Increasing the sustainability of Massachusetts cranberry production through cultural management of the bog habitat

The project participation team met in early 2005 and discussed project design. At that meeting, we confirmed that the primary components of this project would be 1) sanding; 2) pruning; 3) irrigation scheduling; 4) nutrient management; and 5) drainage.

Sanding and pruning. As originally proposed, we compared sanding and pruning in a replicated study of these two practices (see below and Figure 1). A Masters student, Brett Suhayda was hired under the direction of Dr. Justine Vanden Heuvel [Dr. Carolyn DeMoranville took over supervision when Vanden Heuvel relocated to Cornell University at the end of 2006]. He carried out this replicated study as part of his degree requirement. An additional replicated study comparing the interaction of pruning and nitrogen fertilization was conducted by Dr. Hilary Sandler on two commercial farms.

Growers established pruning demonstrations on their farms at intervals 2, 3, or 4+ years after sanding as described below. Over the 10 sites, we had multiple repetitions of the following protocols: 1) sand-no treatment-prune; 2) sand-no treatment-no treatment-prune; 3) four or more years since sanding, then prune. The first and second of these protocols mimic the substitution of pruning in the most common sanding cycles currently used.

At one site we also had a mowed area on the bog that was compared to sanded or pruned areas. This site was established on Matt Beaton's farm. He has been looking at the alternative idea of mowing the 'Stevens' variety periodically to renew the canopy (similar to practices in lowbush blueberry cultivation). Based on these preliminary data, 5 additional grower sites are participating in a grower organization-funded study of mowing as part of 'Stevens' management that began in 2008 under the direction of DeMoranville and Sandler. Sanding, mowing, and pruning and their interaction with nutrient management were the topics of the summer 2008 bogside demonstration field event at a participant farm (Garretson).

Demonstration experiments of pruning after sanding - protocol. These demonstrations were designed to show that at certain intervals after sanding, pruning could replace the next sanding event. Growers identified 10 beds where sand had last been applied 2, 3, or 4+ years previously. If they had a pair of similar beds that had been sanded in the same year, one of the pair was pruned for the demonstration and the other was not. If there was only one bed available, then part of the bed was pruned and compared to the unpruned area. Beds were pruned in 2005 or 2006 and followed through 2008. Yield and vine density were compared in pruned and unpruned areas. Yield was evaluated by collecting and weighing fruit from six 1-ft2 areas (as described in the pruning study below) chosen randomly in each treatment and control pair. Vine density was evaluated by counting uprights collected from six randomly selected areas (using a 6-inch ring template) of each treatment or control (as described below for the sanding/pruning study).

In initial grower surveys in January 2006 many identified pest management as a benefit of sanding. Previous research on dodder (a key weed pest) and tipworm has shown that, unless deep layers of sand are applied, pest control is inconsistent or absent. As part of this project Dr. Anne Averill examined the potential of sanding to control cranberry fruitworm, the key insect pest in Massachusetts cranberry. If, as is the case with previously studied pests, sanding is not effective in fruitworm management, this would remove a barrier to grower adoption of reduced sanding management.

Does sanding suppress cranberry fruitworm? Cranberry fruitworm larvae in hibernacula (this stage overwinters on the bog floor) were placed under different conditions in semi-natural and natural settings to determine the direct effect of how a change in cultural practice (increased bed sanitation, change in sand and trash (fallen leaf duff) levels) might make the system more/less favorable to this key insect pest.
Field trials were set up in September and October 2006 using hibernacula (there were two hibernacula treatments -- formed of either sand or trash) placed in plots covered with 2 cm of sand, cranberry leaf trash, or sand/trash mix (four replicates). We included the trash treatment since we hypothesized that phenolics in the cranberry leaf trash might reduce fungal inoculum and thus enhance overwintering survivorship. The hibernacula were collected in the spring of 2007 and assessed for survivorship and fungal infection. Greenhouse trials were established to evaluate the possible ramifications of eliminating the sanding practice on cranberry fruitworm populations. Sand hibernacula (10 replicates of 10 hibernacula) were placed under various (0, 1.5 cm, 3.0 cm, 4.5 cm, and 6.0 cm) sand layers in 5-inch pots and were held in a cold frame greenhouse over the winter. A portion was assessed for survivorship and fungal infection in the spring of 2007. The remainder was held until May 2007 when moth emergence was assessed.

We conducted two replicated field trials, one looking at the interaction of pruning with nitrogen fertilizer which began prior to the start date of this SARE grant, but was completed during this project and was used as part of the educational effort regarding pruning and the other (comparing sanding and pruning) that was entirely conducted as part of this project.

Pruning and nitrogen interaction field trial. Two commercial 'Stevens' cranberry farms in southeastern Massachusetts were utilized in this study and are identified by location: South Carver (SC, Gilmore) and Rochester (RCH). At each site, the study was set out in a randomized complete block design with treatments arranged in a split-plot with three replicates. Pruning severity (none, low, medium, and high) was the main effect and nitrogen (N) rate (0, 50, 100, and 150 lb/A) was the sub-plot. At each site, any pest or horticultural management practices (e.g., irrigation and frost protection), except fertilization, applied to the commercial bed was applied to the experimental study area.

Pruning treatments. Pruning treatments were conducted each year in early spring (generally April) from 2003 to 2006. Pruning was accomplished with rotating head pruners with multiple slats with ca. 10-12 equally spaced knives per slat. The use of two different pruners, which were owned and modified by the individual growers, necessitated different approaches to create low, medium, and high pruning severities. At SC, the pruning treatments were accomplished by pruning the plots once, twice, or three times (low = one pass, moderate = two passes, and heavy = three passes) while keeping the pruning head at the same height. At RCH, each plot was pruned once but the head was lowered into the vine canopy at three equally spaced settings to prune at three different depths (low = highest head height, moderate = middle head height, and heavy = lowest head height). Vine clippings from each plot were raked, placed into black plastic bags and weighed in the field using a hanging digital scale.

Nitrogen treatments. Granular formulations of fertilizer for all treatments were applied at both sites in four equal portions (equal four-way split) in all years. Each year the fertilizer applications were made at the roughneck, bloom, fruit set, and bud set stages. We varied the amount of N applied to the vines while holding phosphorus (P) and potassium (K) applications equal among treatments. Plots designated as 0 N received applications of 0-25-25 for a total rate of 22 lb/A P and 42 lb/A K per year, applied in an equal 4-way split. N was applied to the low, medium, and high N plots four times at the rates of 12.5, 25, and 37.5 lb/A, respectively. Fertilizer was applied to these N plots using combinations of 20-10-10, 21-0-0, and 0-25-25 in adjusted proportions so that each plot received the N rate and a total seasonal P and K rate matching that in the 0 N plots.

Upright evaluation. To assess the effect of pruning and N rate on upright number and biomass, vine samples were collected yearly at each site. One vine sample was collected from every treatment plot by excising all uprights close to the bog surface within a 28-in2 area. Sampling templates were made by cutting 6-in diameter PVC pipe into 1-in wide bands. The sampling ring was randomly placed into a plot and positioned as close to the bog surface as possible. Using hand clippers, cuts were made around the entire inner perimeter to permit collection of runners that were passing through the area of the ring. The uprights were then held together and clipped as close as possible to the bog surface. Vine samples were evaluated for various yield components including number and biomass of flowering and vegetative (nonflowering) uprights, as well as runner and total plant dry biomass.

Fruit yield. Plots were harvested in the fall each year by collecting all fruit from a randomly selected 1-ft2 area within each. Fruit infected by fruit rot fungi, damaged by insects or physiological causes, or bruised by mechanical means were deemed unusable. Marketable yield was calculated from the weight of all healthy berries collected from the sample area. Potential yield was determined by multiplying the total number of fruit (healthy and unusable) by the average berry weight of the healthy fruit.

Economic analysis. Numbers generated for the economic analysis assumed payment for fruit based on yearly price per barrel (1 barrel = 100 pounds) reports (National Agricultural Statistics Service, 2006; National Agricultural Statistics Service, 2008) and a purchase price for ‘Stevens’ vines of $2,500 per ton. Pruning costs were estimated based on values provided by commercial growers for machine, operator, and laborer costs. Net income was calculated as the revenues generated from the sum of fruit yield plus the cost savings from on-farm generated vines minus the sum of the cost of fertilizer and the cost of pruning.

Statistical analysis. Data were analyzed with Statistical Analysis System software (v. 9.1; SAS Institute, Cary, NC). Model assumptions were tested through residual analysis (Shapiro-Wilk statistic) and no transformations were needed. Since the same measurements were collected annually over a 4-yr period, data were analyzed as a repeated measures experiment in Proc Mixed with an unstructured model. Significant levels that could be legitimately tested for best fit were determined by utilizing partitioning of the sum of squares via Slice option in Proc Mixed. Responses to nitrogen rate and/or pruning severity were determined by evaluating linear and quadratic trends from single degree of freedom analysis. Whenever trends were significant, regression equations were calculated.

Replicated comparison of sanding and pruning. A mature ‘Stevens’ cranberry bed in Myles Standish State Forest, North Carver, MA was used for this experiment. The bed was renovated in 2000 and was never pruned prior to the study. Applications of fertilizers each year were made to the entire bed, including the study area, based on standard nutrition recommendations. The study was established in 2006 as a randomized complete block design arranged in a split-plot with the main effect of either sanding or pruning and the sub-plot was severity level. All treatments were replicated four times. The layout is shown in Figure 1.

Sanding treatments. The sanding treatments were applied on 14 Apr. 2006 using coarse sand, mined on-site. A commercial sander (a small self-propelled vehicle with a hopper and a drop spreader) was used for on-vine sanding. The sander was calibrated to deliver a depth of 1.5 cm sand on each pass. The levels of sanding were determined by the number of times the sander passed over the plot: control (0 passes), light (one pass), moderate (two passes), and heavy (three passes). Each sanding plot was 8 ft (the width of the sander) x 25 ft.

Pruning treatments. The pruning treatments were applied on 17 Apr. 2006 using the same commercial pruner used at the SC site in the experiment above. As with the sanding, the levels of pruning were determined by the number of times the pruner passed over the plots: control (0 passes), light (one pass), moderate (two passes), and heavy (three passes). Each pruning plot was 6 ft (the width of the machine) x 25 ft. Fresh weights of vines excised during the pruning process were collected and weighed in the field.

Upright density, leaf area, and dry weight. A 6-inch-diameter ring (28 inch2) was placed randomly in each plot on 2 June 2006 and again on 4 June 2007. All plant material originating from within the ring was removed. This sampling was repeated twice in each plot. The excised plant material from each ring was evaluated as follows. The uprights were removed at the origin and counted to determine density. The leaves were removed, and leaf area was measured using a leaf area meter. All of the collected plant material was dried and weighed.

Fruiting uprights (Uf) / total upright (Ut) ratio. Once each year in late summer, random samples of approximately 25 uprights were collected from each plot, counted, and evaluated as fruiting (presence of fruit or persistent pedicels) or non-fruiting. The ratio of Uf to Ut was then calculated.

Light penetration. Light penetration into the cranberry canopy was measured in the second week of July and the third week of August in 2006 and 2007 using an Accupar linear PAR/LAI ceptometer. An external, unobstructed sensor was also employed to determine the ratio of below-canopy light to above-canopy light (tau). The ceptometer was placed under the canopy in a south-west direction at four randomly selected locations in each plot; hence 32 readings were taken in each block. Readings were converted to percent light penetration by multiplying tau by 100. In both years, readings were only taken from Blocks 1 and 2 due to weather and time restrictions.

Leaf wetness. Leaf wetness was recorded at 0.5 h intervals using micro dataloggers and leaf wetness sensors. Based on field observations, the dry/wet threshold was set at 20% to control for error in the sensor readings. Data were collected for 4 weeks in each year by deploying eight logger/sensor pairs each time. Block 1 was monitored in the first week of July 2006; the eight logger/sensor pairs were relocated to Block 2 during the second week of July 2006. Using these data, the average number of dry hours per week in early July was determined. Sensors were redeployed similarly in the third and fourth weeks of August, and the average number of dry hours per week in late August was calculated. The procedure was repeated in 2007.

Penetration of chemigation into the canopy. Water-sensitive papers were used to measure spray penetration through the canopy. The papers change color (from yellow to blue) when water contacts the papers. Attached to a stake, papers were placed at the base of the canopy and mid-way between the base and tips of the uprights. Irrigation was applied through a solid-set sprinkler system for 15 min, simulating a typical chemigation event for a pesticide application. The papers were allowed to dry, collected and brought back to the lab for evaluation. Penetration was visually evaluated using a grid system and a scale of 0=no penetration (all yellow) to 100=total penetration (all blue).

Berry yield and marketable yield. In 2006 and 2007, berry yield was estimated as in the previous experiment. Two subsamples were taken from each plot, and the data from the subsamples were averaged.

Total anthocyanin concentration (TAcy). Additional fruit were collected randomly (outside the area sampled for yield) at harvest for TAcy determination. Total anthocyanin concentration (mg per 100 g fresh weight) in harvested cranberry fruit samples was determined with a modification of the protocol of Fuleki and Francis (1968) using an acidified aqueous extractant (0.2N hydrochloric acid).

Economic analysis. The cost of treatments, including equipment and labor costs on a per acre basis, were assigned based on information provided by two commercial cranberry growers (R. Gilmore of AD Makepeace Co. and M. Beaton, personal communication). The costs for light pruning included the pruning machine ($40/h), the buggy to remove cut vines ($40/h), one equipment operator ($75/h), and five laborers ($14/h each). The total cost for 8 h (time to prune 10 acres) was $1,800 or $180 per acre. Costs for the moderate and heavy pruning treatments were calculated by doubling or tripling, respectively, the cost of the light severity treatment as more severe pruning required multiple passes with the pruner.
The costs for the light sanding treatment included one front end loader ($60/h), three sanders ($27.50/h each), four skilled laborers ($30.50/h each), and a daily move-in charge for the equipment of $150. The total cost for an 8-h day (time to sand 7 acres at light severity) was $2,388 or $341/acre for machines and labor. To define a multiplier for the moderate and heavy treatments, it was recognized that fewer acres would be sanded on a daily basis compared to the light treatment and costs would be higher due to longer equipment rental and labor expenditures. Based on grower experience, costs for the moderate and heavy sanding treatments were calculated by multiplying the cost of the light sanding by 1.5 and two, respectively. To these costs was added the cost of sand, $12/yard3 screened and delivered so that the total costs per acre for the sanding treatments were $1,291, $2,415, and $3,536 for light, moderate and heavy severities, respectively. Purchased sand was used in this calculation since not all growers have access to on-farm sand.

The average total cost of production for beds that were not sanded or pruned was $3,200/acre in 2006 and $3,300/acre in 2007 (G. Rogers, AD Makepeace Co., personal communication). The price of cranberries used for the analysis ($38.80 and $43.40 per 100-lb barrel in 2006 and 2007, respectively) was the blended return for fresh and processed fruit to growers in Massachusetts (U.S. Dept. Agr, 2008).

Data analysis. Analyses of variance were conducted on all data using the Proc GLM procedure in SAS (version 9.1; SAS Institute, Cary, NC). Model assumptions were tested through residual analysis (Shaprio-Wilk statistic) and no transformations were needed. Although pruning and sanding severity is referred to in the text with the labels of light, moderate, and heavy, treatment levels were analyzed as continuous variables to reflect the continuity of treatment application (e.g., sequential number of passes) with which the vines were sanded or pruned. Responses to pruning or sanding severity were determined by evaluating linear and quadratic trends from single degree of freedom analysis.

Irrigation scheduling. The grower-scientist team agreed that the most important tool available at the beginning of the project for use in irrigation scheduling was the water level float developed by Dr. Bruce Lampinen in 2000. Growers were reporting that they were interested in its use but were uncertain how to make, install and utilize the device. To address this issue, we held workshops in 2005 and 2006 at the UMass Cranberry Station. Growers were provided with a factsheet of directions and all materials necessary to construct a float. They were then led through the construction during the workshop. After they built their floats, they were taken out onto the research bog and shown how to install and use a float to schedule irrigation. We also constructed extra floats and offered them for sale to growers. This was designed to overcome the financial barrier faced by small growers -- some of the materials needed must be purchased in bulk so that those needing only one or two floats found the cost prohibitive.

By 2006, a local firm had begun to market automated irrigation systems to local cranberry farmers. At the 2007 winter meeting, we held a discussion panel consisting of growers, the automation vendor, and a moisture sensor vendor. During that panel, the idea of using the automated systems with temperature sensors to manage frost protection remotely was discussed as was the use of this technology to conserve water and energy during frost protection (see next topic). Some of the project participants were early adopters of automation and Chris Severance hosted a bogside workshop in the summer of 2007 to show how he was testing the moisture sensors marketed with the system and comparing that technology to the water level float.

In 2007, the Massachusetts Department of Agricultural Resources funded CCCGA, Dr. Peter Jeranyama (Vanden Heuvel's replacement at the UMass Cranberry Station) and DeMoranville to work with growers to conduct on-farm evaluations of next generation sensors and to develop a model for the use of these sensors in irrigation scheduling based on plant needs. That research is ongoing on 6 farms, 4 owned by project participants.

Irrigation cycling in frost management. As the popularity of automation for irrigation began to grow, CCCGA secured a USDA grant to fund installations by growers, including some of the project participants. Matt Rhodes hosted a field day so that others could see how the system was designed. Growers, including project participants, began to experiment with using temperature set points to cycle irrigation on and off during frost nights so as to apply less water and use less energy (for pumping). We incorporated this practice into our project and participants hosted field days (2) and participated on workshop panels to share their methodologies. This topic has become important enough that CCCGA has funded a current study by DeMoranville and Jeranyama to help growers evaluate cycling on-farm to make sure that the various proposed methods for cycling are indeed preventing frost damage.

Nutrient management. At the beginning of this project, two of the grower participants (Beaton and Gilmore) expressed strong views on directions for nutrient management based on their on-farm experiences. Ben Gilmore advocated for the study of slow-release fertilizers for even growth of the canopy. However, these materials were not very popular or affordable (and he has since ceased using them). Matt Beaton had been dealing with the potential degradation of his water source due to phosphorus (P) loading. He had been working with custom fertilizer blenders to design a low-P fertilizer suitable for cranberry. Given the potential for environmental benefit if we could demonstrate sustainable yield when using low-P blends, this became the nutrition focus for our team. Whole bog yields (based on grower records of returns and historic yields) were compared on paired low-P/standard-P bogs and additional growers were encouraged to adopt low-P blends based on the outcomes. Education on how to use P reduction was conducted at winter meetings and in a panel discussion (including Matt Beaton). CCCGA is encouraging adoption of the low-P protocol by funding a 5-year study to compare yield before and after adoption, provide free tissue testing to participating growers (to show adequate tissue P is maintained), and to fund water quality sampling to demonstrate the environmental benefits. Two of the project growers are participating in this ongoing study. Interestingly, Matt Beaton is currently working with a fertilizer company to design a low-P slow release material for use on new cranberry plantings, so the two ideas have come back together. We also looked at the interaction of pruning and nitrogen (Sandler and DeMoranville study, see above) and Garretson hosted a field workshop to discuss the interaction of sanding, pruning, and mowing with nutrient (primarily nitrogen) management.

Drainage. At the beginning of this project, team plant pathologist, Dr. Frank Caruso identified cranberry sites with varied drainage and bed sanitation practices. After two years of sampling, no differences in disease incidence were found so sampling was discontinued. However, growers surveyed strongly believed that improving drainage on their bogs improved productivity and reduced disease incidence. Further, uniform drainage using tiles (pipes) can aid in maintaining a uniform water table as shown when using water level floats. We have also found that in beds with non-uniform soil moisture, vine discoloration (known as yellow vine syndrome) develops in the wettest and driest areas during dry summers. In 2007, CCCGA received funding from the Massachusetts Department of Agricultural Resources to establish a grant program for growers to implement innovative bog renovations (replants). Thirteen of those projects (73 acres) included the installation of drainage tile. Two of those properties were included in a farm tour in the summer of 2008, attended by about 35 farmers. In March of 2008, at their winter meeting (~150 attendees), CCCGA hosted a New Jersey grower to share his experiences in installing drain tiles into established beds. In 2008, CCCGA secured USDA funding for growers to implement various drainage tile protocols (depths, spacing) and for Dr. Jeranyama to assist them in evaluating their effectiveness. These installations are being evaluated at 6 sites, 4 of which are properties of our project participants.

Grower surveys. All attendees at the UMass Cranberry Station winter meetings in 2006, 2008, and 2009 were handed surveys regarding their management practices and were instructed to complete them during the presentation regarding this project (2006) or during meeting breaks (2008, 2009). These three surveys are attached in Appendix 7. In addition, growers were surveyed as part of the Extension follow-up regarding the value of information presented.
A final phone survey was conducted in August 2009 using the script shown in Appendix 7. The growers to be surveyed were chosen based on 2006 respondents who indicated they would be willing to be interviewed for the project or because they were members of the project team.