This study was conducted at the University of Kentucky horticulture research farm in Lexington, Kentucky in the summer of 2018 (plant hardiness zone 6b). This approximately 50-hectare research farm is surrounded by a landscape of urban and rural development. The farm is split into organic and conventional halves where a wide variety of perennial and annual crops are grown. Our research focused on a blackberry plot adjacent to conventional plots of apple orchard and wine-grape vineyards. The plot consisted of six rows, each containing three plants each of mature Triple Crown, Hull, and Chester varieties for a total of nine plants per row. The blackberries were planted with a 1.8 m tall trellis in rows with 1 m spacing within rows and 3.5 m wide alleys between rows. Each row had a length of 25 m. Varieties were planted within row in a randomized complete block design. In June of 2018, each row was hand-weeded, blackberry canes were trimmed back, and were then maintained on a trellis system. We chose to perform this experiment on a small-scale plot in order to imitate the small-scale production of the fruit found in Kentucky. In this state, most blackberry production consists of smallholder farms that grow blackberries in limited quantities for self-harvest (e.g., pick-your-own), local markets, or for use as ingredients in value-added products.
Blackberry Fine-Mesh Netting Exclusion
In late June, we began an experiment with the goal of comparing the effectiveness of a fine-mesh exclusion netting system (ProtekNet) and an organic insecticide spray regime (Entrust™—Dow AgroSciences LLC, Indianapolis, IN, USA) for the control of multiple pests of blackberry. This organic spinosad is the most, and arguably the only, efficacious pesticide available to organic blackberry producers in the United States for the control of Drosophila suzukii
]. We began this experiment after petal fall, when roughly 95% of blackberries had set fruit. This enabled us to exclude fruit-damaging pests without inhibiting pollination. Some fruits on treated blackberry canes had just begun to color, but none had yet begun to turn black, a stage in which they would be vulnerable to attack by insect pests. We netted complete rows to test the effectiveness of this pest control strategy on a scale representative of implementation by small-scale commercial growers. For this reason, and due to the size constraints of the available blackberry plot, we did not subset rows and did not include a no-management control. On alternating rows, we covered all nine blackberry plants in individual rows with a knitted polyethylene mesh ProtekNet (0.85 mm by 1.4 mm mesh size) 6.3 m wide, acquired from Dubois Agrinovation (Saint-Rémi, Quebec, QC, Canada). In each of the three netted rows, netting was held above the blackberry canopy by the existing t-frame trellis system and weighed down at ground level with cement pavers to protect against wind and to prevent intrusion from birds and insects (Figure 1
). It should be known that the integrity of the exclusionary netting in two rows was compromised for a period of time shorter than 12 h due to a severe weather event which partially removed the netting on two rows. Uncovered rows acted as a representation of the standard organic protection regime. Each row was treated with spinosad applied three times in July using a backpack mist sprayer (Stihl SR450 Backpack Mist Sprayer, STIHL Inc., Virginia Beach, Virginia, VA, USA). We aimed to perform each insecticide application in roughly seven-day intervals. However, we allowed flexibility in this schedule and only performed pesticide sprays under low wind-speed-conditions to minimize drift of Entrust™ spray.
To determine the effect of the management systems on SWD, we placed baited traps (40 mL apple cider vinegar and 10 mL pure laboratory grade ethanol) in the center of each row directly within the blackberry canopy and underneath row covers. Baited traps were constructed from a lidded 473-mL red plastic drinking cup (Dart Container Corporation P16R, Mason, MI, USA). Two rows of twelve perforated holes encircled each cup at five and six centimeters from the base to act as semi-selective entrance points for arthropods. As an additional olfactory attractant, one pair of Pherocon SWD Peel-Pak, broad spectrum Drosophila lures was hung inside each trap (Product 5001-1P—Trécé Incorporated, Adair, OK, USA). Traps were attached to trellis posts at a height of 1.5 m and secured from wind-disturbance using zip-ties. Trap contents were collected weekly, then immediately refilled with bait solution. Broad spectrum lures were replaced once after 14 days, and trap data was collected for a total of four weeks across the month of July. The number of SWD, as well as the number of other vinegar flies per trap, were quantified under a stereoscope.
To compare the effects of the two management systems on SWD and all other vinegar flies collected in baited traps, we conducted linear mixed models (LMM) with management treatment (fine-mesh exclusion versus spinosad insecticide) and sampling week as fixed effects with the function ‘lmer’ in the R-package lme4 (Program R 3.5.1). Each individual trap was treated as a random effect to account for multiple repeated measurements across the weeks. Each proximal pair of rows was treated as a random effect to block row pairs together and reduce potential microclimatic biases. Vinegar fly data was then square root-transformed to achieve normality. We also uncovered a significant interaction between management treatment and sampling week by LMM analysis, so we conducted a paired t-test to compare the differences in SWD and vinegar fly presence within each of the four weeks.
To determine SWD infestation of berries during peak harvest, four weeks after initial netting, we collected samples of 20 overripe (non-marketable) berries per variety per row (three samples per row; nine per treatment). These berries were placed in an incubation chamber (Percival I-66VL two-door incubator, Percival Scientific Inc., Perry, IA, USA) inside of rearing containers (16 oz. Dart Solo MicroGourmet 16NW-0007, Mason, MI, USA) with mesh tops. The bottom of the rearing container was covered with sand and a 2.5 cm square sticky trap was suspended from the top of the container to immobilize emergent flies. After 20 days, we counted the number of SWD adults that emerged from the fruits under a stereoscope. To compare the effects of each management system, we conducted an LMM on log-transformed SWD emergence with management treatment as a fixed effect within the model. Each proximal pair of rows was treated as a random effect to block row pairs together. Each blackberry variety was treated as a random effect to account for any differences between varieties.
To document the abundance of Japanese beetles and green June beetles on blackberry canes, we used visual surveys. An observer, starting from one end of each experimental row, moved slowly under the exclusion netting or adjacent control plots, choosing the largest fruit-bearing cane per plant on six total plants per row. This was done in a way to minimize the disturbance of the beetles and to avoid double-counting individual beetles. Each selected fruit-bearing cane was scanned from top to bottom for the number of each beetle species present on fruits, stems, or leaves. This survey was conducted for each row on two occasions on 17 July and again on 25 July. To compare the effects of each management system, we log-transformed beetle abundance and conducted LMM with management treatment and sampling week as fixed effects. Each row was treated as a random effect to account for multiple repeated measurements across weeks. Each proximal pair of rows were treated as a random effect to block row pairs together.
At the completion of the netting experiment, once netting materials were removed, we walked each row and counted the number of bird fecal droppings on all leaves, fruits, and stems to estimate bird activity in rows with fine-mesh exclusions and organic insecticide treatments. To compare the effects of each management system on bird intrusion, we conducted a paired t-test with each proximal pair of rows as the grouping factor and compared the difference in bird droppings between the two management systems.
Following the ripening of fruits, we harvested blackberries each week for six weeks. We denoted unmarketable berries as those with damage or deformation and marketable berries as undamaged, ripe fruits fit for direct-to-market sale. We pooled yield measurements across weeks and across varieties within rows, then compared square root-transformed marketable, unmarketable, and total yields with LMM. Management treatment was our fixed effect within the model. Each proximal pair of rows was treated as a random effect to block row pairs together.
To assess the quality of marketable yield, we compared the sugar content of berries with a PAL-Easy ACID4 pocket sugar and acid meter two times immediately after harvest in early and late July (Atago Co., Ltd., Tokyo, Japan). We compared the sugar content of marketable berries with LMM with management treatment as a fixed effect and row pair, sampling week, and blackberry variety as random effects within the model. For all LMM analyses performed in this study, we confirmed the normality of each set of data with a Shapiro–Wilk normality test of model residuals.