An Assessment of Agricultural Damage Caused by Red Deer (Cervus elaphus L.)and Fallow Deer (Dama dama L.) in Southwest England
Abstract
The impact of deer grazing on agricultural grassland and cereal crops was assessed at two locations with medium to high deer densities in the Exmoor area of southwest England. Red deer impact on early spring grazing was measured at one site by comparing samples of herbage cut from inside and outside deer-proof exclosure cages, just prior to turn out of livestock, on 1 March 1989. Fallow deer impact on first-cut silage grass production was similarly measured using exclosure cages at a second site by cutting samples on 31 May 1990 and on 6 June 1995. In addition, impact of fallow deer on cereal crops was assessed at this site by measuring sample grain yields from areas of the crop used or unused by deer in 1995 (winter wheat), 1996 (winter barley) and 1997 (winter wheat). Significant dry matter yield losses were recorded for red deer impact on spring grazing in 1989 (14.5%) and for fallow deer impact on first-cut silage (15.9%) in 1995 but not in 1990. In 1995 a small but significant yield loss (7.1%) was recorded for winter wheat at the fallow deer site but no loss in cereal yield was recorded in 1996 or 1997. These assessments, carried out in response to complaints about deer damage, suggest that the impacts of deer in this area, where they occur at relatively high density, are only moderate. This highlights the need for careful assessment of cost-benefits when considering deer management strategies to reduce perceived agricultural damage.
Introduction
Deer populations have increased their geographic range in England over recent decades [1,2] as have some deer species in mainland Europe and elsewhere [3,4]. In England this spread has almost certainly been accompanied by an increase in numbers [1,5] and has corresponded with increasing reports of deer damage to agriculture [1,6]. The southwest of England holds the largest population of red deer (Cervus elaphus L.) in the country, estimated in the mid 1990s at ~10,000 [7], and has a number of localised populations of fallow deer (Dama dama L.) with a total population in the area of the Exmoor National Park alone of perhaps around 1000 animals [8]. These deer populations occur in areas of woodland and moorland set within a matrix of largely pastoral farmland, which the deer exploit as a feeding resource.
Historically, the Wildlife Management and Licensing Service (WMLS) in England, and its predecessors, received relatively few enquiries concerning deer problems. Only 11 cases involving fallow deer and one involving roe deer (Capreolus capreolus L.)
were recorded for the period January 2001 to December 2007 (WMLS data).
This, however, does not necessarily reflect the actual impact of deer on agricultural interests since cases were only recorded if the enquirer was seeking statutory authority for actions that would otherwise have been illegal. The Deer Act 1991, which regulates the culling of deer in England, allows landowners or their agents to cull deer outside statutory close seasons without special authorisation and thus authorisation from WMLS need only be sought for use of otherwise prohibited methods, such as night shooting.
There are no recent records of more general enquiries about deer damage in England. However, during the late 1980s, the department kept computerised records of all technical enquiries. Detailed data is only available for January 1987 to March 1989 but over this period 212 deer related enquiries were received (equivalent to about 170 per year); 37 involving red deer, 116 fallow and 59 roe. About 50% of enquiries concerned damage to field crops or grassland. These data are discussed in more detail in Putman and Moore [1].
A more representative assessment of farmers’ perceptions of damage was attempted in the 1990s in four regions of England (Gloucester/Somerset, Essex/Suffolk, lowland Yorkshire and Northamptonshire) using a questionnaire survey [9]. In this study 1192 responses were obtained, with 69% of respondents reporting that they had deer on their farm and 59% of these reporting damage to cereals and other crops. However, only 15% of those reporting cereal damage (54 respondents) claimed that the cost of losses due to deer exceeded £500 (~550 Euros) per year. Although subsequent ground truthing appeared to support reports of cereal yield losses of up to 10% [10], overall the study found “little difference between yield in damaged and undamaged crops” [11].
There have been relatively few attempts to directly assess deer damage to agriculture [1]. One study on roe deer in England found that although the deer foraged on wheat (Triticum vulgare) and barley (Hordeum sativum) from March to May, later feeding appeared to concentrate on broadleaf weeds within the crop and there was no significant yield loss [12]. Similarly, in France, roe were found to have little effect on wheat yield in sample plots [13]. Preliminary findings of a study of red deer damage in the Exmoor area, where medium to high deer densities were reported, suggested yield losses attributable to the deer from 1-12% for winter wheat, 0-8% for oats (Avena sativa) and 5% for barley [6]. However, in the final results of this study the losses caused by deer were not found to be significant [14].
Elsewhere in Europe studies of the impact of deer on agricultural crops have concentrated on roe deer, with estimates of impact on cereal crops inferred from dietary composition [15], comparison of crop yields in areas used or unused by deer [13], recording of damaged maize (Zea mays) plants [16] or mechanical simulation of damage [17,18]. Most of these studies suggest little more than a trivial impact on crop yield, though one study in Sweden, where most deer damage is attributed to European elk (Alces alces L.), reported a 26% increase in yield of oats in areas fenced against deer, compared to unfenced fields [19]. However, a number of studies in North America have found significant yield losses (up to 50% or more) caused by white-tailed deer (Odocoileus virginianus L.) to alfalfa (Medicago sativa) and other forage crops [e.g. 20,21,22]. In a recent study of red deer impact on grassland in Slovenia substantial yield losses were also recorded, averaging 50% dry matter loss, with up to 80% loss reported for the worst affected sites [23]. These losses are much greater than might be expected in England, given the history of relatively low numbers of complaints here and the limited evidence available thus far.
With increasing deer populations in many parts of the world it is increasingly important for decisions about the management of deer for damage prevention to take account of the cost of management relative to the level of damage likely to be suffered [24]. Such decisions rely on the availability of objective data for the deer species, crop and local conditions concerned. Consequently, we report here previously unpublished results from damage assessments that we carried out in the late 1980s to mid 1990s in an area with relatively high deer densities. These assessments were undertaken to help inform decisions on the potential use of statutory powers to require deer culling under the Agriculture Act 1947 and have not previously been published because of the local sensitivity of the information at the time.
Methods
Damage assessments were carried out at two locations in southwest England where there had been long-standing complaints about serious deer damage to crops. Both sites were on the fringes of Exmoor National Park, an area characterised by moorland, pasture and steep-sided wooded valleys. The study sites were both within 250-350 metres above mean sea level and adjoined extensive areas of broadleaf woodland or coniferous forestry plantations which provided the main harbourage for the deer in each area. At both sites, assessments were carried out on the fields reported to be worst affected by deer damage. At Site 1, this was a 2.4 ha rye grass dominated ley (Lolium spp.), adjoining woodland, where grazing by red deer was reported to be causing significant loss of early spring grazing for livestock (cattle and sheep). Red deer density was estimated in this area in the 1990s at about 15-20 deer per km² [25].
At Site 2, assessments were carried out where grazing by fallow deer was reported to be causing losses to grass grown for first-cut silage in a 7.8 ha rye grass dominated ley, adjoining the forest boundary, and to ripening cereals in an adjacent 6.0 ha arable field. Fallow deer density in this area was around 10-25 deer per km² [8 & own unpublished data]. Red deer were also present at Site 2 but in much lower numbers than fallow deer (3% of deer observations recorded during site visits) and the problem reported was considered to be principally associated with fallow.
Grassland impact assessment
Damage to grassland was assessed using deer-proof exclosure cages, 1.8m by 0.9m by 0.5m high, constructed from steel weld-mesh 75mm by 150mm netting. These were designed so that they would not physically exclude rabbits (Oryctolagus cuniculus L.) or hares (Lepus europaeus Pallas) but both species were, in any case, relatively uncommon at the study sites and were not considered likely to affect the assessment of grazing losses.
Damage assessments were carried out at Site 1 in 1989 and at Site 2 in 1990 and 1995. For the assessments in 1989 and 1990, five exclosure cages were placed in early-mid January at random intervals along each of five equally spaced parallel transects across the field reported to be suffering most damage, giving a total of 25 cages. A further five cages were placed within a 15m × 15m deer-proof fenced enclosure in the same field to test if there was any cage-effect on yield.
At Site 1 all exclosure cages were within 200 metres of the field/woodland boundary and at Site 2 all were within 150 metres of the field/forest boundary.
In 1995, at Site 2, cages were similarly placed in the same study field, but because the variance in the data obtained in 1990 suggested that 20 samples would be sufficient to measure the likely levels of loss only 20 cages were used for the assessment. The remaining five cages (one randomly selected on each transect) were individually fenced against deer to test for cage-effect. This method was adopted to avoid any confounding effects of variation in yield across the field area (see results).
At Site 1 the field studied was used for early spring grazing by ewes and lambs. Just prior to turn-out of stock on 1 March 1989 loss of pasture was assessed by comparing yield in samples cut from 0.38m² areas inside and outside the exclosure cages. Samples were labelled and bagged separately and subsequently analysed for dry matter (DM) yield.
At Site 2 the field studied was used for first-cut silage production. Just prior to cutting for silage samples were cut from inside and outside of each cage as for Site 1, but using 0.1m² samples, because of the greater bulk of material resulting from the later cutting date. In 1990 samples were cut on 31 May and in 1995 on 6 June. At both sites, samples were cut from the central area within each cage and from 1 to 2m outside, in the same position relative to each cage for each site and sampling year.
Loss due to deer grazing was assessed by ANOVA comparing DM yields in samples protected from grazing inside cages with those outside. Spatial variation in yield across each study field was tested for by ANOVA using yields inside cages and treating each transect as a separate data set.
Cereal impact assessment
Deer damage to cereals was assessed at Site 2, in the cereal field considered most vulnerable to deer damage, in 1995 (winter wheat), 1996 (winter barley) and 1997 (winter wheat). The whole of this field was within about 400 metres of the farmland/forest boundary and about 120 metres from the boundary, at its nearest point. Yield loss was estimated by comparing yield at harvest in areas of the crop which had been subject to different levels of deer activity. This was assessed by recording faecal pellet groups, evidence of deer grazing on the crop and other field signs in contiguous 15m sections along and within 2.5m either side of a transect comprising of eight evenly-spaced parallel tractor wheelings (‘tramlines’) across the field. Because of the layout of the tramlines, which were the tractor tracks used for routine crop management operations, 89 15m sections were inspected in 1995 and 1996, and 103 in 1997. Deer activity was assessed throughout the growing season by inspecting the transect from crop emergence and at 4-week intervals until harvest. At each inspection, evidence of deer activity was recorded and then removed or destroyed, where possible, to avoid double-recording on subsequent visits.
Prior to harvest sample sections were selected randomly with the aim of selecting sufficient sections to give samples from 10 sections unused by deer, and 10 each from sections with varying levels of recorded deer use.
As close as practicable to harvest in each year 1m² samples were cut from each of the selected sections, from mid-way along each section, between 0.5m and 1.5m to the side from the wheeling away from the tramline, always on the same side relative to the field. These were individually labelled and bagged then subsequently threshed, cleaned and weighed. Yield loss was assessed by ANOVA comparing yield in sections where no deer activity had been recorded with that in deer-used sections. Although this methodology did not employ experimental controls to assess yield where the deer had no access to the crop, it was similar to the method used in an earlier published study [12] and was considered sufficiently reliable for the purpose for which the assessments were primarily carried out. Spatial variation in yield across the study field in each year was tested for by ANOVA using each of the eight tramline/transect sub-sections as a separate data set.
All analyses were carried out and figures drawn in Microsoft Excel.
Results
Grassland impact assessments
Results from the cages inside deer-proof fences did not indicate any consistent cage-effect on yield. We, therefore, did not correct the measured yield figures used in our damage assessments for cage-effect. Mean dry matter (DM) yield was lower outside exclosure cages than inside the cages in each of the three assessments (Table 1).
Yield losses recorded at Site 1 (14.5%) and at Site 2 in 1995 (15.9%), were significant (Site 1: F = 4.514, d.f. = 49, P < 0.05; Site 2, 1995: F = 8.986, d.f. = 39, P < 0.005) but not at Site 2 in 1990 (F = 0.721, d.f. = 49, NS). However, comparing yields protected from grazing inside exclosure cages between transects also indicated that there was significant spatial variation within field area at Site 1 and at Site 2 in 1990 (Site 1: F = 2.897, d.f. = 24, P < 0.05; Site 2, 1990: F = 5.887, d.f. = 24, P < 0.005) (Fig. 1), though not at Site 2 in 1995 (F = 2.291, d.f. = 19, NS).
Table 1. Mean herbage DM yield for grassland damage assessments for each site and year.
Figure 1. Spatial variation between transects in (protected) mean sample yields inside exclosure cages at grassland sampling sites at; (a) Site 1, 1989 and (b) Site 2, 1990.
Cereal impact assessments
The yields recorded for cereals in sections of the trial field with and without recorded deer activity, as indicated by pellet groups, are shown in Table 2 (sections with very low signs of deer activity were omitted). The maximum cumulative number of pellet groups recorded in any one section was eight in 1995, 25 in 1996 and four in 1997.
In 1995, all survey sections had deer tracks or trails, but sections with feeding signs, but no pellet groups, typically only showed a small number of cereal heads bitten off on the last inspection prior to harvest (considered too few to significantly affect recorded yield). Consequently, in all subsequent analyses such sections were treated as undamaged by deer, giving a total of undamaged sections of 37 in 1995, 23 in 1996 and 66 in 1997 (Table 2). Because of the relatively small number of sections with 3 or more pellet groups, as far as possible, all of these sections were sampled. In 1995 and 1996, a small number of samples were lost or damaged in processing, and one ‘undamaged’ sample section in 1995 was found to have been trampled by deer when the visit was made to cut the sample. In 1997 therefore, where possible, 11 samples were collected for each category as a safeguard against such losses.
In 1995 a modest but significant reduction in winter wheat yield was recorded in areas used by deer, as indicated by the presence of pellet groups (F = 8.480, d.f. = 40, P < 0.001). This result appeared to be a particular consequence of reduced yield in the areas most heavily used by deer, based on the pellet group index (Figure 2(a)). Yield for the field as a whole, estimated by extrapolating sample yields in proportion to the number of sections in which each level of deer use was recorded, indicated an overall yield of 5.9 tonnes/ha; a reduction of 7.1% compared to the theoretical potential yield estimated from undamaged sections alone (Table 2). However, there was also significant spatial variation in yield across the field area (F = 2.763, d.f. = 40, P < 0.05) with a 45% lower mean yield in the lowest yielding transect section (A) compared to that with the highest mean yield (C) (Figure 2 (b)).
There was no yield loss recorded in winter barley in 1996. In fact, there was a slight trend for increased yield in areas used by deer, but this was not significant (F = 1.863, d.f. = 49, NS). Again, there was significant spatial variation in yield across the field area (F = 2.908, d.f. = 49, P < 0.05) with a 38% lower mean yield in the lowest yielding transect section (H) compared to that with the highest mean yield (A). The only area of the field unused by deer corresponded to the poorest area of the crop, both in terms of yield and weed encroachment.
There were no significant differences in winter wheat yield in 1997 between areas used and unused by deer (F = 0.032, d.f. = 34, NS) or spatially across the field area (F = 0.825, d.f.= 34, NS). The estimated overall yield for winter wheat in 1997, extrapolated from sample yields, was lower than the estimate of 5.9 tonnes/ha in 1995, even with deer damage, but there was no obvious link between deer activity and yield in 1997 (Table 2).
Table 2. Cereal yields of samples cut from areas of the study field with different levels of deer activity, as indicated by numbers of pellet groups, recorded during each growing season (figures in italics indicate negative loss – i.e. apparent enhanced yield in ‘damaged’ sections).
Figure 2. Mean sample yields of winter wheat at Site 2 in 1995 by; (a) deer use as indicated by pellet group counts (pg = pellet groups); and (b) spatial variation between transect sections.
Discussion
The results reported here indicate that yield loss in grassland attributable to deer grazing at these sites was modest, at up to around 15% DM. However, spatial variability in yield, where yield from one transect within a field was as much as 70% ‘reduced’ compared to the highest yielding transect in the same field, means that any impact caused by the deer could easily be masked by highly localised variations in growing conditions. This result is similar to the findings of another study in southwest England [14] where any impacts of deer foraging were found to be insignificant in comparison with yield variation due to soil fertility levels.
Although deer regularly used the field in which our cereal damage assessments were carried out, the apparent level of use varied between years and was lower than anticipated, in 1997 in particular, given the deer population in the area. In 1995 almost half the field area was ‘unused’ by deer, in 1996 only about one quarter, and in 1997 almost two thirds was ‘unused’.
Although a significant, but relatively low loss to winter wheat was recorded in 1995 (7.1%), a higher level of deer use was recorded in 1996, with cumulative counts of up to 20+ pellet groups, yet there was no apparent yield loss. This might suggest that the signs of deer activity as used were not a reliable indicator of feeding activity, but the results from 1995 do tend to support their use, given that the greatest yield loss in that year corresponded to the areas with most sign of activity. In 1996 however, local spatial variability in yield was greater than any apparent variation due to deer grazing.
Wide variation in deer impacts between years has been reported previously, even when deer continued to be present in relatively high numbers [14,19], and it has been shown that roe deer grazing may not affect yield in cereal crops [12,13]. A possible mechanism suggested for a similar lack of impact of pronghorn antelope (Antilocapra americana Ord) grazing on winter wheat crops in Colorado, is that there is a positive switch away from maturing cereal crops because of the falling crude protein content, and increasing cell wall constituents, relative to alternative forage [26].
Our assessments were carried out in areas with relatively high deer densities and at sites where specific complaints of serious deer damage had been made, yet the recorded losses to grazing or silage grass were modest. Similarly, a minor, but significant, yield loss was only recorded in one out of three assessments in cereal crops. These findings are consistent with several previously published European studies which suggest that deer have limited impact on agricultural crops [6,12-16,19] but are in contrast with a recent study in Slovenia where substantial yield losses from red deer grazing were reported [23]. From the cull density figures provided in the Slovenian study (1.7 to 2.7 deer per 100ha), assuming they corresponded to about 15-20% of the local deer population, it seems likely that the population density was within the range of 8.5 to 18 deer per 100ha, not dissimilar to that in our study area. It might, therefore have been expected that the findings of the two studies would have been more comparable. However, in the USA it has been shown that agricultural fields located in forest-dominated landscapes experience most damage from white-tailed deer [21]. We believe the differences in the levels of impact found in ours and the Slovenian study may also, at least in part, reflect the differences in the extent of woodland cover between the two countries. Slovenia is almost 60% forested [23] whereas England has less than 10% woodland cover [27], resulting in substantially different ratios of deer harbourage to vulnerable agricultural land. This highlights the need for caution in drawing inferences about wildlife impacts between different countries or regions and emphasises the importance of careful assessment of cost-benefits when considering deer management strategies for reducing perceived agricultural damage.
Acknowledgements
We would like to thank the farmers on whose land this work was conducted and former colleagues who assisted with the fieldwork. We would also like to thank two anonymous referees for helpful comments on an earlier draft of this paper.