Changes in moorland and heathland bird abundance in southwest England in relation to environmental change

ABSTRACT

Capsule: Population changes of many moorland and heathland birds in southwest England show associations with environmental change, and the area supports notable breeding populations of species of conservation concern.

Aims: To quantify changes in moorland and heathland breeding bird abundance in relation to changes in environmental variables.

Methods: A two-visit moorland bird survey method was used to estimate the abundance of 23 target bird species in 2008 and 2014 in Exmoor National Park, southwest England. Data on changes in 25 environmental variables over the same period were used to test associations between changes in bird abundance and environmental change.

Results: Fourteen species increased in abundance and nine decreased. Breeding species which wintered on or close to Exmoor were less likely to show negative population changes than species which migrated to winter in southern Europe or Africa. Change in abundance of 18 species was associated with change in at least one environmental variable, and these relationships were broadly consistent with the known ecology of these species.

Conclusion: Although some moorland species have declined or gone locally extinct, this area of moorland and heathland in southwest England remains an important stronghold for several species of high conservation concern. The maintenance of semi-natural moorland and heathland habitats in areas such as Exmoor may provide important refuges for such species, both now and in the future.

Heath habitats are globally widespread, occurring across Mediterranean, cool temperate and subarctic climates, and are particularly common in North America, southern Africa, Australia, Europe and across much of subarctic tundra (Cuff & Goudie 2001, Fagúndez 2013). Heaths are characterized by open, low-growing, shrub vegetation and occur mainly on free-draining, infertile, acidic soils (Fagúndez 2013). They are mainly semi-natural and of anthropogenic origin, created in the long term through centuries of deforestation and maintained through grazing or burning, and generally require management to arrest succession to woodland habitats. Globally, heaths are of high conservation concern, often containing diverse flora and fauna, but widely subject to habitat loss or degradation (Fagúndez 2013).

In northwest Europe, heaths dominated by ericaceous vegetation such as Common Heather Calluna vulgaris occur in both lowland and upland areas, and these are typically recognized as two distinct habitat types. Lowland heaths occur in warmer, drier areas, often overlying sandy soils, and are referred to colloquially as ‘heathlands’. Upland heaths are a more specialized type of heath, occurring in cooler and damper areas in the sub-montane zone, generally above the altitudinal limits of enclosed agriculture and often overlying deep peat, and are known as ‘moorlands’. The UK holds around 32 000 km² of heath habitat, of which around 950 km² is lowland heathland (JNCC 2015a) and the remainder comprised upland moorland. The UK’s heathlands and moorlands are of international importance for their plant and animal communities, with some plant communities virtually confined to the UK (Thompson et al. 1995, Ovenden et al. 1998, JNCC 2015b). The suite of moorland breeding birds is globally unique and of international conservation importance (Thompson et al. 1995, Pearce-Higgins et al. 2009a), while a range of heathland specialists are also of conservation concern in the UK (Hayhow et al. 2015). Both habitat types are subject to degradation, with moorland subject to inappropriate grazing (Fuller & Gough 1999), drainage (Carroll et al. 2011), burning (Glaves et al. 2013), afforestation (Avery & Leslie 1990) and nitrogen deposition (Britton et al. 2008), and heathlands subject to pressure from scrub encroachment, wildfire, habitat loss and fragmentation, recreational and urban disturbance and air pollution (http://jncc.defra.gov.uk/page-5942).

In some parts of the UK, heaths intergrade across an altitudinal gradient from heathland to moorland within the same area. The bird communities of these areas therefore contain specialists of both habitat types, plus generalist species that inhabit one or both of these habitats, among others. One such area is Exmoor National Park in southwest England (Figure 1). The National Park covers 692 km², of which 172 km² is heathland and moorland. In recent decades, large areas of these habitats have been lost, primarily through agricultural activity (Land Use Consultants 2004). The heathland and moorland of Exmoor is important in terms of the upland, coastal heath and blanket bog biological communities, and these areas also make an important economic contribution to the area through livestock grazing, tourism and recreational game hunting, and as a water and carbon storage resource (Land Use Consultants 2004).

Figure 1. Location of study area (unenclosed heathland and moorland; shaded areas) within Exmoor National Park in southwest England.

Previous surveys have shown that Exmoor holds notable breeding bird populations of species including Grasshopper Warbler Locustella naevia (estimate approximately 3% of UK population), Dartford Warbler Sylvia undata (approximately 3–5%), Stonechat Saxicola torquata (approximately 5%) and Whinchat Saxicola rubetra (approximately 3%), with many of these of conservation concern (Stanbury et al. 2008, Eaton et al. 2015). Many species have undergone large changes in abundance on Exmoor over recent decades. For example, between the early 1990s and 2008, Grasshopper Warbler, Reed Bunting Emberiza schoeniclus, Dartford Warbler, Common Snipe Gallinago gallinago, Stonechat and Linnet Carduelis cannabina numbers were coarsely estimated to have increased by at least 50% in the area, while Common Redstart Phoenicurus phoenicurus and Northern Wheatear Oenanthe oenanthe abundance declined by 50% or more over the same period (Stanbury et al. 2008). The reasons underlying such large changes in abundance in this area are unclear, and so there is a need to better understand how population changes in heathland and moorland species relate to environmental change (Pearce-Higgins et al. 2009b).

The moorland of Exmoor is towards the southern edge of the UK range of this habitat, and climate change poses an increasing threat to the persistence of species that are close to the edge of their current altitudinal and geographical ranges (Thomas et al. 2004, Huntley et al. 2006). Indeed, there is evidence of recent climate-related population changes in various moorland bird species (Sim et al. 2005, Pearce-Higgins et al. 2010, Amar et al. 2011). In order to make recommendations for conservation management, it is important to examine changes in heathland and moorland bird abundance and drivers of these changes. Using survey data from 2008 and 2014, collected using directly comparable methods, we examined changes in abundance of a suite of moorland and heathland breeding bird species, and environmental (habitat, land use and topography) correlates of changes between the two surveys.

Methods

Study area

The study area was unenclosed heathland and moorland, ranging from coastal lowland heath to upland wet heath and mire. These habitats are characterized by ericaceous dwarf shrub heath (primarily Common Heather C. vulgaris but also Erica spp), grassland (primarily Purple Moorgrass Molinia caerulea (hereafter Molinia) and Matgrass Nardus stricta) and with patches of scrub (e.g. gorse Ulex spp.) and scattered tree cover, at altitudes ranging from zero to 510 m above sea level. The topography is varied, with areas of steep coastal cliff, steep valley sides and flatter upland plateaux. The main land uses are grazing by sheep and ponies, with some associated management such as prescribed burning to manage grazing habitat. Some areas of upland mire have been subject to restoration management over recent decades, to re-wet habitats and reverse the effects of historic drainage through drain blocking (http://www.upstreamthinking.org/index.cfm?articleid=8699).

Bird surveys

Of the 172 km² of unenclosed heathland and moorland, 166 km² (97%) was surveyed in both 2008 and 2014. The area was divided into 1-km survey blocks along national grid lines and surveyors typically covered 2 km² per day. Breeding birds were surveyed using a modified Brown & Shepherd (1993) method, walking parallel transects spaced at 200 m intervals, such that all areas were surveyed to within 100 m. Surveys were conducted between 05:15 hours and 13:00 hours in both years, coinciding with the peak activity period for many species (Bibby et al. 2000). Two visits were made to each 1-km square, with the first from mid-April to late May, and the second from early June to mid-July. Wherever possible, squares that were surveyed early in the morning on the first visit were surveyed later in the morning on second visits, and vice versa, to ensure that as many squares as possible received one visit in the period of peak activity in the early morning (Harris et al. 2015). All bird locations were recorded to 100 m accuracy, along with standard behavioural codes. We excluded all birds overflying squares (gulls, wildfowl and hirundines). Multiple records considered to involve the same individual bird during a survey were assigned only to the first location.

Environmental variables

A range of environmental variables considered to potentially influence bird abundance were recorded in the field in 2008 and 2014, or calculated using geographical information systems (Table 2). Data were collected at the scale of 200 m × 200 m grid cells across the entire survey area. Field measures were estimated by visual assessment of habitat and other features during bird surveys, generally by viewing from the centre of each grid cell. The collection of field variables was divided between the first and second visits in each year, to enable surveyors to focus primarily on bird recording (Table 2). In addition, data on four topographic measures (mean altitude, mean and standard deviation of gradient, and dominant aspect) were subsequently extracted from digital maps (Table 2).

Data were collected using the same methods in both years, with the following exceptions. Gorse and broadleaved scrub were recorded in eight abundance categories in 2008, but only six in 2014 (Table 2). Categories 2, 3 and 4 in 2008 referred to 1–5, 6–10 and 11–50 scattered gorse/scrub bushes, respectively, recorded as part of the sward. Since these scattered gorse/scrub bushes were likely to sum to less than 1% cover within each grid cell, and thus fall below the lower threshold for category number five in 2008 (1–10%), they were combined into the single category ‘<1%’ in 2014 (Table 2). Thus, gorse and scrub cover were assigned to six categories in both 2008 and 2014.

Statistical analysis

We calculated the abundance of 23 target bird species (Table 1). For most species, abundance per year was calculated as the maximum count of individuals per 200 m × 200 m grid cell across the two survey visits. For Meadow Pipit Anthus pratensis and Skylark Alauda arvensis we used only counts from the first visit per year because many juveniles, indistinguishable from breeding adults, would have been present on the second visits. For Northern Wheatear we used the count of adult birds from late visits only per year, since a large number of non-breeding birds migrate through Exmoor during the first visit period in April and May (Stanbury et al. 2008). The maximum counts per grid cell per species were then summed across all cells per year, yielding a repeatable abundance estimate comparable with national Breeding Bird Survey (BBS) methods (Harris et al. 2015).

Table 1. Abundance of moorland and heathland birds on Exmoor, UK BBS trends (Harris et al. 2015), UK Birds of Conservation Concern (BoCC4, Eaton et al. 2015), European red list status (BirdLife International 2015) and wintering zone of each species (Hayhow et al. 2014). Shown are the maximum (max) count of individuals in 2008 and 2014 across survey area, percentage change in maximum count between 2008 and 2014 and percentage change between 1995 and 2013 from UK-wide BBS data (*significant change at P < 0.05). No BBS trend data were available for Merlin or Dartford Warbler; instead abundance estimates for these species were derived from UK-wide national surveys in 1993–94 (Rebecca & Bainbridge 1998) and 2008 (Ewing et al. 2011), and 1994 (Gibbons & Wotton 1996) and 2006 (Wotton et al. 2009), respectively. Max count figures for Meadow Pipit and Skylark in 2008 and 2014 are counts from early visits only to exclude juveniles, and max count figures for Northern Wheatear in 2008 and 2014 are counts from late visits only to exclude migrants.

Species	2008 Max count	2014 max count	% Change 2008–14	% Change 1995–2013 from UK-wide BBS	UK BoCC4 status	European red list status	Wintering zone
Merlin Falco columbarius	5	1	−80	−13	Red	Least concern	Short distance
Eurasian Curlew Numenius arquata	11	6	−45	−46*	Red	Vulnerable	Short distance
Common Snipe Gallinago gallinago	14	21	50	5	Amber	Least concern	Short distance
Common Cuckoo Cuculus canorus	72	131	82	−46*	Red	Least concern	Humid
Common Kestrel Falco tinnunculus	38	19	−50	−40*	Amber	Least concern	Short distance
Raven Corvus corax	172	72	−58	42	Green	Least concern	Resident
Skylark Alauda arvensis	2693	4096	52	−24*	Red	Least concern	Resident
Willow Warbler Phylloscopus trochilus	871	1014	16	−4	Amber	Least concern	Humid
Common Whitethroat Sylvia communis	127	265	109	38*	Green	Least concern	Arid
Dartford Warbler Sylvia undata	197	24	−88	70	Amber	Near threatened	Resident
Grasshopper Warbler Locustella naevia	327	179	−45	−16	Red	Least concern	Arid
Dipper Cinclus cinclus	10	24	140	−22	Amber	Least concern	Resident
Common Redstart Phoenicurus phoenicurus	68	152	124	47*	Amber	Least concern	Arid
Whinchat Saxicola rubetra	570	554	−3	−54*	Red	Least concern	Humid
Stonechat Saxicola rubicola	764	968	27	17	Green	Least concern	Short distance
Northern Wheatear Oenanthe oenanthe	20	17	−15	−6	Green	Least concern	Arid
Grey Wagtail Motacilla cinerea	74	105	42	−21*	Red	Least concern	Resident
Tree Pipit Anthus trivialis	181	178	−2	13	Red	Least concern	Humid
Meadow Pipit Anthus pratensis	3773	6759	79	−15*	Amber	Near threatened	Short distance
Linnet Carduelis cannabina	946	1145	21	−29*	Red	Least concern	Resident
Lesser Redpoll Acanthis cabaret	369	459	24	40*	Red	Least concern	Resident
Yellowhammer Emberiza citrinella	105	121	15	−15*	Red	Least concern	Resident
Reed Bunting Emberiza schoeniclus	275	410	49	19*	Amber	Least concern	Resident

We analysed changes in environmental variables between 2008 and 2014 (Table 2). Using the value per grid cell in each year, we ran Wilcoxon matched pairs tests, matching values per cell between years, to test for overall change for each measure between years. We then examined whether changes in bird abundance between 2008 and 2014 were associated with changes in environmental variables over the same time period (Table 3). For each bird species we excluded cells that were unoccupied in both 2008 and 2014. The response variable of change in bird abundance per cell was calculated as log ((current count + 0.5)/(previous count + 0.5)), thereby normalizing it and following Fletcher et al. (2010). Explanatory variables were calculated as absolute change in each environmental variable, by subtracting the value in 2008 from the value in 2014. Values of topographic variables remained fixed over time.

Table 2. Environmental variables, variable definitions, recording categories, which visit the data were recorded on, number of 200-m cells in which a change in each variable was recorded, and median values (i.e. summary values of categorical variables) of the 21 environmental variables surveyed in the field in 2008 and 2014. The change 2008–14 column shows whether the change between years in the median value for each variable was significant, using Wilcoxon matched pairs tests. Key: blank = no significant year effect; ---/+++ = decrease or increase at P < 0.001; --/++ = P < 0.01; -/+ = P < 0.05.

Environmental variable	Definition	Recording categories	Visit data recorded	Number of 200-m cells present	Median 2008	Median 2014	Change 2008–14
Short vegetation	Change in % cover of short sward (0–5 cm)	<1%, 1–20%, 21–40%, 41–60%, 61–80%, 81–100%	First	3029	0.50	1.50	+++
Short to medium vegetation	Change in % cover of short to medium sward (6–15 cm)	<1%, 1–20%, 21–40%, 41–60%, 61–80%, 81–100%	First	3059	1.50	2.00	+++
Medium to tall vegetation	Change in % cover of medium to tall sward (16–30 cm)	<1%, 1–20%, 21–40%, 41–60%, 61–80%, 81–100%	First	2940	2.00	2.00
Tall vegetation	Change in % cover of tall sward (>30 cm)	<1%, 1–20%, 21–40%, 41–60%, 61–80%, 81–100%	First	3068	2.50	2.00	---
Variation in vegetation height	Change in number of vegetation height categories present	Scored from 1 to 4	First	2523	3.00	4.00	+++
Ericaceous vegetation (Calluna and Erica spp.)	Change in % cover of ericaceous vegetation	<1%, 1–20%, 21–40%, 41–60%, 61–80%, 81–100%	First	2036	1.50	1.50	---
Molinia caerulea grassland	Change in % cover of M. caerulea grassland	<1%, 1–20%, 21–40%, 41–60%, 61–80%, 81–100%	First	2568	2.00	2.50	+++
Non-Molinia grassland	Change in % cover of non-Molinia grassland	<1%, 1–20%, 21–40%, 41–60%, 61–80%, 81–100%	First	3026	2.50	2.50	+++
Bracken Pteridium aquilinum	Change in % cover of bracken	<1%, 1–20%, 21–40%, 41–60%, 61–80%, 81–100%	First	1675	0.50	0.50
Burnt ground	Change in % cover of ground burnt within the last year	<1%, 1–20%, 21–40%, 41–60%, 61–80%, 81–100%	First	355	0.00	0.00	+++
Tracks, footpaths and bridleways	Change in occurrence of paths or tracks which had received human foot or vehicle traffic	Presence (1) or absence (0)	First	988	0.50	0.50	+++
Medium grass tussocks	Change in % cover of medium (15–25 cm) tussock material, excluding this years growth	0, 1–25%, 26–50%, >50%	First	1887	0.50	1.00	+++
Large grass tussocks	Change in % cover of large (>25 cm) tussock material, excluding this years growth	0, 1–25%, 26–50%, >50%	First	1103	0.00	0.00	+++
Grazing pressure	Change in grazing pressure from sheep and ponies	Light, light to moderate, moderate, moderate to heavy, heavy	Second	2753	3.00	3.00	+++
Gorse Ulex europaeus and U. Gallii	Change in % cover of gorse	Absent, < 1%, 1–10%, 11–30%, 31–50%, >50%	Second	1313	0.50	0.50	---
Scrub	Change in % cover of broadleaved scrub	absent, <1%, 1–10%, 11–30%, 31–50%, >50%	Second	1415	0.50	0.50	---
Wall/bank or bank	Change in occurrence of wall >1 m in height	Presence (1) or absence (0)	Second	814	0.00	0.50	+++
Hedge/woodland edge	Change in occurrence of hedge or woodland edge	Presence (1) or absence (0)	Second	536	0.00	0.00	--
Wire fence	Change in occurrence of wire fence	Presence (1) or absence (0)	Second	589	0.00	0.50	+++
Rush Juncus spp.	Change in % cover of rush Juncus spp.	Absent, <1%, 1–5%, 6–15%, 16–30%, 31–50%, >50%	Second	2510	1.00	1.50	+++
Ground wetness	Change in score of ground wetness	dry, moist or damp, 1–10%, 11–50%, >50% waterlogged	Second	2594	2.00	2.00	+++

Table 3. Summary of GLMs testing for associations between change in bird abundance and change in environmental variables between 2008 and 2014. Data are presented only for environmental variables which showed significant associations with at least one bird species. Key: blank = no significant association; ---/+++ = negative or positive relationship at P < 0.001; --/++ at P < 0.01; -/+ at P < 0.05. No significant associations between Common Cuckoo, Grey Wagtail or Northern Wheatear abundance change and change in environmental variables were found in 168, 134 and 27 200 m sections, respectively.

Environmental change variable	Common Snipe	Common Kestrel	Raven	Skylark	Willow Warbler	Common Whitethroat	Dartford Warbler	Grasshopper Warbler	Whinchat	Stonechat	Dipper	Common Redstart	Tree Pipit	Meadow Pipit	Linnet	Lesser Redpoll	Yellowhammer	Reed Bunting
Short vegetation (0–5 cm)			+	+++		--	-			-
Medium to tall vegetation (16–30 cm)				+++	+									++
Tall vegetation (30+ cm)								++							+	+
Molinia grassland																	---
Non-Molinia grassland				+++										---
Burnt ground								-						--
Tracks, footpaths and bridleways												-
Medium grass tussocks																	+
Large grass tussocks														++
Grazing pressure														-
Gorse Ulex europaeus & U. Gallii					+				-							++
Scrub		+			+			-									++
Ground wetness				---	+													+++
Altitude	+					---	--			---			+				++
Mean slope				++							-
Variation in slope									+
Aspect (N, S, E, W)									++ (S)
Number of 200-m sections	30	54	162	2685	1029	241	161	431	616	894	28	160	268	3234	838	465	146	496

For each bird species we fitted a generalized linear model (GLM) with normal error structure, identity link and explanatory variables as above. To avoid potential over-fitting of the large suite of predictor variables, we followed the two-stage modelling approach of Pearce-Higgins et al. (2009b). Each explanatory variable was first fitted as a univariate predictor variable in separate models. Variables that were significant predictors of change in bird abundance at P < 0.05 were retained. We tested for correlations between significant predictor variables using pairwise Spearman rank tests. Where any pairwise test had an r value > 0.5, the variable that was a weaker predictor of bird abundance change (a lower test statistic in the univariate tests) was excluded. Remaining variables were then fitted in a multivariate model, which was simplified using stepwise modelling to a minimum adequate model. All analyses were carried out using R version 3.0.2 (R Core Team 2013).

Results

Changes in bird abundance

Between 2008 and 2014, 14 species increased in abundance and 9 decreased (Table 1). Increases of greater than 50% were detected for Common Snipe, Common Cuckoo Cuculus canorus, Skylark, Common Whitethroat Sylvia communis, Dipper Cinclus cinclus, Common Redstart and Meadow Pipit, while Merlin Falco columbarius and Raven Corvus corax decreased by more than 50% (Table 1). Two of the four (50%) species wintering in each of the humid and arid zones of Africa decreased, while three of the six (50%) species classed as short-distance migrants decreased (Hayhow et al. 2014; Table 1). However, only two of the nine (22%) species classed as ‘resident’ (i.e. those likely to remain on, or relatively close to, Exmoor during the winter months) decreased (Table 1). Thus, the population changes of resident species breeding on Exmoor during 2008–14 were less likely to be negative than those species which migrated to winter in other areas.

Changes in environmental variables

Of the 21 environmental variables for which data were collected in the field, 15 showed a statistically significant increase, 5 decreased significantly and 1 showed no significant change between 2008 and 2014 (Table 2). There were large increases in the estimated cover of short (0–5 cm) and short-medium (6–15 cm) vegetation, but a decrease in the amount of tall (30+ cm) vegetation. Variation in vegetation height also increased, suggesting an increase in the heterogeneity of vegetation structure. There were increases in the cover of Molinia and non-Molinia grassland, medium and large grass tussocks, Bracken Pteridium aquilinum, Juncus rushes and ground wetness, but declines in the cover of heather, gorse, scrub and woodland edge. There were also increases in the cover of variables associated with anthropogenic activity, including agriculture (i.e. burnt ground, tracks/footpaths/bridleways, walls/banks, fences and grazing pressure).

Associations between bird abundance change and environmental change

Significant associations between abundance change and environmental measures were detected for 18 bird species (Table 3). The most frequent correlate of change in abundance was altitude, with Common Snipe, Tree Pipit Anthus trivialis and Yellowhammer Emberiza citronella abundance change generally more positive, and Common Whitethroat, Dartford Warbler and Stonechat abundance change generally more negative, with increasing altitude. Abundance changes of a range of species were associated with change in vegetation height. Increase in the cover of short vegetation was associated with increases in Raven and Skylark, but decreases in Common Whitethroat, Dartford Warbler and Stonechat. An increase in Skylark, Willow Warbler Phylloscopus trochilus and Meadow Pipit was associated with areas where the cover of medium-tall vegetation increased. Similarly, increases in Grasshopper Warbler, Linnet and Lesser Redpoll Acanthis cabaret were associated with areas where the cover of tall vegetation had increased.

Increases in the cover of grassland were associated with decreases in Yellowhammer and Meadow Pipit abundance, respectively, whereas Skylark increased where the cover of non-Molinia grassland increased. An increase in gorse cover was associated with a decrease in Whinchat abundance, but increases in Willow Warbler and Lesser Redpoll. Where scrub cover increased, Grasshopper Warbler decreased, but Common Kestrel Falco tinnunculus, Willow Warbler and Yellowhammer increased. An increase in grazing pressure was associated with Meadow Pipit declines, and increasing ground wetness was associated with an increase in Willow Warbler and Reed Bunting, but a decrease in Skylark.

Discussion

The data we present here provide two estimates of population size, in 2008 and 2014, for a suite of heathland and moorland birds in a southwest England study area. Since our study represents two ‘snapshots’ in time, it is thus possible that population sizes of some of these species in either year were unusually high or low, and our data should therefore not be considered to represent true trends. Nonetheless, some species considered to be upland specialists, including Eurasian Curlew Numenius arquata and Merlin, showed large declines during this period, while Red Grouse Lagopus lagopus and Ring Ouzel Turdus torquatus, recorded prior to 2008 but not in the present study, are now likely to be extinct as breeding birds on Exmoor (Stanbury et al. 2008). At the UK scale, Eurasian Curlew and Ring Ouzel have shown large population declines and range contractions in recent decades, although Red Grouse and Merlin populations appear to be relatively stable (Sim et al. 2010, Ewing et al. 2011, Balmer et al. 2013, Harris et al. 2015). The Eurasian Curlew is now considered globally ‘Near Threatened’, due to population declines across much of its world range, whereas Ring Ouzel, Red Grouse and Merlin are classed as of Least Concern (iucnredlist.org). However, 6 of 8 (75%) amber-listed, and 7 of 11 (64%) red-listed species showed more positive population changes on Exmoor than more widely across the UK (Table 1).

Effects of environmental change on bird abundance

Assessments of environmental change were estimated in the field by eye, at a relatively coarse scale (200 m × 200 m grid cells), and using largely different observers between the two years. There was, therefore, the potential for a lack of precision in estimating environmental change. Our data indicate that the extent of short vegetation, vegetation height heterogeneity, grass and Bracken cover, and grazing pressure increased, while the extent of gorse, scrub, heather and tall vegetation decreased on Exmoor between 2008 and 2014. Decreases in the cover of vegetation such as heather, and reductions in vegetation height, are consistent with increases in grazing pressure (Stevenson & Thompson 1993, Thompson et al. 1995, Hartley & Mitchell 2005). This suggests that our estimates were able to detect changes in habitat that were plausibly related to changes in management, also assessed by eye.

For Skylark, the increase in the cover of vegetation less than 15 cm tall appears to have been beneficial, likely providing habitats with an open structure and short vegetation which are known to be favoured by this species (Pearce-Higgins & Grant 2006). Skylarks also increased in relation to increases in non-Molinia grassland in particular, suggesting a positive response to more diverse grass swards. In contrast, Meadow Pipits declined where non-Molinia grassland increased, and Yellowhammers declined where cover of Molinia grassland increased. Further research may help to determine the importance of different upland grass sward types for these species.

Some upland mires on Exmoor have been subject to re-wetting over recent years (http://www.upstreamthinking.org/index.cfm?articleid=8699), and this may explain the increases in Reed Bunting (Gregory & Baillie 1998, Brickle & Peach 2004) and Willow Warbler (Tiainen et al. 1983) in areas where ground wetness increased. However, both Common Snipe and Eurasian Curlew showed declines in areas specifically targeted for re-wetting between 2008 and 2011–12 (Boyce 2012), and Skylark decreased in areas where ground wetness increased, consistent with previously demonstrated reduced usage of wet heaths (Chamberlain et al. 1999).

Several farmland passerine species (e.g. Linnet, Reed Bunting and Yellowhammer) which increased on Exmoor between 2008 and 2014 were formerly largely associated with lowland agricultural areas (Fuller et al. 1995). However, agricultural intensification and/or habitat loss has negatively influenced farmland passerine abundance across the UK and more widely across Europe (Fuller et al. 1995, Donald et al. 2001, Inger et al. 2015), and the maintenance of semi-natural habitat mosaics in areas such as Exmoor may provide important refuges for these species, both now and in the future.

Across Europe, heathland and moorland habitats have suffered from a severe reduction in their range over recent decades (García et al. 2013). Within the UK, moorlands have been subject to major land use changes, including afforestation (Avery & Leslie 1990), large increases and decreases in livestock grazing (Fuller & Gough 1999) and increases and decreases in management for Red Grouse shooting, including burning and predator control (Robertson et al. 2001, Yallop et al. 2006, Douglas et al. 2015). These changes have likely impacted upon moorland bird populations (Thompson et al. 1995, Pearce-Higgins et al. 2009a). Recent Eurasian Curlew declines in the UK uplands are associated with increased indices of predation pressure (Grant et al. 1999, Fletcher et al. 2010, Douglas et al. 2014) and this might potentially explain the declines on Exmoor, although we have not examined this in the current study. The causes of population declines of the Ring Ouzel on Exmoor are also unclear, although there is evidence from elsewhere in the UK of range contraction to areas at higher altitudes, with greater heather cover and more distant from plantation woodland (Buchanan et al. 2003, Sim et al. 2007).

Effects of range-edge and climate on bird abundance change

Exmoor lies towards the southwestern edge of the UK breeding range for moorland species including Red Grouse, Merlin, Eurasian Curlew and Ring Ouzel, thus any negative effects of climate change on local populations of these species might be expected to be especially apparent (Thomas et al. 2004, Huntley et al. 2006). Although Ring Ouzel declines have been widespread throughout the UK in recent decades, they appear to have been more pronounced towards the southern edge of the range, with large declines on Dartmoor, Devon and local extinction on the Long Mynd, Shropshire (Sim et al. 2010). In addition, climate change has been proposed as a potential cause of Ring Ouzel declines in the UK, with warmer and drier late summers possibly reducing availability of invertebrate and berry food sources (Beale et al. 2006).

Climatic effects may also be apparent through changes in the altitudinal distribution of different bird species. Dartford Warbler, Stonechat and Common Whitethroat showed declines at higher altitudes on Exmoor. The severe winters of 2009–10 and 2010–11 in the UK (Taws et al. 2011) probably reduced Stonechat and Dartford Warbler numbers directly through known effects on increased mortality, especially at higher altitudes where conditions would be expected to be harshest (Dobinson & Richards 1964, Wotton et al. 2009). Potential reasons for the decline in Common Whitethroat at higher altitude are less clear and warrant further research. Conversely, Common Snipe, Tree Pipit and Yellowhammer increased at higher altitudes. The increase in Common Snipe might potentially reflect a response to re-wetting of mire areas, which mainly took place on higher ground. Tree Pipit and Yellowhammer increases at higher altitudes may reflect the spread of preferred scrub habitat into such areas in recent years (Cramp 1988, Fuller et al. 2006).

Effects of migratory strategy on bird abundance change

Population changes of resident species breeding on Exmoor during 2008–14 were less likely to be negative than those of species which migrate to winter in other areas. These findings are partly in contrast to the UK migratory bird indicator, which found that only short-distance migrants showed positive trends, while resident species and longer distance migratory species that winter in the arid zone were generally stable, and humid/tropical zone species showed the largest declines during 1970–2012 (Hayhow et al. 2014). However, the suite of species for which such data were available was much smaller on Exmoor than for the whole of the UK, so these results need to be treated with some caution. Nonetheless, increases in abundance for long-distance migrant species such as Common Cuckoo, Willow Warbler, Common Whitethroat and Common Redstart, plus the relative stability of Whinchat and Tree Pipit, suggest that Exmoor continues to provide habitats of considerable importance for these species of high conservation concern.

Acknowledgements

We are very grateful to the following for their advice and support during the project: Ali Hawkins and Matt Sully (ENPA), David Smith (Exmoor Mires Project), Flemming Ulf-Hansen (NE) and Helen Booker (RSPB). We are grateful to all landowners for access during surveys. Fieldwork was conducted by Lee Dark, David Jarrett, Mary Richardson, John Tayton and Nick Unwin. David Ballance (Somerset Ornithological Society) kindly shared his knowledge of the birds of the Exmoor area. Greg Miller produced figure maps, Alex Sansom assisted with project setup and Staffan Roos provided statistical advice.

Additional information

Funding

This work was funded by Exmoor National Park Authority (ENPA), Exmoor Mires Project/South West Water and Natural England (NE).