Final Report for FNE12-762
The goal of this research was to test the environmental and economic impacts of converting a northern hardwoods stand into a silvopasture when compared to treating the stand as a managed forest or converting it to an open pasture. Additionally this project addressed the establishment and productivity of six forage treatments and their influence on soil properties within the silvopastures and recently cleared open pastures. An analysis of variance was utilized to quantify forage productivity and changes in site characteristics such as soil properties. During the first year of establishment (2013), open pastures produced significantly more forage per acre than silvopastures, 365 lbs DM/acre and 228 lbs DM/acre respectively. Of all planted forages, orchard grass consistently produced the greatest amount of dry matter per acre during 2013 and 2014. Bulk density of the soil increased across all treatments but was the greatest in open pastures when compared to silvopastures and the forest treatment (having the lowest increase in bulk density). Potentially minerizable nitrogen, iron, and phosphorus in the soil increased across the site between 2012 and 2014, while aluminum decreased. Understory plant diversity (Shannon-Wiener Diversity Index) increased on all the sites between 2012 and 2014, as did species richness although no differences were found between silvopasture, open pasture, and forest treatments. While basal area and volume of residual trees increased between 2012 and 2014 on the silvopasture and open pasture treatments, no epicormic branching was witnessed in sample plots and no significant differences between treatments were found for tree growth. Based on the financial analysis of this specific case, it makes the most financial sense to retain trees in a silvopasture setting when converting a pole size northern hardwood forest to pasture as this option yielded a total net present value and internal rate of return of $566/acre and 7% after being projected out 30 years.
North Branch Farm in Saranac, NY comprises 78 acres of open pasture and forestland. Historically, the property was a dairy farm, but farming operations were abandoned in the early 1980’s. When I purchased the property in April 2010 many of the pastures had reverted to shrubs and trees. Since then, I have been bringing the pastures back into production through mechanical and manual improvements and the use of pastured livestock. Approximately 2/3 of North Branch Farm is forested; comprised of early successional northern hardwoods and mixed conifer stands. It is my intention to continue incorporating agroforestry practices into some of the forest land, primarily in the form of silvopasture. Due to the lack of research related to silvopasture in the northeast, this project was aimed at determining the tradeoffs and benefits of forest conversion to silvopasture. Applied research also complemented the educational outreach nature of North Branch Farm.
Pastured livestock raised on the farm includes modest numbers of pigs, poultry, and beef cattle (approximately 20 head). Livestock is sold direct to the consumer through farmer’s markets. I also sell produce from the farm and market it locally. In the last few years I have also expanded the farm into production of heirloom figs from my family’s original tree. More information on the farm can be found at www.adkfigs.com. I am also a faculty member at Paul Smith’s College and I utilize my farm as a demonstration and teaching site in forestry and agriculture courses. I also use the farm to host classes from other educational institutions.
University of New Hampshire Professor Richard Smith, technical advisor to this grant, has been active in providing advice and some lab equipment to this project. His support, Professor John Carroll’s, and that of other faculty at the University of New Hampshire has been instrumental in the success of this project.
The objective of this project was to investigate the system productivity, environmental effects, and economics of converting an unmanaged northern hardwoods forest stand into a silvopasture when compared to open pasture and managed woodlot conversions. Specifically the following objectives were accomplished:
- Documented the differences in production of multiple forages under silvopastures when compared to open with consistent management and history of each system.
- Documented differences in tree growth and stem quality between sites managed as silvopastures and woodlots after a uniform harvest.
- Described the ecological characteristic divergence, e.g. understory plant communities and soil properties, between silvopastures, open pastures, and woodlots.
- Estimated the financial aspects of a northern hardwood forest converted to silvopasture, open pasture, and thinned woodlot.
Site: This research took place on North Branch Farm, located in the town of Saranac, New York, part of the Adirondack Mountain region. Prior to manipulation of the 6.75 acre site for this experiment, the area was a 50 year old mid-succession northern hardwood forest with an average soil pH of 4.68. Species composition was dominated by pole size northern hardwoods including red maple (Acer rubrum), paper birch (Betula papyrifera), white ash (Fraxinus Americana), black cherry (Prunus serotina), quaking aspen (Populus tremuloides), American elm (Ulmus Americana), apple (Malus spp.) and American basswood (Tilia Americana). Overstory basal area averaged 82 ft2 per acre and was comprised of 985 stems per acre with a quadratic mean diameter of 4 inches. The site’s history includes charcoal production in the 1800’s followed by cattle pasture until the 1960’s when use ceased and the site reverted back to forest.
Experimental Design: The experimental design was a split-plot randomized complete block (Figure 1). The main plot factor was overstory conditions and there were three treatments: crop tree thinning (woodlot), silvopasture, and open pasture (all trees removed). Each main plot treatment was replicated three times using ¾ acre plots, the locations of these plots was randomized within each of the three blocks. The thinning treatment allowed a comparison of silvopasture productivity and economics to that of a forest without cattle and planted forages. The open treatment allowed a comparison of silvopasture productivity and economics with a newly established pasture; it also allowed for a determination of the effects of trees on forage productivity. Response variables for the main plots included financial costs and returns, forage production (woodlot treatments excluded), tree production (open treatments excluded), soil properties, and understory plant community diversity.
Main plot treatments were established in July of 2012. The woodlot and silvopasture treatments had the forest overstory basal area reduced to 29% of full stocking basal area, a level similar to stocking levels on silvopastures from other regions (Devkota et al., 2001; Garrett et al., 2004; Walter et al., 2007). Dominant and well-formed stems were favored; American elm and white ash retention was discouraged due to threats from invasive alien species. Black cherry and red maple comprised the majority of stems favored. In open treatments, 100% of trees were removed. Due to challenges with farm owned machinery, a logging contractor was hired to harvest the site using whole-tree felling and skidding techniques. Harvested trees were chipped off-site and sold by the logging contractor for biomass, removal volume data was retained. Cost of harvesting to the landowner was the wood removed plus $429 per acre on a total of 7 acres. These costs may have been revenues if a larger area was harvested, but due to economies of scale harvesting was a cost.
The split-plot factor in this experiment was forage establishment (Figure 1). Forage treatments were established in the silvopasture and open pasture main plot treatments. The forage establishment treatments included: none, loose hay depositing, orchardgrass-white clover, perennial ryegrass-white clover, Kentucky bluegrass-white clover, and smooth bromegrass-white clover. Forages were broadcast seeded using a hand operated seeder on August 15th and 16th of 2012. The broadcast seeder was calibrated off-site for each species prior to on-site use. The loose hay treatment had locally-sourced hay bales spread thinly over their respective split-plots to simulate on-pasture feeding of livestock. High quality, first cut, timothy hay with seed heads was selected with the intention of ensuring a quality and weed free seed source. For forage seeding mixtures white clover was broadcast seeded at a rate of 5 lbs pure live seed per acre. Grasses were broadcast seeded at rates of 20 lbs pure live seed per acre. Immediately following seeding, cattle (4,900 pound herd weight) were let in to all open pasture and silvopasture plots for one day to trample in seed.
In 2013 and 2014 all silvopasture and open treatments were grazed with beef cattle to a forage height of two inches. The intensity of grazing was dependent upon the overall condition of forages within the plots, but grazing intensity remained constant between plots and no preferential grazing of forage treatments by cattle was observed. The first round of grazing took place in August of 2013 and each plot was grazed by cattle (herd weight 9,400 pounds) for two consecutive days. During the first week of June 2014 and the middle of August 2014 a cattle herd weight of 12,000lbs was grazed on each plot for two and 1.5 days respectively.
Data Collection and Analysis: In May and July of 2012, prior to any treatments on the site an inventory of all trees over 2 inches DBH and ground cover flora was conducted. A total of 36 nested inventory plots were established. Overstory plots were a fixed 1/20th of an acre and composite understory plots using the same plot center were 1/1000th of an acre. Following treatment in the summer of 2012 an inventory of residual overstory trees was conducted. The tree and understory plant inventory was be repeated in the summer of 2014 to address changes in each treatment. Composite soil samples to a depth of 15cm (five samples per split-plot = 1 composite sample) were collected prior to any treatments, totaling 54 soil samples in the establishment year. Soil samples were analyzed for pH, extractable bases, total C and N, and available N and P. Composite soil measurements related to soil bulk density and particle size distribution were also collected. Soil sampling took place in main and split plots in July of 2012 and were repeated two years post treatment in July of 2014, comprising a total of 108 soil samples over two years.
Prior to each grazing period, understory plants and forages were clipped above 2 inches in composite sample plots then oven dried to determine dry matter yield in silvopasture and open treatments. Data on forage production was collected within the center portions of the main and split plots so as to minimize potential edge effects (Figure 1). The edge buffer was narrower on the north side of main plots due to the southern aspect of the site and forest canopy light dynamics related to the sun’s location in the northern hemisphere. Forage samples in each strip were an aggregate of five square foot subsamples distributed evenly through the center of each strip.
A split-plot ANOVA model was used in Program R to test for significant differences between main treatments and split plots (Team, 2014). Prior to analysis, forage data was tested for normality and then log10 transformed. Total forage production was analyzed as a combination of planted grasses, planted clover, and volunteer grasses dry matter per acre. Soil properties and understory plant richness and diversity were normally distributed so not transformed, but a split plot ANOVA was utilized to test differences between years and treatments. A least significance difference (LSD) test was used for pairwise comparisons.
Time, labor, equipment, and other real costs (such as seed) were recorded throughout the process of establishing and managing each treatment. Financial costs related to all aspects of silvopasture, open pasture, and crop tree management were assessed in net present dollars at multiple interest rates. Livestock production revenues were extrapolated from dry ton per acre forage production in the different treatments. Financial comparisons were made between main treatment groups and forage establishment treatments.
Value of standing timber was extrapolated based on regional stumpage prices published by the New York Department of Environmental Conservation. Rates of unintended tree mortality and changes in epicormic branching were compared between silvopasture and woodlot treatments. Tree diameter measurements were recorded prior and two years post treatment, although due to the slow growth of trees and potential thinning shock, analyses of these short-term data did not yield results which would inform long-term tree productivity.
During the first year after establishment, 2013, total forage production was significantly greater in open pasture treatments than in silvopasture treatments with 365 lbs DM/acre produced in open pastures and 228 lbs DM/acre produced in silvopastures (p = 0.0476), these samples were collected in mid-August. Of planted forages, orchardgrass produced the greatest amount of forage at an average of 579 lbs DM/acre but this was not significantly different from perennial ryegrass or smooth brome, 318 lbs DM/acre and 172 lbs DM/acre respectively. Table 1 shows these results but one important one to point out from 2013 was that in hay and none (no grass planted) strips, significantly less forage was produced. The take home message from the first year of forage production was that in strips where no forages were seeded, very little forage was produced. No significant difference was found between non-forage plant dry matter yield between treatments, these non-forage plants were primarily Rhubus (brambles) and Carex (sedge) species.
In 2014, also seen in Table 1, no significant diffence in total forage production was seen between open pastures and silvopastures with June respective values of 280 lbs DM/acre and 166 lbs DM/acre and August values of 238 lbs DM/acre and 203 lbs DM/acre. Orchardgrass again stood out as the highest producer in mean dry matter production at 212 lbs DM/acre in June and 291 lbs DM/acre in August, although these values were only significantly different than the none and hay forage treatments, as well as the smooth brome yield in August. Figure 2 shows the relative dry matter production of different forage groups between silvopastures and open pastures in August 2013, June 2014, and August 2014. Figure 4 shows the production of each forage strip relative to its counterpart in the open/silvopasture treatment. Note the relative success of orchardgrass compared to other treatments. Additionally, note the success of orchardgrass and perennial ryegrass in 2014 in both silvopastures and open pastures. The treatment where no forages were planted consistently produced much less forage than all other treatments, especially in the silvopastures.
Table 2 shows the understory plant dynamics on the site between 2012 and 2014. Across the site, understory plant richness and diversity increased between 2012 and 2014. This is likely due to an increase in available resources, primarily sunlight, due to harvesting activities in 2012. Additionally, species richness was greatest in silvopastures in 2014 than all other treatments, although the forest treatment saw the greatest increase in species richness. No significant differences were found between treatments using the Shannon-Wiener Diversity Index (Spellerberg and Fedor, 2003). The most abundant understory plant genera post treatment were Rhubus, Carex, Solidago (brambles, sedge, goldenrod). In terms of grasses, both povertygrass (Aristida dichoroma) and crabgrass (Digitaria spp) were common on the site in 2014 but not in 2012. Table 2 also shows the species richness and diversity of woody plants in the understory of all treatments, with no significant changes between years or treatments.
Table 3 shows the dynamics of soil properties across the site between 2012 and 2014. In terms of significant differences, bulk density, potentially minerizable nitrogen, iron, and phosphorus in the soil increased across the site between 2012 and 2014, while aluminum decreased.
Most notable from the soils data is the lack of many significant differences between treatment groups, suggesting that differences in soil properties are not rapidly diverging between silvopastures, open pastures, and the woodlot treatments. One exception to this is the bulk density interaction with year is significant at the alpha = 0.1 level. Figure 3 shows the change in bulk density between treatments over the two year period. It is important to note that the degree of bulk density increased across the site from 2012-2014, likely due to compaction from timber harvesting. What is interesting in Figure 3 is that the greatest change in bulk density was in the open pastures as compared to the silvopasture and forest treatments. Soils were also compared between forage treatment strips and it was found that the change in pH increased the greatest in orchardgrass strips when compared to other forage treatments, Figure 5.
Costs for forage were calculated based on a grass seeding rate of 20lbs of seed per acre with an added cost of $17 for 5lbs of white clover per acre. Although hay was a cost in this study its cost in the financial analysis was assumed to be 30% of its actual, as it was meant to simulate waste hay after out-feeding livestock at a rate of 30% waste. Value of forage yield per acre was set to equal the cost of purchasing an equivalent amount of dry hay locally. This was determined to be $0.08 per lb as one 600lb round bale of hay at 17% moisture content (thus 498lbs of dry matter) costs $40 delivered to the farm. Table 4 shows the net present value (NPV) of forage production and costs on the site between 2012 and 2014. It is clear the orchardgrass stands out for its high yields and low establishment costs. It is especially important to note that orchardgrass pays for its establishment in less than two years.
Financial analysis of main treatments were compared assuming that revenue from pastures would be valued at that of dry hay, therefore labor for, and revenue from, livestock management was not included as this would be occurring on the farm regardless of where the animals are pastured. One challenge to excluding livestock values in this analysis is that it did not take into account any additional benefit toward livestock production that may be derived from a shaded/sheltered/climate controlled pasture. It is a fair assumption that livestock under heat or cold stress will have lesser yields, and this is backed by some research which suggests that tree cover moderates temperature and that cattle yields benefit from shade (Ferrez et al., 2011; McDaniel and Roark, 1956). What is not well understood is how much production will be increased by shelter from trees nor how this potential benefit is tied to variable climates. Therefore, while the shelter benefits from silvopasture may have real monetary yields in terms of livestock production, including them in this analysis would have added an unknown degree of variability due to the lack of research quantifying these benefits.
Fencing and watering costs were established based on what was actually used on the site, in this case a poly tank and portable polywire fence on permanent cedar posts. This costs will likely be unique to each farm and situation. Forage establishment values were based on the cost and production of orchardgrass, as this was the most successful of all forage treatments in both silvopastures and open pastures.
When comparing the three main treatments in this study (open pasture, silvopasture, and woodlot) it was determined that a fourth treatment could also be assumed for the purpose of financial analysis. This was a no-management treatment where timber revenues were projected from the pre-harvest inventory of the site in 2012 to the end of a 30 year rotation using NED-2, an Ecosystem Management Support Program available from the US Forest Service (Twery et al., 2005). For all treed treatments, the forest vegetation simulator northeast variant growth model (Dixon, 2002) was utilized to project standing timber volumes in 2042, the 2014 inventory was used to project silvopasture and woodlot treatments. These volumes were then given stumpage values based on 2014 timber prices by species and product. Forage revenues were based on mean orchard grass total production in 2014 for both silvopastures and open pastures as their production was not significantly different that year. Reforestation costs were not included in this analysis but it is important to note that they would exist but be variable depending on management goals for the next 30 years. Table 5 demonstrates the cost and revenue structure used in this financial analysis. While basal area and volume of residual trees increased between 2012 and 2014 on the silvopasture and open pasture treatments, no epicormic branching was witnessed in sample plots and no significant differences between treatments were found for tree growth. Timber quality and tree growth were therefor kept at similar values between silvopastures and woodlot treatments for the purpose of financial analysis.
It is clear from this financial analysis that both the silvopasture and no management scenarios are the highest yielding in both internal rate of return (IRR) and net present value (NPV) at a 3% discount rate. Table 6 shows these IRRs and NPVs. The no management scenario, assuming timber is harvested at the end of 30 years, had the second highest IRR and NPV because of its high yields of timber value at the end of the 30 year rotation and no initial establishment costs. The woodlot treatment yielded the least because future timber yields did not warrant a high up front cost of the heavy thinning which was conducted. The open pasture treatment performed well but was surpassed by silvopasture because of the long term value of timber remaining in silvopastures. Based on the analysis of this specific case, it makes the most financial sense to retain trees in a silvopasture setting when converting a pole size northern hardwood forest to pasture.
- Table 3: This table shows soil property changes across the site between 2012 and 2014.
- Table 1: This table provides forage production values and ANOVA results for forage production in 2013 and 2014.
- Table 6: This table provides the total internal rate of return (IRR) and net present value (NPV) after a final harvest in 30 years of four main options: open pasture, silvopasture, heavily thinned woodlot, and unmanaged forestland.
- This picture shows cattle grazing an open pasture treatment in June 2014.
- Figure 4: This figure shows the production of forage among each forage treatment in both silvopastures and open pastures.
- Figure 5: This figure shows the significantly higher increase in pH in the orchardgrass strips across the site between 2012 and 2014 when compared to other forages established
- Table 2: This table shows understory plant richness and diversity between 2012 and 2014
- Table 4: This table provides the net present value of forage treatments over two years in 2012 dollars.
- Table 5: This table shows the cost/revenue structure utilized to analyze NPV and IRR of the main treatments.
- Figure 2: This figure shows mean dry matter production of forages and non-woody plants over the course of the study.
- This picture shows cattle grazing a silvopasture treatment in June 2014.
- Figure 3: This figure shows the relative change in bulk density between treatments from 2012-2014.
Since the beginning of this project interest in silvopasture in the northeast US has increase greatly. Many workshops and presentations on the topic have been given in the last three years. Being the only formal study on silvopasture establishment in the region, this project garnered quite a bit of interest. Some farmers who were interested in the study and came to visit did begin to implement silvopastures on their farms. The demonstration of silvopasture in practice through this study and others silvopastures in the region are the main driver of adoption. It is important for farmers to be able to visualize how silvopasture was adapted in a specific case so they can modify the system to meet their resource availability and goals.
The most unsuspected occurrence that arose from this project was publicity of North Branch Farm through interest in silvopasture research. At least three articles were written on the project by authors who sought out the farm, two notable ones being in Farming Magazine and Country Folks.
Education & Outreach Activities and Participation Summary
The results of this project will continue to develop over the next few years, and it is my intention to share them both in scientific and practitioner circles. Peer reviewed publications have not yet been developed but will be over the next year. In line with what was proposed for this grant, the following outreach efforts were conducted as part of this project between 2012 and 2014:
- 136 students have visited the site as part of 10 Paul Smith’s College and Plattsburgh State classes
- Two farm open houses were held and attendees had the opportunity for a silvopasture tour
- A tour was given to the Adirondack Mountain Club in August 2014, 8 member were in attendance
- In September 2014 a silvopasture field tour was hosted in conjunction with Cornell Cooperative Extension, 18 farmers and extension professionals were in attendance
- February 2013, North Country NY Pasture Meeting, Chateauguay NY
- March 2013, Vermont Society of American Foresters Meeting, Lake Morey VT
- April 2013, Sustainable Living Festival, Canton NY
- September 2013, Paul Smith’s college, Paul Smiths NY
- October 2013, Society of American Foresters National Convention, Spokane WA
- November 2013, Maine Organic Farmers an Growers Association, MOFGA Farmer to Farmer Conference, Maine
- January 2014, Northeast Silvopasture Conference, Latham NY
- February 2014, Cornell Cooperative Extension Meeting, Westport NY
- April 2014, Algonquin chapter of the ADK Mountain Club, Plattsburgh NY
This project is valuable as an example of pasture establishment. Being a case study it is important that people understand the results of this project are not replicated between farms and may vary depending on starting forest conditions. Additionally, management practices play a key role in the success of silvopastures and these pastures saw very few days of grazing between long rest periods. With the long-term nature of silvopasture, this project is likely to continue to inform farmers about silvopasture production following forest conversion. Based on the results I feel very strongly that silvopasture is a viable practice in the region both financially and ecologically.
More research is surely needed to determine the benefits and tradeoffs of converting forests to silvopasture. Specifically, work should be conducted to investigate the conservation properties of silvopasture and how they relate to other options. For example, in this study silvopasture seemed to compare well to both open pasture and woodlot options, but there are many factors that speak to conservation and they were not all addressed in this study. Additionally, more work could be done on forage establishment mechanisms. This study used broadcast seeding after whole tree harvesting, but what would have occurred if there were slash left?
One aspect of this project that should receive more attention in the future is the change in bulk density between treatments. Figure 3 seems to show a trend between greater increases in bulk density in pastures than non-grazed areas. It also suggests that the presence of trees reduces the degree of bulk density in silvopastures when compared to open pastures. It is possible that tree roots in silvopastures are partially mitigating compaction due to livestock, but further work should be done to investigate this question.
Silvopasture is an exciting practice with a strong potential to increase the value of pastures through our region in many ways. However, we are currently at the beginning stages of this practice in the region and how to maximize the benefits from the system is highly variable between farms and sites. This project should be used to direct future analyses of silvopasture and inform practitioners of potential impacts but the silvopastures describe should not be directly mimicked as they no-doubt could be improved. For example, while silvopastures will still be created from forests on North Branch Farm, they will likely contain a higher density of trees. One thing is for certain however, all new pastures created will be silvopastures, for timber or orchard production.