Experimental set-up
Twelve plots (Fig1, 2) were used in the present study, eight of which were created for a previous buffer strip experiment established in 2008 and four of which were newly created in the same location for the work described herein. Each plot was hydrologically-isolated from the neighbouring plots via the installation of gravel filled French drains on the upper and side edges with waterproof membrane on the bottom and side of the drain adjacent to the plot. The plots measured 34 m x 10 m, with nine of the plots having an extra 12 m x 10 m area which constituted the buffer strip (i.e., a total area of 46 m x 10 m). The three plots lacking the extra buffer strip areas were treated as the experimental controls. Three replicate plots were set up as buffer strips for each of three different types of vegetation cover which was established in 2016 in the uppermost 10 m of the buffer section. The remaining bottom 2 m of the buffer section of each replicated plot with a buffer treatment was left as an uncut grass strip (to replicate the minimum requirements of farmers for buffer strips in agricultural policy for England at the time the experiment was initiated). The three types of buffer strip vegetation consisted of a deep rooting grass (Festulolium cv. Prior), short rotation coppice willow, and native broadleaved woodland trees. Prior to sowing theFestulolium grass seed, the existing grassland sward was removed using glyphosate herbicide and the ground rotovated. An initial sowing in October 2017 failed to establish sufficient Festulolium , and a second sowing was undertaken in September 2018. The willow treatment comprised 200 stems in the 10 m x 10 m buffer area (i.e., equivalent to 20,000 stems ha-1) equally split between five cultivars (Endurance, Terra Nova, Cheviot, Hambleton and Mourne). The area was pre-treated with glyphosate herbicide to remove the existing grassland sward prior to planting willow, with stems of 30 cm length inserted flush to ground level in May 2017. The layout of the planting consisted of five pairs of lines 0.75 m apart, with a gap of 1.5 m between pairs of rows. Willow stems were randomly inserted into the ground at 0.5 m spacing along the rows, which ran perpendicular to the slope of the field. Four-strand electric fencing was erected around the outer edge of each buffer strip with willow to provide protection from browsing deer. The native broadleaf tree treatment consisted of six species – hornbeam (Carpinus betulus L.), sweet chestnut (Castanea sativa Mill.), hazel (Corylus avellana L.), pedunculate oak (Quercus robur L.), small-leaved lime (Tilia cordata Mill.) and wych elm (Ulmus glabra Huds.). Five bare-root trees of each species were planted (i.e., equivalent to 3000 per hectare) in each 10 m x 10 m buffer strip area in December 2017. Five rows running perpendicular to the slope of the field, each with six trees, were randomly planted 1.75 m apart, with each row offset by approximately 0.85 m to the neighbouring row. Prior to planting, the existing grassland sward was removed using a glyphosate herbicide spray. The trees each had a 1.2 m tall green plastic tree guard held in place via a 1 m wooden stake, with a four-strand electric fence surrounding the outer edge of the buffer strip to protect the trees from browsing deer.
The upper part of each of the 12 replicate plots was treated as an agriculturally managed area, with a grass silage crop being cut twice in 2017 and three times in 2018, and once in 2019. In April 2019, the existing sward in this agriculturally managed area was removed using glyphosate herbicide, ploughed, rotovated and a maize crop (Garni CS cultivar with methiocarb and fludioxonil coatings) established to test buffer performance with a higher risk crop upslope. A post emergence herbicide was applied six weeks after drilling (Nico Pro 4SC (Nicosulfuron) at 1L/ha). Details of fertilizer applications on the agriculturally managed areas are shown in Table 2.