Heritability and Correlations of Agronomic and Fiber Traits in an Okra-Leaf Upland Cotton Population

In cotton (Gossypium hirsutum L.), the cost and time to develop and evaluate appropriate genetic populations have limited the number of intensive and complete heritability studies. Herein, three agronomic and 17 fiber quality traits were assessed for heritability and correlation analyses on progeny rows in an okra-leaf cotton population of 208 families. Progenies were advanced in succeeding generations by a single-seed descent. Comparison between F2:3 and F2:6 generations for individual traits and individual progeny by trait revealed significant differences between the two generations. Heritability estimates (h . 0.60), and correlations within and between (r . 0.55) F2:3 and F2:6, generations have practical applications for the simultaneous improvement of multiple fiber traits. Fiber strength was positively correlated to 2.5 and 50% fiber span length and negatively correlated to short fiber content. Number of neps was positively correlated to number of seed coats, and short and immature fiber content, and negatively correlated to mean fiber fineness and maturity ratio. The genetic potential for improving agronomic and fiber traits may exist in populations with this alternative leaf morphology, okra-leaf type. Mass selection may be effective for improving most of the above traits (h . 0.60). However, pedigree, sibs, and progeny tests need to be used to achieve higher genetic progress. Selection may be applied as early as the F3 when selection units can be replicated. Thereafter, antagonistic trait correlations may become neutral or favorable in later generations, facilitating improvement of fiber quality. COTTON is produced as a raw material for the textile industry and is a high value crop. Marketing of this crop is based on measurable quality properties in an industry where manufacturing technology changes are being implemented rapidly (Sawhney et al., 2003). The most widely planted current cotton cultivars are well yielding, day-length neutral and early maturing with easily ginned and abundant fiber. These improved characteristics resulted from human selection from perennial ancestors with shorter and sparser fiber (Fryxell, 1984). The continuing demands for better quality for consumer goods and the recentmovement from the preponderance of ring spinning to faster, less labor intensive andversatile spinning methods have been driving research programs to look for alternatives to genetically improve lint and fiber quality (Meredith et al., 1991). All the changes in spinning technology have in common the requirement of unique and often greater cotton fiber quality, especially fiber strength for processing (Deussen, 1992). Many cotton research programs require measurements of agronomic and fiber quality traits such as lint percentage, boll weight, 2.5 and 50% fiber span length, fiber bundle strength, and fineness (micronaire reading, fiber maturity, fiber perimeter, etc.). The cotton research community has established fiber testing methods (Breeder, Spinning, Areolometer, Sticky, and HVI) for the above traits, which are run in-house, or through public or private institutions such as the International Textile Center (Lubbock, TX) and Starlab Inc. (Knoxville, TN). Fiber span length at 50% (SL 50%) and 2.5% (SL 2.5%) can be measured with a digital Fibrograph instrument. SL 2.5% estimates the length of the longest 2.5% of fibers scanned in a sample, and the distance is presented in millimeter (mm). Fiber strength (T1) is the strength of a bundle of fibers measured by the stelometer. Elongation (E1) is an estimate of the elasticity of the bundle sample. Micronaire reading (Mic) is a measure of fiber fineness and maturity. A relatively new fiber testing method that has been incorporated rather slowly due to the lack of access to some research programs is called the Advanced Fiber Information System (AFIS). This method measures neps, fiber length and diameter, and trash for fibers (Bragg and Shofner, 1993; Hossein et al., 1994). Correlations among traits can be useful in developing selection criteria, but correlations can also present difficult scenarios for interpretation of the association for trait responses. Mic and SL 2.5% length, which both influence lint percentage, are a good example of components of a more complex trait, fiber yield (Ulloa and Meredith, 2000; Ulloa and Meredith, 2002). In addition, multiple traits can be correlated due to linkage or pleiotropy (Miller and Rawlings, 1967; Meredith and Bridge, 1971; Culp et al., 1979). In cotton, several studies have reported negative correlations between fiber quality and agronomic traits, particularly fiber strength and lint percentage (Miller and Lee, 1964; Worley et al., 1976), but other studies did not detect such correlations (Benedict et al., 1999). For negative correlations, several generations of intermating in an isolation block with approximately 50% self-fertilization changed the genetic correlation between lint yield and fiber strength within a population from antagonistic to favorable (Meredith and Bridge, 1971). The negative correlation between lint yield and certain yield components (boll size, number of fiber per seed, and seed index) demonstrates the problem often associated with breeding for specific yield component e.g., increasing one component often results in decreasing another component(s) due to balanced compensation. Breeders face great difficulties in enhancing fiber traits while mainUSDA-ARS-WCIS Res. Unit, 17053 N. Shafter Ave., Shafter, CA 93263. Received 23 Aug. 2005. *Corresponding author (mulloa@