SciELO - Scientific Electronic Library Online

vol.21 issue4Genetic variance and covariance components related to intra- and interpopulation recurrent selection in maize (Zea mays L.)A new method to estimate genetic gain in annual crops author indexsubject indexarticles search
Home Pagealphabetic serial listing  

Services on Demand




Related links


Genetics and Molecular Biology

Print version ISSN 1415-4757On-line version ISSN 1678-4685

Genet. Mol. Biol. vol. 21 n. 4 São Paulo Dec. 1998 

Screening cotton genotypes for seedling drought tolerance


Julio C. Viglioni Penna1, Laval M. Verhalen2, M.B. Kirkham3 and Ronald W. McNew4
1 Departamento de Agronomia, Universidade Federal de Uberlândia, Caixa Postal 593, 38400-902 Uberlândia, MG, Brasil. Send correspondence to J.C.V.P.
2 Department of Agronomy, Oklahoma State University, Stillwater, OK 74078, USA.
3 Department of Agronomy, Kansas State University, Manhattan, KS 66506, USA.
4 Agricultural Statistics Lab., University of Arkansas, Fayetteville, AR 72701, USA.




The objectives of this study were to adapt a screening method previously used to assess seedling drought tolerance in cereals for use in cotton (Gossypium hirsutum L.) and to identify tolerant accessions among a wide range of genotypes. Ninety genotypes were screened in seven growth chamber experiments. Fifteen-day-old seedlings were subjected to four 4-day drought cycles, and plant survival was evaluated after each cycle. Three cycles are probably the minimum required in cotton work. Significant differences (at the 0.05 level or lower) among entries were obtained in four of the seven experiments. A "confirmation test" with entries previously evaluated as "tolerant" (high survival) and "susceptible" (low survival) was run. A number of entries duplicated their earlier performance, but others did not, which indicates the need to reevaluate selections. Germplasms considered tolerant included: `IAC-13-1', `IAC-RM4-SM5', `Minas Sertaneja', `Acala 1517E-1' and `4521'. In general, the technique is simple, though time-consuming, with practical value for screening a large number of genotypes. Results from the screening tests generally agreed with field information. The screening procedure is suitable to select tolerant accessions from among a large number of entries in germplasm collections as a preliminary step in breeding for drought tolerance. This research also demonstrated the need to characterize the internal lack of uniformity in growth chambers to allow for adequate designs of experiments.




Most land plants are subjected periodically to a greater or lesser degree of water deficit. Heatherly et al. (1977) stated that the development and yield of many species are repressed severely even by moderate water stress. Drought is a major limiting factor in cotton (Gossypium hirsutum L.) production in the northern region of Minas Gerais, Brazil, where this crop is of great economic and social importance. Cotton is normally not classified as a drought-tolerant crop, and it is not very efficient in water use, as are other plants such as sorghum (Sorghum bicolor (L.) Moench) (Briggs and Shantz, 1914), which is cultivated in areas normally too hot and dry to grow other crops (Poehlman, 1986). Nevertheless, cotton does have mechanisms that make it well adapted to semi-arid regions, such as its deep-penetrating and extensive root systems, leaves and fruits that can be shed when plants are stressed and a flexible fruiting period (Ray et al., 1974). Roark et al. (1973) reported differences in stomatal behavior and distribution among cotton cultivars grown on the high plains of Texas.

Variation in drought resistance among and within species has been reported in the literature. Gomathinayagam et al. (1988) found differences in root, shoot and leaf traits among rice (Oryza sativa L.) germplams subjected to a pot screening test for drought tolerance and pointed out some advantages of the system: growth and physiological parameters were easily studied, trials could be grown in any season, and it was relatively inexpensive. However, fewer genotypes could be studied than in the field, and only early vegetative growth could be assessed. Tischler et al. (1991) evaluated a soil-tray system used in a growth chamber to determine drought tolerance among hybrids of lovegrass (Eragrostis curvula var. conferta Nees x E. curvula (Schrad.) Nees and plant introductions of kleingrass (Panicum coloratum L.)). These plants showed a broad range of tolerance. The system successfully identified drought-tolerant germplasm in lovegrass, and results agreed with field performance of genotypes with known drought tolerance. Erb (1993) described a breeding scheme that would allow blueberry (Vaccinium section Cyanococcus) breeders to efficiently access existing variability for mineral soil adaptation, based on an initial screening for drought tolerance, followed by selection for root development and shoot growth on "mineral" soil. Wright (1964) employed a growth chamber to evaluate six species of range grasses (three species of Eragrostis and three of Panicum). Seedlings were grown in trays and water stressed for up to 24 days. Eragrostis spp. were markedly more drought tolerant than Panicum spp., and species within the genera also differed and were ranked for that trait. The rankings agreed with known range performance. Nour et al. (1978) screened 9-day-old sorghum seedlings in a growth chamber where they were rewatered after seven to eight days of water stress. Mean percent survival was recorded after each of four drought cycles. They found this to be a simple and effective method of screening for drought resistance among unknown genotypes.

According to Hurd (1976), root mass under drought conditions is important in breeding for drought resistance. Cregg (1994) screened 27 families of ponderosa pine (Pinus ponderosa Dougl. Ex Laws), representing nine geographically diverse origins, for drought tolerance based on survival and growth under imposed drought. Seedlings from South Dakota and Nebraska survived longer under drought than those from Montana and New Mexico. The relation between survival and shoot/root ratio suggested an optimum pattern of allocation between them for survival. Nour (1975) grew sorghum plants in pots filled with washed sand for a 3-week period after which wet root weights, lengths, volumes and root/shoot ratios were taken. The more drought-resistant cultivars (as determined in a previous screening test) had the greatest values for all four variables.

The objectives of this study were to adapt a drought screening method (previously used to assess seedling drought tolerance in cereals) for use in cotton and to identify tolerant accessions among a wide range of genotypes.



All tests were conducted in a walk-in growth chamber with automatic temperature and light controls set for a 14-h light period (32oC and 45% relative humidity) and a 10-h dark period (18oC and 90% relative humidity). A mixture of fluorescent and incandescent bulbs was used, giving approximately 150 watts/m2 at 76 cm below the light source, where the seedlings were located. Air was kept in motion by means of six large fans mounted in two groups of three on opposite ends of the chamber and by six small fans located above the light source.

A preliminary trial was conducted to test the growth chamber environment. Ten metal trays (50 by 35 by 9 cm) were filled with a 1:1:1 mixture by volume of soil, vermiculite and peat moss. Cultivars `Westburn M' and `Stoneville 213' were grown in alternate 35-cm long rows spaced 4 cm apart with plants spaced 3 cm apart. Four-day drought cycles separated by an irrigation were applied four times, starting 15 days after planting. Survival counts were made two days after each irrigation to allow stressed plants a chance to recover (data not shown). After the fourth cycle, sections within the chamber were identified as having plants with significantly more severe drought injuries than others.

Based on these results, a randomized complete-block experimental design with four replications was utilized in the subsequent experiments. Also based on those results, 12 plants were left per row (after thinning) and 12 rows were planted per tray. Two plants on each extremity of each row and two rows on each side of each tray were disregarded in an attempt to negate possible border effects. Holes were punched in the trays in all outside bottom edges to permit more complete and even water drainage because outside plants in the trays appeared to suffer less drought stress. Each of the screening experiments consisted of 16 entries (primarily cultivars and/or race stocks). Each block consisted of two trays planted with eight entries per tray; 30 seeds of each were planted in single rows and thinned after emergence to the desired spacing. Seeds were treated with fungicides (a mixture of carboxin and thiram) to avoid seedling diseases. Trays within blocks were rotated daily to reduce possible variation within blocks. Plants were irrigated on alternate days with 1.5 l of water per tray until 15 days after planting, when irrigation was suspended for four days. The number of surviving plants was counted two days after rewatering with 1.25 l of water, and four cycles were used. Seven experiments were conducted, and 90 genotypes were screened (Table I). In the sixth experiment, a group of entries which performed well, together with a group of poor performing entries from previous trials, were selected, and a "confirmation test" was conducted. In the seventh trial, 11 primitive race stocks from the Texas collection were tested together with five selected cultivars (four previously tested and a new entry). The survival data (expressed as percentages) were transformed to arcsines of square roots, due to the wide range of values, as recommended by Steel and Torrie (1980), with a further correction suggested by Bartlett (cited by the same authors) employed for percentages of 100 and zero. Statistical analyses were conducted as split-plots in time on the transformed data.

21n4a23t1.GIF (32459 bytes)




Figure 1 illustrates the response of four entries (the two with the highest percent survival and two of the three with the lowest after the third-fourth drought cycles) in experiment 1, over successive drought cycles. Differences in survival among some entries were obvious after the first cycle, whereas differences among others appeared more dramatically after the third. The pronounced effects in many entries in the third and fourth cycles may indicate a threshold effect, i.e., some stress can be tolerated, but prolonged stress overcomes what tolerance some entries may have.

21n4a23f1.GIF (9024 bytes)

Figure 1 - Differential seedling survival of four selected cotton entries in experiment 1 across four cycles of water stress.


The drastic shifts in performance for some entries from one cycle to the next suggested interactions between entries and cycles, so two types of statistical analyses were conducted. In the first, all cycles of an experiment were analyzed together. As expected, some of these experiments (3, 4, 5 and 6) exhibited significant (P < 0.05) entry by cycle interactions (results of analyses not shown). Based on these results and since the last two drought cycles displayed more severe symptoms across experiments, statistical analyses were made based only on those last two cycles. None of these analyses detected a significant entry by cycle interaction at the 0.05 level. These results imply that the last drought cycle was unnecessary. Omitting it would permit greater efficiency in screening by saving almost a week per experiment.

All comparisons made in the following discussion are significant at the 0.05 or lower probability levels. Mean percent survival data for each genotype over the last two drought cycles were determined for the seven experiments (Tables II and III). Means from those experiments with significant differences among entries were compared according to a protected LSD test. The "confirmation test" (experiment 6, Table III) partially confirmed, partially conflicted with results of the previous experiments. Some entries (e.g., `IAC-13-1', `IAC-RM4-SM5', `Minas Sertaneja' and `Acala 1517-75') appeared resistant as they had previously; others (e.g., Stoneville 213 and `Minas Dona Beja') again appeared susceptible. Yet others (e.g., `AC 307', `Allen 333-61', `Deltapine 61', and `M4') reacted differently from the earlier experiments.

21n4a23t2.GIF (34329 bytes)



21n4a23t3.GIF (19845 bytes)


Little unequivocal information on the field performance of cotton cultivars under drought conditions is available; however, subjective observations of several of the above cultivars over time suggest a consistency between these results and field performance. For example, the cultivars `Paymaster Dwarf', `Stoneville 213' (Table I, Figure 1) and `Minas Dona Beja' generally produce relatively low yields under dryland or drought conditions, although all do very well under high rainfall or irrigation (data reported by McCall et al., 1980 and 1981; Bayles et al., 1983 and 1984). Furthermore, some indication of drought tolerance in the field exists for the cultivar `Minas Sertaneja', for it was developed and recommended for areas subjected to drought spells in the State of Minas Gerais, Brazil (Castro, 1978).

Generally, this technique shows promise for screening a large number of genotypes for drought tolerance, especially if confirmation tests are run to verify earlier performance. Mistakes in classification will be made, but all selections made by breeders are subject to such errors. Confirmation tests should reduce the number of those mistakes. Also, having a tolerant and a susceptible check in common for all experiments would allow one to adjust for differences among experiments. Germplasms considered tolerant included: `IAC-13-1', `IAC-RM4-SM5', `Minas Sertaneja', `Acala 1517E-1' and `4521'.

Since this technique has demonstrated usefulness for selecting seedlings for drought tolerance, the next step in developing drought tolerant cultivars is to correlate seedling and mature plant performance. The selection procedure objetive of this study is suitable to select tolerant accessions from among a large number of entries in germplasm collections as a preliminary step in breeding for drought tolerance.

Of paramount importance before such studies begin is a complete evaluation of the environment within the growth chambers to be used. Some control of the variation present can be attained by grouping like areas into blocks and eliminating plants on the borders of the trays.



The authors gratefully acknowledge the help of D.W. Weibel (deceased), J.E. Quisenberry, B. Roark and G.H. Hartman. J.C.V.P. is recipient of CNPq fellowship and also expresses his gratitude to EPAMIG and Embrapa for their financial support.




Os objetivos desta pesquisa foram adaptar para o algodoeiro (Gossypium hirsutum L.) técnica de seleção previamente usada para avaliar tolerância à seca em plântulas de cereais e identificar acessos tolerantes entre uma gama de germoplasmas. Noventa diferentes genótipos foram avaliados em sete experimentos em câmara de crescimento. Plântulas de 15 dias de idade foram submetidas a ciclos de estresse hídrico de 4 dias e os índices de sobrevivência foram avaliados para cada ciclo. Três ciclos são o mínimo para se trabalhar com algodoeiro. Diferenças significativas (P < 0,05%) entre acessos foram obtidas em quatro dos sete experimentos. Um teste de confirmação foi executado, composto de entradas previamente classificadas como "tolerantes" (alta sobrevivência) e "suscetíveis" (baixa sobrevivência). Um número de entradas duplicou seu desempenho anterior mas outras não, o que indicou a necessidade de reavaliar as seleções. Em geral, a técnica é simples, embora exigente em tempo, e tem valor prático para avaliar grande número de genótipos. Os resultados dos testes avaliatórios geralmente coincidiram com informações subjetivas de campo. Entre outros, os germoplasmas considerados tolerantes foram: `IAC-13-1', `IAC-RM4-SM5', `Minas Sertaneja', `Acala 1517E-1' e `4521'. Esta pesquisa também demonstrou a necessidade de se caracterizar a falta de uniformidade nos ambientes das câmaras de crescimento e assim se escolherem delineamentos experimentais adequados. O procedimento de seleção descrito é adequado para selecionar entradas tolerantes dentre grande número de acessos em coleções de germoplasma e como passo preliminar em programas de melhoramento para tolerância à seca.




Bayles, M.B., Verhalen, L.M., Avis, J.D., Foraker, R.W., Thomas, N.B., Greenhagen, B.E., Baker, J.E. and Cain, L.K. (1983). Cotton variety tests, Oklahoma-1982. Agric. Exper. Sta., OSU. Res. Rep. P-837, pp. 18.         [ Links ]

Bayles, M.B., Verhalen, L.M., Foraker, R.W., Avis, J.D., Greenhagen, B.E., Thomas, N.B., Boman, R.K., Baker, J.E., Cain, L.K. and Ha, S.B. (1984). Cotton variety tests, Oklahoma-1983. Agric. Exper. Sta., OSU. Res. Rep. P-845, pp. 21.         [ Links ]

Briggs, L.J. and Shantz, H.L. (1914). Relative water requirements of plants. J. Agric. Res. 3: 1-64.         [ Links ]

Castro, E.M. (1978). Variedades recomendadas para o Estado. Inf. Agropec. 41: 11-14.         [ Links ]

Cregg, B.M. (1994). Carbon allocation, gas exchange, and needle morphology of Pinus ponderosa genotypes known to differ in growth and survival under imposed drought. Tree Physiol. 14: 883-898.         [ Links ]

Erb, W.A. (1993). Improved drought tolerance and root development as components of a scheme to breed blueberries for mineral soil adaptability. Euphytica 70: 9-16.         [ Links ]

Gomathinayagam, P., Ingram, K.T. and Maguling, M.A. (1988). Pot screening for drought tolerance in rice. Int. Rice Res. Newsl. 13: 19.         [ Links ]

Heatherly, L.G., Russell, W.J. and Hinckley, T.M. (1977). Water relations and growth of soybeans in drying soil. Crop Sci. 17: 381-386.         [ Links ]

Hurd, E.A. (1976). Plant breeding for drought resistance. In: Water Deficits and Plant Growth, IV. ( Kozlowski, T.T., ed.). Academic Press, New York, pp. 317-353.         [ Links ]

McCall, L.L., Schmidt, S.L., Verhalen, L.M., Oswalt, E.S., Foraker, R.W., Kruska, P.D., Greenhagen, B.E., Mapes, J. and Martin, J.K. (1980). Cotton variety tests, Oklahoma-1979. Agric. Exper. Sta., OSU. Res. Rep. P-793, pp. 17.         [ Links ]

McCall, L.L., Verhalen, L.M., Oswalt, E.S., Thomas, N.B., Avis, J.D., Kruska, P.D., Foraker, R.W., Greenhagen, B.E., Greenhagen, B.S. and Foltz, M.N. (1981). Cotton variety tests, Oklahoma-1980. Agric. Exper. Sta., OSU. Res. Rep. P-806, pp. 15.         [ Links ]

Nour, A.M. (1975). some aspects of drought resistance in grain sorghum. Ph.D. thesis, Oklahoma State University, Stillwater, OK, USA.         [ Links ]

Nour, a.E.M., Weibel, D.E. and Todd, G.W. (1978). Effect of repeated drought periods on the survival of sorghum seedlings. Agron. J. 70: 509-510.         [ Links ]

Poehlman, J.M. (1986). Breeding field crops. 3rd edn. Van Nostrand Reinhold, New York, pp. 724.         [ Links ]

Ray, L.L., Wendt, C.W., Roark, B. and Quisenberry, J.E. (1974). Genetic modification of cotton plants for more efficient water use. Agric. Meteorol. 14: 31-38.         [ Links ]

Roark, B., Quisenberry, J.E. and Fest, G. (1973). Behavior, distribution, and frequency of stomates on leaves of selected varieties of Gossypium hirsutum L. In: Beltwide Cotton Prod. Res. Conf., National Cotton Council, Memphis, USA, pp. 51 (Abstract).         [ Links ]

Steel, R.G.D. and Torrie, J.H. (1980). Principles and procedures of statistics: A biometrical approach. 2nd edn. McGraw-Hill Book Co., Inc., New York, pp. 633.         [ Links ]

Tischler, C.R., Voigt, P.W. and Young, B.A. (1991). Tray system for measuring drought tolerance of forage grasses. Crop Sci. 31: 1696-1699.         [ Links ]

Wright, L.N. (1964). Drought tolerance - Program-controlled environmental evaluation among range grass genera and species. Crop Sci. 4: 472-474.         [ Links ]


(Received April 8, 1997)

Creative Commons License All the contents of this journal, except where otherwise noted, is licensed under a Creative Commons Attribution License