**4.1. Gene expression profiles from Subtractive Libraries of cDNA and RT-qPCR**

Long-term global climate changes have conducted to an increase in the occurrence of drought episodes in different locations around the globe [57, 58]. This fact concurrently with agriculture expansion into marginal areas have led to increasing environmental instability, a limiting factor for crop yielding with potential negative impact on food stocks worldwide. This problem is especially aggravated by the rapid human population growth and consequent augmented food demand, especially in developing countries. Therefore, drought has been considered one of the main abiotic constraints that affect agriculture [59].

Plant responsiveness to drought stress can be affected by different factors; it mainly depends on the severity of the event, including the extension of the water-deficit period, and if the plant has already been exposed to a previous regime of acclimatization to this condition [60]. Acclimatization to drought results from a series of integrated events that comprehend the perception of the stress by the plant, translation of the signal, the regulation of the expression of specific genes, and the consequent shifts at metabolic level [61].

Drought perception often leads to a reduction in the photosynthetic rates of the plant, affecting its growth, which is directly related to shifts in carbon and nitrogen metabolism [62]. This reduction on the photosynthetic net is a result of a series of coordinated events such as stomatal closure and the reduction on photosynthetic enzymes activity [63, 64]. At cellular level, drought stress results in the accumulation of the chemically reactive molecules containing oxygen termed as ROS (reactive oxygen species), which ultimately can also drive to the oxidative stress of the photosynthetic apparatus [65,66], thus ROS-efficient removal for avoid oxidative stress can be used as a measure for drought stress tolerance in plants [67]. These molecules act inside cells as secondary messengers involved in signaling transduction that leads to specific stress responses [65]. At molecular level, some specific sets of genes can undergo different processes of regulation of their expression (mainly through cycles of induction and repression of expression) determining new protein synthesis profiles, therefore changing their biological functions [61]. Several genes have been both collectively and individually implicated in drought stress response in plants, but the identification of which ones would be more useful for adoption at breeding and transformation approaches aiming the improvement of drought stress tolerance remains a great challenge [68, 69].

Strategies for plant transformation and genetic breeding usually focus on the transfer of a single or a small set of genes that can codify for specific biochemical pathways or for final targets of the signal transduction pathways that usually are controlled by constitutively active promoters [70]. These gene products protect the plant against the damages caused by drought stress and are divided into different classes: osmoprotectors (amino acids, dimethyl-sulfonyl compounds, mannitol, sorbitol, complex carbohydrates); enzymatic and non-enzymatic ROS scavengers; LEA proteins; heat-shock proteins; ion transporters; fatty acid desaturases; aquaporins; signaling components (homologous to histidine kinases, MAP kinases, Ca+2 dependent protein kinases, protein phosphatases, Ca+2 sensors, inositol kinases); transcription factors (EREBP/AP2, bZIP, ABRE, NAC, MYB); and growth regulators (ABA, cytokines, brassinosteroids) [60–71, 72].

At the transcriptional level, expressed sequence tags (EST) sequencing has been widely used to discover and identify genes potentially involved in drought stress response [73, 74]. Therefore, by using a great amount of transcriptome profiling methods, researchers are being able to contrast genotypes with different potential for drought tolerance, thus increasing the already large datasets of candidate genes for using in studies regarding the improvement of drought stress in plants.

Suppressive subtractive hybridization (SSH) method has been successfully used to construct cDNA libraries enriched in transcripts that are differentially expressed in target tissues, developmental stages, and specific treatments in various biological systems [74,75]. The SSH method [76] consists on the hybridization of one cDNA population (*tester* – sample whose genetic profile is of interest, e.g., drought-tolerant genotype), with an excess of cDNA from a control population (*driver* – usually drought-susceptible genotype or well-watered control), followed by the separation of the nonhybridized molecules (*target genes* – the ones of interest) from the hybridized ones (what is common for both samples). In this session, we are aiming to present some of the results obtained by our group during the construction of a SSH library contrasting populations of cDNAs extracted from root tissues of two common bean genotypes, BAT 477 (tester – drought-tolerant) and Carioca 80SH (driver – drought-susceptible), both submitted to a 192 hours of water-deficit regime at the R5 developmental stage [77].

The sequencing of the SSH library consisting of a BAT 477 cDNA population enriched for transcripts exclusively expressed by this drought-tolerant genotype under 192 hours of waterdeficit generated 1,572 valid reads that were grouped into 189 contigs and 931 singletons (total of 1,120 unigenes). Public green plant EST databases (available at the National Center for Biotechnology Information: http://www.ncbi.nlm.nih.gov/) and bioinformatics tools were used for initial trimming, clustering formation, gene annotation. Final functional annotation was achieved using the Gene Ontology Consortium database (http://geneontology.org/) combined to the CS model (*CombinedScheme*) developed by [78] (http:// www.biochem.ucl.ac.uk/~rison/FuncSchemes/) (for further details on adopted bioinformatics tools and analysis specifications, see [77]).

Gene annotation based on homology search using the BLASTX tool and redundant sequences with E-value ≤ e-5 generated putative information on 896 reads: 315 reads displayed similarity with sequences with not yet assigned putative or hypothetical functions, and 259 reads had good quality control but had no similarity with sequences available in public databases. Table 1 lists the most abundant contigs annotated via BLASTX tool and classified under the biological process that they might be involved in the plant. Final functional annotation classification of the 896 reads is summarized in Figure 3. The six main functional classes are described as follows: 1. Cellular Metabolism (Energy, Macro/ Micronutrients); 2. Biological Process (Cell Division, Regulation, Signaling, Cell Death, Signal Transduction, and Nuclear Cycling); 3. Transport of Compounds; 4. Structural Organization (Membrane, Cell Wall, Nucleus, Organ‐ elles, and Nodules); 5. Information Pathways (DNA, RNA, proteins, and transposons); and 6. Stress Response (Biotic and Abiotic Stresses).


factors (EREBP/AP2, bZIP, ABRE, NAC, MYB); and growth regulators (ABA, cytokines,

At the transcriptional level, expressed sequence tags (EST) sequencing has been widely used to discover and identify genes potentially involved in drought stress response [73, 74]. Therefore, by using a great amount of transcriptome profiling methods, researchers are being able to contrast genotypes with different potential for drought tolerance, thus increasing the already large datasets of candidate genes for using in studies regarding the improvement of

Suppressive subtractive hybridization (SSH) method has been successfully used to construct cDNA libraries enriched in transcripts that are differentially expressed in target tissues, developmental stages, and specific treatments in various biological systems [74,75]. The SSH method [76] consists on the hybridization of one cDNA population (*tester* – sample whose genetic profile is of interest, e.g., drought-tolerant genotype), with an excess of cDNA from a control population (*driver* – usually drought-susceptible genotype or well-watered control), followed by the separation of the nonhybridized molecules (*target genes* – the ones of interest) from the hybridized ones (what is common for both samples). In this session, we are aiming to present some of the results obtained by our group during the construction of a SSH library contrasting populations of cDNAs extracted from root tissues of two common bean genotypes, BAT 477 (tester – drought-tolerant) and Carioca 80SH (driver – drought-susceptible), both

submitted to a 192 hours of water-deficit regime at the R5 developmental stage [77].

The sequencing of the SSH library consisting of a BAT 477 cDNA population enriched for transcripts exclusively expressed by this drought-tolerant genotype under 192 hours of waterdeficit generated 1,572 valid reads that were grouped into 189 contigs and 931 singletons (total of 1,120 unigenes). Public green plant EST databases (available at the National Center for Biotechnology Information: http://www.ncbi.nlm.nih.gov/) and bioinformatics tools were used for initial trimming, clustering formation, gene annotation. Final functional annotation was achieved using the Gene Ontology Consortium database (http://geneontology.org/) combined to the CS model (*CombinedScheme*) developed by [78] (http:// www.biochem.ucl.ac.uk/~rison/FuncSchemes/) (for further details on adopted bioinformatics

Gene annotation based on homology search using the BLASTX tool and redundant sequences with E-value ≤ e-5 generated putative information on 896 reads: 315 reads displayed similarity with sequences with not yet assigned putative or hypothetical functions, and 259 reads had good quality control but had no similarity with sequences available in public databases. Table 1 lists the most abundant contigs annotated via BLASTX tool and classified under the biological process that they might be involved in the plant. Final functional annotation classification of the 896 reads is summarized in Figure 3. The six main functional classes are described as follows: 1. Cellular Metabolism (Energy, Macro/ Micronutrients); 2. Biological Process (Cell Division, Regulation, Signaling, Cell Death, Signal Transduction, and Nuclear Cycling); 3. Transport of Compounds; 4. Structural Organization (Membrane, Cell Wall, Nucleus, Organ‐ elles, and Nodules); 5. Information Pathways (DNA, RNA, proteins, and transposons); and 6.

brassinosteroids) [60–71, 72].

130 Abiotic and Biotic Stress in Plants - Recent Advances and Future Perspectives

drought stress in plants.

tools and analysis specifications, see [77]).

Stress Response (Biotic and Abiotic Stresses).


**Table 1.** List of most abundant contigs containing the original ID of SSH library, number of reads assigned, NCBI identification number (GI) of the EST used for gene putative annotation inference, EST description and correspondent species, e-Values. ESTs are organized according to the functional class Biological Process [77].

**Access code in library**

*Unclassified*

**Number**

**of reads GI number Description/ Species e-value**

[Mesembryanthemum crystallinum]

Contig80 4 |255587991| cation:cation antiporter [Ricinus communis] 1e-39

[Ricinus communis]

[Ricinus communis]

[Populus trichocarpa]

[Arachis hypogaea]

[Arabidopsis thaliana]

Contig11 5 |255584772| histone h2a, putative [Ricinus communis] 2e-27 Contig19 3 |57013900| NitaMp027 [Nicotiana tabacum] 6e-33

Contig154 3 |187940303| NAC domain protein [Glycine max] 8e-84 Contig51 4 |20138704| eIF-5A [Manihot esculenta] 7e-40 Contig52 4 |255646048| transferase activity [Glycine max] 2e-58 Contig162 3 |155212489| N3 protein [Glycine max] 1e-47

Contig72 3 |255626205| unknown [Glycine max] 3e-78 Contig87 3 |255639776| unknown [Glycine max] 3e-71 Contig98 3 |255647862| unknown [Glycine max] 8e-55 Contig145 3 |255646578| unknown [Glycine max] 5e-47 Contig6 4 |224101339| predicted protein [Populus trichocarpa] 5e-30 Contig64 4 |255637247| unknown [Glycine max] 2e-38 Contig77 4 |255637264| unknown [Glycine max] 2e-10 Contig82 6 |255629893| unknown [Glycine max] 7e-27

**Table 1.** List of most abundant contigs containing the original ID of SSH library, number of reads assigned, NCBI identification number (GI) of the EST used for gene putative annotation inference, EST description and correspondent

species, e-Values. ESTs are organized according to the functional class Biological Process [77].

Contig64 4 |255637247| calcium ion binding [Glycine max] 2e-38

1e-61

8e-30

8e-31

1e-24

2e-23

1e-59

Contig164 3 |61651606| plastidic phosphate translocator-like protein1

Contig142 3 |255549412| Vesicle-associated membrane protein, putative

Contig2 3 |255552798| ATP binding protein, putative

132 Abiotic and Biotic Stress in Plants - Recent Advances and Future Perspectives

*Structural Organization (Membrane, Cell Wall, Nucleus, Nodulation and Organelle)*

Contig137 3 |146233385| abscisic acid ABA receptor

Contig148 3 |194466205| putative L24 ribosomal protein

Contig83 4 |30682545| ARF3 (ADP-Ribosylation factor 3)

*Information Pathways (Processing of DNA, RNA and proteins/ Transposons)*

**Figure 3.** Final functional annotation classification of the 896 reads with positive putative gene description information (E-value ≤ e-5) obtained after blastx homology search using NCBI EST green plant public database (http:// www.ncbi.nlm.nih.gov/). Each sector contains the relative EST numbers, in parenthesis the representation percentage in relation to the total number of ESTs successfully annotated. (Extracted from [77]).

The most abundant functional class was Cellular Metabolism (218 ESTs), something that was already expected since, as mentioned before, plants that undergo long periods of water deprivation tend to reduce its photosynthetic rates due to shifts in carbon and nitrogen metabolism, therefore needing to adjust its basal metabolic rates in order to keep homeostasis. Such elevated number of ESTs may be related to a more efficient mechanism of metabolic adjustment present in the drought-tolerant genotype BAT 477 that allows these plants to better adapt during the drought period, thus achieving better survival rates. And, 148 reads were grouped at the Response to Stress and some of them may be directly linked to drought stress tolerance: transcription factors (NAC, DREB, ABRE, WKRY, bZIP, MYB), transmembrane transporters like aquaporins, K+ /H+ pumps and Ca+2 transporters, osmoregulators (LEA proteins, dehydrins, proline-rich peptide chains), and proteins associated with protection (heat-shock proteins, chaperones) and degradation (ubiquitins) [77].

A common bias usually associated with the SSH library construction technique combined with the traditional Sanger-based sequencing technique [79] is the possibility of obtaining falsepositives. Recently, the use of SSH library technique combined with new high-throughput NGS-sequencing technologies [74–80, 81] has provided evidence for solving this issue since they are more able to achieve sample saturation. In RNA-Seq technologies, saturation could be reached when an increment in the number of reads does not result in additional true expressed transcripts being detected or in more features called as differentially expressed when two or more conditions are compared [82]. However, the elevated costs usually associ‐ ated with NGS-sequencing technologies make further experiment validation a more attractive option for researchers. The validation experiments consist of taking the same RNA samples initially used for cDNA library construction and re-analyzing them using a complementary technique, usually microarrays (for those species who already have this platforms available) [83,84] or RT-qPCR (quantitative reverse transcription PCR) [85].

**Figure 4.** Relative gene expression profile of 10 ESTs selected from the BAT 477 drought stress-related SSH library [77]. Captions: BAT – BAT 477 drought-tolerant common bean genotype; CAR – Carioca 80 SH drought-susceptible com‐ mon bean genotype; W.D. – 192h of water-deficit treatment; ctr. – control plants. BATWD/BATctr (BAT 477 under stress compared with BAT 477 control plants); CARWD/CARctr (Carioca 80SH under stress compared with Carioca 80SH control plants); BATWD/CARWD (BAT 477 under stress compared with Carioca 80SH under stress – SSH library initial condition); and BATctr/CARctr (BAT 477 control plants compared with Carioca 80SH control plants). (Extracted from [77]).

For the BAT 477 drought stress SSH library, it was selected as a set of 10 ESTs among those with most abundant contigs: *LEA5, Sina, histone h2a, methionine adenosyltransferase, NAC protein,* *N3 protein, EF-hand – calcium binding motif, S-adenosylmethionine decarboxylase, malate dehydro‐ genase-like protein, cation:cation antiporter*. For each of the ESTs, a specific pair of primers for RTqPCR analysis was designed [77] and gene relative expression quantification was obtained for the same tester and driver samples used for the SSH library construction (Figure 4). These results served well for the SSH library validation since all the selected transcripts revealed to be upregulated in BAT 477 plants under drought stress. Besides, for some of the transcripts (*LEA5, NAC protein, N3 protein, Ef-hand – calcium binding motif*, *and S-adenosylmethionine decarboxylase*), although they are expressed in lower concentrations on Carioca 80SH 192h drought-stressed plants, when compared to Carioca 80 SH controls, they undergo an even greater upregulation in relation to BAT 477 (Figure 4). This not only confirms the relevance of these transcripts on drought stress response regulation in common beans but also reveals that the drought-tolerant genotype BAT 477 may already keep a basal level expression of some important drought-related transcripts, thus stress perception by this drought-tolerant geno‐ type may trigger more efficient signaling mechanisms that leads to a more discreet gene expression upregulation allowing the plant not to dislocate resources that otherwise may be saved for keeping homeostasis and therefore secure development and growth during the stress period.

expressed transcripts being detected or in more features called as differentially expressed when two or more conditions are compared [82]. However, the elevated costs usually associ‐ ated with NGS-sequencing technologies make further experiment validation a more attractive option for researchers. The validation experiments consist of taking the same RNA samples initially used for cDNA library construction and re-analyzing them using a complementary technique, usually microarrays (for those species who already have this platforms available)

**Figure 4.** Relative gene expression profile of 10 ESTs selected from the BAT 477 drought stress-related SSH library [77]. Captions: BAT – BAT 477 drought-tolerant common bean genotype; CAR – Carioca 80 SH drought-susceptible com‐ mon bean genotype; W.D. – 192h of water-deficit treatment; ctr. – control plants. BATWD/BATctr (BAT 477 under stress compared with BAT 477 control plants); CARWD/CARctr (Carioca 80SH under stress compared with Carioca 80SH control plants); BATWD/CARWD (BAT 477 under stress compared with Carioca 80SH under stress – SSH library initial condition); and BATctr/CARctr (BAT 477 control plants compared with Carioca 80SH control plants). (Extracted

For the BAT 477 drought stress SSH library, it was selected as a set of 10 ESTs among those with most abundant contigs: *LEA5, Sina, histone h2a, methionine adenosyltransferase, NAC protein,*

from [77]).

[83,84] or RT-qPCR (quantitative reverse transcription PCR) [85].

134 Abiotic and Biotic Stress in Plants - Recent Advances and Future Perspectives

#### **4.2. DREB transcription factors as candidates for drought-tolerance improvement**

Finding candidate genes and investigating their functional role and association with droughttolerance traits and mechanisms have been of prime interest for many crop plants such as common bean. The DREB transcription factors subfamily has been studied in depth as candidate genes for breeding of abiotic stress tolerance. This group comprises a series of genes intermediating the regulation process to cope with abiotic stresses effects such as drought. They were originally described by [86], which identified a *cis*-acting regulatory element, DRE (dehydration responsive element), present in the gene promoter COR78/RD29A and involved in the response to drought, high salinity, and low temperature, further named as DREB (DRE-Binding). These proteins are capable of binding to DRE to activate the expression of genes of the stress signaling pathway. DREB transcription factors are unique to plant species and so far several genes have been described in *Arabidopsis* and other plants [87, 88].

The primary feature of a DREB transcription factor is the presence of a highly conserved protein domain, the EREBP/AP2. It was discovered within *APETALA2*, which plays an important role in flowering and seed development in *Arabidopsis.* Several proteins have been found containing this domain along their amino acid chain, consisting of a repeated motif of approximately 60 amino acids [89–91]. All these proteins are comprised in the larger superfamily EREBP/AP2 divided into three families referred as AP2, ERF, and RAV, based on their sequence similarity and the number of EREBP/AP2 domains [92]. The ERF protein family contains only one EREBP/ AP2 domain and is subdivided into two main subfamilies, CBF/DREB and ERF [91]. The amino acids 14 and 19 of the EREBP/AP2 domain distinguish DREBs (valine and glutamic acid, respectively) from ERF (alanine and aspartic acid, respectively) [91]. In addition, ERF genes are involved primarily in responses to biotic stresses such as pathogenesis while DREB genes have main role in abiotic stresses responses.

DREB genes can be divided into six subgroups (A-1 to A-6). This categorization was based on phylogenetic trees as well as particular features related to their induction. The two most studied groups have been A-1 and A-2. Genes *DREB1/CBF* belong to subgroup A-1 and have been characterized as induced by low temperature in *Arabidopsis* [93], but other studies revealed some inducibility under drought and salinity as well [91, 94]. *DREB2* genes are primarily involved in responses to osmotic stress (dehydration and salinity) [91, 95].

Most of DREB findings have been associated with *Arabidopsis*; however, many studies have been performed with other species as well, revealing several new orthologs and different inducibilities for each one of the six DREB subgroups. Some of these findings have been done with legumes such as *Medicago truncatula* and *Glycine max*, close relatives to com‐ mon bean.

Few studies have been published so far for common bean DREB genes, and they were mostly related to polymorphic sites identification along gene sequences. Ref. [96] categorized two orthologs *DREB2A* and *DREB2B* and identified polymorphisms between some Mesoamerican and Andean genotypes. Further investigation of these genes has been done to identify polymorphism patterns across wild and domesticated common beans. An attempt for phenotypic associations with drought-tolerance traits has been performed as well, but no clear patterns were obtained [18].

The research team of University of São Paulo, Brazil, has been studying DREB genes in depth. A pre-categorization study of the *PvDREB* gene subfamily has been done [97], showing putative DREB representatives for the species. Several genes have been isolated and their expression profiles determined under several abiotic stresses, including drought. One particular gene showed strong induction under many abiotic treatments, such as drought, salinity, and cold [98]. Some genes have been selected for a deeper molecular basis under‐ standing as well as for their functional role in improving drought tolerance as well as other abiotic stresses.

Some other studies have found DREB genes in whole transcriptome profiles, such as in one experiment contrasting the drought-tolerant cultivar Long 22-0579 and the sensitive Naihua, in which a RNA-seq analysis was performed for samples under drought and control condi‐ tions. DREB transcription factors were identified to be differentially expressed and RT-qPCR analyses showed one transcript had the relative number of transcripts increased during the drought period [99]. Moreover, not only drought treatments have been analyzed but also one transcriptome profile has been done for a salt-tolerant bean cultivar named Ispir. It revealed several AP2/EREBP genes differentially expressed when contrasting a saline hydroponic solution with control conditions. Nevertheless, authors have not performed further categori‐ zation to identify which of those genes fitted *PvDREB*-specific characteristics [100].

Much more has to be done with DREB genes in common bean. Isolating and characterizing DREB genes for the species seems to be an important step toward the improvement of beans for abiotic stresses tolerance, especially for drought.

#### **4.3. Phenotyping for drought tolerance in common bean**

DREB genes can be divided into six subgroups (A-1 to A-6). This categorization was based on phylogenetic trees as well as particular features related to their induction. The two most studied groups have been A-1 and A-2. Genes *DREB1/CBF* belong to subgroup A-1 and have been characterized as induced by low temperature in *Arabidopsis* [93], but other studies revealed some inducibility under drought and salinity as well [91, 94]. *DREB2* genes are

Most of DREB findings have been associated with *Arabidopsis*; however, many studies have been performed with other species as well, revealing several new orthologs and different inducibilities for each one of the six DREB subgroups. Some of these findings have been done with legumes such as *Medicago truncatula* and *Glycine max*, close relatives to com‐

Few studies have been published so far for common bean DREB genes, and they were mostly related to polymorphic sites identification along gene sequences. Ref. [96] categorized two orthologs *DREB2A* and *DREB2B* and identified polymorphisms between some Mesoamerican and Andean genotypes. Further investigation of these genes has been done to identify polymorphism patterns across wild and domesticated common beans. An attempt for phenotypic associations with drought-tolerance traits has been performed as well, but no clear

The research team of University of São Paulo, Brazil, has been studying DREB genes in depth. A pre-categorization study of the *PvDREB* gene subfamily has been done [97], showing putative DREB representatives for the species. Several genes have been isolated and their expression profiles determined under several abiotic stresses, including drought. One particular gene showed strong induction under many abiotic treatments, such as drought, salinity, and cold [98]. Some genes have been selected for a deeper molecular basis under‐ standing as well as for their functional role in improving drought tolerance as well as other

Some other studies have found DREB genes in whole transcriptome profiles, such as in one experiment contrasting the drought-tolerant cultivar Long 22-0579 and the sensitive Naihua, in which a RNA-seq analysis was performed for samples under drought and control condi‐ tions. DREB transcription factors were identified to be differentially expressed and RT-qPCR analyses showed one transcript had the relative number of transcripts increased during the drought period [99]. Moreover, not only drought treatments have been analyzed but also one transcriptome profile has been done for a salt-tolerant bean cultivar named Ispir. It revealed several AP2/EREBP genes differentially expressed when contrasting a saline hydroponic solution with control conditions. Nevertheless, authors have not performed further categori‐

zation to identify which of those genes fitted *PvDREB*-specific characteristics [100].

for abiotic stresses tolerance, especially for drought.

Much more has to be done with DREB genes in common bean. Isolating and characterizing DREB genes for the species seems to be an important step toward the improvement of beans

primarily involved in responses to osmotic stress (dehydration and salinity) [91, 95].

136 Abiotic and Biotic Stress in Plants - Recent Advances and Future Perspectives

mon bean.

patterns were obtained [18].

abiotic stresses.

The identification of genomic regions or candidate genes, their functional role, and association with drought tolerance in common bean are fundamental aspects to understand the molecular signatures involved in acquiring such tolerance. However, in that purpose phenotyping methods are essential to effectively proving the effect of those genes on traits of interest. Thereby, it is important establishing and standardizing a phenotyping methodology to compare and select genotypes with different levels of stress tolerance in the studies one might be conducting. Furthermore, bringing data from the lab and greenhouse to the field is a big challenge, but of great importance for successfully applying the knowledge obtained about the genes, genotypes, and phenotypes of interest.

Phenotyping techniques have been developed to differentiating common bean accessions and cultivars for their levels of drought tolerance. Greenhouse trials have been applied to pheno‐ type several shoot and roots traits and a common method employed has been the soil tube screening system assay that has been developed at CIAT [101]. Ref. [102] points out several traits that might be measured through such system, including many photosynthetic traits (photosynthetic efficiency, total chlorophyll content – SPAD, stomatal conductance, transpi‐ ration rates, leaf temperatures, leaf water potential), shoot and root biomass at the time of harvesting, leaf area and root traits (length, diameter, specific root length, and dry weight). Determination of root length might be done by image analysis system (WinRHIZO, Regent Instruments Inc.) [102] or might be manually determined by following root development on a graded plastic transparent tube in which plants were grown, all placed in PVC tubes.

The tube system developed by [101] was used to evaluate the effect of drought stress on root growth and distribution and compare different genotypes. Due to the difficulties of pheno‐ typing roots in the field, this method has been shown to be a good complementary strategy applied in greenhouse conditions [102]. Examples in this sense are the studies of [103, 104] that analyzed the rooting patterns in greenhouse conditions with PVC soil cylinders and photo‐ synthetic and yielding traits in different field areas. A population of recombinant inbred lines (RIL) from the crossing between the deep-rooting genotype BAT 477 and the small red-seeded and drought-susceptible DOR 364 was evaluated in both conditions. The greenhouse experi‐ ment showed that BAT 477 had significant larger root system based on root volume and deeper rooting ability, larger and thicker root, wide root diameter and biomass, under well-watered and progressive drought stress treatments [103].

For experiments conducted at the field, several traits can be evaluated since initial plant growth still harvesting. Ref. [102] made a very elaborated list with many parameters such as plant biomass at mid-pod filling and at harvesting time, seed yield, harvest index (HI), pod harvest index (PHI), drought intensity index (DII), and drought susceptibility index (DSI). The latter is based on the mean yields of a given genotype in drought stress and under no stress [102]. It assumes that one genotype will be more drought tolerant if the yielding is not so much reduced by the stress treatment in comparison to other genotypes. Pod harvest index has also been shown as a good indicator of drought tolerance, as shown by a field study in Ethiopia with the population from the crossing SXB 405 (breeding line) × ICA-Bunsi (white pea bean). Sensitive lines presented significant reduction on PHI while no differences were observed for the most resistant lines [105].

Despite the availability of traits that might be evaluated in field conditions, the environment turns out to be a critical component interfering with results from one site to another. Drought field trials performed with the RIL population of the crossing BAT 477 × DOR 364, previously referred to the greenhouse experiment, showed significant variability across four locations evaluated [104]. A QTL analysis associating the field traits to a previous set of molecular markers disposed in a linkage map [106] showed significant QTL–environment interactions. Therefore, determining if one cultivar is tolerant to drought does not necessarily mean it will respond well to all environments, in a sense that it must be tested in multiple environments to check for its performance.

**Figure 5.** Strategies for phenotyping common beans for drought tolerance. Greenhouse experiments and field trials are complimentary strategies for evaluating several parameters and defining drought-tolerance levels. Recurring to wild beans based on environmental data might also be useful to select for beans adapted to drought episodes. Those evalua‐ tions altogether may be used toward the development of new varieties and the identification of genomic regions asso‐ ciated with the phenotypes evaluated as well as the functional role of candidate genes that are under characterization.

Although greenhouse and field methods have been developed to identify drought-tolerant genotypes and gene markers associated to such parameters, recent efforts have also been focused on the identification of sources of drought tolerance in wild beans spanning the natural area of distribution of *P. vulgaris* [107]. However, reliable estimations of drought tolerance in wild beans are not easy to establish, and attempts toward the development of new methods have been in course. Potential evapotranspiration models coupled with precipitation regimes were used to define a drought index for a series of wild bean accessions. Considering this factor along with the population structure might be a useful tool to analyze the levels of drought tolerance and use these materials for introgression of alleles of interest [107].

All these methods might be useful to carefully understand the phenotypic basis of drought tolerance variation in common bean genotypes. With standardized methods for the traits one might be interested, the accuracy between the association of molecular data and phenotypes might be much higher. It may be applied to QTL and association mapping studies, which link genome-wide molecular markers such as microsatellites, SNP, and gene-specific markers to drought-related traits (103, 104, 106, 108]. On the other hand, standard greenhouse parameters can be used to test transgenic lines for determined candidate genes to verify their performances under imposed drought stress. Figure 5 shows a scheme of how greenhouse, field, and wild environment phenotyping studies might be useful for association and functional genomic studies in common bean.
