Click on author name to view affiliation information
Pawpaw (Asimina triloba) is a native tree fruit that is in small-scale production across much of the eastern and southern portions of the United States. There is some evidence that pawpaw requires cross-pollination; however, there is also evidence of self-compatibility or self-fruitfulness in some cultivars. The objective of this study was to determine the incidence of self-compatibility and its impact on fruit and seed set in pawpaw. Two pawpaw cultivars, Sunflower and Susquehanna, were examined in this study at the Kentucky State University Harold R. Benson Research and Demonstration Farm. Approximately 1000 crosses, including 250 self- and cross-pollinations for each cultivar, were carried out with open flowers in Apr 2016 and 2017. Competing flowers were removed after pollination, and the number of fruit clusters and fruit number within each cluster were recorded three times a year during May, July, and August. The fruit were harvested in August, followed by seed extraction. After stratification, the seeds were sown in a greenhouse. The leaves of parent material and the newly germinated seedlings from the seeds in each treatment were collected and DNA was extracted. Primers for three pawpaw simple sequence repeats (SSRs) for the loci Pp-B3, Pp-B103, and Pp-G124 were selected. These primers generated unique SSR fingerprint patterns for each parent and allowed the determination of whether a seedling was the result of selfing or crossing between the two parents. This study is the first to report that selfing can occur in pawpaw. The DNA fingerprinting results confirmed some self-fruit set in ‘Sunflower’ in 2016 and 2017. Both cultivars served as pollinizers for the other; however, genetic yield potential by the maternal tree could be more important for determining the ultimate fruit set than outcrossing between genotypes.
Pawpaw (Asimina triloba L. Dunal) is a tree fruit native to North America with a tropical-like flavored fruit similar to a blend of mango, pineapple, and banana (Duffrin and Pomper 2006). It thrives in the forest understory in large patches, often due to root suckering along riversides in the Midwest and southeastern United States (Kral 1960). There are small commercial orchards of selected cultivars throughout these regions in the United States (US Department of Agriculture National Agricultural Statistics Service 2022). Pawpaw fruit are valued for their fresh-market and processing potential, nutritional benefits, and the tree’s ornamental appeal to homeowners (Pomper and Layne 2005). Its popularity is increasing; however, nurseries struggle to meet the growing public demand for pawpaw cultivars (Behrends et al. 2024).
Although pawpaw is thought to require outcrossing for successful fruit set, the extent of pollination as a limiting factor remains unclear. Pawpaw flowers are maroon, have a fermented scent, and bloom from mid-April to late May, with individual trees flowering for 3 to 4 weeks (Goodrich et al. 2006). The flowers are hermaphroditic, dichogamous, and protogynous, thus encouraging cross-pollination (Willson and Schemske 1980). Flies (Diptera) and beetles (Coleoptera) serve as primary pollinators (Willson and Schemske 1980). However, large clonal patches of genetically identical trees are thought to lead to limited cross-pollination often resulting in less than 1% fruit set in wild patches. Hand-pollination can increase the pollination success rate and increase fruit set in pawpaw, with rates ranging from 4% to 17% compared with 0.41% to 0.63% under open-pollinated conditions (Peterson 1991; Willson and Schemske 1980). In central Florida, two pawpaw species (Asimina obovata and Asimina pygmaea) exhibited fruit set rates of up to 16% with open pollination, whereas hand-pollination increased fruit set to 32% in A. obovata and 60% in A. pygmaea (Norman and Clayton 1986).
Despite increased fruit set from improved pollination, pawpaw trees often experience significant fruit drop during early development (commonly referred to as June drop), likely a result of resource limitations. As an understory tree, shading can restrict photosynthesis, further exacerbating competition for assimilates among developing fruit (Moore 2015). There is likely a genetic component that limits the yield of particular genotypes or cultivars (Pomper et al. 2008b). Although hand-thinning can improve fruit size and reduce resource competition, this practice is rarely used in pawpaw cultivation (Crabtree et al. 2010). In fruit crops, fruit size and development can be affected by the number of seeds that develop in the fruit. Lang and Danka (1991) reported that cross-pollination increased fruit size by 14% and seed count by 27% in southern highbush blueberries (Vaccinium corymbosum L.); however, fruit set was unaffected, but a higher seed count did increase earliness of ripening. Pawpaw fruit are round, oblong, or cylindrical, depending upon the cultivar. The average fruit weighs ∼150 to 200 g, although this varies widely among cultivars, and each pawpaw fruit contains ∼6 to 15 seeds. However, the relationship between seed count and fruit size or quality in pawpaw has not been explored, particularly under self- and cross-pollination scenarios.
There is little evidence to indicate whether A. triloba is self-incompatible or self-compatible (Willson and Schemske 1980); however, there are anecdotal observations that one of the cultivars, Sunflower, may be self-compatible (Pomper et al. 2008b). In other Asimina species, a study with A. obovata in central Florida reported that bagged flowers of this particular cultivar occasionally self-pollinate (4% in 1981), whereas A. pygmaea showed no sign of self-compatibility (Norman and Clayton 1986). Similarly, another study with Asimina parviflora (Michx.) Dunal indicated that bagged flowers occasionally produce fruit (Norman et al. 1992). The same study also noted that the seeds of the self-pollinated fruit were significantly smaller compared with those from cross-pollinated fruit, and their germination rate was zero.
Pawpaw (A. triloba) is diploid (2n = 2x = 18) (Bowden 1948) and, like other plants, it contains a unique and conserved set of DNA sequences. Simple sequence repeats (SSRs), or microsatellites, are the short (1–6 bp) tandem repeats of DNA sequences and they vary among individual plant genotypes in number and type of repeats. SSR markers have been developed for pawpaw to fingerprint different cultivars and assess genetic diversity in the Kentucky State University (KSU) repository collection (Pomper et al. 2010). Using SSR DNA fingerprinting, progeny resulting from self-pollination (selfing) or out-crossing could be determined. Therefore, the main objective of our study was to determine whether pawpaw are self-compatible. The second objective of our research was to evaluate the impact of self-compatibility on pawpaw fruit and seed set, and to confirm crossing and selfing using SSR fingerprinting.
This study was conducted in 2016 and 2017 in a 12-year-old pawpaw orchard with several pawpaw cultivars, including uniform ‘Sunflower’ and ‘Susquehanna,’ and seedling border-row trees in a completely randomized design. The trees were planted with a spacing of 8 ft between trees and 18 ft between rows in May 2004 at the KSU Harold R. Benson Research and Demonstration Farm in Frankfort, KY, USA (Pomper et al. 2021). Trees were pruned minimally and irrigated as needed with drip irrigation using two 3.78 L·h–1 emitters per tree. Fertilization was performed annually in Spring with urea (46–0–0) at 141.5 g/tree per year. Glyphosate was used for weed control as needed during the growing season at the label rate. Trees selected for this study were uniform in size, bud load, and flowering characteristics to minimize variability in experimental conditions.
In Apr 2016, open maroon flowers on branches were hand-pollinated using cotton-tipped applicators or Q-Tips (Unilever, Englewood Cliffs, NJ, USA). Uniform ‘Sunflower’ and ‘Susquehanna’ trees were selected within the orchard for controlled pollination treatments. A total of 1000 pollination events were conducted, with 250 flowers from three ‘Sunflower’ trees and 250 flowers from five ‘Susquehanna’ trees subjected to self-pollination. In addition, 250 flowers from four ‘Susquehanna’ trees were cross-pollinated with ‘Sunflower’ as the pollen donor, whereas another 250 flowers from nine ‘Sunflower’ trees were cross-pollinated with ‘Susquehanna’ as the pollen donor as a result of fewer available flowers on “Sunflower’ after a spring frost event. The flowering branches were tagged with colored flagging tapes representing the four treatments: 1) ‘Sunflower’ × ‘Sunflower’ (‘Sunflower’ pollen recipient/’Sunflower’ pollen donor), 2) ‘Susquehanna’ × ‘Susquehanna’ (‘Susquehanna’ pollen recipient/‘Susquehanna’ pollen donor), 3) ‘Sunflower’ × ‘Susquehanna’ (‘Sunflower’ pollen recipient/‘Susquehanna’ pollen donor), and 4) ‘Susquehanna’ × ‘Sunflower’ (‘Susquehanna’ pollen recipient/‘Sunflower’ pollen donor).
Any competing flowers that bloomed on labeled branches after the hand-pollination were removed to prevent undesired crosses. The resulting fruit and clusters were assessed at three intervals between hand-pollination and harvest: 24 May, 22 Jul, and 22 Aug (fruit harvest date). Data on the number of fruit clusters present were determined to assess the pollination success rate or flowers producing cluster and the number of fruit per cluster. The fruit were harvested at physiological maturity, while they were still green and attached to the tree, to ensure accurate treatment identification.
In 2016, the total number of seeds per fruit for each treatment was recorded to determine the average number of seeds per fruit. Mature seeds were counted, whereas visibly immature seeds incapable of germination were disregarded. Seeds collected from each treatment were mixed with moist peatmoss and stratified at 4 °C in a refrigerator for ∼110 d to fulfill their chilling requirement. Three lots of 20 stratified seeds from each treatment were then placed randomly in the growth chamber on 6 Jan at 21 °C to promote germination. Germination was unsuccessful, likely because of inadequate chilling. To address this, 20 additional stratified seeds from each treatment, which were still in a refrigerator, were selected and sown into 1-gal tree pots filled with a medium composed of two thirds Pro-Mix® peatmoss (Premier Horticulture, Inc., Red Hill, PA, USA) in a greenhouse on 18 Apr 2017. Germination was observed 12 Jun 2017.
In 2017, hand-pollination treatments were repeated on two trees of each cultivar, with 50 flowers per treatment, resulting in 200 pollination events. Data on pollination success rate through cluster count, and fruit set were collected on three dates: 31 May, 13 Jun, and 21 Aug. Seeds were extracted a day after fruit harvest, and both the total seed count per fruit and the individual fruit weights were recorded to calculate the percentage seed data. Later, the total seeds of each treatment were mixed with moist peatmoss and refrigerated at 4 °C to satisfy their chilling requirements.
Young leaves were collected from parent trees during the flowering season in 2016 and 2017, and were stored at –15 °C until DNA extraction. DNA extraction from leaves of parents and seedlings was carried out by following the protocol described by Pomper et al. (2010). Polymerase chain reaction conditions were performed as outlined by Lu et al. (2011); however, plates were evaluated with a 3500 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). Three specific pawpaw primers, developed and validated previously, were used to compare genotypes (Lu et al. 2011; Pomper et al. 2010). These primers included loci B3, B103, and G124 (Table 1). The individuals were genotyped or DNA fingerprinted with GeneMapper Software ver. 4.1 (Applied Biosystems).
Offspring were classified based on allele size across the three loci. Individuals with all alleles matching a single parent across loci B3, B103, and G124 were identified as self-pollinated offspring. Offspring exhibiting alleles from both parents were designated as cross-pollinated. Those containing unknown alleles were labeled as unknown cross-pollinated offspring. The percentage of self- and cross-pollinated offspring was calculated for further analysis. Data on fruit and cluster numbers were analyzed using CoStat software (Version 6.45; CoHort Software, Monterey, CA, USA), and data were subjected to analysis of variance at a least significant difference of P < 0.05.
In our study, individual flowers served as the replicate for each pollination treatment. Analysis of variance showed that tree effect was not significant. Therefore, based on this finding, the same approach was applied in 2017; flowers served as the replicate for pollination treatment.
All pollination treatments, including selfs (‘Susquehanna’ × ‘Susquehanna’ and ‘Sunflower’ × ‘Sunflower’) and crosses (‘Sunflower’ × ‘Susquehanna’ and ‘Susquehanna’ × ‘Sunflower’), resulted in cluster formation during both study years (Table 2). In 2016, the cluster set was comparable across all selfs and crosses, but ‘Susquehanna’ × ‘Sunflower’ had the lowest cluster set consistently throughout the study (Table 2). Following a typical June drop, cluster retention was greater in treatments in which ‘Sunflower’ served as the maternal parent. In 2017, the crosses demonstrated an initially higher cluster set than the selfs in May; however, this advantage was offset by a greater June drop in crosses. In contrast, selfs exhibited stable cluster retention across the season (Table 2). Fruit set per cluster was initially greatest in treatments in which ‘Susquehanna’ was the maternal parent regardless of the treatment, but it became comparable across all treatments from July onward. Final fruit count averaged ∼2.1 and 2.8 fruit per cluster in 2016 and 2017 respectively, with June drop contributing to this uniformity (Table 2).
In 2017, the average fruit weight was comparable across the treatments, although the fruit from the crosses tended to be slightly larger than those from the selfs (Table 3). The number of seeds per fruit in 2016 and 2017 indicated that ‘Susquehanna’ fruit had fewer seeds than ‘Sunflower’ (Table 3). In addition, the average seed weight in treatments in which ‘Susquehanna’ was the maternal parent was also significantly less compared with treatments in which ‘Sunflower’ was the maternal parent (Table 3). A notable observation was that seeds from ‘Susquehanna’ fruit were immature at harvest, likely because of their later maturation period compared with ‘Sunflower’. This aligns with previous findings that ‘Susquehanna’ fruit have a lower seed percentage and greater pulp content than ‘Sunflower’ (Pomper et al. 2008b, 2021). Seed viability was high across all treatments, with germination rates ranging from 80% to 100%, indicating robust seed quality for both cultivars.
DNA was extracted from the seedling leaves that were grown in the greenhouse and were fingerprinted with three SSR primers. The fingerprinting revealed distinct DNA profiles for offspring resulting from selfs and crosses. The parent DNA product profiles, along with those of selfs and crosses offspring, are summarized in Table 4. The data also identified contributions from unknown parental genotypes, marked as unknown cross-pollinated offspring (Table 5). The results indicated that only ‘Sunflower’ exhibited self-pollination at a very low rate both in 2016 (7.1%) and 2017 (4.7%), and that many of the fruit that were initially categorized as self-pollinated (based on fruit set data) were found to be the result of cross-pollination between the two parents (Tables 2 and 5). No self-pollinated offspring were detected in treatments in which ‘Susquehanna’ was the maternal parent. Interestingly, within the presumed cross-pollinated offspring of ‘Sunflower’ × ‘Susquehanna’, a few were confirmed to be self-pollinated offspring of ‘Sunflower’. In addition, DNA fingerprinting revealed evidence of foreign pollen sources contributing to unknown cross-pollination. These unknown crosses likely originated from other cultivars or genotypes in the KSU orchards, further complicating the pollination dynamics (Table 5).
This is the first study to demonstrate selfing occurs in pawpaw, as determined by DNA SSR fingerprinting. Notably, this was observed only in the pawpaw cultivar Sunflower, which appears to support the observations of some pawpaw growers, as stated by Pomper et al. (2008b). However, no self-pollinated treatments resulted exclusively in self-pollinated offspring. Instead, self-pollinated offspring were detected within the cross-pollination treatments of ‘Sunflower’ × ‘Susquehanna’, likely a result of inadequate protection of stigmatic surfaces or pollinator activity during or after pollen transfer. The presence of unknown crosses further suggests that pollen from other pawpaw genotypes in the orchard contributed to fruit set. These findings are consistent with results in other fruit crops such as avocado (Persea americana Mill.), in which cross-pollination is predominant, although self-pollination can occur (Davenport 1986; Schnell et al. 2009). In studies of the commercial cultivar Hass, with adjacent ‘Bacon’ pollinizers, results showed that 75% of the ‘Hass’ offspring were the result of outcrossing whereas the remaining 25% of seedlings were from selfing of ‘Hass’ as confirmed by DNA SSR fingerprinting. Similarly, 45% of ‘Bacon’ offspring were the result of outcrossing and 55% were the result of selfing of ‘Bacon’ (Schnell et al. 2009). Similarly, pawpaw demonstrates a primarily cross-pollinated reproductive strategy, with low levels of selfing in specific cultivars.
The DNA SSR data confirm that seeds from most fruit on a pawpaw tree are the result of cross-pollination. The cross-pollination between ‘Sunflower’ (recipient) to ‘Susquehanna’ (donor) consistently yielded the highest cluster and final fruit set in 2016 and 2017, supporting the hypothesis that the genetic makeup of the maternal cultivar plays a critical role in determining yield, as is the case in many fruit crops. A previous study comparing the yield of various pawpaw cultivars in Kentucky indicated that ‘Sunflower’ (16.1 kg of fruit per tree at 6 years) produced significantly greater yields compared with ‘Susquehanna’ (8.8 kg of fruit per tree at 6 years), although both cultivars had a similar number of fruit per cluster (two fruit per cluster), thus confirming ‘Sunflower’ to be a high-yielding cultivar despite its pollination type (Pomper et al. 2008a, 2008b). This highlights the importance of cultivar selection for maximizing fruit set and yield.
Fruit drop remains a significant challenge in pawpaw production, which occurred twice during the season (Table 2). The initial drop in May likely resulted from improper fertilization or failure of embryo development, whereas the June drop may have stemmed from excessive fruit load exceeding tree canopy capacity. In particular, there was a marked reduction in the number of fruit per cluster on ‘Susquehanna’ × ‘Sunflower’ cross- and ‘Susquehanna’ × ‘Susquehanna’ self-treatment trees in 2016, as well as a decrease in the number of fruit per cluster for ‘Susquehanna’ × ‘Sunflower’ cross trees in 2017. This likely reflects the greater fruit load capacity of ‘Sunflower’ trees compared with ‘Susquehanna’ trees, as previously reported by Pomper et al. (2008b). Although it is difficult to assess whether selfing resulted in a reduction in the number of fruit per dropped per cluster, because there were significant numbers of seeds that developed from selfs, it would seem likely that selfing did not result in at least the June drop of fruit, and that limited resources by fruit in the tree were primarily responsible for June drop and nonparthenocarpic fruit that failed to develop seeds.
Our study demonstrates that ‘Sunflower’ is self-compatible, although outcrossing remains crucial for optimal fruit set and seed development. Among the four pollination treatments, the cross-pollination of ‘Sunflower’ (recipient) to ‘Susquehanna’ (donor) outperformed others in fruit set, seed set, yield, and germination percentage, indicating a strong pollinizer relationship between these two cultivars. However, the genetic yield potential of the maternal tree appears more influential in determining final yield than the pollination type. Further studies should investigate the pollinizer relationships in pawpaw to optimize fruit set, including whether pawpaw exhibits gametophytic or sporophytic self-incompatibility. DNA fingerprinting proved invaluable in identifying the genetic inheritance of markers and determining parentage, revealing that ∼7% of the fruit from cross-pollination treatments were the result of selfing. Although ‘Sunflower’ demonstrated low self-compatibility, cross-pollination accounted for the majority of fruit set, emphasizing the importance of developing effective pollinizer strategies in pawpaw cultivation. In conclusion, this study highlights the genetic and reproductive dynamics of pawpaw, providing a foundation for improving yields through strategic pollinizer development and better understanding of self-compatibility in pawpaw cultivars.
Contributor Notes
This project was supported by US Department of Agriculture Evans Allen Capacity Grant Funding. KSU Ag Experiment Station Publication number KYSU-000144.
