Marker Assisted Breeding to Develop Multiple Stress Tolerant Varieties for Flood and Drought Prone Areas

Background Climate extremes such as drought and flood have become major constraints to the sustainable rice crop productivity in rainfed environments. Availability of suitable climate-resilient varieties could help farmers to reduce the grain yield losses resulting from the climatic extremities. The present study was undertaken with an aim to develop high-yielding drought and submergence tolerant rice varieties using marker assisted introgression of qDTY1.1, qDTY2.1, qDTY3.1 and Sub1. Performance of near isogenic lines (NILs) developed in the background of Swarna was evaluated across 60 multi-locations trials (MLTs). The selected promising lines from MLTs were nominated and evaluated in national trials across 18 locations in India and 6 locations in Nepal. Results Grain yield advantage of the NILs with qDTY1.1 + qDTY2.1 + qDTY3.1 + Sub1 and qDTY2.1 + qDTY3.1 + Sub1 ranged from 76 to 2479 kg ha− 1 and 396 to 2376 kg ha− 1 under non-stress (NS) respectively and 292 to 1118 kg ha− 1 and 284 to 2086 kg ha− 1 under reproductive drought stress (RS), respectively. The NIL, IR96322–34-223-B-1-1-1-1 having qDTY1.1 + qDTY2.1 + qDTY3.1 + Sub1 has been released as variety CR dhan 801 in India. IR 96321–1447-651-B-1-1-2 having qDTY1.1 + qDTY3.1 + Sub 1 and IR 94391–131–358-19-B-1-1-1 having qDTY3.1 + Sub1 have been released as varieties Bahuguni dhan-1′ and ‘Bahuguni dhan-2’ respectively in Nepal. Background recovery of 94%, 93% and 98% was observed for IR 96322–34-223-B-1-1-1-1, IR 96321–1447-651-B-1-1-2 and IR 94391–131–358-19-B-1-1-1 respectively on 6 K SNP Infinium chip. Conclusion The drought and submergence tolerant rice varieties with pyramided multiple QTLs can ensure 0.2 to 1.7 t ha− 1 under reproductive stage drought stress and 0.1 to 1.0 t ha− 1 under submergence conditions with no yield penalty under non-stress to farmers irrespective of occurrence of drought and/or flood in the same or different seasons. Electronic supplementary material The online version of this article (10.1186/s12284-019-0269-y) contains supplementary material, which is available to authorized users.


Background
Increasing incidences of abiotic stresses under changing climate are major constraints to meet the ever-growing demand for food for rapidly escalating population and attain the global food security (Lesk et al. 2016). Drought and flood are the two most prevalent abiotic stresses reducing rice yield in the rainfed environments.
Worldwide, drought and flood stresses have been reported to affect approximately 40 million hectares of total rice area at different crop stages producing negative impacts on plant growth, development and yield (Barnabás et al. 2008;Neeraja et al. 2007). Rainfed rice ecosystems in South Asia and Southeast Asia are the key hotspots for the occurrence of combination of drought and flood stresses (Dilley et al. 2005). High rainfall over short period during crop growth or low rainfall or early withdrawal of monsoon rains may bring flood or prolonged dry spell causing substantial reduction to crop yields (Lobell et al. 2011). Many times, both flood and drought may occur in the same season at different crop growth stages. In the coming years rainfed shallow lowland areas will face heavy precipitation during early crop growth stage leading to flood, and then dry period leading to drought at terminal stages. Variability in the pattern, intensity and frequency of rainfall with the changing climate are several of the factors leading to unpredictable occurrence of drought and flood conditions. These adverse conditions are causing crop failures, volatility in economic growth and making it harder for the small and marginal farmers to move up from the persistent poverty (Mottaleb et al. 2015).
In Nepal, about 50% and in India more than 33% of the total cropland is dedicated to cultivation of rice (Pandey and Bhandari 2007;Gumma et al. 2011). Large area in the rainfed rice-growing ecologies India and Nepal are vulnerable to submergence and drought (Dar et al. 2014). More than 7.3 and 0.27 million hectares of rainfed lowland rice ecologies in India and Nepal respectively, are affected by drought stress (Pandey and Bhandari 2008). Over 5 million hectares of rice land in India is prone to submergence, leading to a paddy loss of 4 million tons per year which is otherwise enough to provide food to 30 million people (Mottaleb et al. 2015).
The rice varieties such as Swarna, Samba Mahsuri, IR64 and MTU1010 are popular among the farmers in India because of their high yield, preferred grain quality traits and higher market value. However, most of these rice varieties are extremely sensitive to drought and submergence, leading to high yield losses every year in regions of their cultivation. The traditional varieties cultivated before the development of semi-dwarf green revolution varieties are less sensitive to drought and flood but poor in yield and grain quality. Introgression of drought and flood tolerance into existing popular rice varieties has been an effective approach to cope with the effects of drought and submergence and reduce yield losses under drought and flood.
In past few decades, efforts have been devoted at the International Rice Research Institute (IRRI) in identifying major genes/QTLs , developing selection strategies (Kumar et al. 2018) and understanding the genetics of grain yield under drought in rice . Major effect drought QTLs explaining large proportion of the phenotypic variance for grain yield such as qDTY 1.1 (Vikram et al. 2011;Ghimire et al. 2012;Sandhu et al. 2014), qDTY 2.1 Sandhu et al. 2014) and qDTY 3.1 Venuprasad et al. 2009) have been identified. These QTLs in synergistic combinations of two to three together can provide grain yield advantage of 0.8 to 1.0 t ha − 1 under reproductive stage drought stress .
A major effect QTL Sub1 from a landrace FR13A explaining phenotypic variation of 69% (Xu and Mackill 1996;Septiningsih et al. 2015) providing tolerance to two weeks of complete submergence has also been identified. The submergence tolerance of many mega varieties such as Swarna (Neeraja et al. 2007), Ciherang Toledo et al. 2015) and PSB Rc18  were improved using marker-assisted backcross breeding of Sub1.
The rice varieties combining drought and submergence can provide yield insurance to farmers in regions exposed to occurrence of drought or submergence or both in the rainfed ecosystems. In this study, the strategy of marker assisted backcross selection in the early stages combined with phenotypic selection at later stage of development has been followed. The present study was conducted with the aim to develop the drought-submergence tolerant near-isogenic lines (NILs) in the background of Swarna combining high yield under non-stress, drought and submergence conditions with preferred grain quality traits. The objectives of the present study include (1) to introgress qDTY 1.1 , qDTY 2.1 and qDTY 3.1 with Sub1 in the background of Swarna (2) to study the interactions between the drought QTLs and Sub1 in terms of performance under drought or submergence when introgressed together (3) to study the performance of breeding lines with different combinations of DTY QTLs and Sub1 on grain yield under reproductive stage drought stress (RS), submergence stress (Sub) and non-stress conditions (NS).

Single trial analysis
The developed NILs were screened at IRRI and in national trials in India (Hyderabad, Sabour, Faizabad, Madhepura, Dhangain, Patna, Varanasi, Tripura, Cuttack and Raipur) and Nepal (Hardinath and Nepalgunj). A total of 60 multi-locations experiments were conducted (Additional file 1: Table S1). The mean grain yield ranged from 2037 to 9848 kg ha − 1 under NS (non-stress), 175 to 6739 kg ha − 1 under RS (reproductive stage drought stress) and 1278 to 8068 kg ha − 1 under submergence conditions (Additional file 1:  Table S1).
The NILs with days to 50% flowering less than that of Swarna and Swarna-Sub1 under NS and RS were identified (data not shown). Most of the NILs showed similar plant height as that of Swarna and Swarna-Sub1 under NS and less under RS (data not shown).

QTL × environment
The variance component for QTL × environment interaction was significant at all stress levels; non-stress, moderate drought stress and severe drought stress. A significant environment variance component was observed in the case of non-stress experiments. The QTL main effect was significant only under severe drought stress (Table 1). Fig. 1 a The morphological differences between Swarna and introgressed NILs; b field view of introgressed NILs under reproductive stage drought stress condition; submergence tolerance of introgressed NILs compared to Swarna (c) one day after draining; d six day after draining; e recovery of introgressed NILs compared to Swarna one month after draining. S: Swarna, A: introgressed NIL with qDTY 2.1 + qDTY 3.1 + Sub1 B: introgressed NIL with qDTY 1.1 + qDTY 2.1 + qDTY 3.1 + Sub1 The selected NILs showed significant yield advantage over recipient parent Swarna under reproductive stage drought indicating capture of positive interaction between QTL × QTL or QTL × genetic background utilizing strategy combining genotypic and phenotypic selection for grain yield under NS and RS.

Performance of Swarna NILs with different QTL combinations across different locations
The tested NILs across all location and trials (Additional file 1: Table S1) were pooled together and categorised based on QTL combinations into three different stress levels; non-stress, moderate drought stress and severe drought stress. The NILs with qDTY 1.1 + qDTY 2.1 + qDTY 3.1 + Sub1, qDTY 1.1 + qDTY 3.1 + Sub1 and qDTY 3.1 + Sub1 had shown consistent grain yield advantage across different locations and seasons ( Table 2).

Discussion
Farmers in drought and submergence prone areas are mainly cultivating the abiotic stress susceptible rice varieties primarily due to their good grain yield potential and market-driven grain quality traits. The performance of these varieties is generally good during the nondrought-submergence affected years (Dar et al. 2018).
However, during the natural calamities, large reduction in grain yield is observed in these varieties due to their inability to survive and yield under drought or submergence or both. Some of the rice growing regions are equally vulnerable to both drought and submergence. There is a need to look at the problem of farmers in a holistic way to improve the resilience of farmer's livelihood so that food and other basic needs can be met on a sustainable basis. Development of dual flood and drought tolerant rice varieties lead to the overall increase in the farm output, farmer's income and will improve the livelihood systems of rice farming communities.
The introduction of marker-assisted breeding in agriculture has provided new opportunities towards the introgression of identified trait associated genes/ QTLs in several popular rice varieties (Dixit et al. 2017;Shamsudin et al. 2016). The grain yield advantage over Swarna and Swarna-Sub1 under RS drought stress and submergence, of those NILs possessing earlier identified major and consistent-effect QTLs, qDTY 1.1 (Vikram et al. 2011;Ghimire et al. 2012;Sandhu et al. 2014), qDTY 2.1 Sandhu et al. 2014), qDTY 3.1 Venuprasad et al. 2009) and Sub1 (Neeraja et al. 2007) in the current marker-assisted backcrossing breeding programs indicates the suitability of these loci in improving drought-submergence tolerance in the Swarna background. In addition, the selected NILs showed significant grain yield advantage under NS as the selection for grain yield under both NS and RS was made across generation advancement. IR 96322-34-223-B-1-1-1 showed grain yield advantage of 2 to 54% under NS (Table 3) and 1 to 17% under moderate to severe drought stress (Table 4) over the combined    (Table 3) and 8 to 24% under moderate to severe drought stress (Table 4) over the combined mean across seasons and locations. Similarly, IR 94391-131-358-19-B-1-1-1 showed grain yield improvement of 3 to 48% under NS (Table 3) and 4 to 38% under moderate to severe drought stress (Table 4) Sandhu and Kumar 2017) as well as in the current study validates the success of QTL introgression in improving grain yield and tolerance to multiple stresses. The developed NILs were previously shown to perform well across different severity/intensity of drought and in regions with different conditions/ soil types . Some of the selected NILs showed early days to flowering than Swarna under NS (data not shown) and this may have QTL × environment interaction is a phenomenon in which QTL effects may significantly differ across environments. In the present study, the significant QTL × environment interaction for all the stress levels and the release of varieties with different combinations of QTLs in different ecosystems indicating the role of environment in influencing the effect of QTLs. The expression of introgressed genomic loci affecting grain yield and yield contributing components is influenced by the environment and the same loci may have differential effects in different ecosystems, signifying strong QTL × environment interaction (Xing et al. 2002). In the present study, the differential performance of NILs with same QTL combination in same or different ecosystem could be due to unknown QTL x genetic background interactions. There is urgent need to identify such epistatic gene interactions which complicates the genotype phenotype relationship of complex trait (Carlborg and Haley 2004) such as drought and submergence.
The phenotypic screening of NILs under different stresses; non-stress, drought stress and submergence conditions revealed huge variation among different combinations of QTLs. The selection of NILs outperforming under different intensities of multiple stresses allows the selection of lines with desirable characteristics. The selected lines with significantly high grain yield advantage over Swarna under both NS as well as RS captured positive interaction due to both genotypic and phenotypic selection practised in the present study, a strategy suggested to be followed in marker assisted breeding for abiotic stress tolerance till all such unknown interactions are identified. Revealing of such epistatic interactions, between introgressed QTL and genetic background sheds light in understanding the differential phenotypic expression of introgressed/pyramided lines for quantitative traits that can be useful in future marker-assisted selection programs.
In our study, across ecosystems in India and Nepal, Swarna and Swana-Sub1 showed similar performance under drought at majority of the locations as against Swarna-Sub1 reported to show higher performance over Swarna under drought (Fukao et al. 2011).
The release of marker-assisted breeding product for drought earlier in IR64 background ) and now in the present study in Swarna-Sub1 background are successful examples that should encourage breeders to use QTLs in the breeding programs targeting grain yield improvement under abiotic stresses. This is one of the first studies in rice developing stable and high-yielding varieties combining both drought and submergence tolerance in a high yielding variety background through marker-assisted backcrossing breeding successfully released as varieties for cultivation by farmers. In the present study, we observed improved performance of selected NILs over Swarna and Swarna-Sub1 under drought and submergence but we are yet to evaluate the lines under varying incidences of subsequent submergence and drought in the same season and observe plants adaptation. Even after affected by submergence under which duration of varieties increases by 10-12 days, the developed NILs showed an average of 10 days earlier flowering than Swarna-Sub1. This will allow the timely planting of second season crop hence enhancing sustainable productivity. The yield improvement across three different backgrounds, TDK-Sub1 (Dixit et al. 2017), IR64-Sub1 (data not published) and Swarna-Sub1 (present study) possessing both QTLs combining drought (DTYs) and submergence (Sub1) clearly indicates that the drought and submergence tolerance can be efficiently combined even though both have different molecular and physiological regulatory mechanisms. QTLs on grain yield under drought has been reported to effect water-uptake, stomatal conductance, canopy temperature, transpiration and root growth at depth (Henry et al. 2014(Henry et al. , 2015 whereas the physiological evidence for the submergence tolerance points towards the proper balance between the production and consumption of plant assimilates Kretzschmar et al. 2015), fast coleoptile elongation,

Conclusions
Over last one-decade marker assisted breeding for abiotic stress has moved from just improving yield under drought or submergence to combining high yield potential with good yield under multiple stresses. The nine selected promising NILs with different QTL combination showed an average grain yield advantage of 0.2 to 1.7 t ha − 1 under RS and 0.1 to 1.0 t ha − 1 under submergence conditions with no yield penalty under NS. The three drought-submergence tolerant varieties CR dhan 801, Bahuguni dhan-1 and Bahuguni dhan-2 have been released for cultivation in India and Nepal. The marker-assisted derived drought and submergence tolerant rice varieties will help to reduce the yield losses associated with farming in drought-flood prone rainfed lowland areas, provide farmers with insurance of good yield and shall encourage markerbreeding programs developing better varieties tolerant to multiple abiotic and biotic stresses.

Description on experiments and agronomic management
The breeding lines were developed and screened under lowland transplanted control (non-stress; NS); lowland reproductive-stage drought stress conditions (RS) and seedling stage submergence condition (Sub). For evaluation of the introgressed lines and to identify the superior lines, a total of 60 experiments were conducted in Philippines (IRRI), Nepal (Hardinath, Nepalgunj) and India (Hyderabad, Sabour, Faizabad, Madhepura, Dhangain, Patna, Varanasi, Tripura, Cuttack, Hazaribagh and Raipur) from 2014DS to 2016WS. In all experiments each plot was 1 to 4 or more (advanced national trials) rows of 3 to 5 m plot length, with 0.20 m row-to-row spacing and 0.15-m plant to plant spacing. Nursery bed was raised and 21 to 25-days old seedlings were transplanted. The agricultural management practices were followed as Vikram et al. (2011).

Screening for reproductive stage drought stress
The non-stress experiments were conducted under irrigated, flooded, puddled, transplanted, and anaerobic conditions with no drought and submergence stress. The reproductive-stage drought stress experiments referred to the experiments maintained as described by Sandhu et al. (2014). Depending on the site's ability to measure data, tensiometers (at 30 cm depth), water pipes (1.1 m length) were installed and rainfall data was recorded across different years (Additional file 1: Figure S1). These measurements were collected from 50 to 100 days after seeding (DAS), which approximately represents the reproductive stage of the NILs evaluated under reproductive stage drought stress. For the reproductive stage drought stress, the stress was initiated at 50-52 days after seeding, after which the drought stress treatment was maintained depending on the rainfall. When the tensiometers reading ranged from − 50 to − 70 kPa, and the water table in PVC dropped to 100 cm from the soil surface and wilting and drying of leaves were observed (data not shown), the plots were re-irrigated. This cyclic screening at reproductive stage allows the precise screening of breeding lines with wide range of growth duration (Lafitte et al. 2004). Screening for submergence tolerance The screening for submergence tolerance was carried out in Nepal (Hardinath, Nepalgunj) and India (Faizabad, Madhepura, Dhangain, Patna). The protocol for the screening for submergence tolerance was as described in Dixit et al. (2017). Selected NILs were planted in the nursery beds along with Swarna, Swarna-Sub1 and susceptible check IR42. The field was submerged for about two weeks from 14 days after seeding (DAS) to 27 days after seeding and then field was drained. The final recovery was recorded seven days after draining (Dixit et al. 2017). The tolerance to submergence was recorded on 1-9 scale of increasing order of susceptibility based on the standard evaluation system for submergence tolerance (IRRI 2002).
In dry season, the submergence screening was conducted in a concrete tank facility. The selected lines were seeded in the seeding trays and at 14 DAS the trays were submerged along with IR42. The concrete tanks were drained on 30 DAS depending upon the survival of IR42, the susceptible check. The numbers of seedlings survived per line were recorded before submergence and at 2, 7, 14, and 21 days after draining the tanks, and percentage survival was calculated (Dixit et al. 2017). Survival percentage = (number of seedling survived after submergence/total number of seedlings planted)*100.

Phenotypic evaluation
Key traits such as days to 50% flowering (DTF, days), plant height (PHT, cm), grain yield (GY, kg ha − 1 ) were measured and submergence was scored based on 1-9 scale. Days to 50% flowering referred to the day when more than 50% of the plants in plot showed panicle exertion. Plant height from root-shoot base to the highest panicle was recorded from three plants per plot and averaged. Grain yield data was recorded per plot and normalized for moisture content to 14% before final yield computation in kg ha − 1 . Plant and panicle selections were made to get grain type similar to Swarna.

Genotyping
Genotyping work was carried out at GSL (Genotyping Service Laboratory); IRRI Genomic DNA was extracted from leaves of 21 days old seedling using modified CTAB method (Murray and Thompson 1980). The polymorphic markers linked to qDTY 1.1 (RM315, RM11943, RM431, RM12023, RM12091, RM12146 and RM12233; Vikram et al. 2011), qDTY 2.1 (RM5791, RM327, RM521, RM3549, RM324, and RM6374, RM424; Venuprasad et al. 2009) and qDTY 3.1 (RM15791, RM416, RM16030, and RM520; Venuprasad et al. 2009) and Sub1 (ART5; Septiningsih et al. 2009) was used for foreground selection in Swarna background. A total of 156 polymorphic markers out of a total of 600 were used for the background study. PCR (polymerase chain reaction) amplification was carried out to check and confirm the introgressed loci and the amplicon size was checked on non-denaturing 6% or 8% PAGE (polyacrylamide gel electrophoresis) depending on size of amplicon. SYBR SafeTM was used to stain the gel, viewed after 20 min, and allelic profile was recorded. Stepwise and precise selection involving both phenotyping and genotyping strategies were used to select and advance the plants with desired introgressed loci in every generation. The background genotyping of selected Swarna-Sub1 NILs was also done at GSL-IRRI using 6 K SNP Infinium chip.
In both the years and at all locations in both stress and normal situations, the experiment was laid out in three replications in the randomized complete block design (RCBD), following the spacing of 20 × 15 cm in a plot size of 15 m 2 . Sowings under submergence were taken up in the last week of June and plantings in the last week of July to first week of August. Under normal irrigated conditions, sowings were taken up in the last week of June to first week of July and plantings in the last week of July to first week of August. In case of drought stress conditions, sowings were taken up from last week of June to second week of July and plantings from last week of July to second week of August. Fertilizers were applied as N:P 2 O 5 :K 2 O (90:30: 30) kg ha − 1 in control trials. In drought trials, fertilizer dose of 75:30:30 N:P 2 O 5 :K 2 O was applied. Observations were recorded for days to flowering, days to maturing, plant height (cm), biomass yield kg ha − 1 , grain yield kg ha − 1 , and harvest index. Under drought stress conditions, data was recorded on number of rain free days during crop growth season and stage of crop under drought stress was noted along with the severity of the stress. Protective irrigations were given at critical crop growth stages depending on the severity of the drought imposed as observed from the performance of susceptible checks and the recurrent parents. Under submergence stress, water stagnation of 30-80 cm was maintained.
A total of 46 Swarna-Sub1 + drought QTLs introgressed NILs were evaluated in three environments namely as control, reproductive drought stage stress and submergence at National Rice Research Program (NRRP, Hardinath, Nepal) during 2014 and 2015. The seeding date was June 19 in 2014 and June 18 in 2015. The twenty-one days old seedlings were transplanted in each trial. Two selected breeding lines, IR 96321-1447-651-B-1-1-2 and IR 94391-131-358-19-B-1-1-1 were evaluated in the National Coordinated Varietal Trials (NCVT) at Regional Agricultural Research Station, Tarahara, Parwanipur, Nepalgunj, and Lumle, National Maize Research Program, Rampur, National Wheat Research Program, Bhairahawa, under rainfed condition in 2015WS and 2016WS. All the experiments at NRRP Hardinath were laid out in alpha lattice with three replications where as in NCVT, the experiments were laid in RCBD design with two replications. The spacing was 20 cm between rows and 20 cm between plants. The plot size was 5 × 2 m 2 . Fertilizers were applied as N:P 2 O 5 :K 2 O (90:30: 30) kg ha − 1 . In drought trials, fertilizer dose of 75:30:30 N:P 2 O 5 :K 2 O was applied. The submergence trial was subjected to 15 days complete submerge and then field was de-submerged. Seeding of drought stress trail was delayed by 25 days from normal planting so that rainy season terminated with the onset of reproductive stage of the crop. Trench was constructed around trial having the drought experiments to prevent seepage of water from other fields as well as for efficient drainage for drought imposition. In drought trials, tensiometer was installed to monitor soil moisture during the crop in the drought field. Pizzometer was also installed to monitor water table from the drought field. The observations on heading days, plant height (cm), and grain yield were recorded.

Statistical analysis
Step 1: Single-trial analysis Experimental designs across trials varied. For all the experiments with alpha lattice design, the effects of replications and blocks within replication were considered as random and lines as fixed. The model used for alpha lattice (AL) design was: For the Randomized Complete Block Design (RCBD) was: For the augmented RCBD, the model used was: where μ represents overall mean, g i represents the effect of the i th genotype, r j represents the effect of the j th replicate, b lj represents the effect of the l th block within the j th replicate, b l represents the effect of the l th block and e ijk is the error.
Step 2: Mean comparison of qDTY and Sub1 combinations The performance of the breeding lines nested within the QTL class in the block within the replicate is modeled as follows: where μ is the population mean, r k is the effect of the k th replicate, b(r) kl + q i is the effect of the l th block within the k th replicate, q i is the effect of the i th QTL, g(q) ij is the effect of the j th genotype nested within the i th QTL and e ijkl is the error. The effects of QTL and genotypes within QTL are considered fixed while the replicate and blocks within replicate effects are considered random. ANOVA and F test using SAS v9.2 were used to see whether the QTL classes differed significantly from each other.
Step 3: Q × E interaction model based on genotype means Based on the trial mean grain yield, each experiment was re-classified for the observed drought stress intensity based on the yield reduction compared to non-stress (control) as per Kumar et al. (2009). The trials were classified as non-stress (control); moderate stress (30% to 65% yield reduction); severe stress (greater than 65%); and over-stressed (greater than 85%). Experiments which were classified as being overstressed were excluded from the analysis due to poor expression of their genetic variability.
The linear model used to study the Q × E interactions for the breeding lines tested across environment was: y kmj ¼ μ þ q k þ g q ð Þ kj þ l m þ ql km þ g ql ð Þ kmj þ e kmj Where y kmj is the yield of the j th breeding line nested in the k th QTL class in the m th environment, μ is the overall mean, q k is the effect of the k th QTL class, g(q) kj is the effect of the j th line nested in the k th QTL class, l m