Management of cover areas may increase numbers of breeding Corncrakes Crex crex

Abstract

Capsule Density of Corncrakes is greater in areas of positively managed non‐crop cover vegetation which occur alongside fields managed to enhance breeding success.

Once an iconic farmland bird of Britain, Corncrakes Crex crex began declining in the late 19th century and are now largely restricted to the Hebridean and Orkney Isles of Scotland (Norris 1945, Hudson et al. 1990, Williams et al. 1991, Stowe et al. 1993, Green & Gibbons 2000). Intensification of agriculture and changes in management from the 1940s onwards are considered to have contributed substantially to the rapid decline in numbers (Williams et al. 1991, Fuller et al. 1995, O’Brien et al. 2006). The causes of the species’ decline include direct mortality of adults and chicks by mowing machinery, reduction in the quality and availability of breeding grounds through conversion from hay to silage‐making, earlier mowing dates and increased number of cuts per year (Hudson et al. 1990). Considerable effort has been made to determine more precisely the causes of these declines and to design and implement management plans in order to halt and reverse them. Management prescriptions rolled out since the 1990s thus specify non‐destructive and suitably timed mowing regimes in hay meadows and silage fields, and provision of unmown areas (Tyler et al. 1998, Green & Gibbons 2000).

Corncrake numbers have risen in recent years from the lowest recorded level of 480 singing males in 1993 to 1236 in 2007, although a slight reduction to 1140 was recorded in 2008 (RSPB unpubl. data, O’Brien et al. 2006). While there has been much literature documenting the decline of the species, there is a lack of evidence assessing which elements of the current management prescriptions are most effective in increasing breeding numbers. Here we analyse quantitatively the effects of Corncrake‐orientated habitat management on choices of breeding habitat and, thus, their effects on population change.

During the summer Corncrakes require tall vegetation of at least 20 cm in height, which is dense enough to provide cover from predators, but not so dense that it restricts movement (Koffijberg & Schaffer 2006). These vegetation features conceal young birds and eggs, and adults during moulting (Williams et al. 1991). Areas of ‘cover’, i.e. strips and pockets of suitably structured vegetation where Corncrakes can seek refuge (Koffijberg & Schaffer 2006), can thus be imperative to the birds both early in the season when grass in hay/silage fields is too short and later in the season after fields are cut (Williams et al. 1991).

Three farms on the Isle of Islay, Scotland were examined. These varied in both the level and extent of Corncrake‐orientated management: high management (Loch Gruinart: NR285665), intermediate management (Ardnave Farm: NR282731) and minimal management (Craigens Farm: NR298672). Loch Gruinart is an RSPB reserve and is one of three such reserves on the island, which has been carrying out management to benefit Corncrakes for approximately 15 years; areas of cover habitat managed specifically for Corncrakes are situated across a network of silage fields, covering around 38 ha (2.2%) in total. Ardnave has been under management agreements with Scottish National Heritage and the RSPB over the last 10 years (RSPB unpubl. data) and has around 0.2 ha (0.2%) managed as cover. Craigens is managed for agricultural purposes, with some management prescriptions for Corncrakes included within a Rural Stewardship Scheme (RSS) agreement, but the farm manager states that such management is applied minimally.

Data collection was conducted during 2008. We focussed on three aspects of vegetation structure known to affect choice of breeding habitat by Corncrakes: height, density and species composition. Both cover and silage fields are used as Corncrake breeding habitat, so a paired cover–silage approach was used, with two silage sites and two cover sites sampled within each management category (farm). Each two silage and cover sites per farm were chosen as being representative of the habitat type at each farm, with the exception of the cover areas of Craigens Farm. Here, two areas of abandoned rough land where Corncrakes are occasionally found breeding were analysed as no Corncrake‐managed cover areas existed at this farm. As Corncrakes require areas of tall grass throughout the breeding season, cover areas were sampled twice, once in June and again in August, while silage fields were sampled once in June. Silage fields were not surveyed during August as fields had already been cut at this time. Within each sample site, 36 × 0.5‐m quadrat sample points were created across a 100‐m² area, each yielding one measure of vegetation density, five measurements of vegetation height and the percentage cover of each plant species present. Sample points were created every 20 m along six 100‐m transects spaced 20 m parallel to each other (including beginning and end points, thus yielding 36 samples in total), except in two sites where the shape of the site did not conform to a grid pattern. In these sites the length of transects was altered so that measurements could still be taken every 20 m in parallel. The sampling grids were located immediately within the entrance to the sites so as to minimise surveyor bias. Vegetation density was estimated as the percentage volume occupied by plant material within a hypothetical cuboid of 0.5 × 0.5 × 0.2 m² area to the nearest 5% (the upper limit of 0.2 m was used because Corncrakes are known to preferentially occur in vegetation of 20 cm or more in height). Estimates could thus range from 0% (bare earth) to 100% (entire area occupied by plants). Corncrake counts were carried out across each of the three farms as part of the RSPB’s annual surveys across the island in June and early July, following Hudson et al. (1990) and Green & Gibbons (2000). The lack of replicate farms within each management category imposes some limitations on the scope of the analysis and interpretation of the results; however, we believe that while taking this consideration into account, the results of the study nonetheless offer potentially useful insights for Corncrake management.

A factorial repeated‐measures anova design was used to test for differences across farms and field type in June sward height and density datasets while a one‐way anova design was tested for August datasets. None of the four data sets fully satisfied anova assumptions of normality and homogeneity of variances, even after various transformations of the data, with the exception of the sward height August dataset with Log₁₀ transformation. anova can be reliable in cases where assumptions are violated (see review by Thode [2002]), but to support the analysis non‐parametric Friedman tests were performed for all four datasets. A two‐tailed Pearson moment correlation coefficient was calculated to investigate correlations between choice of breeding habitat by Corncrakes and differences in habitat management across farms and habitat types.

Eighty‐one singing male Corncrakes were recorded on Islay during the breeding period of 2008, of which 16 (19.8%) were located at Loch Gruinart, 8 (9.9%) were found at Ardnave and 2 (2.5%) at Craigens. Of those found across the island, 36% were located within the three RSPB reserves, which collectively account for 6.6% of the island area.

Of the 16 singing Corncrakes located at Loch Gruinart, 12 were within areas of cover habitat, compared with four within silage fields. Three Corncrakes were in silage fields at Ardnave compared with only one in cover habitat (the remaining four were located in other types of grassland); at Craigens the two Corncrakes were within one silage field. Overall, 13 Corncrakes were found in cover areas, while nine were in silage fields. Both number and density of birds was highest at Loch Gruinart RSPB Reserve (0.94 birds km⁻²) and lowest at Craigens Farm (0.15 birds km⁻²), with Ardnave achieving intermediate values (0.74 birds km⁻²; see Endnote a). Where density of Corncrakes was analysed independently within cover and silage areas (calculated as the total number of birds within each habitat type per farm divided by the total area of distinct silage/cover areas where birds were located), it increased with increasing availability of managed cover habitat, with Loch Gruinart achieving the highest density of birds (Fig. 1). In silage areas the opposite trend was found: density of Corncrakes was lowest at Loch Gruinart and highest at Craigens: more birds were utilising silage fields where less managed cover was present. Total areas of farms, areas of land under silage production and cover habitat can be found in Table 1. Density of birds in Table 1 refers to density on a farm scale rather than in terms of areas utilised by the birds as in Fig. 1.

Figure 1 Density of singing male Corncrakes located within cover and silage habitat types in Loch Gruinart, Ardnave and Craigens Farms (Islay, Scotland) during the breeding period of 2008.

Table 1. Numerical data concerning areas of land and Corncrake occurrence across the three farms under analysis in 2008.

	Loch Gruinart RSPB reserve	Ardnave Farm	Craigens Farm
Total area of farm (km^2)	16.99	10.82	13.61
Number of Corncrakes recorded within each farm area	16	8	2
Within cover areas	12	1	0
Within silage fields	4	3	2
Density of Corncrakes across total area of farm (birds km^−2)	0.942	0.740	0.147
Total area of fields under silage production (km^2)	1.269	0.874	0.877
% of total farm area	7.47	8.08	6.44
Density of Corncrakes across total area of fields under silage production (birds/ km^2)	3.152	3.434	2.281
Total area of ‘cover’ actively managed for Corncrakes (km^2)	0.377	0.022	0
% of total farm area	2.22	0.2	0
Density of Corncrakes across total area of areas managed as cover habitat (birds/ km^2)	31.76	44.96	0
Total area of silage and cover area per farm combined (km^2)	1.647	0.896	0.877
Density of Corncrakes across silage and cover areas combined (birds km^−2)	9.715	5.58	0.439
Density of Corncrakes refers to density across the whole farm area; density of Corncrakes across silage and cover areas combined (birds km^−2) is calculated as the number of birds per farm divided by the total area of silage and cover area per farm combined.

One mean sward height score and one mean density score per farm for each habitat category were calculated (Fig. 2). Differences in June sward height across farms, field type and the interaction between these two variables were all found to be highly significant (F _{1.69, 1.2} = 71.1 P < 0.001, F _1,71 = 17.0 P < 0.001 and F _2,142 = 76.7 P < 0.001 respectively; farm F‐ratio corrected using Greenhouse–Geisser correction). The significant main effect of farm (P = 0.001) reflects differences between Loch Gruinart and Craigens, and Ardnave and Craigens; as analysed using a Bonferroni adjustment (Abdi 2007). Differences in June density across farms, field type and the interaction between these two variables were also all found to be highly significant (F _{2, 142} = 43.0 P < 0.001, F _1,71 = 1.3 P < 0.001 and F _2,142 = 72.2 P < 0.001 respectively). Again, a significant difference (P = 0.001) between Loch Gruinart and Craigens and Ardnave and Craigens was found, whereas no significant difference was found between Loch Gruinart and Ardnave. The significant interaction results indicate that both sward height and density respond in different ways across management type (farm), according to field type. Results of Friedman’s tests to address the limitations of the data supported this analysis (see Endnote b).

Figure 2 (a) Sward height; (b) mean density of vegetation within cover areas Phase 1, cover areas Phase 2, and silage sites (mean ± se), across habitat types and farms (n = 72), in Loch Gruinart, Ardnave and Craigens Farms (Islay, Scotland) during the breeding period of 2008.

For August data (cover only), differences across farms of sward height (Log₁₀) and density were both found to be highly significant (F _{2, 213} = 9.21 P < 0.001 and F _{2, 213} = 13.43 P < 0.001 respectively). The Bonferroni post‐hoc test of the Log₁₀ mean sward height showed the differences between Loch Gruinart and Craigens, and Craigens and Ardnave were highly significant (P = 0.001), while the differences between Loch Gruinart and Ardnave were not. For density, differences between Loch Gruinart and Craigens were highly significant (P = 0.001), while the differences between Loch Gruinart and Ardnave, and Ardnave and Craigens were significant (P = 0.023 and P = 0.04, respectively).

Species composition within cover sites (June and August) and silage sites were analysed by calculating the mean percentage cover of ‘favourable’ species (most preferable to Corncrakes [Cadbury 1980; Green et al. 1997; Green 1996]) across farms and habitat type, namely: Stinging Nettle Urtica dioica, Yellow Flag Iris Iris pseudacorus, Reed Canary Grass Phalaris arundinacea, and Common Reed Phragmites australis (grouped as ‘Phalaris spp.’ in this paper), Rushes Juncus spp. and Hogweeds Heracleum spp. These species have been found to provide the most suitable breeding habitat for Corncrakes due to their physical structure (Green et al. 1997). No benefit from diversity of such species within cover areas has been described.

Craigens had the greatest cover of Juncus spp., I. pseudacorus, Phalaris spp. and Heracleum spp. in both June and August cover areas. In June cover U. dioica was in greatest proportion in Ardnave, but in August cover it was greatest in Loch Gruinart. Overall, Juncus and Phalaris spp. (in June cover) and Juncus spp. (in August cover) were less abundant in managed cover, while U. dioica (in August cover) was present in larger proportion in managed cover areas; the latter perhaps owing to disturbance and nutrient enrichment from winter grazing cattle (Pigott & Taylor 1964) as a result of management prescriptions. In silage fields, three favourable species were found at Loch Gruinart ( Juncus spp., U. dioica and Heracleum spp.) and two were found at Ardnave ( Juncus spp. and U. dioica), while there were none at Craigens. Friedman’s test identified significant differences between farms in cover of several species (see Endnote c).

For June cover areas, August cover areas and silage areas, the mean sward height, mean density, and mean overall percentage cover of favourable species (calculated by taking the mean of all six favourable species) were tested for correlation with Corncrake density. Two correlations were significant: mean sward height of June cover areas was inversely correlated with bird density (r = −1.0, P = 0.002, n = 3), as was the overall mean percentage cover of favourable species (r = −0.998, P = 0.043, n = 3), indicating that as mean sward height and cover of favourable species in early cover increased, bird density decreased. Correlations calculated using count data of Corncrakes and by combining all mean sward height data, all mean density data and all favourable species cover data, irrespective of habitat type, were not found to be significant.

Corncrake density increased with increasing amounts of managed cover areas across the farms in our study. As sward height and overall vegetation cover within cover areas increased, bird density decreased. Interestingly, in both June and August cover, the mean sward height of vegetation in cover areas across all three farms was greater than the advised minimum of 20 cm; however, no Corncrakes were found in the areas of Craigens sampled as cover areas, although Corncrakes have nested there historically (RSPB unpubl. data). Craigens cover areas were, on average, taller and denser than the other two farms’ cover areas, in what were Juncus and Phalaris‐dominated cover habitat. Where such vegetation becomes very tall, it is typically denser at ground level and consequently becomes unsuitable for Corncrakes. Green (1996) noted that Juncus‐dominated habitats tend not to be used by Corncrakes, as clumps may be too dense for the birds to penetrate. The density and height of vegetation within positively managed cover areas at Ardnave and Loch Gruinart were more suitable for Corncrakes owing to differing species composition as well as physical structure. In addition, Loch Gruinart had a greater number of patches and total area of cover habitat than the other two farms; around 19 times more land was under management as cover areas at Loch Gruinart than at Ardnave. Thus the fact that Corncrake density was considerably higher in cover areas at Ardnave than at Loch Gruinart, despite comprising a much smaller area, strongly suggests that cover habitat as acting as a limiting factor at Ardnave.

The presence of two Corncrakes in silage sites at Craigens, while none were found in Craigens ‘cover’ (abandoned land) suggests that silage fields provide a more attractive habitat than the cover areas at this farm. As silage fields at Craigens are cut early in the season with a second cut in August in most years and without the use of ‘Corncrake‐friendly mowing’ (Tyler et al. 1998), Corncrake nests, young and adults are at risk from mowers when the grass is cut. The absence of suitable cover across the farm suggests that the silage fields at this farm may be acting as an ecological trap (Schlaepfer et al. 2002).

The density of Corncrakes at Loch Gruinart and Ardnave was far greater in cover areas than in silage fields, indicating that where good cover habitat is provided, Corncrakes are more likely to breed in cover areas than in silage (or hay) fields. The opposite trend apparent at Craigens suggests that where good quality cover is not available, Corncrakes are more likely to utilise silage sites. Berg & Gustafson (2007) found that in Swedish sites, the majority of Corncrakes were found in tall vegetation, produced without subsidies to farmers, rather than managed meadows. Our results suggest that in addition to the use of Corncrake‐friendly mowing, the most beneficial management practice is the provision and maintenance of good‐quality cover sites. When managed to limit appropriately height and density of the vegetation, these cover areas can provide breeding habitat for Corncrakes throughout the season and without risk of destruction by mowing.

At present, many of the Corncrake management prescriptions focus on late cutting of silage fields and the practice of Corncrake‐friendly mowing for Corncrake conservation (O’Brien et al. 2006). The results of this study suggest, however, that subsidy payments would benefit from a greater focus towards the creation and management of cover areas for Corncrakes. It is the authors’ observation that the current focus on silage management in conservation strategies appears somewhat at odds with the needs of most farmers; delaying the timing of silage cuts into August increases the risk of a poor‐quality silage crop and, thus, this risk may be considered too great for most farmers to take, limiting the uptake of subscriptions and thus the range of the species. Greater focus on the provision of sufficient cover areas, managed to have a suitable physical structure in subsidy payments may, therefore, not only bring greater benefit to Corncrake populations, but may also attract more farmers to Corncrake conservation, with potential to increase the range, as well as the population of this species.

ACKNOWLEDGEMENTS

Many thanks to James How, Tom Epps, Craig Archibald, Cath Fotheringham, Michael Coppleston, Michael Sur, Eion Brown, Andy Schofield, Dr Patrick Doncaster and John Jones. Many thanks to two reviewers, whose suggestions greatly improved the manuscript.

Notes

a. Number and density of Corncrakes was calculated from the number of birds found across the entire area of each farm and the entire area of cover and silage habitat within each farm (Table 1).

b. Differences in June density in cover habitat across farms were not found to be significant (χ² = 3.0, P = 2.0), while differences in silage habitat across farms and differences across farms irrespective of habitat type were found to be highly significant (χ² = 90.0, P = 0.001 and χ² = 32.9, P = 0.001, respectively). Differences in June sward height in cover habitat across farms (χ² = 18.6), in silage habitat across farms (χ² = 93.8) and across farms irrespective of habitat type (χ² = 73.1) were all found to be highly significant (P = 0.001). Differences in August density across farms (χ² = 17.7) and sward height across farms (χ² = 15.5) were also found to be highly significant (P = 0.001).

c. Significant differences between farms in mean percentage cover of several species were identified by Friedman’s test: June and August cover areas of U. dioica, (χ² = 21.3, P = 0.001 and χ² = 10.2, P = 0.006, respectively); Heracleum spp. (χ² = 10.8, P = 0.005 and χ² = 8.9, P = 0.012, respectively), Phalaris spp. (χ² = 18.0, P = 0.001 and χ² = 18.5, P = 0.001), Cynaeae spp. (χ² = 7.28, P = 0.026 and χ² = 17.6, P = 0.001) and of Juncus spp. August cover and silage (χ² = 7.4, P = 0.025 and χ² = 14.9, P = 0.001, respectively).