Unraveling the genetic basis of drought tolerance in potato CURRENT STATUS: POSTED

Background: Potato is highly valued as a food security crop but is sensitive to drought stress. A relatively small reduction of the optimum amount of water can already result in a significant reduction in tuber yield. Therefore, unravelling the genetic basis of drought tolerance is important in order to enhance tolerance to drought in potato. We have evaluated a diploid potato backcross mapping population for drought tolerance under field conditions. Water application was withheld at the stage of tuber initiation. At the end of the drought stress period, we collected data on a number of morphological and physiological traits. Results: The potato population showed significant reduction in several growth traits measured, as well as in tuber yield. We applied quantitative trait locus (QTL) analysis to find genomic regions underlying these drought tolerance traits. We found 60 QTL for several of the traits measured under well-watered and drought stress conditions. From the total number of QTLs identified, 21 were detected under water stress conditions while 39 QTLs were identified under well-watered conditions. Drought specific QTLs for tuber fresh yield was detected on chromosome 7, 9 and 12. Conclusions: There is genetic variation in potato for drought tolerance and QTLs discovered for drought tolerance traits in this study can aid in improving potato for drought tolerance.

drastic effect on tuber yield and makes recovery even more difficult [8]. Many authors have studied drought responses in potato in order to understand the underlying mechanisms of drought stress tolerance [9]. The effect of water scarcity in potato ranges from reduced photosynthetic rate to induced canopy senescence with negative consequences on tuber yield and quality [10,11].
The availability of genetic variation in wild potato germplasm allows for drought stress tolerance improvements [12,13]. Such landraces are better adapted to harsh environments including water scarce conditions [12], and genes from this germplasm could help improving cultivated potatoes that are considered susceptible to drought stress [14]. Drought tolerance is a quantitative trait controlled by a number of genes and/or gene families. The mechanisms underlying drought tolerance in cultivated potato can be studied using progenies derived from interspecific crosses between drought tolerant and drought sensitive potato genotypes.
However, there are only few studies describing the dissection of the genetic basis of drought tolerance in potato [13,15,16]. Drought tolerance studies under in-vitro and green house conditions have led to the discovery of QTL for drought tolerance traits in diploid mapping populations, and some of the QTL were drought specific [13,15]. In the study of [16] QTL were detected for drought tolerance traits both under green house and field conditions with a potential to contribute to potato improvement. respectively) in response to water stress, showing the significant impact of water shortage on tuber yield. Similarly, biomass production was reduced by half in response to drought stress. Plant height was reduced by 26.1 and 22.6% in 2010 and 2011 experimental years, respectively. The lowest relative reduction in both experimental years due to drought stress was observed for stem numbers, indicating drought stress has little effect on stem number once stems are formed.
In two successive field experiments, analysis of variance revealed highly significant variation for almost all traits measured under drought stress and control conditions ( Table 1). Highly significant variation (P<0.001) among genotypes was observed for the above ground growth traits stem number, plant height, shoot fresh and dry weight under drought stress and well-watered conditions. Statistically significant variation was observed for below-ground traits root fresh and dry weight, and root length for both treatment conditions. Total fresh and dry biomass produced under drought stress conditions showed highly significant genotypic differences. Genotypes under water-limited conditions were significantly different for tuber number and tuber fresh and dry weight. The CxE genotypes were significantly different for harvest index based on dry weight only under well-watered conditions. Genotypes also showed significant variation for chlorophyll content measured at 19, 29 and 34 DAS, under both treatment conditions. Table 1. Mean values of traits measured at harvest both under drought stress and well-watered conditions in two different years with their respective significance value, relative reduction (RR), and broad sense heritability (H2). conditions. Root to shoot ratio was significantly correlated to harvest index under control conditions, but not anymore under water-limiting conditions, also indicating that the investment in roots did not result in higher tuber yield. The differential response of the root to shoot ratio was illustrated by the relatively poor correlation between control and drought for this trait (0.29). Harvest index had the lowest correlation between drought and control treatment (0.12), indicating a strong treatment by genotype effect. Another trait with relatively poor correlation between drought and control is tuber number (0.31), which was lower than fresh and dry tuber weight (0.71 and 0.56, respectively), which may indicate a specific effect of drought on tuberization while tuber bulking is much less affected.

QTL mapping
A QTL analysis was done to unravel the genetic basis of the phenotypic variation observed in the CxE population under drought stress and well-watered conditions. In two successive field experiments, a total of 60 QTLs were detected for agronomical, morphological and physiological traits ( Table 2; contain QTL under drought stress and well-watered conditions in a greenhouse experiment [13], and an expression QTL hotspot (40-50 cM) was identified on chromosome 5 for drought stress conditions [17]. and total dry biomass with LOD values of 18. 16 were detected under water stress treatment. For tuber fresh weight, QTLs located on chromosome 5, 7, 9 and 12 were detected for stress condition.
The highest phenotypic variation (28.6%) was explained by a QTL located in the 26.2-44.0 cM interval on chromosome 5. QTLs found on chromosome 7, 9, and 12 were specific to the drought stress condition.
For the well-watered treatment, we detected QTLs on chromosome 5 for stem number, plant height, shoot fresh weight and dry weight, root fresh and dry weight, root length, total fresh and dry biomass, tuber fresh and dry weight, and tuber number. All these QTLs fall in the interval range of 26-47.3 cM.
A QTL for harvest index based on dry weight falls in the interval of 52.4-62.9 cM (Fig.2). This suggests there may be two independent QTL regions located on chromosome 5, in agreement with [13,17].
Twelve significant QTLs were detected on other chromosomes than chromosome 5 for some of the traits measured under well-watered treatment, specifically on chromosome 1, 3, 6, 7, 9, and 11. A QTL on chromosome 1 was identified for root dry weight with LOD value of 4.71 and accounting for 7% of phenotypic variation. On chromosome 3 and 7 QTLs for tuber number were identified that explained 11.8 and 15.3% of variance, respectively. QTLs for root dry weight, tuber number and tuber fresh weight co-localized on chromosome 8. QTLs for plant height were detected on chromosome 6 for both experimental years. Chromosome 9 had a QTL for stem number with LOD value of 4.72 that explained 15.5% of the observed variation. QTLs for shoot fresh weight and total fresh biomass were co-located on chromosome 11, explaining 10.6 and 9.3% of phenotypic variation, respectively.
Several genomic regions responsible for the variation in chlorophyll content were found on chromosome 1, 2, 3, 7, 10, and 11, for different time points and both under drought stress and wellwatered conditions. The QTL detected on chromosome 2 was present across years and treatments.
This QTL falls in the interval from 79.9 to 109.3 cM and the phenotypic variation explained ranged from 14.3-27.7%. In addition, a QTL expressed across treatments and years was identified on chromosome 10, located between 20 and 46 cM. For well-watered conditions, a QTL on chromosome 1 was found expressed across years. QTLs on chromosome 3 and 7 were identified for chlorophyll content measured under drought stress conditions, and these explained 13.9 and 9.3% of the phenotypic variation, respectively.

Response to drought stress
In the evaluation of 104 individuals of the diploid potato CxE mapping population for morphological, physiological and agronomical parameters for drought tolerance, we observed a significant effect of water shortage on tuber yield production, in agreement with several other drought tolerance studies [3,13,18,19]. All measured growth traits showed a considerable reduction due to water limitation stress, with largest effects on tuber yield and shoot parameters. [7] indicated that the response to drought in potato can be influenced by the severity of water limitation stress and the developmental stage of potato. Since our drought tolerance evaluation was done by completely with-holding water (severe stress) starting from tuber initiation, our discussion will be in view of this scenario. Moreover, how potato responds to water shortage condition can be influenced by the maturity type [13,20] [21]. This indicates that short photoperiod has accelerated the development process through early onset of senescence.
However, under both short and long day conditions early genotypes were observed to be early and late genotypes were also observed to be late, indicating earliness behavior of the genotypes were similar under different day light conditions. Under short day (Ethiopia) conditions genotypes with longer senescence period were reported to have higher tuber yield [21].
Drought stress occurring at the stage of tuber initiation can significantly reduce photosynthesis, biomass production and tuber yield [10]. The strong positive correlation of above ground biomass with tuber yield and tuber dry weight in this study indicates that shoot biomass is an important indicator for yield both under well-watered and drought conditions. [22] indicated that potato genotypes that can maintain above ground biomass under stress conditions are able to produce higher yields. The amount of shoot biomass produced can be influenced by maturity type, as we have observed significant positive correlations with plant maturity under both stressed and non-stressed conditions.

QTLs
Genetically dissecting drought tolerance in the CxE population has resulted in the discovery of 60 QTLs responsible for controlling agro-morphological and physiological traits under drought stress and well-watered conditions. The QTLs discovered under drought stress conditions can be used to enhance potato for tolerance to drought as we define tolerance the ability of genotypes to give higher yield under drought stress condition. As in the other drought studies [13,16], Unlike the current study, [13] reported only one QTL on chromosome 5 for well-watered conditions.
Although both experiments were done using a similar set of diploid potato genotypes from the same population and similar genetic markers, the difference in the detected QTLs between these two studies could be attributed to the different experimental conditions (field and greenhouse). Besides a genetic component of a trait controlling its expression, environmental factors do have great influence on the expression of quantitative traits [24]. This confirms that selection for drought tolerance is complex. Furthermore, a major QTL for earliness is located on chromosome 5 [25]. The underlying gene of this major QTL was identified as a CDF gene with a pleiotropic effect on many agromorphological traits of potato [25,26] . The CDF alleles controls tuber initiation in potato, and also affects plant maturity. Other studies have identified QTLs on chromosome 5 for maturity, vigour and tuberization [27][28][29].
The co-localization of QTLs for shoot and root traits with tuber yield on chromosome 5 indicates that the underlying gene(s) have pleiotropic effects (Fig. 2). A QTL for tuber yield was found on chromosome 5 and explained 28.6% of observed phenotypic variation under water stress conditions.
Similarly, a QTL for tuber yield on the same locus was reported under stress conditions in a greenhouse experiment [13]. These findings suggest that chromosome 5 harbors important QTL that can be used in marker assisted breeding to improve potato for drought tolerance. QTLs for root length, root fresh and dry weight co-located on chromosome 5 with plant height, shoot fresh and dry weight. The amount of variation explained by these QTLs ranged from 30.9 to 54.2% However, measuring root traits can be difficult and tedious thus focusing on shoot traits that showed high correlation with root traits as an indirect selection criterion for root traits is very practical. Under stress conditions we found that plant height, shoot fresh and dry weight had showed high correlation with root length and root fresh and dry weight. Thus these traits can be used in enhancing potato for tolerance to drought.
A QTL associated with plant height located on chromosome 5 was found under water stress conditions and it had a significant high correlation with tuber weight and tuber dry weight. Although we cannot establish this QTL as drought specific as it was also detected under well-watered conditions, it still can be useful as a selection criterion for higher yields under water-limiting conditions. Similarly, a QTL for plant height was reported on chromosome 5 under stress conditions (that where imposed at the later developmental stage of potato) in a field experiment with the diploid mapping population [16]. This may suggest a locus located on chromosome 5 controls drought responses imposed both at early and late developmental stages of potato. [31] have shown the correlation of plant height with tuber yield under drought stress conditions suggesting this as a morphological trait that can be used as a good indicator of tolerance to drought in potato. Unlike in our study, [13] reported QTL for plant height under stress conditions on chromosome 2 and 7. However, it is important to account for the environmental difference (temperature, relative humidity, soil compositions, etc.) of the experiment that could result in significant differences of QTL detected for similar traits, as the expression of a quantitative trait is not only determined by its genetic component but also by the environment in which it is growing. In our study, the maximum temperature recorded was 29.9 0 C and 29 0 C for the two successive year experiments (Table 3) while [13] reported 33 0 C and 37.9 0 C for their experiment period. Besides, in our experiment other factors such as day length (which was short) and growing medium (which was the field) where quite different compared to [13] experiments where day length was long and a pot was used as growing medium. It has been reported that shorter day length could restrict plant height growth compared to longer day length under normal field conditions [21].
Therefore, such differences may account for the differences in QTLs expressed under different environmental conditions. Shoot traits showed high correlation with tuber fresh and dry weight under water stress conditions, which may put them as potential drought tolerance indicator traits that can be used in a potato improvement program. Furthermore, measuring shoot traits is easy, direct, and inexpensive. This may suggest that canopy measurement can be used as selection criteria for genotypes under water limited conditions. The first morphological effect when potatoes are exposed to drought stress is shown in the reduction in leaf size that can affect the canopy architecture [18]. [32] has reported the usefulness of canopy to be used as a selection criterion for nitrogen use efficiency. This study has shown the relationship between canopy development and nitrogen use efficiency. The genomic region controlling shoot fresh and dry weight under stress conditions were found on chromosome 5.
Similarly, [13] had reported QTL for shoot fresh weight on chromosome 5 under water stress and recovery conditions. QTLs for harvest index dry, total fresh and dry biomass were also identified on chromosome 5 under drought stress conditions. These findings suggest that the locus on chromosome 5 does not only influence plant maturity but also how genotypes respond to water stress conditions, indicating the effect of maturity on yield under well-watered and drought stress conditions. In our research, late genotypes both under stress and well-watered conditions had higher yield than early maturing ones. This suggests that late maturing genotypes can benefit from longer periods of photosynthesis that would allow production of more assimilates to be partitioned into tubers.
However, under drought stress conditions biomass production is tightly linked to transpiration indicating higher yield is associated with increased transpiration efficiency [33]. Therefore breeding for traits such as root dry mass that increases soil moisture capture for transpiration is important.
The relationships that we have seen between tuber yield and physiological and growth traits under drought stress conditions indicate that there is not one trait that can be used as a drought tolerance indicator; rather it is the aggregated effect of different traits. However, there is a difference in the amount of variation explained by the QTLs linked to these different traits. Moreover, the heritability of a trait is important to consider since response to selection for drought is efficient if traits show correlation to yield and have a high heritability [34,35]. Heritability in the current study ranged from moderate to high for several of the traits measured under drought stress conditions. Heritability estimates provide a basis for predicting the response to selection in drought tolerance improvement programs. The higher the heritability estimate, the better the response to selection. Similar heritability ranges for drought tolerance experiments were reported by [13,16] . Since direct selection for tuber yield under drought stress conditions is difficult in most cases, agro-morphological, and physiological traits that have correlation with tuber yield and higher heritability can be used as drought tolerance indicators for an indirect selection.
Drought specific QTLs were detected for tuber yield on chromosome 7, 9, and 12. Also, on chromosome 8 we have found a QTL for plant height specific to stress conditions. These QTLs were expressed only in one experimental year ( Table 2; Fig. 2). This implies that the expression of the QTLs may be influenced by the difference in environmental conditions (temperature, relative humidity, etc.) between different years ( Table 3). The main difference between the two experimental years was the minimum temperature recorded, 10.8 0 C in year 2010 and 4.5 0 C in 2011. The stability of these QTLs being expressed under water stress conditions might need to be confirmed in a further field trial.
However QTL detected on chromosome 12 for tuber fresh weight under drought stress conditions showed an overlap with QTL detected for onset and inflection point of plant height under normal growing conditions in [23]. Also, QTL identified on chromosome 8 for plant height specific to drought stress condition co-located with QTL identified (single trait QTL analysis) on chromosome 8 for a parameter controlling inflection point of senescence under short photoperiod conditions [23]. The QTL detected on chromosome 7 for tuber yield under water stress conditions co-located with a QTL found for tuber number under well-watered conditions indicating the same locus controls tuber number under both treatment conditions. QTL for tuber number under well-watered conditions was also detected on chromosome 3. Similarly, QTL for tuber number under short photoperiod conditions was reported at the same chromosome in [23].
A QTL for harvest index based on dry weight was detected on chromosome 5 only under well-watered conditions, [16] reported QTLs for harvest index based on fresh and dry weight on chromosome 5 for water stress and well-watered conditions. Their harvest index data from fresh weight was collected from both green house and field trials, while data of harvest index from dry weight was collected from green house only. On chromosome 8 a QTL for root dry weight co-located with tuber number and tuber yield under well-watered conditions. Chromosome 8 was reported to contain QTLs for tuber fresh weight, tuber dry weight, harvest index fresh weight under well-watered conditions of a greenhouse experiment [16]. However, we were not able to compare the co-localization of the QTLs on the same position since different genotypes of the population and markers were used.
Nevertheless, this finding indicates that QTL located on chromosome 8 is expressed under different environments (greenhouse and field) and different mapping population. For two of the experimental years QTL for plant height measured under well-watered conditions was detected on chromosome 6, indicating the stability of the QTL being expressed in different years.
Chlorophyll content (SPAD measurement) was indicated to have a close correlation with leaf photosynthetic capacity [36,37]. Stay green traits are related with maintaining higher chlorophyll content under drought stress conditions and this is associated with higher photosynthetic capacity  (Table 1)  34DAS for both treatment conditions. Chromosome 10 was also reported to contain QTL for chlorophyll content measured on 3 and 8DAS under stress conditions of a greenhouse experiment [13], moreover this QTL was reported to co-locate with an eQTL belonging to the functional classes of global transcription factors group and putative DNA-binding protein [13]. Under well-watered conditions, QTL for chlorophyll content co-located with shoot fresh weight and total fresh biomass on chromosome 11, suggesting a relationship between chlorophyll content and growth traits however the correlation observed between these traits is weak. QTL for chlorophyll content were detected on chromosome 1 under well-watered conditions however it did not co-locate with QTL identified for root dry weight under well-watered conditions.

Conclusions
All these QTLs found under drought stress conditions indicate that there is genetic variation in this population that can be used in the improvement program of potato for drought tolerance.
Furthermore, studies like these give a better insight in drought tolerance traits that can be used to enhance potato for drought tolerance. However, it is important to confirm stability of drought related QTLs in order to be implemented in marker assisted breeding program. Finding stable QTL for drought tolerance traits, that are expressed across environments can be complex since response to drought is

Plant materials
A set of 104 diploid potato mapping population from crosses of two diploid potato clones, USW53373.3 coded C and 77.2102.37 coded E was used. Clone C is a hybrid between S. phureja PI 225696.1 and the S. tuberosum dihaploid USW42. Clone E is a cross between clone C and the S.
vernei -S. tuberosum backcross clone VH 3 4211. The full description of the population can be obtained from [43]. The selected population largely overlaps with the set used by [13].

Experimental design
A split plot design with two blocks was used and the 104 CxE diploid potato genotypes together with Parents C and E were randomized within a block. The water regime was allocated to the main plot and genotypes were assigned to subplots with two replications. In each replication eight plants per genotype were planted and planting was done manually. The distance between and within row were 75 cm and 30 cm, respectively. A split application of Urea at planting and flowering was done at a rate of 165kg/ha and Di-ammononium phosphate (DAP) was applied at planting at a rate of 195kg/ha. All the field cultivation practices were done manually. The field was kept free of weeds by hand weeding whenever necessary. During the growing periods ridging of the field was done four times.

Statistical analysis
All the data collected for both treatments were statistically analyzed using Genstat 15 th edition [45].
For correcting drought tolerance traits for disease effect, the disease incidence scored at 44DAS was used as a covariate in analysis of variance (ANCOVA), since it coincides with the timing of drought trait data collected at the end of stress. Broad sense heritability (H 2 ) was computed as H 2 =, where is genetic variance, is environmental variance, and is number of replications. Relative reduction of traits was calculated as a difference between control and drought which then divided by the control mean and expressed in terms of percentage.

Genetic map construction
The marker data used to construct the genetic map used in this study is described in [13]. The marker data includes Amplified Fragment Length Polymorphism (AFLP), Simple Sequence Repeats (SSR), Cleavage Amplified Polymorphism (CAP) and Single Nucleotide Polymorphism (SNPs). The genetic map order and positions developed by [13]were used to reconstruct a genetic map for the CxE progeny used in this study. Joinmap 4.1 [46] was used to construct an integrated CxE genetic map.
We have used 529 markers in total and markers with LOD value above 5 were included for map construction.

QTL analysis
QTL analysis was done separately for control and drought stress treatments using MapQTL 6 [47].
Interval mapping was done first to identify and locate QTL on the linkage group. A permutation test was performed to determine significant QTL and a genome wide LOD threshold level of P<0.05 was used to declare presence of significant QTLs. Following this, restricted multiple QTL mapping (rMQM) was done and markers for QTLs detected by interval mapping were used as a co-factor. This procedure was continued until a stable list of cofactors was obtained. The integrated maps and QTLs were drawn using MapChart 2.2 [1] .