DEVELOPING AND PROMOTING WOODLAND PAWPAW PRODUCTION PRACTICES TO IMPROVE FRUIT YIELD AND QUALITY

Question 1 ~ Pollen and genetic incompatibility as limitations on production

2022: We piloted our reciprocal hand-pollination experiment at two separate sites in contrasting growing zones in Ohio. A third location will be added in 2023. Experiments revealed significantly enhanced pollination and fruit set where hand pollination was used. Importantly, for a tree long-thought to be self-sterile, we detected higher rates of fruit set in trees pollinated from other plants in their potentially clonal patch. We will investigate this more thoroughly in 2023 and explicitly investigate evidence for self-fertility. Additionally initial inspection of pawpaw flower morphology revealed significant differences in pistil numbers between wild and varietal pawpaws with the former having 1-3 pistils and the latter 3-many. This reveals an important constraint on the maximum number of fruit many wild trees can produce.

2023: We worked across 4 sites with 3-5 plots per site, looking at pollination efficiency for hand pollination within plants, between trees within the same plot and between trees in different plots. We also compared these with the effectiveness of using cultivar pollen and relying on open (i.e. natural) pollination. Our results were somewhat impacted by frost where a number of flowers were damaged.  Based on the surviving flowers, hand pollination was more effective than open pollination. Based on seed number, sufficient pollen appears to germinate and is capable of fertilization. Importantly, and in contrast differences may affect pollination and eventual fruit production we counted the stigmatic surfaces on 300 flowers including from wild and open grown trees. This confirmed that wild trees under cover have significantly fewer stigmatic surfaces.

2024: In this year, we focused on monitoring the success of controlled pollinations at two relatively adjacent sites (i.e., within a mile of each other), one an orchard setting and the other a group of four woodland stands.  Bulk collections of pollen from both orchard and wild trees were used to cross pollinate a population of flowers at each. In addition, a series of controlled self- pollinations were initiated in both locations.  Lastly, flowers were tagged but left uncovered within the orchard and woodland plots to ascertain the frequency of natural pollination and fruit set. Cross pollinations either in the orchard or woodland with either bulk pollen source were generally successful (49-57%). Success of self-pollinations within both the orchard and woodland varied among genotypes with some individuals apparently entirely self-sterile (perhaps due to the presence of incompatibility alleles). Others, however, developed very limited fruit loads suggesting that incompatibility may not be complete. Fruit development from tagged open flowers in either location was almost non-existent suggesting that the natural portion of flowers needed to establish a “full crop” is low (note: this is normal for most fruit crops; in apples a full crop requires fertilization of only 5% of the flowers that bloom). In addition, we initiated a study of pollen viability differences with respect to germination conditions, storage conditions or source. Our initial findings suggest that cultivar and woodland pollen grains behave similarly, each germinating at about 50% (normal for other members of the Annonaceae). Also, pollination can occur evenly over a wide range of temperatures.  However, successful storage of the pollen for future crossing efforts may require dehydration prior to freezing and then rehydration prior to use.  In 2025 we will repeat or expand upon the studies of 2024. In addition, we will initiate a detailed study documenting self-pollen germination on stigmatic surfaces in genotypes that differ in capacity to form self-pollinated fruit. This study will involve microscopic examination of pollen tube growth (or the lack of it) through stylar tissue.

2025 (Final): The initial Critical Question 1 of the grant proposal was – “Can fruit set, production and quality be improved by hand pollination?” Through the 2022-2025 flowering seasons we have answered this question by initiating and monitoring 1201 controlled self- and cross-pollinations in orchard and woodland trees using pollen gathered from flowers on the same tree (geitonogamy), from flowers within the same woodland stand, from flowers from adjacent or distant woodland stands, or bulk collections of woodland or orchard pollen representing extensive genetic diversity. We compared our hand pollination success rate to the frequency of fruit formed by naturally-occurring insect vector pollinations in tagged but unbagged flowers (i.e., open) in both orchard and woodland settings.   In addition, we contrasted orchard-collected and woodland-collected pollen viability, investigated differences in yield and yield parameters associated with wild and cultivar trees, ascertained differences in woodland and cultivar flower morphology and explored the physiological underpinnings of self-incompatibility in cultivar and wild trees.

The rate of naturally-occurring insect vectored pollinations forming fruit compared to the efficacy of controlled self- and cross-pollinations initiated throughout the study is evinced in Tables 1 and 2 respectively.

Table 1

Table 2

Our overall success rate among controlled crosses with all pollen sources was 29.5% (i.e., 354/1201) while we observed that only 5.0% (20/402) of tagged, open-pollinated flowers formed fruit. Clearly, controlled hand-pollination of flowers in either woodland or orchard stands would increase fruit set. However, according to Westwood (1993), fruit set from as few as 5 – 10% of apple flowers formed in a given season results in a balanced crop load yielding leading to fruit of acceptable size and quality. To our knowledge, detailed pawpaw crop load management studies have yet to be published, but Pomper and coworkers (2008a) reported mature trees of eight cultivars to range from 0.7% to 13.8% and to average 5.2% fruit set over a three year period without fruit thinning intervention. It is interesting to note that orchard tagged flowers in 2024 did not yield fruit in our study. In 2024, this orchard was in its first full cropping season and flowering abundantly. The number of flowers present in the orchard in comparison to those in woodland plots would have lowered the statistical probability that a flower chosen randomly would have been successfully insect-pollinated.    

Perhaps a more intriguing question to ask might be – “Is crossing success in the woodlands affected by pollen source and more specifically by its genetic relationship to maternal plant”?  According to Pomper and Layne (2005) protogyny (i.e., a phenological floral pattern where stigmatic surfaces are receptive before pollen is shed) and putative self-incompatibility are main factors favoring outcrossing within feral pawpaw stands. Research reports suggest that trees in some wild pawpaw patches may exhibit a degree of genetic diversity, However, Pomper et al. (2009) found determined through genetic analysis that three of six woodland stands in Kentucky were comprised entirely of clonal material via rhizomatous asexual reproduction. In a recent study, this research group demonstrated that self-incompatibility, at least in some cultivars, is not absolute (Magar et al., 2025b), with 4.0 – 12.4% of controlled self-pollinated flowers of “Susquehanna’ and ‘Sunflower’ producing clusters of harvestable fruit, respectively. Moreover, genetic analysis revealed that 4.7 – 7.1 % of seedlings resulting from open-pollination of ‘Sunflower’ were autogamous progeny. As reviewed by Saunders (2020), multiple evolutionary biologists studying the Annonaceae affirm that in the absence of allelic mechanisms of incompatibility (i.e., gametophytic or sporophytic), self-pollination is possible, but that flower structure and pollination mechanisms overwhelming promote xenogamy (cross-fertilization) including the production of pollinator rewards (scent, heat, nectar from stigmatic surfaces and brood sites), protogyny, herkogamy (physical separation of staminate and pistillate structures) and in some species, floral synchrony (compressed phenology of pistil receptivity), and dicliny (formation of flowers that are structurally or functionally unisexual).  

Between 2022 and 2025, we initiated 464 controlled crosses in woodland stands using pollen sources that were putatively dissimilar to maternal recipient (i.e., pollen collected from other woodland stands or bulked pollen collected widely in the woodland or orchard).  From these crosses we harvested fruit from 140 clusters representing a success rate of 30.2% (Table 2). Our success rate was reduced to 20.5% (57/278) when stigmatic surfaces were coated with pollen collected from other trees within the plot. As we did not genetically determine the level of homogeneity among genotypes within each of our seven wild plots (i.e., the level of clonal reproduction within stands), it was not possible to ascertain whether these pollen sources were more or less similar to the female parents than bulk sources obtained from other plots or sites.

Controlled crosses using bulked pollen sources were performed in both woodland and orchard settings in the 2024 and 2025 seasons which allowed direct comparison female and male performance in each. Our overall crossing success rate using either wild or orchard bulk pollen was 49.0% (103/210) in the orchard but only 38.9% (77/198) in the woodland plots, suggesting a reduced fecundity rate in feral germplasm.  Reduced performance was also noted for bulk pollen obtained from wild in comparison to the orchard collected bulk sample. When the former was used to pollinate either woodland or orchard flowers, the success rate was 39.8% (82/206) whereas the use of bulk orchard pollen resulted in the formation of clusters from 98 of 202 crosses (48.5%). The superior performance of both orchard flowers and pollen is likely an outcome of breeding and selection during the domestication process. However, the exact physiological nature of these improvements was not evident in this study’s findings.

We reorganized data collected in 2023 to reflect differences in fecundity rates among the 4 feral sites where controlled crosses were initiated (Table 3).

Table 3

The pollination success rate at Columbus Site 2 was substantially higher than at the other three sites. Situated on the banks of the Olentangy River, Site 2 is the only riparian site in our study and is perhaps the most urban as it borders on a pedestrian path. However, light levels do not differ substantially from values we report below for the 4 plots at the other Columbus site (see data below).  Soil samples taken from Site 2 indicate an elevated organic matter content and carbon to nitrogen ratio, higher cation exchange levels, higher calcium levels associated with this site, but leaf nutritional data collected at the two sites is nearly identical (see data below). We did not monitor soil moisture content throughout the season so we cannot claim trees at Site 2 had greater access to water. Our Increased crossing success rate at this site may simply result from better tree genetics.  

We conducted studies in 2024 and 2025 to ascertain if discrepancies exist in woodland and orchard pollen germinability. The studies were performed at the USDA/OSU Ornamental Plant Germplasm Center on the University Campus approximately 1.5 miles from both the Columbus woodland and orchard sites. Approximately 100-150 flowers with dehiscing anthers were collected. Anther balls were removed manually in the field and pollen was collected in Petri dishes. The Petri plates containing pollen tetrads and associated filaments were then transported on ice to the laboratory. Pollen plating began no later than 2 hours after the initial flowers were harvested. The experiments, experimental methodology, and results are displayed in Table 4.

Table 4

Pollen germination percentages in the mid-range of temperatures were very similar to those report ed by Lora and coworkers (2012) and somewhat higher than those described by Pereira et al. (2014) for species of Annona germinated at 25°C. In general, pollen collected from woodland and cultivar flowers behaved similarly across a wide range of germinating temperatures. As pawpaw flowers reach anthesis in mid- to late-April in Columbus, daily temperature minimums range from 5.5 – 8.9°C whereas afternoon highs are typically 16.7 – 20.6°C (Weatherspark, 2026). Therefore, the germination of pollen from either location may be delayed in early mornings but be fully functional by mid-day when pollinators are likely most active. We also measured tube length of 100 germinated pollen grains after 2 hours of growth for each pollen type at each temperature as they appeared in images that were captured using our Leica microscope camera attachment and using associated software. The data were analyzed as a linear model with pollen type and temperature as fixed effects. Temperature was significant (p=>0.0001) whereas pollen type was not (Figure 1).

Figure 1

Our data therefore suggests that the superiority of cultivar pollen at affecting crosses in the orchard or woodland does not lie in differences in ability to germinate per se, but rather somewhere downstream in the pollination and/or fertilization process.

In addition to cross-pollination trials, we self-pollinated 184 woodland genotypes and 110 cultivars flowers using pollen-shedding flowers on the same tree as a pollen source (i.e., via geitonogamy) (Table 2). Our self-pollination success rate of 2.7% was identical among feral and cultivar trees refuting the long-held belief that pawpaws are completely self-incompatible and supporting the work of Magar et al. (2025b) demonstrating that self-pollinations are possible but occur at very low frequencies. To verify that the impediments to self-fertilization were not linked to one of the common allelic systems in angiosperms (i.e., gametophytic or sporophytic incompatibility) as maintained by Saunders (2020) and others, we conducted a controlled crossing study in 2025 to examine self- and cultivar bulk pollen performance on stigmatic surfaces and in pseudo styles 90 minutes post pollination. For this study we chose 8 woodland trees, 4 of which demonstrated the ability to set some self-pollinated fruit in the 2024 season and 4 that did not. Seven flowers on each tree were bagged approximately 7 days before anthesis to control pollination events. After anthesis, when stigmatic surfaces were receptive, three of the flowers were pollinated with pollen from the same tree, three were pollinated with bulk cultivar pollen and one was left unpollinated as a control. In the orchard, 4 trees of ‘KSU Benson’, a cultivar demonstrating some propensity to self-pollinate, were treated similarly. Flowers were harvested 90 minutes after pollination, the petals and sepals were removed and the pistillate portion of the flower along with the anther ball were immediately fixed in a formaldehyde-based preservative (Lora et al., 2018). After 48 hours, the floral structures were rinsed and transferred carefully to 70% ethanol for storage. After storage, pistils were separated from anther balls under a dissecting scope. In preparation for viewing, pistils were rinsed, softened in a solution of NaOH, rinsed again, stained with aniline blue, placed on a microscope slide, covered with a cover slip and “squashed”. We examined one pseudo style from each of 54 self-pollinated flowers, 53 flowers pollinated with orchard pollen and 17 unpollinated controls. Figure 2 depicts images of floral fractions from self-incompatible woodland trees.    

Figure 2

Although the data from this study is still being processed, visual observations suggest that pollen from either source readily germinates on the stigmatic surface and then begins (at least) to transverse the style. As we did not examine multiple timepoints we were not able to ascertain that pollen grains reached the ovary. Therefore, the operation of a sporophytic incompatibility system could not be discounted based on our data. However, as self-pollination is possible in pawpaw, we hypothesize that a genetic malfunction shortly after fertilization or early in seed development likely causes most fruitlets resulting from self-pollination to abort.

Given the general functionality of both woodland and orchard pollen, a second intriguing question to ask might be – “whether a sufficient cadre of known pollinators are present to transfer pollen from flower to flower”?  In their study of tree architecture with respect to flower and fruit production, Willson and Schemske (1980) posited that pawpaw yield in southern Illinois may be “limited by the availability or abundance of pollinators”. Therefore, we conducted an informal study of pollinator identity and diversity by hand-capturing pollen-bearing insects found within open pawpaw flowers during the 2024 and 2025 season, storing them frozen in screw-capped plastic centrifuge tubes, and then having them identified by Dr. Kayla Perry, an OSU forest entomologist. The list of captures and their identities are shown in Table 5.

Table 5

According to Saunders (2012, 2020) common pollinators of the Annonaceae include insects belonging to the Coleoptera (small and large beetles), Hymenoptera (bees and ants) Diptera (flies), Thysanoptera (thrips) and less frequently Blattodea (cockroaches) orders. Apparently, small beetles were the ancestral pollinators within the family and still predominate, although several insect species are likely to visit flowers within a site and dominance of a pollinator type is likely to vary geographically and perhaps seasonally. In a formal study employing traps in trees of ‘Sunflower’ and ‘Susquehanna’ Magar et al (2025a) found beetles most closely associated with flowering branches whereas flies were captured most frequently in branches devoid of open flowers and that the frequency of both varied among years. Goodrich and coworkers (2023) reported the yeast-like, fermented volatile profiles produced by pawpaw flowers to attract flies predominantly, but Coleoptera and Hymenoptera species were also found in flowers. Our findings predominantly agree with previous reports. In addition, we report visitation by three species of Hemiptera (true bugs), typically regarded as plant or insect pests with piercing-sucking mouthparts. Whether the latter three species were involved with pollination or were present only as plant or insect predators was not established. We and our commercial collaborators also saw flies frequenting open flowers, but these species were difficult to capture by hand, so they did not appear in our list. Admittedly, our study was less than extensive, but the results suggested that a normal population of pollinators was present in both the orchard and woodland sites we observed.

In addition to monitoring pollination success as a driver of yield discrepancies between orchards and woodland stands, it is important to ask – “whether the yield determinants differ between feral and domesticated trees”? 

From 2022 -2025 we collected all fruits (those hand-harvested or drops gathered from the ground) produced in the four woodland plots of Columbus Site 1. In addition, we hand-harvested orchard fruit from trees used in the pollination experiments in 2024 and 2025. Each fruit was weighed and dissected to remove seeds which were counted and weighed. The findings are displayed in Table 6.

Table 6

The number of fruit obtained from woodland plots or the orchard varied substantially from year to year. The “on- year, off-year” pattern exhibited in both the woodland and orchard sites suggests that they have entered a biennial (alternate) bearing pattern common in feral and domesticated temperate fruit tree species (Campbell and Kalcsits, 2024). In alternate bearing, differences in carbohydrate and hormonal balances influence the year-to-year floral initiation. Imbalances can be triggered or maintained by excessive or meager crop loads reducing or increasing flower bud development determining the following season’s cropping potential, respectively, or by adverse environmental conditions such as temperature extremes or drought. We have witnessed the potential for pawpaw to be biennial bearing in crop load studies we have conducted in other grant programs and Pomper and coworkers (2008b) report biennial bearing indices of greater than 0.6 (i.e., strong potential) for three clones in a regional multi-year variety trial. They attribute the pattern in their trial to have been initiated by a spring frost in 2003 that resulted in crop failure for that season. In our study, a spring frost in 2023 may have reduced the number of developing fruit throughout the summer resulting in carbohydrate/hormonal balances allowing for excessive flower bud development and spring bloom in 2024. Likewise, in both the orchard and woodland, extreme drought conditions in 2024 could have curtailed flower initiation for the 2025 season.

As expected, mean fruit weights from woodland trees were only half as massive (51.2%) as those harvested from clones in 2024 and 2025. Except for the 2022 harvest, mean fruit weights from woodland plots were consistently low. The number of seeds per fruit and seed weights per fruit did not vary substantively from year to year or between the orchard and woodland sites. The fruit weight, seed weight and seed numbers from individual fruits are significantly correlated to each other (Table 7). Understandably, seed number and seed weight are strongly related with coefficients of determination (r² values) ranging from 0.49 to 0.60. Fruit weights are more predictive of seed weight (r² values from 0.46 to 0.48) than seed number (r² values from 0.20 to 0.22). The strength of these relationships seem to be consistent across a number of studies we have performed under the auspices of other grants.

Table 7

Admittedly, fruit yield per tree or per unit of land area was not calculated in our study. However, based on data presented in Table 6 we hypothesize that the discrepancy in productivity between orchard and woodlands is due in part to genetic improvements in germplasm made by breeders during the domestication process. 

Improvements during domestication may have included alterations of floral structure or function. Therefore, the fourth question that may be important to ask is – “Did pawpaw cultivar breeders intentionally or coincidentally select germplasm with genetic improvements in floral structure or function that improved yield potential”?

According to Losada et al., (2017) the pistillate portion of the pawpaw flower is composed of “simple, uniloculate carpels forming a multicarpellate, apocarpous gynoecium” indicating that pawpaw fruits arise from a single carpel but that multiple fruits can be obtained from an individual flower because they possess several carpels that are not fused (as in apple). The collection of fruits derived from a single flower are referred to as a cluster. Early in this study (see report for 2022 above) we observed that woodland flowers routinely possessed fewer carpels than flowers obtained from the orchard. Because carpels lack spatial separation prior to fertilization, it was visually easier to determine the number present by counting stigmatic surfaces (one per carpel). Formal counts in 2,951 flowers were recorded during the 2023 – 2025 seasons. The results are displayed below in Table 8 and Figure 3.

Table 8

Figure 3

From this data we can confirm that the yield potential of cultivars based on the number of possible fruit per flower is greater than that of genetically-unimproved woodland trees. It would be necessary to converse with breeders to determine if carpel number was increased intentionally or serendipitously. 

Question 2 ~ Relative importance of growing conditions and genetic identity for production and quality

2022: To initiate this experiment we completed reciprocal grafting between orchard and wild trees at two different sites. Graft take was minimal at one site due to unusually stressful conditions but was close to 50% at the second site. We will be tracking the flowering and fruiting effort by the grafts in 2023 and completing a second round of grafting where there was initial failures.

2023: We were unable to draw strong conclusions regarding fruit production due to a state-wide crop failure associated with severe late frosts that killed most flowers. We continued to evaluate the survival of our reciprocal grafting experiments. Graft survival averaged around 50% and initial analysis suggests it may have been slightly higher in the orchard than in woodlands. All non-surviving grafts were replaced. In our orchards, we additionally experimented with use of the bark inlay grafting method, as opposed to the more frequently used cleft-graft technique. We will continue to evaluate the outcomes but survival seemed to be substantially better for the graft inlay method when we needed to graft larger-diameter trees.

2024: Currently about 25% of the original grafts are still developing. Grafted scions in the orchard or in the woodland failed to flower in 2024. We do not consider this rate of failure or the continued juvenility of the surviving grafts to be unusual. We are hopeful that the surviving grafts will produce fruit in 2025.

2025 (Final): Critical question 2 from the grant proposal was – “Are wild trees inherently unproductive even when light is not limiting” Our approach to answering this question was to graft cultivar scions onto woodland trees and vice versa. Our hypotheses were that if the yield of wild germplasm grafted into the orchard improved, impediments to yield in the woodlands were predominantly environmental and conversely, if the yield of cultivar germplasm was maintained in the low light levels of the forest, productivity was mostly predicated on genetics.  Although our initial graft success rate and subsequent losses were greater than we anticipated, and we have been hampered by the extended juvenility of grafted material, our approach to answering the question remains sound. Many of our woodland scions grafted into the orchard bloomed for the first time in 2025. Unfortunately, a late frost prevented half of the flowers from developing fruit.  Stigma counts made on flowers killed by the frost and on those remaining viable averaged 3.4, suggesting that an improved environment is not likely to enhance yield potential. However, these grafts are still considered to be immature and answering the question above will inherently be a long process that continues in subsequent seasons. Cultivar material grafted into wild stands did not bloom in 2025, perhaps reflecting a delay in development related to low light levels.

Question 3 ~ Light availability as a critical control of production and quality

2022: Shade enclosures were established around trees at two different orchards. Fruit yield did not differ significantly in 2022 but was not expected to. We will continue to track production in 2023 when, having experienced 18 months of enhanced shading, we anticipate differences in production will become noticeable. Our treatments did lead to substantial differences in ripening time for fruit under different levels of shade. We are thus investigating effects on phenology and are excited to have revealed a potential means for growers to extend the fruiting season.

2023: We continued to maintain and monitor trees within our shade enclosures including assessment of flowering effort and light availability. Evaluation of fruit production and quality was limited by this season's crop failure. Significant effort was dedicated to the repair of several of our shade-enclosures that were damaged by 60 mph windstorms.

2024: As in 2023, we continued to monitor the effects of shading on fruit production in orchards as a model of light level effects in understory plots within the woodlands. Once again, a late frost caused a 50% loss of the flowers. Fruit production from the remaining viable flowers did not differ with shading levels. Once again, fruit ripening was substantially delayed in the high shade treatment (see above). Unfortunately, the severe decline in orchard health at one of the original sites precluded erecting the shade structures in 2024. Wind damage to the shade structures at the remaining site continued to be an issue in 2024. 

2025 (Final): Critical question 3 from the grant proposal was – “How does light availability control pawpaw productivity and quality” Our original approach to answering critical question 3 was to construct and install shade-cloth structures covering entire trees in two cooperator commercial orchards located respectively in Kentucky and southern Ohio simulating natural woodland environments that were 0, 30, 50 and 80% shaded.  The effectiveness of this approach was severely hampered by late spring frosts, violent summer windstorms that toppled structures and by orchard decline at the Ohio location. As stated above, 2024 fruit yields of a single cultivar at the Kentucky site were similar under all levels of shade (approximately 100 fruit per tree) whereas only 32 fruit were harvested from control trees of the same cultivar. The control trees did however contain a myriad of vestigial peduncles indicating previously harvested or dropped fruit supporting our early observation that shade significantly delays fruit ripening. We have observed similar patterns in our Columbus sites as both flowering and fruit ripening seem to be hastened in the orchard and delayed in the woodland. In 2024, we measured photosynthetically active radiation (PAR) levels under the shade cloth structures in Kentucky (Table 9)

Table 9

As a follow-up, we measured PAR levels in the four woodland plots of Columbus Site 1 and at Columbus Woodland Site2 at approximately two-week intervals throughout the 2025 season. The seasonal pattern is displayed in Table 10.

Table 10

The levels of shade found in Columbus woodland plots during the fruiting season most closely resembled values associated with 80% shade. After leaf fall of most species, plots still maintained levels of shade close to 50%. Conventional wisdom and many scientific studies proport that reduced PAR levels will limit photosynthetic activity, chlorophyll content and productivity. Although well established for most agronomic crops, some fruit and leafy vegetable species benefit by cultivation under levels of shade (Tekie et al, 2025). For example, pineapples cultivated under mango trees providing 48.6% shade (51.4% PAR level), respectively improved plant biomass, chlorophyll content and leaf nutritional status, but produced fruit with lower °Brix levels (Kishore et al., 2020). Tropical crops produced in multispecies agroforestry systems may compensate for low light intensities by increasing light use efficiency (Charbonnier et al., 2017). Very little has been written about low light effects in woodland pawpaw stands, but Young (1987) found evidence that vapor pressure deficits were lower and stomatal conductance rates were reduced at PAR levels <150 µmol/m²/s in pawpaw patches in central Virginia, which may serve to protect trees from water deficit stress. Although we have presented evidence for the determinants of low yield in woodland plots, our study did not entirely unravel the effects of light levels.

To document further potential woodland environmental parameters that may affect pawpaw yields, we collected soil and leaf samples from Columbus Sites 1 and 2 and from the Columbus Orchard. After drying and sample preparation the samples were submitted to OSUs College of Food, Agricultural and Environmental Sciences Agricultura and Environmental Testing Core for analysis. Soil fertility values are listed in Table 11 whereas leaf nutrient levels are presented in Table 12.

Table 11 and Table 12

In general, soil fertility parameter levels associated with woodland sites equal or exceed those found for the Columbus orchard. As expected, soil organic matter in woodland plots is ample and soil pHs approach neutrality. Although they are located several miles apart and differ in microenvironments (i.e., isolated woodland and urban riparian pathway) the two sites display similar levels of macro and micronutrients although Site 1 may be higher in aluminum and iron whereas Site 2 may have elevated levels of calcium, perhaps effecting the somewhat higher soil pH. In contrast, orchard phosphorus levels exceed those in the woodland by ten-fold, reflecting the history of the orchard site as an experimental site for agronomic crops.

Nutrient levels in the leaves of woodland trees confirm that soil nutrients are available and readily absorbed by tree roots. Leaf fertility parameter values in the two woodland sites and in the orchard vary less than soil fertility. Published leaf nutrient sufficiency ranges are not yet available for pawpaw, but when the leaf values are compared with adequate ranges for apple, deficiencies are not indicated. Therefore, we conclude that differences in yield between orchards and woodlands are not predicated on differences in soil or plant nutrition.  

Question 4 ~ Management intervention effects on fruit quality

2022: Work on Question 4 will commence in 2023 once we have a first crop of fruit from trees manipulated in experiments 1-3

2023: Fruit yields across Ohio were significantly affected by late frosts which led to widespread crop failure. For instance, despite abundant flowering in our experimental orchards sites, each containing several hundred trees, we only recovered a handful of fruit. These significant challenges reduced our ability to recover meaningful fruit samples from across our experiments. We will thus be focusing on fruit collection and quality evaluation in 2024.

2024: This season, we re-envisioned our efforts to document woodland pawpaw quality within the scope of this study. In addition to obtaining routine measures of quality (e.g., firmness, °Brix) we now intend to upgrade the relevance of our efforts by incorporating woodland plot samples into a larger effort to document the content and accumulation of acetogenins among pawpaw cultivars and throughout the fruit developmental process. There is considerable interest in these secondary product compounds among consumers as they demonstrate both health-beneficial and potentially health-detrimental properties. To date there is nothing known about acetogenin levels in woodland-grown fruit. We also intend to examine methods to ensure the microbiological safety of woodland pawpaw fruits which are routinely harvested by collectors after they fall to the forest floor. Information about safety procedures will be of great value to producers of pulp from woodland fruits.

2025 (Final): Critical question 4 from the grant proposal was – “How do interventions to enhance woodland production affect pawpaw fruit quality characteristics”?  This objective was originally included in the proposal in anticipation that fruit from reciprocal grafts in the orchard and woodland and or fruit from the shading experiment would be available for quality analysis. As an alternative, we have been developing laboratory procedures to measure acetogenins in woodland fruits as part of a larger study monitoring their levels in 50+ cultivars collected from our Columbus or cooperating orchards, as fruit develop throughout the season, how they change during fruit ripening and post-harvest storage and how they may differ in cultivar fruit harvested from different locations.  

In September, we hand-harvested fruits from each of six woodland plots across Ohio representing different climatic, edaphic, topographic, and environmental (e.g., percent shade, water availability) conditions to include in the acetogenin study. Fruits were transported to Wooster where they were “after-ripened” under refrigeration for a period of two weeks. The stored fruit were brought to room temperature and evaluated tactilely for firmness (an indication of stage of ripeness, 1 = under-ripe, 3 = ripe, 5 = over-ripe).  Twenty-four fruit (3-5 fruit from each site) at ripeness stages 2-4 (edible) were processed by separating, weighing, and freeze-drying peel, pulp and seed fractions. The freeze-dried material has been powdered (40 mesh) for inclusion in the ongoing acetogenin study described above.

In the interim to characterize additional important quality characteristics of woodland fruit while acetogenin analysis procedures are finalized, the freeze-dried pulp was analyzed for sugar content by the anthrone assay of Clegg and for total phenolic content by the Folin-Chiocalteu reagent procedure of Singleton and Rossi. The results are displayed in Table 13.

Table 13

Mean fruit weights were highest in the Belmont sites and lowest in the Columbus site (Columbus Site 1). In general, these woodland fruit weights were higher than those presented in Table 6, reflecting the selective nature of our harvest for the acetogenin study as opposed to the inclusive and much larger collection of fruit that resulted from hand pollination. The average seed weight/fruit and seed number/fruit for Sites 1-4 were similar to those reported above, whereas those associated with the Belmont, Ohio sites were somewhat elevated due to the size of the fruit. Correlations between fruit weight and seed weight, fruit weight and seed number, and seed weight and seed number were very similar to those found in the pollination study across years (i.e., r = 0.686, r = 0.460; and r = 0.775, respectively) (Table 7).

To our knowledge, we are the first to report sugar content in pawpaw on a dry weight basis. However, given that combined totals of glucose and fructose average 14 g/100 gfw, Brix values among 10 cultivars average 18%, and fruit moisture contents average 74 g/100 gfw (Brannan, 2025; Francino, 2019) our values indicating sugars to comprise between 57 and 75% of the fruit dry weight are credible. In general, fruit from Sites 4 and 5 had higher total sugar content than those from other sites for reasons that are yet undelimited. However, sugar level averages in Sites 1-3 may have been depressed by the inclusion of a Stage 2 (under-ripe) fruit harvested from each location (values 44.4, 48.5 and 52.9% sugar).

Phenolic compounds in fruits act as antioxidants and therefore are phytochemicals that have health beneficial properties. In addition, phenolic compounds act as important as substrates for the processes of enzymatic and non-enzymatic browning (Sathya et al., 2024; Tilley et al., 2023) and as such confer negative quality attributes to the fruit pulp. As much of the woodland fruit that is commercially harvested (i.e., not for personal use) is processed into fruit pulp, frozen and sold to food manufacturers for the formulation of value-added products, control of pulp browning is essential. The mean phenolic levels found in our woodland fruit are consistent with cultivar values reported in the literature (Brannan, 2025) and found within our own studies.  Phenolic contents, similar to sugar levels, are particularly elevated in fruit harvested from Site 5, which is on a hillside that is perhaps less shaded and more well-drained.

Perhaps of more concern to commercial harvesters of woodland fruit is the potential microbial load that they may bear, especially if dropped fruit is retrieved from the forest floor (the traditional and cultural approach to harvesting as, in response to low light levels, woodland tree architecture is often tall with limited trunk cross-sectional area). With this in mind, we collected fourteen dropped fruit from seven woodland sites along with two fruit from our Columbus research orchard. Pulp from each fruit was manually expressed through mesh bags without peeling allowing sufficient epidermal contact to ensure the transfer of microbes on the fruit surface into the pulp if present. Processed pulp was held for 24 hours under refrigeration then transported on ice to the Consumer Protection Laboratory (CPL) at the Ohio Department of Agriculture in Reynoldsburg, OH for analysis of yeast, mold, aerobic bacteria, coliform bacteria and E. coli O157:H7. All samples were found to contain <10 E. coli O157:H7 colony forming units (CFUs) per gram of pulp (which is below the limits of quantification). Levels of other microbes present are listed in Table 14.

Table 14

Coliform bacteria are commonly of fecal origin, some of which can cause serious illness (e.g., E. coli). Only three of the sixteen fruit sampled evinced quantifiable but low coliform levels.  Of potentially greater concern is the presence of aerobic bacteria in most fruit samples and at very high levels in some. Aerobic bacteria are ubiquitous and are generally benign, but some (e.g., Brucella spp., Listeria spp., Bacillus spp.) can be harmful or fatal. The mold test employed at CPL is non-specific. We assume that most of the positive tests for mold result from spores of Phylosticta asiminae an extremely common pawpaw fruit and leaf pathogen. Inoculation of fresh pulp with native yeasts has been noted to limit storability and use due to unwanted fermentation. Our study herein neither affirms nor negates the potential threat of pathogenic microbial contamination of processed pawpaw pulp. However, the potential for an outbreak of a food borne illness that is detrimental to the individuals infected and the industry value chain cannot be discounted. Woodland pulp processors should undertake a sanitary regimen that complies with USDA’s Good Manufacturing Practices (GMP) to protect consumers, their livelihood and the expansion of the industry as a whole.

Question 5 ~ Efficacy and efficiency of patch establishment methods

2022: We established ten new pawpaw patches in woodland currently lacking any known pawpaw populations. Plots were cleared of invasive species and thinned to reduce understory competition. We logged the time and resources required to impose the treatments. Thirty bare-root pawpaw seedlings were planted in each plot half with and half without tree guards. Three container-grown grafted varietals were planted in each plot. Grafted varietals showed significant mortality with most of this attributed to disturbance by racoons post-planting. Deer browsing pressure was high and counter-intuitively particularly damaging on trees with guards (guards were removed by the deer in many plots). Survival of seedlings was assessed in summer 2022 and will be monitored again in spring, summer and early fall 2023. Initial survival of the seedlings was good (80-90%) despite deer damage. In response to queries from growers and collaborators we established two additional planting trials including - 1) invasive removal and underplanting in existing/mature black walnut plantations; 2) development of chestnut and black walnut / pawpaw interplanting trials. Pawpaw survival was strong in all of these experiments. Severe drought and browsing damage limited the survival of chestnut trees. All dead trees will be replaced in spring 2023.

2023: We continued to monitor and maintain our orchard and woodland plantings throughout 2023. Overall survival remained high (> 80%) and the replanted grafted varietal showed better survival than the initial cohort as more attention was paid to their thorough protection from wildlife. Many plants had been affected by significant browsing damage during late winter and early spring. Damage to our plots was not evenly distributed and greatest in more open woodland plots and in open orchard plantings. The effects of deer rubbing on, and destroying, mesh tree-guards and stakes has been very significant and remains and costly and time-consuming task to recover from. We will be analyzing whether these effects actually outweigh damage from browsing from which most plants seem able to recover. We noted relatively significant drought stress among trees planted in our orchards and experienced some mortality from this. Such effects were greatest for the chestnut trees. In our direct seeding trials we planted nearly a thousand pawpaw seeds in our experimental plots at Pomerene Forest Laboratory. Emergence was slow and patchy but most plots showed several emerging seedlings in 2023. A full inventory of establishment will be made in 2024.

2024: With the relocation of the project’s PI to the Falkland Islands, the monitoring of these sites planned for 2024 did not occur.  Through discussions with the original PI (G.M. Davies, forest ecologist) the current PI (J.C. Scheerens, horticulturist), and two additional colleagues (S. Power, faculty and K. Packer, staff, School of Environment and Natural Resources) we have developed a plan to rekindle efforts in this portion of the study. Both additional individuals are forest ecologists, and both are enthusiastic about becoming more involved with evaluating the efficacy of pawpaw plot establishment parameters. Ms. Packer assisted in the original plot establishment efforts. The site will be visited first in spring and then data will be collected throughout the 2025 season.

2025 (Final): Critical question 5 from the grant proposal was – “How does pawpaw establishment vary as a function of plant material and environmental conditions”? 

As mentioned above, progress on this grant objective was re-invigorated in 2025. The pawpaw establishment sites were visited by the current project P.I. and were routinely monitored by Dr. Simon Power and others, primarily Dr. Kirsten Packer, the principal scientist now undertaking this study. What is offered below as a final report is Dr. Packer’s synopsis of establishment plot progress from 2022-2025. The following text and graphics should be attributed to her.

In autumn 2025 we monitored and maintained the woodland pawpaw plots at Pomerene Forest Laboratory, including surveying for growth of live bare root stems and grafted varietals, and seed germination/establishment.

In 2022, two months after initial planting in May, overall survival of bare root pawpaw stems (no variety) remained high (ca. 98%, Figure 4). In the years that followed (2023 – 2025) some stem death was recorded, however, overall survival remained high (ca. 80%) indicating a good degree of success in the outplanting of bare root stock. Moreover, surviving stems showed a significant increase in growth between 2022/2023 and 2025 (Figure 5), with some individuals reporting a 200 cm increase in height over the period. Unsurprisingly, anecdotal observations of environment-growth relationships indicated a tendency for greater growth in wetter, bottomland plots compared to drier upland or south-facing plots. Although we saw good evidence of establishment, we did not note the occurrence of any flowering/fruiting maturity.

By 2025, survival of the container-grown grafted varietals was 33.3 ± 24.8 % (mean ± SD), with a mean height limited to 66.1 ± 20.9 cm, compared to 84.2 ± 27.6 cm for bare root stock. Neither original (2022) or replanted (2023) grafted varietals showed much success, with several instances of the graft dying back and the stem resprouting from the non-varietal root stock.

The efficacy of tree protection (mesh guards) remains unresolved. Due to the extensive damage deer caused to the guards, often ripping them off and crushing pawpaw stems in the process, we removed many and did not replace any. Thus, we were unable to accurately calculate percent survival of stems with and without guards.

In May and June 2022 direct seeding of pawpaw was implemented at each woodland plot. From the center of the circular plots, a 1m² quadrat was placed in the northwest and southeast quarters and 100 seeds were planted per quadrat (Figure 6). By 2025 we found 7.3 ± 15.2 % (mean ± SD) survival in the northwest quadrat and 10.6 ± 16.5 % in the southeast. The mean height of the tallest seedling per quadrat was 34.8 ± 10.8 cm in the northwest quadrat and 31.4 ± 3.9 cm in the southeast.

Figure 4

Figure 5

Figure 6

 Question 6 ~ Pawpaw patch management and forest health

2022: All pawpaw patches and planting sites were monitored for woodland structure and composition prior to the application of invasive control and understory thinning treatments. Evaluation of the effects of thinning and invasive control treatments will commence in summer 2023

2023: We remonitored all our woodland plots across multiple experimental sites in Ohio. We recorded the abundance and composition of all seedlings and saplings regenerating in the plots. Several plots that were heavily invaded prior to our treatments showed dramatic changes in their light environment and this promoted vigorous responses from the herbaceous flora. In some settings establishing species were mostly native but Japanese stiltgrass emerged as a significant secondary invasive in a number of plots particularly where these were located adjacent to existing tracks. Analysis of our data is on-going and forms part of an MS project to be completed in the summer of 2024. Initial results suggest good initial control of invasives but that in most locations there was some substantive regeneration from the existing seedbank and resprouting.

2024: Data for the MS project has been gathered and is under review. If appropriate, it will be analyzed in conjunction with data taken on the same sites prior to the onset of this study. It is our goal to ascertain the ecological condition of woodland plots prior to intervention, to measure the immediate effects of intervention, and then to assess the longevity of intervention efforts and its effect on plot productivity and forest health.

2025 (Final): the 6^th critical question in the original proposal was – “Does manipulating pawpaw patches to enhance productivity improve woodland health”?  With the departure of Dr. Davies (forest ecologist) and his graduate student assigned to this project, we made no additional progress on completing this objective in 2025. We are, however, in possession of all raw data collected and in collaboration with Dr. Davies we will revisit the study with the aim of answering the critical question as time allows.  

Question 7 ~ Woodland pawpaw as a premium product

2022: Questionnaires and surveys were distributed to consumers and growers at the Ohio Pawpaw Conference and the Ohio Pawpaw Festival to evaluate how pawpaw certification schemes and growing conditions (e.g. organic v. conventional) would affect consumers willingness to pay for the crop. This work will continue through 2023 focusing on preferences for woodland/wild versus orchard grown fruit.

2023: We completed a discrete choice experiment with 182 pawpaw consumers in 2023. This experiment focused on ice cream formulations and sought to converge the sensory and consumer behavior literature. The 2023 experiment complements the efforts exerted in 2022 when 237 consumers were asked to evaluate alternative pawpaw extract products along with extrinsic product attributes. A scientific manuscript summarizing and discussing the findings is currently being prepared. Preliminary results suggest that locally produced items originating from independent growers or growers’ alliances tend to be valued higher compared to large-scale enterprises. Certification is also a relevant attribute. We have not observed statistically significant differences in consumers’ perceptions between value-added items produced from foraged fruit versus orchard-grown fruit.

2024: The data collected via surveys during the Ohio Pawpaw Conference and Ohio Pawpaw Festival (2022) and during the Ohio Pawpaw Festival (2023) was cleaned and submitted to statistical analysis following the Discrete Choice Experiment methodology.

Results from the pawpaw purée choice experiment study were consolidated, reinforcing the preliminary results obtained and reported above for the 2023 calendar year. The representative sample of pawpaw purée consumers (n=237) indicated a preference for sweeter pawpaw extracts versus natural (unsweetened) purée. Results also suggest statistically significant preferences for commercial pawpaw purée products that adhere to verified sustainability standards. Consumers’ preferences for the entity responsible for verifying the sustainability standards showed no statistically significant differences among NAPGA (North American Pawpaw Growers Association), USDA, or an independent third-party certifier. Furthermore, results show that consumers are less likely to buy pawpaw purée products if produced by large-scale enterprises. This finding suggests business opportunities for independent pawpaw growers or growers’ alliances to develop and market pawpaw purée products. The representative sample of consumers showed no preference distinction for products from independent growers versus growers’ alliances. Finally, we failed to reject the hypothesis of statistical differences in stated preferences for purée produced using fruit from woodland patches versus orchards. A scientific publication documents these findings (Hoffman et al. 2025). The final results of this study were reported at the 2024 Ohio Pawpaw Conference, led and organized by the project PD and Co-PIs.

The second discrete choice experiment focused on consumers’ preferences for different formulations of pawpaw-based frozen desserts in addition to extrinsic product attributes. To allow future comparisons, a similar set of extrinsic attributes was used in this experiment: verification of sustainability claims (levels: absent and verified), verification entity (levels: NAPGA, USDA, and Third-party certifier), organic certification (levels: yes or no), and price (levels: $3.99/pint, $5,99, and $7.99). Results showed that the ice cream dessert formula was consistently preferred versus the alternative sorbet formula. That is, the estimate for the sorbet formula level was negative and statistically significant versus the baseline ice cream formula. Price was also negative and statistically significant, reinforcing the notion that consumers’ welfare decreases as product prices increase. All other attributes included in the experiment were not statistically significant, suggesting that consumers emphasize taste when products resemble hedonic foods (Raghunathan et al. 2006, Okada 2005). Preliminary results from the dessert experiment were also reported at the 2024 Ohio Pawpaw Conference. The results from this latter experiment are yet to be documented in a scientific manuscript.

2025 (Final): The seventh and final critical question in the original proposal was – “Can woodland-grown pawpaw command a premium”?  Work on this objective was completed in 2024 so last year’s report will also serve as the final report.

The data collected via surveys during the Ohio Pawpaw Conference and Ohio Pawpaw Festival (2022) and during the Ohio Pawpaw Festival (2023) was cleaned and submitted to statistical analysis following the Discrete Choice Experiment methodology.

Results from the pawpaw purée choice experiment study were consolidated, reinforcing the preliminary results obtained and reported above for the 2023 calendar year. The representative sample of pawpaw purée consumers (n=237) indicated a preference for sweeter pawpaw extracts versus natural (unsweetened) purée. Results also suggest statistically significant preferences for commercial pawpaw purée products that adhere to verified sustainability standards. Consumers’ preferences for the entity responsible for verifying the sustainability standards showed no statistically significant differences among NAPGA (North American Pawpaw Growers Association), USDA, or an independent third-party certifier. Furthermore, results show that consumers are less likely to buy pawpaw purée products if produced by large-scale enterprises. This finding suggests business opportunities for independent pawpaw growers or growers’ alliances to develop and market pawpaw purée products. The representative sample of consumers showed no preference distinction for products from independent growers versus growers’ alliances. Finally, we failed to reject the hypothesis of statistical differences in stated preferences for purée produced using fruit from woodland patches versus orchards. A scientific publication documents these findings (Hoffman et al. 2025). The final results of this study were reported at the 2024 Ohio Pawpaw Conference, led and organized by the project PD and Co-PIs.

The second discrete choice experiment focused on consumers’ preferences for different formulations of pawpaw-based frozen desserts in addition to extrinsic product attributes. To allow future comparisons, a similar set of extrinsic attributes was used in this experiment: verification of sustainability claims (levels: absent and verified), verification entity (levels: NAPGA, USDA, and Third-party certifier), organic certification (levels: yes or no), and price (levels: $3.99/pint, $5,99, and $7.99). Results showed that the ice cream dessert formula was consistently preferred versus the alternative sorbet formula. That is, the estimate for the sorbet formula level was negative and statistically significant versus the baseline ice cream formula. Price was also negative and statistically significant, reinforcing the notion that consumers’ welfare decreases as product prices increase. All other attributes included in the experiment were not statistically significant, suggesting that consumers emphasize taste when products resemble hedonic foods (Raghunathan et al. 2006, Okada 2005). Preliminary results from the dessert experiment were also reported at the 2024 Ohio Pawpaw Conference. The results from this latter experiment are yet to be documented in a scientific manuscript.

Final Report References:

Brannan, RG. 2025. A review of the postharvest biochemical, biophysical and biological properties of 61 cultivars of North American pawpaw (Asimina triloba) fruit. J. Sci. Food Agric.

https://doi.org/10.1002/jsfa.70115

Brewbaker J, Kwack B. 1963.The essential role of calcium ion in pollen germination and pollen tube growth. Amer. J. Bot. 50:859-865.

https://doi.org/10.2307/2439772

Campbell T, Kalcsits L. 2024. Strategies to overcome biennial bearing in apple – A review. Eur. J. Agronomy. 158:127213.

https://doi.org/10.1016/j.eja.2024.127213

Charbonnier F, Roupsard O, leMaire G, Guillemot J, Casanoves F, Lacointe A, Vaast P, Allinne C., Audebert L, Cambou A, Clement-Vidal A, Defrenet E, Duursma RA, Jarri L, Jourdan, C, Khac E, Leandro P, Medlyn BE, Saint-Andre L, Thaler P, Van Den Meersche K, Aguilar AB, Lehner P, Dreyer A. 2017. Increased light-use efficiency sustains net primary productivity of shaded coffee plant in agroforestry system. Plant Cell and Env. 40:1592-1608.

https://doi.org/10.1111/pce.12964

Clegg KM. 1956. The application of the anthrone reagent to the estimation of starch in cereals. J. Sci. Food Agric. Jan. 1956. https://doi.org/10.1002/jsfa.2740070108

Francino SE. Variation in Pawpaw (Asimina triloba) Cultivar Productivity and Quality Across a Biogeographic Gradient in Ohio, M.S. Thesis, Ohio State University, 2019.

Goodrich KR, Ellis I, DeHaas A, Senski R, Savage, J. False advertising with fermented scents: Floral mimicry in pawpaw (Asimina triloba: Annonaceae) pollination. Intl. J. Plant Sci. 184:485-497.

https://doi.org/10.1086/725107

Hoffman S, Signorini G, Soldavini AM, Simons CT. 2024. Implementing a discreet choice experiment within consumer sensory evaluation to better understand purchase intention. J Sensory Studies 40:e70006.

https://doi.org/10.1111/joss.70006

Kishore K, Rupa TR, Samant D. 2021. Influence of shade intensity on growth, biomass allocation, yield and quality of pineapple in mango-based intercropping system Scientia Horticulturae 278:109868.

https://doi.org/10.1016/j.scienta.2020.109868

Lora J, Herrero M, Hormaza JI. 2012. Pollen performance, cell number and physiological state in the early-divergent angiosperm Annona cherimola Mill. (Annonaceae) are related to environmental conditions during the final stages of pollen development. Sexual Plant Reproduction 25: 157-167.

https://doi.org/10.1007/s00497-012-0187-2

Lora J, Laux T, Hormaza JI. 2018. The role of the integuments in pollen tube guidance in flowering plants. New Phytologist 221:1074-1089.

https://doi.org/10.1111/nph.15420

Losada JM, Hormaza JI, Lora, J. 2017. Pollen-pistil interaction in pawpaw (Asimina triloba), the northernmost species of the mainly tropical family Annonaceae.

https://doi.org/10.3732ajb.1700319

Magar ST, Lowe J, Pomper KW, Crabtree S. 2025a. Diversity and abundance of insect pollinators in pawpaw (Asimina triloba), an emerging commercial fruit crop. HortScience 60(7):S490. Abstract.

https://doi.org/10.21273/HORTSCI.60.7S.S1

Magar ST, Pomper KW, Lowe J, Crabtree S. 2025b. Impact of Self- and Cross pollination on fruit and seed set in pawpaw in Kentucky. HortScience 60:771-774.

https://doi.org/10.21273/HORTSCI18437-24

Nielson GH, Neilson D. 2003. Nutritional Requirements of Apple. Ch.12, pp. 267-302 in Apples: Botany, Production and Uses (DC Ferree and IJ Warrington, eds.) CAB International, Cambridge MA.

Okada EM. 2005. Justification effects on consumer choice of hedonic and utilitarian goods. J. Marketing Res. 42:43-53.

https://doi-org/10.1509/jmkr.42.1.43.56889

Pereira MCT, Crane JH, Montas W, Nietsche S, Vendrame WA. 2014. Effects of storage length and flowering stage of pollen influence its viability, fruit set and fruit quality in ‘Red’ and “Lessard Thai’ sugar apple (Annona squamosa and ‘Gefner’ atemoya (A. cherimola X A. squamosa) Sci. Hortic. 178:55-60.

https://dx.doi.org/10.1016/j-scienta.2014.08.004

Pomper KW, Layne DR. 2005. The North American pawpaw: botany and horticulture.  Horticultural Rev. 31:351-384.

https://doi.org/10.1002/9780470650882.ch7

Pomper KW, Crabtree CB, Lowe DR, Peterson RN. 2008 Flowering and fruiting characteristics of eight pawpaw [Asimina triloba (L) Dunal] selections in Kentucky. J. Amer. Pomological Soc. 62:89-97.

https://doi.org/10.71318/apom.2008.62.3.89

Pomper KW, Crabtree CB, Lowe DR, Peterson RN, Masabni J, Wolfe, D. 2008b. The Kentucky Pawpaw Regional Variety Trial.  J. Amer. Pomological Soc. 62:58-69.

https://doi.org/10.71318/apom.2008.62.2.58

Pomper KW, Lowe JD, Crabtree SB, Collins LA. 2009. Clonality of pawpaw (Asimina triloba) patches in Kentucky.  J. KY Acad. Sci. 70:3-11.

https://doi.org/10.3101/1098-7096-70.1.3

Raghunathan R, Naylor RW, Hoyer WD. The unhealthy = tasty intuition and its effects on taste inferences, enjoyment and choice of food products. J. Marketing 70:170-184.

https://doi-org/10.1509/jmkg.70.4.170

Sathya R, Rasane P, Singh J, Kaur S, Bakshi M, Gunjal M, Kaur J, Sharma K, Sachan S, Singh A, Bhadariya V, Mahato DK. 2024. Strategic advances in the management of browning in fruits and vegetables. Food and Bioprocess Technol. 17: 325-350. https://doi.org/10.1007/s11947-023-03128-8

Saunders RMK. 2012. The diversity and evolution of pollination systems in Annonaceae. Bot. J. Linnean Soc.169:222-244

https://doi.org/10.1111/j.1095-8339.2011.01208.x

Saunders RMK. 2020. The evolution of key functional floral traits in the early divergent angiosperm family Annonaceae. J. Systematics and Evolution 58:369-392

https://doi.org/10.1111/jse.12645

Singleton VL, Rossi JL.  1965.  Colorimetry of total phenolics with phosphomolybdic-phosphtungstic acid reagents.  Amer. J. Enol. Vitic. 16:144-158. https://doi.org/10.5344/ajev.1965.16.3.144

Tekie S, Zainali S, Zidane TEK, Lu SM, Guezgouz M, Zhang J, Amaducci S, Dupraz C, Campana PE. 2025. Unraveling the crop yield response under shading conditions through the deployment of a drought index: A meta-analysis. Energy Nexus 19:100523.

https://doi.org/10.1016/j.nexus.2025.100523

Tilley A, McHenry MP, McHenry JA, Solah V, Bayliss K. 2023. Enzymatic browning: The role of substrates in polyphenol oxidase mediated browning. Current Research in Food Science 7: 100623. https://doi.org/10.1016/j.crfs.2023.100623

United States Department of Agriculture 2022.  Good Manufacturing Practices (GMP) Audit Standard https://www.ams.usda.gov/sites/default/files/media/GMPAuditStandard.pdf

Weatherspark 2026.

https://weatherspark.com/m/17263/4/Average-Weather-in-April-in-Columbus-Ohio-United-States#google_vignette

Westwood MN. 1993. Temperate-zone pomology and culture (3rd ed). Timber Press, Portland, Oregon.

Willson MF, Schemske DW. 1980. Fruit production and floral display in pawpaw (Asimina triloba). Bull. Torrey Bot. Club. 107:401-408.

https://doi.org/1032307/2484160

Young DR. 1987. Daily and seasonal variations in the water relations of the understory tree, Asimina triloba. Acta Oecologica – Oecologia Plantarum 8:59-68 abstract