Field Methods

 Agroforestry systems

     Two most promising agroforestry systems are compared: an agro-silvicultural system and an agro-silvo-pastoral system.  An agro-silvicultural system includes growing trees and producing hay (improved four-species forage mix).  An agro-silvo-pastoral system involves grazing on the improved forage pasture after trees reach (at about four years) the cow-resistant height of 10 feet (since the results of this project are intended to be demonstrated to and applied by farmers and foresters who are more comfortable with the English systems of measures, the experimental design and results are presented in English rather than SI units).

     This study is established on a 30-acre area located at the Southwest Research and Extension Center of  the Arkansas Agricultural Experiment Station near Hope, AR.  Twenty-five and a half acres are used for the main study, one and a half acres are used for the two comparative studies (Regular 8 x 10 plantation, and the Wide spacing trial), while the remaining three acres are used for borders and other miscellaneous functions.  The main study area is divided into two equal portions, one for hay and another for beef production.  On each portion are three replications.  Thus, the area for each treatment block is 25.5/(2 x 3) = 4.25 acres.  Due to the size of this study, a single location could not be found and it had to be established in four different areas.  The cluster study was located in all four areas while the two comparative studies were located in only three.  A third comparative study, Competition mortality, was established in 1998 on a 1.05 acre area about 50 feet north of area 3.  However, due to a severe drought in 1998 and the study being placed on a fragipan, survival was only about 40%.  The study was re-established in 1999 about 100 feet north of area 3.  Once again a severe drought occurred and survival was only about 10%.  The study has been discontinued.  All field work implementing the study, subsequent maintenance, and annual inventories were done by the Hope Station personnel under the field manager Robert Colvin.

Loblolly pine seedlings that were originally planted in 1998 were obtained from the International Paper Co. nursery at Bluff City, AR.  The shortleaf pine seedlings were obtained from a nursery in Missouri operated by the Missouri Forest Commission.  Replanted seedlings in 1999 came from the Arkansas Forest Commission's Baucum nursery located in Little Rock.

Figure 3.  Planting of one year old loblolly pine and shortleaf pine tree clusters in area 3 during February of 1998 using a planting grid designed by Bob Colvin (in white shirt).  Loblolly at the center of the grid and shortleaf planted on the adjacent sides and center line.

   Metal fence posts were placed every 144 feet outside the outer lines of the three-line rows.  Besides preventing the mowing of trees, the posts help to identify tree clusters alleviating the need to nail tags to the trees or paint numbers on them.

Figure 4.  Metal fence posts are used for both cluster identification using a small metal numbered tag (left), and to prevent clusters from being mowed when harvesting the hay alleys (right).  The stake (to the right of the hay bale) is not straight because the tractor hit it when harvesting the hay alley (both pictures were taken in November of 2000).

The posts were positioned across every seventh loblolly pine cluster (24 x (7-1) = 144).  A tag was attached to indicate the cluster number in a row: 1, 7, 13, 19, ... .The other outer line of the same row has the posts with numbers 4, 10, 16, 22.

     This project maintains or modifies as necessary a single level for the following factors: (1) crop tree species; (2) nurse tree species; (3) distance between tree rows; (4) distance between crop trees within rows; (5) length of rotation; (6) forage species; and (7) density of animals.  Several levels are tested for the following variables: (1) row structure (number of lines in a row and distance between them); (2) distance between nurse trees; (3) thinning schedule, and (4) pruning options.  All testing is compared in three replications for the two agricultural systems.

     Some of the listed variables such as density of pines and forage production involve a tradeoff between benefits to the forestry and agricultural components.  Others such as forage fertilization and wide tree spacing intend to boost growth of agricultural plants, pines, and indirectly, animals.

Tree species
     Loblolly pine has proven to be the most successful species for productive forest management in the southeastern United States.  This species provides crop trees to harvest at the end of the rotation.  Nurse trees, which surround the crop trees in each cluster, get more light.  To avoid overshadowing of crop trees, slower growing shortleaf pines are used as nurse trees.  Since this species does not grow as fast as loblolly pine, both species can be planted at the same time.

Distance between tree rows
     The alleys between pine rows are 30 feet wide to facilitate tree growth and agricultural operations.  Heavy thinning will increase the width of alleys from 30 to 34-40 feet.  A 1.5-foot buffer area has been left on both sides of the alleys to insure that the tree clusters are not damaged during forage harvesting (decreasing the harvested portion of the hay alleys to 27 feet wide or 31-37 feet at later ages).

Figure 5.  Cluster rows and alleys between clusters in area 3 on April 28,  1998 shortly after planting.

Distance between crop trees
     The distance between the centers of loblolly pines within rows is 24 feet.  It is expected that each cluster will produce one loblolly pine tree with a diameter of 15-16" at the end of the rotation.  

Length of rotation
     It will be the age that maximizes the combined annual financial return (approximately 20-25 years).

Choice of crops
     A four-species mix that includes forages tested for this area (AU Triumph tall fescue, common bermuda grass, Marion lespedeza, and Osceola ladino white clover) was established in October of 1997.  Osceola ladino white clover was planted, while common bermuda grass, Marion lespedeza, and AU Triumph tall fescue had previously seeded in naturally in the meadows.  Additionally, Kentucky Tall 31 fescue was planted along with Osceola ladino white clover in September and October of 1998.  It is expected that the species composition will respond to the current level of shading.  Legumes will be reseeded in later years as necessary.

Density of animals
     Cross-bred steers will be grazed from weaning until around 700 - 800 lbs.  The initial stocking rate for each season will be based on the average A.U.M. (Animal Unit Month) forage production of the soil series corrected for estimated shading effects.  Average A.U.M. values are available from the U.S.D.A.-N.R.C.S.  Forage fertilization, supplemental feed and herd health practices will be based on current recommended procedures.

Row structure
     Tree rows consist of one or three lines of trees.  The crop trees are planted in the inner (central) line.  Nurse trees are also planted in lines on the two adjacent sides.  In three-line rows, two distances between lines were established: 4 and 6 feet.  These rows are referred to as 3L4 and 3L6.  Accordingly, the widths of three-line rows are 8 or 12 feet.  The width of single-line rows (referred to as 1L) is 0.  Agricultural alley (AL) widths are always 30 feet.  Large inter- and intra-row distances between crop trees make buffer zones between clusters unnecessary.  Each cluster is a sample unit.  The crop trees in the adjacent rows were staggered.  The rows run in the north-south direction to increase sunlight for forage.

Figure 6.  A single row (1L) loblolly pine cluster (center), and forage alleys in area 3 in January of 2000 during the third growing season of the seedlings.
 

     To facilitate studies of forage production and hydrological regime within and between the rows and at the same time randomize the location of clusters in order to minimize the effect of site variability, the layout was a combination of random and systematic designs.  For each row, one distance between trees was selected at random from the chosen values (see below).  The rows alternate regularly.  The east-west cross-section of the field have the following dimensions:

 This sequence of rows with the associated alleys (3L4, AL, 1L, AL, 3L6, AL) is referred to as a strip.

Distance between nurse trees
     Four distances between nurse trees and the center of loblolly pine crop tree clusters were established: 3, 6, 8, and 12 feet.  A single distance was used for each strip.  These distances were selected at random.

Thinning of crop trees
     Four seedlings of loblolly pine crop trees were planted one foot apart in the center of each cluster to provide for mortality and selection.  They were thinned to two trees per cluster in March of 2000, and thinned to one tree per cluster in December of 2000.

Thinning of nurse trees
     Three levels of thinning intensity will be tested:
0 - no thinning;
1 - thinning half of the nurse trees as soon as they can be sold as long pulpwood (minimum requirement is 4" with bark at the height of 25');
2 - thinning all nurse trees as soon as they reach the size of long pulpwood.

     These levels were assigned to the entire strip at random.

Pruning of crop trees
     Pruning of pine grown for sawtimber is profitable for regular plantations.  This operation promises to be even more important in the agroforestry context because of low stand density that encourages production of large branches and good accessibility of trees. Nurse trees will not be pruned.  Based on previous experience, including plots established at Hope, three levels of pruning will be tested:

     The ages are approximate and pruning will be based on the average tree height. The pruning levels were assigned to the entire strip at random.

Cluster codes
     Each cluster can be identified by a code such as 3L4-D8-T1-P2 describing:
1.  Row structure (1L, 3L4, and 3L6).
2.  Distance between nurse trees (D3, D6, D8, and D12 corresponding to the distances of 3, 6, 8, and 12 feet).
3.  Thinning code (T0, T1, and T2).
4.  Pruning code (P0, P1, and P2).

Forage samples

    Subsample clippings of vegetation in the hay alleys are collected to obtain dry weight yields and nutrient contents.  Seven strips 12 feet long and 4 feet wide, running parallel to the tree clusters, are completely clipped across the hay alleys.  The strips run the edge of 3 row tree clusters to 1 row tree clusters.  When the clipping is finished, a rectangle 12 feet by 28 feet is completely clipped in the hay alleys.  Scales which are placed on tripods are used to weigh the vegetation.  The clipped vegetation is placed into a blue tarp and is weighed to obtain green weight.  A subsample of the weighed vegetation is then collected to determine dry weight and nutrient content.

Figure 7.  Bob Colvin weighing forage samples in July of 2000.  Harvested forage is placed into the blue tarp that Bob is holding, and weighed using scales hanging on the tripod to obtain green weight.

Regular 8 x 10 plantation
     To compare our design with regular forestry practice, we planted three control blocks (areas 1, 2, and 3) of loblolly pine.  Each block has 9 trees in 9 rows.  Rows are 10 feet apart.  Within rows, trees are 8 feet apart.  There are 81 trees in each block.  The total combined area for all three blocks is around 0.4 acres.  No replanting was conducted for this study.

Figure 8.  The area 1 Regular 8 x 10 plantation in June of 2000 during the third growing season of the loblolly pine seedlings.

Wide spacing trial
     Growing crop trees without nurse trees will be tested separately.  The spacing is 24 x 34 feet.  Nine clusters (four seedlings per cluster) per treatment is sufficient.  Within each site a total of 27 planting locations, or clusters, were created.  Similar to the agroforestry stud with nurse trees, a total of four seedlings were planted at each cluster.  Replanting was conducted in 1999 if a cluster had less than two surviving seedlings.  For these clusters, seedlings were replanted so that the cluster had four seedlings.  They were thinned to two seedlings per cluster in March of 2000, and thinned to one seedling per cluster in December of 2000.  Treatment will include the three levels of pruning.  For three replications we will need 9 x 3 x 3 x 4 = 324 seedlings.  The total combined area for all three blocks is around 0.9 acres.

Figure 9.  Wide spacing trial of area 1 in June of 2000 during the third growing season of the loblolly pine seedlings.

Competition mortality
     Most of the experience gained in Hope was about growth of spaced trees.  To expand on this experience, this project included a small study of competition mortality, which is an important component of stand dynamics.  For this purpose, 1296 loblolly pine seedlings were planted at the closest distance (18 inches apart) that permits tree measurements.  Due to repeated drought, mortality wiped out twice this study. 

Figure 10.  One-year old loblolly pine seedlings planted for the competition mortality study.

Economic analysis
     Comparison of investment alternatives related to the agroforestry alternatives will be performed in a number of different ways.  Present net worth (PNW) is the most commonly recognized method.  Normally, PNW values are computed by analyzing the stream of discounted cash flows that are expected to occur within the life of the investment.  Two different types of investments are considered in this project: agricultural returns from beef, hay and forage, and a forest-based investment from pine pulpwood and sawlogs.

    Investment valuation of agricultural scenarios will be done as follows: net annual income streams are evaluated on a PNW basis for a defined period of  time.  In dynamic scenarios, when periodic changes are made in the production base, such as crop rotation, or phasing out one crop in favor of another, calculations are more difficult, but still relatively straightforward.  If there is a permanent election to remain in agricultural production, rather than converting land to a different opportunity, such as a shopping mall, equivalent annual income streams are perhaps the best of the available valuation techniques to use across alternatives.  PNW values can be easily converted to equivalent annual income values.  Internal rates of return are slightly more difficult to calculate, but can be accomplished without too much difficulty.

     In this analysis, the economic returns from a pasture only and forestry regime, and a pasture-beef-forage and forestry regime will be contrasted.  The principal economic problem in this analysis is one of establishing equivalency of net returns (evaluation at equivalent point in time).  For the pasture-beef-forage and forestry example, the present value of the costs and returns over the 25 year scenario would consist of annual costs for lime, fertilizer and mowing and annual returns from hay sales for the first five years, plus the annual net returns for the cattle-forage investment for years six through 25.

    The net returns from the thinning, projected to be at year 16, is discounted to time zero, as is the net return of the final harvest at year 25.  Costs include the initial establishment cost at time zero, as well as the discounted costs of pruning twice, at an estimated ten and 18 years.  Records for each tree cluster will be maintained throughout the analysis period; cluster type will be the unit of analysis.

     Additional considerations that weigh on the nature of forest investment are the various incentives available to non-industrial land-owners.  Currently, early amortization of investment expenses is available at up to $10,000 per year.  Investment expenses may be recovered in as little as seven years using this scheme.  Additionally, one time investment credits and capital gains treatment of returns are available.  Other incentive programs, such as the Stewardship Incentive Program (SIP) and the Conservation Reserve Program (CRP), both of which are federally-subsidized cost-share programs, provide additional relief from initial high investment costs.  All investment analysis should be done on an after-tax basis.

     To be viable, agroforestry, in addition to its environmental benefits, should be profitable. This is a guiding idea of this project. All considerations presented above allow us to predict that, given the range of crops, enhanced efficiency of land and labor use, the proposed system will increase return at least 25-30% as compared with either forest management or agriculture alone.

Competitiveness of forestry investments
     Forestry investments are often criticized as being "non-competitive" with agricultural or financial instrument investments.  This is not the case for several reasons.  First, in the southern U.S. rich agricultural soils are, in most cases, ideally suited for pine growth.  Silvicultural growth rates of 12 to 20% per year give rise to value growth rates of 9 to 15% during the "wonder years" of pine establishment and early growth.  This period of high growth is the same period that we propose to analyze in our agro-forestry design.  Given the relatively low stocking rates planned for this study, volume and value growth on the study plots will be maximized.

     The second reason underscoring the competitiveness of forest investments in the south is that long-term real price appreciation, spurred by continued strong demand, has created extremely favorable market conditions for pine thinnings and harvest sawlogs.  And, finally, investment tax credits coupled with early amortization of forestry investment expenses works with existing incentive programs (Stewardship Incentive Program and Conservation Reserve Program) to reduce investment requirements. Additionally, capital gains treatment for timber proceeds provides an additional incentive for timber investments.

Statistical Methodology
     The use of fixed effects in regression (dummy or categorical variables) and the ANACOVA models (continuous variables in ANOVA) makes our design statistically manageable. The fundamental premise of both techniques is decomposition of the error term extant within the general linear model.  While the error term in these mixed models becomes quite complex, its reduction is not difficult with modern computing capabilities.

 Ecological energetics and resource partitioning
     In addition to economic values that depend on the current market, this project will evaluate an unchanging metric - energy. All inputs in agroecosystems have energetic values and their use should be weighed against the energy values of the agroforestry outputs. The energy balance resource partitioning between the system's inputs and outputs will be calculated. The inputs will include the energy contained in or required to deliver water, fertilizer, pesticides, machinery and fuels, seed, seedling and animal stocks. The outputs will be calorific contents of hay, food products, and timber.  The calorific values of the inputs will be compared to the calorific value of the outputs to obtain an energetics efficiency value.  Such an evaluation will complement the economic evaluation by considering energy costs and benefits.

     Availability and partitioning of resources such as water, solar radiation, and nutrients will be assessed by "the bottom line" - the yield of timber, hay, and beef. The data to be collected will allow us to estimate the amounts of various substances needed to produce a unit of harvested products. In our view (Zeide 1996) sustainable flow of goods and services from a given ecosystem is the best indicator of its structure, functions, and what is called ecosystem health.