Metabolite and transcript markers for the prediction of potato drought tolerance

Abstract

Potato (Solanum tuberosum L.) is one of the most important food crops worldwide. Current potato varieties are highly susceptible to drought stress. In view of global climate change, selection of cultivars with improved drought tolerance and high yield potential is of paramount importance. Drought tolerance breeding of potato is currently based on direct selection according to yield and phenotypic traits and requires multiple trials under drought conditions. Marker-assisted selection (MAS) is cheaper, faster and reduces classification errors caused by noncontrolled environmental effects. We analysed 31 potato cultivars grown under optimal and reduced water supply in six independent field trials. Drought tolerance was determined as tuber starch yield. Leaf samples from young plants were screened for preselected transcript and nontargeted metabolite abundance using qRT-PCR and GC-MS profiling, respectively. Transcript marker candidates were selected from a published RNA-Seq data set. A Random Forest machine learning approach extracted metabolite and transcript markers for drought tolerance prediction with low error rates of 6% and 9%, respectively. Moreover, by combining transcript and metabolite markers, the prediction error was reduced to 4.3%. Feature selection from Random Forest models allowed model minimization, yielding a minimal combination of only 20 metabolite and transcript markers that were successfully tested for their reproducibility in 16 independent agronomic field trials. We demonstrate that a minimum combination of transcript and metabolite markers sampled at early cultivation stages predicts potato yield stability under drought largely independent of seasonal and regional agronomic conditions.

Keywords: drought tolerance; machine learning; metabolite markers; potato (Solanum tuberosum); prediction models; transcript markers.

Figure 1

Drought tolerance of 31 potato cultivars (Table S1) based on six field experiments (F1–F5 and F7; Table S2). Drought tolerance was calculated as deviation of relative starch yield from the experimental median (DRYM). DRYM values represent mean values across experiments, and error bars represent the SE of the means. Zero indicates average tolerance, negative values indicate sensitivity, and positive values indicate tolerance.

Figure 2

Expression plots for the selection of reference genes. (a) Relation between log₂ FPKM mean and log₂ FPKM variance measured by RNA‐Seq. Vertical lines indicate the expression range from 5 to 50 FPKM by an interval of 5. Selected candidates as reference genes are highlighted in red. (b) Expression of 15 candidate genes measured as C _t value by qRT‐PCR. The final selection of four reference genes is indicated in grey.

Figure 3

Results of qRT‐PCR using 88 selected marker candidates for drought tolerance. Correlation between gene expression measured by qRT‐PCR (log₂($2^{-\mathrm{\Delta }{C}_{\mathrm{t}}}$)) and RNA‐Seq (log₂ FPKM) from Sprenger et al. (2016).

Figure 4

PCA scores plots of metabolite (a) and transcript (b) data of samples from field experiments and agronomic trials. PCA results indicating the difference between well‐watered control (blue) and drought‐stressed plants (red) as well as 2 years of agronomic trials (2011: green, 2012: orange) are shown for PC1 and PC2.

Figure 5

Plots illustrating the metabolite marker selection. (a) Plot of out‐of‐bag (OOB) error rate and its standard deviation (dashed lines) of the Random Forest model in relation to number of metabolite markers (predictors). The model was based on field training data. The least important predictors were eliminated successively from the model resulting in a set of 24 predictors (red diamond) according to the ‘1 SE rule’. (b) Importance of the selected 24 metabolite markers measured as mean decrease in Gini index for Random Forest models of field trial data.

Figure 6

Plot of out‐of‐bag (OOB) error rate and its standard deviation (dashed lines) of the Random Forest model in relation to number of transcript markers (a). Equivalent plot of OOB error rate of the Random Forest model for combination of metabolite and transcript data (b). The models were based on field training data. The least important predictors were eliminated successively from the model resulting in a set of 14 transcripts (a) and 27 transcripts/metabolites (b), respectively (indicated by red diamond).