Implications of Mating Behavior in Watermelon Breeding

Understanding the natural mating behavior (selfor cross-pollination) in watermelon is important to the design of a suitable breeding strategy. The objective of this study was to measure the rate of selfand cross-pollination in watermelon [Citrullus lanatus (Thunb.) Matsum. & Nakai] using the dominant gene Sp (Spotted leaves and fruit) as a marker. The experiment consisted of two studies and was a split plot in a randomized complete block design with 3 years (2009 to 2011) and four locations (Clinton, Kinston, Oxford, Lewiston, NC). For the intercrossing study, whole plots were the two spacings (1.2 3 0.3 m and 1.2 3 0.6 m) with four replications in 2010. For the inbreeding study, whole plots were two equidistant spacings (3 3 3 m and 6 3 6 m) with four replications in 2009 to 2011. Cultivars Allsweet and Mickylee were subplots within each whole plot. In the inbreeding study, spacing and year had a significant effect on the rate of self-pollination, which was moderate (47% and 54%, respectively) when watermelon plants were trained in a spiral and spaced 3 3 3 m or 6 3 6 m apart. Spacing and cultivar did not have a significant effect on cross-pollination in the intercrossing study. Closely spaced watermelon plants (1.2 3 0.3 m and 1.2 3 0.6 m) had low natural outcrossing rate (31% and 35%, respectively) and was not adequate to intercross families. However, breeders should consider the amount of self-pollination in watermelon to calculate the estimates of component of genetic variances. Watermelon [Citrullus lanatus (Thunb.) Matsum. & Nakai var. lanatus] has been improved for yield and other traits as part of the process of plant breeding. Knowledge of the rate (percentage) of selfor cross-pollination is useful for watermelon breeders interested in planning isolation distances, estimating components of genetic variance, or selecting among progenies produced through openpollination. In cross-pollinated crops, it is often assumed that individuals produced from a single parent are half-sib families and those genetic variances should be calculated on that assumption. However, variances may be improperly estimated if there is natural self-pollination (inbreeding). In addition, knowledge of the rate of natural selfor crosspollination in crops is useful in designing experiments for genetic studies, crop improvement, and for maintaining elite inbred lines (Chowdhury and Slinkard, 1997). Studies of several crops, including barley (Hordeum vulgare L.), lima bean (Phaseolus lunatus L.), wild oat (Avena fatua L.), and rose clover (Trifolium hirsutum All.), all assumed to be predominantly self-pollinating species, have shown that even low outcrossing rates of 1% to 10% had a significant effect on the genetic structure of the populations (Harding and Tucker, 1964; Jain, 1976). Crop improvement methods for self-pollinated crops are different from those of cross-pollinated crops (Fehr, 1993). Common methods for crop improvement used in watermelon are pedigree breeding and recurrent selection (Wehner, 2008). Breeders often work with large F2 populations to recover improved trait combinations for individual plant selections using pedigree breeding. In cross-pollinated crops, controlled self-pollination is made on individual plants by covering flowers before they open, requiring resources for each population and family to be advanced. If the rate of self-pollination can be increased in the field, watermelon breeders can harvest openpollinated seeds from individual plant selections in early generations. On the other hand, if the rate of natural outcrossing can be increased, watermelon populations can be improved by recurrent selection by using natural intercrossing of selected families in isolation blocks (Kumar and Wehner, 2011). Intercrossing can play an important role in genetic gain (Wehner and Cramer, 1996). Crop species are classified as autogamous, allogamous, or mixed mating types. Watermelon is predominantly an allogamous species with monoecious or andromonoecious flowering habit (Ferreira et al., 2002). The a locus determines sex expression in watermelon, producing monoecious (AA) or andromonoecious (aa) sex expression (Guner and Wehner, 2004; Martin et al., 2009; Rhodes and Dane, 1999; Rhodes and Zhang, 1995). Monoecious sex expression promotes allogamy, whereas andromonoecious sex expression can promote autogamy (Martin et al., 2009). Cucurbits often grow as single plants or small populations in the wild. That leads to inbreeding and selection against inbreeding depression as well as the elimination of deleterious recessive alleles (Allard, 1999). In previous studies, there is no significant inbreeding depression measured in watermelon (Wehner, 2008). Further support for that was provided by Ferreira et al. (2000, 2002) who reported an inbreeding coefficient as high as 0.41 and outcrossing rate of 65% in andromonoecious families of watermelon and 77% averaged over monoecious and andromonoecious families. Thus, although watermelon is assumed to be a cross-pollinated crop, there is significant self-pollination that breeders should be aware of. Natural selfpollination can be of some use in a low-resource breeding program. Pollination in watermelon is mediated by honeybees (Apis mellifera L.) and bumblebees (Bombus impatiens Cresson) that visit flowers to collect pollen and nectar (Delaplane and Mayer, 2000; Free, 1993; McGregor, 1976). The movement of honeybees and bumblebees among flowers in a field is directional, within rows rather than across rows (Cresswell et al., 1995; Handel, 1982; Walters and Schultheis, 2009; Zimmerman, 1979). The directional movement of pollinators within rows may reduce the revisits of flowers and maximize foraging efficiency (Collevatti et al., 2000). Watermelon breeders often use 3 · 3-m spacing when working with single-plant hills in their breeding program (Neppl and Wehner, 2001). The 3 · 3-m plant spacing provides good separation of vines for pollination and selection. Walters and Schultheis (2009) reported that watermelon plants were mostly self-pollinated when spaced more than 10 m apart. However, 10-m hill spacing may not be economical in breeding programs that handle thousands of plants per year. It might be possible to manipulate the mating behavior (pollination) of watermelon plants by optimizing plant spacing. Furthermore, close plant spacing may be used to enhance crosspollination, thus facilitating intercrossing among families in a recurrent selection program. Recurrent selection is predominantly used to improve quantitative traits (Hallauer and Miranda, 1988). On the other hand, increasing the plant spacing might increase the amount of self-pollination for individual plant selection where required in methods such as pedigree breeding. The environment affects pollen flow in cucurbit crops which, in turn, affects the rate of selfor cross-pollination (Gingras et al., 1999; Stanghellini and Schultheis, 2005). Variation in wind velocity, humidity, light intensity, temperature, and other environmental factors over years and locations may influence pollinator behavior and sex expression in watermelon and thus affect the rate of self-pollination (Kalbarczyk, 2009; Robinson and Decker-Walters, 1997). Self-pollination has been reported to vary from 23% to 77% over locations in cucumber (Wehner and Jenkins, 1985). Jenkins (1942) reported 30% to 35% natural self-pollination in cucumber. Received for publication 26 Apr. 2013. Accepted for publication 14 June 2013. We gratefully acknowledge Ms. Tammy Ellington for assistance with the field tests. To whom reprint requests should be addressed; e-mail [email protected]. 960 HORTSCIENCE VOL. 48(8) AUGUST 2013 Hence, it is useful to study the rate of selfor cross-pollination over multiple years and locations. In tomato and watermelon, cultivars differ in their ability to produce pollen (Lesley, 1924; Stanghellini and Schultheis, 2005). Pollination may be affected by the amount of pollen produced by the flowers of a particular cultivar. The availability of more pollen to the pollinators may increase the rate of outcrossing. The mating behavior in watermelon depends on environment, cultivar, and flight pattern of pollinators. However, we were interested in those aspects of mating behavior (pollination) that are under the control of the plant breeder. The objective of this study was to determine the rate of selfor crosspollination in watermelon as affected by spacing, year, location, and cultivar. Materials and Methods The experiment was conducted at research stations in North Carolina: Horticultural Crops Research Station, Clinton; Cunningham Research Station, Kinston; Peanut Belt Research Station, Lewiston; and Oxford Tobacco Research Station, Oxford, in 3 years (2009 to 2011). Standard horticultural practices were used as recommended by the North Carolina Extension Service (Sanders, 2004). Treatment plots. The experiment consisted of two studies: inbreeding study in 2009 to 2011 and intercrossing study in 2010. The objective of the inbreeding study was to determine if wide spacing increased the amount of self-pollination sufficiently for single-plant selection in early generations for methods such as pedigree breeding. In contrast, the goal of the intercrossing study was to determine if close plant spacing increased the rate of cross-pollination for use in intercrossing families for recurrent selection. The inbreeding study was a split plot in a randomized complete block design with two blocks of two repetitions (within block) at each of two locations (Kinston and Clinton, NC) in 2009 to 2011 (Fig. 1). Spacings of 3 · 3 m (3-m row spacing with 3-m hill spacing) and 6 · 6 m (6-m row spacing with 6-m hill spacing) were used in this experiment as whole plot treatments. The 3 · 3-m spacing is often used by breeders in the United States (Neppl and Wehner, 2001). We were interested in the common 3 · 3-m spacing as well as wider plant spacing, 6 · 6 m. Wider spacing than 6 · 6 m might not be too economical to use in breeding program. The intercrossing study was also a split plot in a randomized complete block design planted at Oxford (north side), Oxford (south side), Rocky Mount, and Lewiston in North Carolina in 2010 (Fig. 2). Each of the four locations had one replication. Whole plot treatments were the two in-row spacings: 1.2 · 0.3 m (1.2-m row spacing with 0.3-m hill spacing) and 1.2 · 0.6 m (1.2-m row spacing with 0.6-m hill spacing). The objective of this study was to determine if close spacing leads to adequate cross-pollination among plants for intercrossing families in a recurrent selection program. An additional objective of both studies was also to generate information about self-pollination of watermelon at different plant spacings. Watermelon is assumed to be a cross-pollinated crop; hence, estimates of genetic variances are calculated on that assumption. Estimates are biased if there is inbreeding (selfpollination) in the breeding population. In both studies, ‘Allsweet’ and ‘Mickylee’ were subplot treatments in the whole plot treatment (spacings). The seeds (progeny) harvested from ‘Allsweet’ and ‘Mickylee’ were planted to measure the rate of selfor cross-pollination. ‘Moon and Stars’ was planted next to plants of ‘Allsweet’ and ‘Mickylee’ as the pollen donor because it had the spotted (Sp) marker gene. The in-row spacing was defined as the distance between pollen acceptor, ‘Allsweet’ and ‘Mickylee’, and the pollen donor, ‘Moon and Stars’. ‘Moon and Stars’ has large, elongate fruit, a dark green rind with yellow spots, and firm, sweet flesh with dotted seeds. The dominant trait of bright yellow spots on the rind and leaves is dominant to the recessive trait of uniformgreen rind and foliage color and is the result of a single dominant gene, Sp (Guner and Wehner, 2004; Poole, 1944; Rhodes, 1986). The Sp gene was used as a marker to measure outcrossing, as observed in the progeny of ‘Allsweet’ and ‘Mickylee’. ‘Allsweet’ has large, elongate fruit with a striped rind and ‘Mickylee’ has small, round fruit with a gray rind (Wehner, 2002). Both of these cultivars have uniform green foliage (sp) and represent two different groups of cultivars. Fig. 1. Experiment design showing an arrangement of a treatment plot in inbreeding study in 2009–11. The experiment was a split plot in a randomized complete block design with two blocks of two repetitions (within block). Whole plot treatments were the two in-row spacings: 3 · 3 m (3-m row spacing with 3-m hill spacing) and 6 · 6 m (6-m row spacing with 6-m hill spacing). ‘Allsweet’ (A) and ‘Mickylee’ (M) were two subplot treatments within each in-row spacing. ‘Moon and Stars’ (S) was planted next to plants of ‘Allsweet’ and ‘Mickylee’ as the pollen donor because it had the spotted (Sp) marker gene. Fig. 2. Experiment design showing a replication of treatment plots in intercrossing study in 2010. The experiment was a split plot in a randomized complete block design with four replications (four locations had one replication each). Whole plot treatments were the 2 in-row spacings: 1.2 · 0.3 m (1.2-m row spacing with 0.3-m hill spacing) and 1.2 · 0.6 m (1.2-m row spacing with 0.6-m hill spacing). ‘Allsweet’ (A) and ‘Mickylee’ (M) were two subplot treatments within each in-row spacing. ‘Moon and Stars’ (S) was planted next to plants of ‘Allsweet’ and ‘Mickylee’ as the pollen donor because it had the spotted (Sp) marker gene. HORTSCIENCE VOL. 48(8) AUGUST 2013 961 Transplants were grown in 72-cell polyethylene flats in the greenhouse of North Carolina State University in Raleigh, NC. A 4P Fafard soilless mix (Conrad Fafard Incorporated, Agawam, MA) was used in the flats. The transplants were moved to coldframes when they were 4 weeks old and transplanted to the field after 1 week of acclimation. Transplants were planted on raised, shaped beds covered with black plastic mulch. Fertilizer was incorporated before planting as a mix including ammonium nitrate at a rate of 90 kg·ha nitrogen (N), 39 kg·ha phosphorus, and 74 kg·ha potassium with an additional 40 kg·ha N (as sodium nitrate) applied at the vine tipover stage. Soil was fumigated with mixture of 1,3-dichloropene and chloropicrin applied at a rate of 60 L·ha. Irrigation was applied in a drip irrigation system for a total of 25 to 40 mm per week. Rows were 1.2, 3, or 6 m apart (center to center). For the inbreeding study, plants were trained in a spiral arrangement each week starting when the vines reached the edge of the raised bed and ending at the time of fruit set (Gusmini and Wehner, 2007). For the intercrossing study, no spiral training was practiced to encourage cross-pollination. Honeybees were placed in the field at the stage of first flower opening using the recommended rate of two active hives per ha. No disease problems were observed. Progeny evaluation plots. Progenies were evaluated from seeds obtained from treatment plots grown in the spring season each year and planted in the summer season. One ripe fruit was harvested from each singleplant hill of ‘Allsweet’ and ‘Mickylee’ in each treatment combination. Progeny evaluation plots were 1.5 · 5.2 m. Progeny were evaluated at the four true-leaf stage using 100 plants per plot to calculate the rate of selfand cross-pollination. A second evaluation was done 2 weeks after the first to confirm the results. All seeds in the fruit that were set on ‘Allsweet’ and ‘Mickylee’ were produced through a combination of selfand crosspollination. Progeny with bright yellow spots on their leaves were the result of crosspollination by ‘Moon and Stars’ carrying the Sp allele. The rate of self-pollination was measured as the percentage of uniform (nonspotted) plants out of the total. The data were analyzed using the MEANS and GLM procedures of SAS (SAS Institute, Inc., Cary, NC). Means were separated using least significant difference (P # 0.05) for those factors having a significant F ratio in the analysis of variance. Data were means of four locations in the intercrossing study and two locations and two blocks within each location in the inbreeding study. Data were analyzed separately for the inbreeding study and the intercrossing study.