Effects of Perimeter Trap Crop Pollen on Reproduction in Butternut Squash (Cucurbita moschata) (2024)

Experimental design.

This experiment was conducted at the University of Massachusetts at Amherst greenhouses from February through Aug. 2005. On 16 Feb., 240 butternut squash plants were planted (cultivar ‘Waltham Butternut’; Johnny's Selected Seeds, Winslow, ME) and transplanted after 1 month to 22-L pots containing Metromix 360 soil (The Scotts Co., Marysville, OH) with 9.5 g per pot of slow-release fertilizer (Osmocote 14N–4.2P–11.6K; Scotts-Sierra Horticulture Products Co., Marysville, OH). Twenty blue hubbard plants were grown as pollen sources.

The experiment had 12 treatment combinations with 20 replicates each for 240 plants total. Butternut squash were randomly assigned to three pollen composition treatments (butternut pollen, blue hubbard pollen, or mixed pollen) crossed with four pollen amounts [low (two applications), medium (four applications), high (six applications) and saturation]. In the low, medium, and high treatments, pollen was applied to stigmas using a thin wire applicator (Impex Systems Group, Miami, FL) with a 1-mm diameter. One application contained 109 ± 8 butternut pollen grains (X ± se, n = 50) or 58 ± 4 blue hubbard grains (n = 40). Because the pollen requirements of C. moschata were not known, we chose these pollen amounts for two reasons. First, they closely approximate the pollen deposited by one, two, or three visits by a single squash bee (250 grains) or honeybee (Apis mellifera; 232 grains) to the congener C. pepo in a field study (Tepedino, 1981). Second, these amounts are comparable to those used in previous studies examining effects of pollen load on fruit maturation in C. pepo (Winsor et al., 1987). For the saturation treatment, pollen was applied to one stigma lobe using a paintbrush depositing an average of 4088 ± 711 butternut grains (n = 10) or 2651 ± 657 blue hubbard grains (n = 8). Although butternut stigmas have three lobes, only one lobe was covered because this applied nearly an order of magnitude more pollen than the next highest treatment and nearly 40 times more pollen than the maximum seed production (Fig. 1). The differences in the number of pollen grains applied in the blue hubbard compared with butternut treatments were likely the result of the larger size of blue hubbard pollen grains (butternut 130.15 ± 0.87 μm, blue hubbard 150.35 ± 1.78 μm, n = 20 grains, P < 0.0001). Although different pollen amounts may introduce a treatment bias in terms of pollen number, field pollen deposition per stigma area by bees may also be lower for blue hubbard compared with butternut as a result of pollen size. Thus, our experimental treatments control for pollen deposition area rather than number.

Treatments began on 2 May 2005 and continued until 29 July 2005. Butternut and blue hubbard squash are both monoecious cucurbits and produce actinomorphic, self-compatible unisexual flowers that last a single day. Three newly opened butternut and blue hubbard staminate flowers were collected between 0600 and 0700 hr every morning to ensure pollen freshness. Applicators and paintbrushes were cleaned with 100% ethanol between applications to prevent contamination. Up to two flowers were pollinated per plant. We concentrated our efforts on treating more plants rather than more flowers per plant because plant is the unit of replication and so pollinating additional flowers per plant would not increase our statistical power to test treatment effects. We also note that in the field, the congener C. pepo (pumpkin) produced an average of only 3.4 and 5.4 female flowers per year across a 2-year study (Stapleton et al., 2000), and potted butternut in the greenhouse produced 4.4 + 0.52 female flowers over a 3-month period (n = 15; Hladun, unpublished data). Thus, low female flower and fruit production is not unusual. Fruits were collected on 5 Aug. 2005. Of 240 butternut plants, 123 plants had at least one female flower treated; 39 of these plants produced one fruit and three plants produced two fruits. The responses measured were fruit set (fruit or aborted), fruit fresh weight, seed number, proportion developed seeds, and dry weight per seed.

Data analysis.

Two-way multivariate analyses of variance (MANOVAs) and subsequent univariate analyses of variance (ANOVA) using SAS (2003; SAS Institute, Cary, NC) were used to determine the effects of pollen composition, pollen amount, and their interaction on butternut squash fruit and seed production. All data were averaged within plant to use plant as the unit of replication. Logistic regression was used to predict the probability of fruit set, including the continuous variable pollination date as a covariate. Because fruit set was relatively rare, plants that set fruit in either flower were counted as having fruit. Although fruits have an inhibitory effect on subsequent fruit production in the congener C. pepo (Stephenson et al., 1988), in our study, only three plants produced multiple fruits. Removing these plants did not change our results, and we report results including all plants that produced fruit.

Pollen composition.

Pollen composition affected butternut reproduction (MANOVA, P < 0.0001), largely through seed production. Butternut pollen and mixed pollen produced, respectively, 53% and 38% more seeds and 75% and 83% higher seed weights than blue hubbard pollen (ANOVA, seed number: P < 0.006; seed weight: P < 0.0001; Fig. 1A, C). Pollen composition did not affect fruit weight or the proportion of developed seeds per fruit (ANOVA, P > 0.13 for both) and did not affect fruit set [logistic regression, P = 0.57; butternut pollen: 16 of 48 plants produced fruit (33%), mixed pollen: 14 of 33 (42.4%), blue hubbard pollen: 11 of 42 (26.2%)].

Pollen amount.

Pollen amount affected butternut reproduction (MANOVA, P < 0.0001), again driven by seed production. Saturating amounts of pollen produced two to six times more seeds than lower pollen amounts (ANOVA, P < 0.0001, Fig. 1B). Pollen amount did not affect fruit weight, weight per seed (Fig. 1D), or the proportion of developed seeds per fruit (ANOVA, P > 0.10 for all). Pollen amount had a marginally significant effect on fruit set with the lowest pollen amount producing half as many fruit as the other treatments [logistic regression, P = 0.08; low: six of 34 plants produced fruit (17.6%), medium: 11 of 24 (45.8%), high: 10 of 29 (34.5%), saturation: 14 of 36 (38.9%)].

Interaction between pollen composition and pollen amount.

Pollen amount and composition had additive effects on butternut reproduction (MANOVA, interaction term: P = 0.93).

Pollination date.

Plants pollinated later in the season were significantly less likely to produce fruit (logistic regression, P < 0.0001).