The Henfaes experimental site is one of a national network of six sites established across the country investigating the potential of silvopastoral agroforestry on UK farms (Sibbald and Sinclair, 1990). It was established in 1992 on 14 ha of agricultural land at the University of Wales, Bangor farm (Henfaes), which is located in Abergwyngregyn, Gwynedd, 12 km east of the city of Bangor.  The climate is Hyperoceanic, with an annual rainfall of about 1000 mm.  The soil is a fine loamy brown earth over gravel (Rheidol series) classified as a Dystric Cambisol in the FAO system.  The parent material consists of postglacial alluvial deposits from the Aber river, comprising Snowdonian rhyolitic tuffs and lavas, microdiorites and dolerite in the stone fractions and Lower Paleozoic shale in the finer fractions (Teklehaimanot and Sinclair, 1993). Topography consists of a shallow slope on a deltaic fan of approximately 1-2^° and the aspect is northwesterly, at an altitude of 4-14 m above sea level. The depth of the water table ranges between 1 and 6 m (Teklehaimanot and Sinclair, 1993).  The entire site was sown to a mixture of perennial ryegrass (Lolium perenne L.) and white clover (Trifolium repens L.) in April 1992 at a seed rate of 12.5 kg ha^-1 of L. perenne var. Talbot, 12.5 kg ha^-1 of L. perenne var. Condessa, 2 kg ha^-1 of T. repens var. Gwenda and 2 kg ha^-1 of T. repens var. S184.

The network sites have a common set of core treatments described by Sibbald and Sinclair (1990).  These comprise sycamore (Acer pseudoplatanus L.) planted at 100 and 400 stems ha^-1 into grazed pasture and at 2500 stems ha^-1 without grazing as a farm woodland control, and pasture without trees as an agricultural control (Sibbald et al., 2001).  Farm woodland plots were managed according to normal forestry practice for farm woodlands in the UK (Hibberd, 1988) and are therefore equivalent to farming system controls sensu Dupraz (1999), in as much as normal recommended practice for farmers was applied but implemented on a research site owned and managed by the University of Wales rather than by a farmer.  The size of the farm woodland control plots (0.1 ha) is the smallest area of trees defined as small woods in the UK Forestry Commission National Inventory of Woodland and Trees (Wright, 1998).  There are three additional treatments at the Henfaes site: red alder (Alnus rubra Bong.) planted at 400 stems ha^-1 into grazed pasture and at 2500 stems ha^-1 without grazing as a farm woodland control, and, sycamore planted at an overall density of 400 stems ha^-1 but in clumps (described below) rather than as individual trees (Figure 1).  Treatments are replicated three times in a complete randomised block design.  Water table varies across the site and blocks were identified so that plots in each block had similar water table depth.  Blocks were defined and mapped and allocation of treatments within blocks was done by ordering the treatments in a random sequence and then working from the south-eastern corner of each block demarcating the treatment plots on a field plan.  Orientations of individual plots were adjusted to fit with field boundaries and to avoid a gas pipeline running under the site.  All grazed plots, except those planted with alder (see below), receive standard fertilization for productive pasture at 160 kg N ha^-1 a^-1.  All grazed plots are managed according to the network protocol (Sibbald et al., 2001) and are grazed from spring until autumn by a core group of Welsh Mountain ewes with single lambs (12 ewes ha^-1) with additional stock from a buffer flock added or removed to maintain a threshold sward height of 5 cm.

The clump treatment was introduced at Henfaes to investigate an alternative planting pattern to widely spaced individuals dispersed throughout fields. Clumps have several potential advantages in terms of production and environmental impact over individual tree planting.  The cost of tree protection may be lower for clumps than for individually protected trees.  Within clumps it is possible to select high quality trees, as is done in conventional forestry, by progressive thinning to leave a small number of final crop trees in each clump.  Furthermore, shading amongst trees within the clump may have silvicultural benefits of enhancing tree height growth and self pruning and in exposed conditions the outer trees may shelter inner trees.  From an environmental perspective, a micro-woodland habitat may be created in the clump with a richer wildlife value than that associated with single trees in fields.  From a landscape perspective, clumps resemble copses, which are a traditional feature in the UK countryside.  Copses were historically dense groups of coppiced trees or bushes, many of which have subsequently grown into larger groups of trees in fields after coppicing ceased.

Each clump comprised 13 trees and there were 31 clumps ha^-1 with approximately 20 m between clump centres.  The term clump refers to a group of trees and has been variously applied in forestry.  What are termed clumps at Henfaes fall within the definition of groups of trees in the UK Forestry Commission National Inventory of Woodland and Trees (Wright, 1998), distinguishable from individual trees on the one hand and small woods on the other.

The widely spaced trees were protected against sheep browsing using 4 m of stock-proof fence around each individual tree whereas clumps only required 24 m of fence surrounding the entire clump, equivalent to 1.8 m tree^-1, less than half of what was required for individual tree protection. Trees within clumps were planted in a diamond pattern, 1.5 m apart, with the fence a further 1.5 m from the outer trees.

Red alder was also introduced at Henfaes to investigate the use of biological nitrogen fixation as an alternative to chemical fertilizer. This could have on-farm economic advantages by reducing fertilizer costs and, if part of an organic production system, by attracting price premiums, as well as environmental benefits brought about by reduction in the amount of N leached. Red alder was chosen because it grows rapidly when young, can tolerate wet sites and produces a wide range of quality wood products (McIver, 1991). Alder was managed at the site on a low input basis (no inorganic N was applied), to encourage biological nitrogen fixation associated with the alder and with clover in the pasture.

The trees were planted in November 1992 in pits 30 cm in diameter by 30 cm deep, with 1 m² of black polythene mulch laid around each tree to control weeds and a spiral guard added for rabbit protection.  Seedlings of both species were obtained from the Forestry Commission nursery at the Northern Research Station, Bush Estate, Roslin, Midlothian, UK.  The alder was a North Pacific USA provenance and was inoculated with Frankia spp. at the nursery.  The sycamore was of UK origin.  Any dead trees were replaced on an annual basis for the first three years after planting.