba variety (zero trichomes) was polymorphic in this marker compared to other samples.

Trichome analysis

Since previous studies have revealed a relationship of thrips resistance with the presence of non-glandular trichomes on the adaxial surface of the leaves (Nugaliyadde et al., 1984), and thrips reported to be fed on the mesophyll cells of the young leaves through the adaxial surface (Gimhani, 2010), our study focused on inspection of the trichome density on the adaxial surface of Dahanala, which is known as a thrips-resistant variety.

Even though all examined accessions were conserved in genebanks under the same variety name, we detected a significant variation among trichome densities as well as in the trichomes' nature and locality among accessions (Figure 7; Figure 6).

In a first attempt at identifying QTLs for thrips resistance in rice, Gimhani (2010) identified both thrips resistance and trichome density as quantitative traits based on bulk segregating analysis followed by a screening of an F₂ population between thrips-resistant Dahanala and susceptible Suduru Samba. Gimhani (2010) found a significant negative correlation between thrips damage and the trichome density values (r = -0.378, P<0.05), revealing the damage score is low when the trichome density is high. They also showed that 14.3% of phenotypic variation (thrips resistance) in F₂ segregants could be explained by a factor of presence of trichomes, suggesting that the presence of trichomes could play a significant role in defending the plants against thrips, in agreement with the findings of Nugaliyadde et al. (1984), who reported that morphological features of the foliage and the presence of allomones in resistant plants contribute to their defence against thrips. In addition,Ananthakrishnan (1979) found that leaf age and thickness, and the nature and distribution of trichomes and silica cells in the substrate are some important biophysical factors that influence thrips oviposition. On the other hand, Wickramasinghe et al. (2007), showed that the removal of trichomes on the adaxial surface of the leaf of Dahanala seedlings did not result in a highly susceptible reaction. This indicates that besides trichome density there might be several other factors influencing thrips resistance. Based on the analysis of this study, the nature along with the position of trichomes on the leaf can also be one of those factors.

Although an association of trichrome density with RM277 and RM279 markers was identified previously (Gimhani, 2010), the results of the current project were not able to provide clear evidence of this relationship, as the Dahanala accessions used in this study were not polymorphic for RM277 and the allelic patterns for RM279 of both accessions with the highest (003386; 30.83 trichomes/mm²) and lowest (006357; 6.70 trichomes/mm²) trichome density were identical. Additionally, as our current study only analyzed selected representative samples of Dahanala using RM279 and RM277 putative markers, the results are insufficient to predict a clear association. Hence additional studies will be required to further test this hypothesis.

Although marker RM277 did not show polymorphism between selected Dahanala accessions and the Bg 360 variety, the highly thrips-susceptible Suduru Samba varieties with zero trichomes were polymorphic for this marker. Therefore, RM277 can be suitable for use in future analysis to detect Suduru Samba against the Dahanala variety. Additionally, since significant allelic variation was discovered within all selected Dahanala accessions as well as among Dahanala, Suduru Samba, and Bg 360 varieties with RM279 marker, this evidence can be used to select among these groups in future studies.

One reason for not detecting a strong association between trichome density and RM279 or RM277 markers in this research might be some limitations associated with the study, such as the low number of accessions studied. Therefore, before selecting a donor parent for thrips resistance among Dahanala accessions, further investigations, including disease assays, are required to corroborate the relationship between trichome density, RM277 and RM279 markers and a possible thrips resistance.