Hydrobiologia 279/280: 297-307, 1994. J. J. Kerekes (ed.), Aquatic Birds in the Trophic Web of Lakes. © 1994 Kluwer Academic Publishers. Printed in Belgium.
297
Breeding bird communities and environmental variable correlates of Scottish peatland wetlands A. D. Fox & M. C. Bell The Wildfowl & Wetlands Trust, Slimbridge, Gloucester GL2 7BT, UK
Key words: waterfowl, breeding, wetlands, peatlands, Scotland
Abstract The majority of lochs and water courses in an area of northern Scotland were visited during July 1988. Breeding waterfowl and details of environmental variables were recorded for each site and subjected to multivariate classification techniques. Non-hierarchical classification identified seven habitat types on the basis of environmental parameters. These corresponded well with, and further sub-divided, categorisation using conventional trophic-level habitat type classifications. A hierarchical approach, using TWINSPAN, classified twelve waterfowl groupings based on the presence of indicator species. This approach identified characteristic community types and groupings with high diversity and hence conservation importance. Correspondence between habitat classifications and waterfowl communities was reasonable, but failure to identify key features of wetland complexity was probably the reason for the inability to identify clear relationships. Induction analysis was used to show that waterfowl species with the most restricted distributions characterised the main groupings, with Black-throated Diver and Common Scoter showing preferences for mesotrophic/eutrophic waters with islands, where peaty waters drain onto base-rich sandy substrates. Greylag Geese occurred on large acidic oligotrophic lochs and Wigeon on base-rich streams. The need for catchment-wide site-safeguard and management programmes to safeguard the wetlands of greatest significance is discussed.
Introduction
Anser anser (L.), Wigeon Anas penelope L. and
Common Scoter Melanitta nigra (L.) (Stroud The peatland areas of Caithness and Sutherland in Northern Scotland support northern bird communities not encountered elsewhere and are of considerable conservation importance in British, European and world terms (Stroud et al., 1987). Although most previous studies have centred on the nesting wading birds of the region (e.g. Reed et al., 1983), the breeding waterfowl are also of considerable significance, supporting nationally important numbers of breeding Red-throated Diver Gavia stellata (Pontoppidan), Blackthroated Diver Gavia arctica (L.), Greylag Goose
et al., 1987; Fox et al. 1989).
Commercial afforestation of bogland biotopes in Britain has been shown to alter the nature of the water bodies of a catchment through increasing sediment load (Robinson & Blythe, 1982; Burt et al., 1983; Batterbee et al., 1985; Francis & Taylor, 1989), increasing nutrient status through runoff (Hornung & Newson, 1986) or fertiliser loss (Harriman, 1978) and increasing acidification (Stoner & Gee, 1985). Extensive planting of the Sutherland and Caithness peatlands during the 1980s threatened their important wetland bird
298 communities, including those associated with rivers and lochs. As part of the current review of their wildlife importance, there is a clear need to identify and safeguard the most important wetlands and their breeding waterfowl communities. In an earlier analysis, we used ordination techniques and univariate analysis to look at habitat selection by waterfowl in this important region (Fox et al., 1989). In the present analysis, we attempt to describe species and habitat associations using multivariate classification techniques to identify the range of variation and wetland types present in the study area. In addition, we have applied induction techniques to identify habitat features which define the presence/ absence of the waterfowl species of highest conservation importance.
Methods All lochs and major river courses in the peatland complex of Caithness and Sutherland within 17 National Grid 10 km squares were visited during 9 to 30 July 1988 (see Fox et al., 1989 for details). One hundred and eighty-six sites were visited at least once, most twice and a few on more occasions. Physical features of the lochs were recorded in the field (namely number of islands, water colour, water clarity, dominant shore substrate, emergent vegetation, and habitat type) or derived from 1:10 000 maps (area of waterbody and altitude) and geological maps (basal geology). A water sample from each site was taken during the survey and analysed for pH and conductivity using a Whatman 2500001 portable pH/ temperature probe and Whatman conductivity sensors. Environmental data from 136 sites for which there were complete data were included to classify each site into habitat types, using a nonhierarchical classification technique (Webster & Oliver, 1990). This dataset included sites without any waterfowl present and was based on eight variables: altitude, number of islands, conductivity, pH, water colour, water clarity, substrate type and logarithmically transformed site area. Varia-
ble scores used in the analysis are shown in Table 1. Using this technique, the criterion for optimisation was taken as the within-class sums of squares, which makes no assumptions of multivariate normality or equal within-class dispersion (Genstat 5 Committee, 1989). Classifications were derived for between two and twelve groupings. The optimal number of groups was identified by the inflection point on the plot of optimisation criteria against number of groups and the classification based on seven groupings chosen in this way. Bird abundance was recorded as the maximum count of each species from all visits to each site. A classification of bird community types based on these data was derived using the divisive clustering technique embodied in the programme TWINSPAN (Hill et al., 1975; Hill, 1979). This technique, known as two-way indicator analysis, divides groups of sites into two on the basis of the first axis of an ordination. Groups are further split by successive ordinations of the sub-groups so defined. At each level of division, the species contributing most to the ordination are defined as indicators, characterising the groups defined at that level. Greater weight was placed on the presence/absence of a species (pseudospecies 1 see Hill et al., 1975) than on absolute numbers (pseudospecies 2, defined at a threshold level of 5 birds). Minimum group size was set at 15 sites and the analysis was limited to five divisions. We then attempted to describe the features of wetlands which best determine their suitability for Table 1. Scores used in the analysis of habitat variables relating to water courses and lakes in Caithness and Sutherland during July 1988. All other variables were recorded as logical continuous variables. Parameter score
Water colour
Water clarity
Substrate types
1 2 3 4 5 6
Colourless Pale brown Light brown Brown Dark brown -
Clear Slightly cloudy Turbid Peaty Very peaty -
Boulders Stony Sandy Silty Organic mud Peat
299 Table 2. Summary table of environmental variables characterising the seven groupings of waterbody sites from Caithness and Sutherland, July 1988. Classification is based on a non-hierarchical classification technique described in the text. + indicates high values of each variable, - indicates low values. See Table 3 for characterisation using mean variable scores for each grouping. Group
n
1 2 3 4
20 11 24 5
5 6
12 29
7
35
Area
Altitude
-
+
+
Islands
Conductivity
pH
Colour
Clarity
Substrate
+ +
+
+
+
+ +
+ + +
+
species of high conservation interest (Blackthroated Diver, Common Scoter, Greylag Goose and Wigeon). The rule-induction programme PC/ BEAGLE (Forsyth, 1987) was used to examine the full set of environmental variables and find a set of rules which best discriminated between sites with and without a given species. The programme is based on a machine-learning technique, applying an evolutionary strategy for improving on a random initial set of rules.
The mean values of the environmental variables are shown in Table 3, there were significant differences between groups (shown with confidence limits in Fig. 1). The seven groups can be characterised as follows. Group 1 are midaltitude, peaty sites with highly coloured water, similar to the upland dubh-lochans with relatively low pH comprising Group 3. Group 2 are small lowland water bodies with a high pH, generally in peaty catchments and hence with highly coloured peaty water, but often on sandy substrates. Group 4 is a well defined set of large lakes with islands and Group 5 represent lowland sites with high conductivity measurements, colourless water (i.e. not draining from large peatland catchments) generally on coarse substrates. Group 6 differs from Group 5 in higher altitude situations and lower relative conductivity. Finally, Group 7
Results Non-hierarchicalclassification of habitat types Cluster analysis of environmental data produced the seven groupings shown in Table 2.
Table 3. Mean values for each environmental variable in each of the seven groupings of waterbody sites from Caithness and Sutherland, July 1988. Classification is based on a non-hierarchical classification technique described in the text. See Table 1 for specific variable scores. Analysis of variance amongst all environmental parameters for different groups was carried out and F-ratio and probability values given below. Group
n
Area
1 2 3
20 11 24
8.1 0.1 2.5
120 98 154
0.0 0.0 0.0
4
5
27.6
146
2.6
5
12
9.4
89
0.0
6 7
29 35
9.8 16.8
167 142
0.1 0.1
88.4 85.7
F6,135= p<
29.6 0.001
Altitude
17.0 0.001
Islands
98.8 0.001
Conductivity
Colour
Clarity
5.62 6.96 4.67
3.2 3.2 2.5
2.1 1.1 1.0
4.5 3.0 5.8
97.0
5.52
2.0
1.2
2.0
141.4
6.54
1.6
1.0
2.1
6.04 6.26
1.5 3.5
1.0 1.0
2.4 2.2
13.5 0.001
16.9 0.001
149.2 0.001
53.5 0.001
95.6 95.0 81.2
19.1 0.001
pH
Substrate
300 120.
LAKE AREA 9,
100. 80
87.
60
6 5.
40-
4.
.
.
i
5
6
7
20 j
. 1 200
2
3
4
*
$
321
ALTITUDE
1
2
1
2
I
3
I
I
I
4
5
6
7
WATER COLOUR 4-
180'
t
f
160 140-
f
I
3. 0
= 120-
2
100-
l1
II
80-
1
hn
1
2
.
.
.
4
5
6
.
.
.
3
.
7
1
NO. ISLANDS
2
3
4
5
6
7
a
i
i
5
6
WATER CLARITY 2.0
3.
2
'a
.s
0.
1.0on
: -
:
:·
.
1
2
3
4
5
6
6
140'
5'
120-
0 U
80-
f 1
tt 2
3
4
7
SUBSTRATE
160
100-
2
1
7
CONDUCTIVITY
-
a
·
3
\f jt 4
5
Cluster Group
6
7
o
4.
VI
3.
a
a
.11 1
2
3
4
5
6
7
Cluster Group
Fig. 1. Characterisation of environmental factors from each group of wetlands produced by the non-hierarchical cluster analysis. Bars indicate confidence intervals about mean values.
301 comprises waters which are large, highly coloured (i.e. peatland-draining) on coarse substrates. We compared this classification with a field assessment of the nutrient status of each waterbody, based on simple categories of dystrophic, oligotrophic peaty, other oligotrophic, mesotrophic and eutrophic (after the criteria of Nature Conservancy Council 1989). A comparison of the two classifications is shown in Table 4, showing the high degree of correspondence between the field determinations and the analysis classifications.
Hierarchicalclassification of bird associations TWINSPAN classified the sites based on bird data through five division levels, defining twelve groupings as shown in Fig. 2. Internal homogeneity within the groupings was extremely high and the groupings were biologically meaningful. Three low diversity groups were characterised by the presence of Common Scoter, a) with no other constant species (000), b) with Black-throated Diver (0010) and c) with Mallard Anas platyrhynchos L. (0011). Scoters were also present in two other groups with high diversity of duck species, both of considerable conservation importance (01000 and 01001). Mallard and Mallard/Teal Anas crecca L. waterbodies were low diversity assemblages with few associated species (01011, 01010), and Tufted Duck Aythya fuligula (L.) and Red-breasted Merganser Mergus serratorL. characterised two further groups comprising three sites (011, 101). The remaining sites fell into two groups with Red-throated Diver and Greylag
Geese as constants (1000, 1001). The mean habitat variables are shown for each group in Table 5, there being significant differences between the groups in area, altitude, pH and conductivity. The distribution of sites between the TWINSPAN groups and the non-hierarchical habitat classification are shown in Table 6.
Determinants of species richness In an attempt to find the determinants of species richness, we analysed the mean number of species in each TWINSPAN grouping against the mean environmental variables for each set of sites. The resulting scattergrams show little correspondence between species richness and environmental parameters except for pH and conductivity, where each shows a positive correlation (pH, r12 = 0.57,
p<0.01; conductivity,
r
2
= 0.34,
p>0.05; removing the two groups which have only 3 sites in each group and correspondingly high variance in water chemistry determinations improves these to r 10 =0.68, p< 0 . 0 0 1 and r 0 =0.51, p<0.01 respectively).
Rules for determination of habitat requirements of rare species The dendrograms derived from the induction analysis for each of the four species are shown in Table 7, with the success rate of using these rule sets to correctly predict presence/absence of each species shown.
Table 4. Comparison of the classification of waterbodies from the Caithness and Sutherland survey of July 1988 based on the Nature Conservancy Council classifications of water habitat type and the results of the non-hierarchical analysis summarised in Table 2. Integers indicate the number of lakes falling within each category of the two classifications. NCC category
Dystrophic Oligotrophic peaty Other oligotrophic Mesotrophic Eutrophic
Group 3
1 20 3 0 0
Group 1
Group 7
Group 6
0 10 5 5 0
8 16 10 0
2 19 7 0
1
Group 5
Group 4
Group 2
0 3 6 3 0
0 0 3 2 0
0 0 0
9 2
302 TWINSPAN
SPECIES CODES
Caithness and Sutherland Waterfowl Survey 1988
Aa - Greylag Goose Ac - Teal Af - Tufted Duck A_
mrl:
...
Speci
er
11
101
1001
1000
011
01010 01011 01000 01001
0011
0010
000
Fig. 2. TWINSPAN hierarchical classification of site groups based on bird species from Caithness and Sutherland wetlands, based on survey data from July 1988. Species presence/absence characterising the groupings are indicated.
303 Table 5. Mean values for teach environmental variable in each of the 12 groupings of waterbody sites from Caithness and Sutherland, July 1988. Classification is based on a TWINSPAN hierarchical classification as described in the text. See Table 1 for specific variable scores. Analysis of variance amongst all environmental parameters for different groups was carried out and F-ratio and probability values given below, although comparisons between islands, clarity and substrate were not possible because of uniformity of several groups. Group
Constant species
Species richness
n
Area
Altitude
Islands
Conductivity
pH
Colour
Clarity
Substrate
000 0010 0011 01000 01001 01010 01011 011 1000 1001 101 11
Ga, Mn Mn Ap, Mn Ape Ap Ac, Ap Ap Af Gs Aa Ms Ga
2.75 1.20 2.20 3.56 6.25 1.92 1.00 2.33 1.31 2.73 1.33 1.22
4 10 5 17 4 13 7 3 13 11 3 9
17.9 16.6 10.4 17.3 57.2 7.5 3.3 9.9 218.0 25.4 21.2 62.8
144.0 152.5 153.8 119.4 122.5 150.0 110.0 157.9 109.5 120.0 152.9 142.9
0.2 0.3 0.0 0.1 1.3 0.3 0.0 0.4 0.0 0.0 0.0 0.0
100.4 83.8 91.0 92.9 101.8 84.5 93.3 81.8 81.3 102.7 87.9 95.9
6.18 6.53 6.19 6.52 7.25 5.35 6.40 5.45 5.75 6.77 6.59 5.57
1.8 2.0 2.8 3.0 3.0 2.3 2.3 2.8 3.2 1.3 2.9 1.6
1.4 1.8 1.4 1.8 1.0 1.7 1.0 1.5 1.8 1.0 1.4 1.0
4.2 3.0 3.4 3.0 2.8 3.9 4.0 3.0 2.6 2.3 1.7 3.0
2.56
3.94
1.89
-
-
0.001
0.001
0.052
F6,135=
3.45
2.22
p<
0.001
0.021
-
Species codes as follows: Aa - Greylag Goose; Af- Tufted Duck; Ape - Wigeon; Gs - Red-throated Diver; Ms - Red-breasted Merganser; Ac - Teal; Ap - Mallard; Ga - Black-throated Diver; Mn - Common Scoter.
Table 6. Comparison of the classification of waterbodies from the Caithness and Sutherland survey of July 1988 based on the TWINSPAN classification of water habitat type based upon breeding waterfowl and the results of the non-hierarchical analysis summarised in Table 2. The groups are ordered approximately from oligotrophic (Group 3) to eutrophic (Group 2).
Group Group Group Group Group Group Group
3 1 7 6 5 4 2
1001
0010
000
1 2 6 1
1 3 1
1 2
0011
1000
1
1 2 2 4 1 1
1 1 1
1
Discussion The use of TWINSPAN proved extremely useful in summarising the type of assemblages present and was very effective in identifying twelve waterbody types based on the constant presence of one or two waterfowl species. Since indicator species included waterfowl with high conservation importance (Black-throated Diver, Red-throated Diver, Common Scoter and Greylag Goose), such
01000
11
5 2 2
1 4 3
4
01011
01010
101
1
4 1
1
5
1 1 1
2 4 1 1
1
01001
011
1
1
3
2
a classification gives a firm basis for site safeguard based on these assemblages for single species conservation. Perhaps more important, the analysis has also identified three community types with very high waterfowl species richness, communities remarkable for the high number of species present. Some of these lochs also support the rarer species and hence contribute to the overall strategy for site safeguard of single species. The classification of habitat types based on the
304 Table 7. Signature Tables summarising induction rules generated by the PC Beagle (Forsyth 1987) program predicting presence/absence of four rare waterfowl species based on environmental variables gathered in Caithness and Suterland during July 1988. The program generates rules as indicated, if the rules are satisfied, score a one, if not, score zero. Based on these responses, the combinations of zeros and ones give rise to presence absence considerations as indicated. 'No combination' indicates that this combination of characters was not represented amongst the lakes studies. 1. Black-throated Diver Rules: A. Large site (> 32 ha) with high pH (> 5.54) B. Islands present and fine substrates (sandy or finer) Signatures: i - present 10 - present 01- no combination 00 - absent Prediction success against observed data: 86.8% 2. Greylag Goose Rule: A. Large site (> 38 ha) with low conductivity (< 100) Signatures: 1- present 0 - absent Prediction success against observed data: 91.2% 3. Wigeon Rules: A. River site B. High pH (> 5.68) and fine substrate (sandy or finer) Signatures: 11 - present 01 - absent 00 - absent Prediction success against observed data: 89.7% 4. Common Scoter Rule: A. High pH (>4.82), fine substrate (sandy or finer), conductivity low (< 132), large area (> 8 ha), low altitude (< 154 m) and water lightly coloured Signatures: 1- present 0 - absent Prediction success against observed data: 89.7%
environmental variables measured in the field showed a clear correspondence to traditional classifications used in Britain. It proved extremely effective in sub-dividing the simple trophic states into meaningful ecological units in the context of the peatland waters. In turn, the habitat classification gave some basis for explaining the distribution of the different bird communities identified by TWINSPAN groupings. However, some of the birds' groupings straddled both dystrophic and more eutrophic waters and it became clear that nutrient status in the broadest sense was not the only determinant of species diversity or species composition within the waterfowl communities of the Caithness and Sutherland wetlands. The reasonable but limited correspondence between habitat and bird classification may partly be due to the nature of the variables measured, some of which perhaps fail to be of sufficient importance to waterfowl to produce obvious linkage. This analysis supports the earlier ordination approach in identifying elevated pH and conductivity levels as being associated with high ornithological conservation interest, either through high waterfowl species richness or the presence of rarer species (e.g bird groups 0010, 01000, 01001, 011). However, it is clear that the structural diversity of the habitat is also important. Hence, while the sites with the greatest waterfowl diversity have the highest pH and conductivity measurements, they are also streams or lochs with islands or areas of flood-plain associated, always at low altitude with clear peaty water over a sandy substrate. These lakes therefore combine features attractive to the common and ubiquitous species with high pH and conductivity (known to be important for species such as Black-throated Diver and Common Scoter) as well as supporting rarer species (for Caithness and Sutherland) such as Wigeon, Gadwall Anas strepera L. and Tufted Duck. The induction analysis proved useful in determining the precise environmental features characterising the distribution of the rarest species, but clearly offers little opportunity for extrapolation. The very high rate of success in application to the original dataset is in part due to
305 the relatively low numbers of sites with each of the rarer species, but nevertheless gives confidence for the use of the technique. In specific cases, it is clear that Black-throated Divers do select large non-peaty lakes with high pH and islands on which to nest, not just in Caithness and Sutherland but throughout Scotland (Campbell & Talbot, 1987). In considerable contrast, Greylag Geese breed on large oligotrophic lochs by virtue of their feeding on acidic peatland plant species (Fox et al., 1989, 1990). Common Scoter are conspicuously associated with waters of a high pH and conductivity draining acidic peaty waters over sandy substrates. However, the species also utilises very large lochs with low conductivity levels, a feature of the species distribution elsewhere in Scotland (unpubl. data). Wigeon select mainly riverine sites, particularly those of high base status on sandy substrates, features which ensure the presence of deer-grazed grasslands along their floodplains, favoured as nursery areas for the ducklings. Few waterbodies in the Caithness and Sutherland peatlands show high pH and conductivity levels, and most of these are associated with sedimentary bedrock. From the apparent constancy of water chemistry, it would appear that acidic, base-poor, peat-stained water running offbogland catchments is buffered in such sites by contact with the basal bedrock. This seems especially true if the waterbody lies on a sandy substrate, the combination giving rise to flora and fauna (including rare invertebrates with highly restricted British distributions, Lindsay et al., 1988) which contrasts with adjacent acidic waterbodies. These waters are also less base-enriched than the lochs of the agricultural lowlands of Caithness (the survey results of which are not presented here) which are considerably more eutrophic than those of the peatland areas. It has been widely shown that eutrophic and mesotrophic waters support more diverse and productive phytoplankton and zoobenthos compared to oligotrophic systems (e.g. Holopainen & Paasivirta, 1977; Roff & Kwiatkowski, 1977). Since waterfowl during the breeding season tend to represent top predators of such trophic webs (both as growing young and as
adults), diversity and density of waterfowl communities tend to correlate with complexity and biomass of benthic communities (e.g. Danell & Sjoberg, 1978; Desgranges & Darveau, 1985). From our survey, we observed a similar situation, with rarer species associated with lochs of high pH and conductivity situated within oligotrophic peatland ecosystems; high diversity and presence of the rarest species was associated with highest water quality. Equally significant are the species-poor waterfowl communities of the acidic waters, some species of which (e.g. Red-throated Diver and Greylag Goose) are locally common, but nationally rare. The composition of breeding waterfowl communities of Caithness and Sutherland peatland wetlands seem therefore to be determined primarily by water quality. However, with rapid landuse change in the region through wide-scale afforestation, it is likely that extensive water quality changes will occur. Species such as Wigeon and Common Scoter nest along river courses and use these corridors as nursery areas for raising young. It is well known that changes in water chemistry resulting from afforestation in upland areas has a detrimental effect on existing invertebrate and fish communities and aquatic plants (Milner et al., 1981; Barko & Smart, 1983; Ormorod etal., 1988), hence changes in sediment loads, nutrient status and acidity even in remote areas of a catchment will directly affect species of high conservation status. We are not able to predict the effects of afforestation around individual water bodies on their water chemistry, but it is likely that long term changes will occur. Indirectly, the provision of cover is likely to increase the densities of potential nest predators of wetland birds (Andren etal., 1985) although this effect has not been clearly demonstrated for wading bird species (Stroud & Reed, 1986; Avery, 1989; Stroud et al., 1990). The Caithness and Sutherland peatland waters are internationally important for their high species richness and the presence of rare breeding waterfowl. In particular, the peculiar properties of waters in rivers and lochs draining peatland catchments onto sandy substrates support the less
306 common and widespread species. The survey and classification confirm the importance of sites with high rarity scores and species richness. However, because of the particular nature of the water quality, adequate site safeguards to maintain species diversity and the existing assemblages will rely on protection and management on a catchment basis rather than through individual waterbody protection.
Acknowledgements This survey could not have been carried out without the active support of the owners and occupiers of the wetlands which we surveyed. We thank them all for not just allowing access, but'also giving considerable background information. Enormous thanks go to all who helped with the survey, namely M. Ayress, G. Clarkson, T. Drew, A. Fox, H. Gitay, B. Hughes, N. Jarrett, S. Laybourne, D. Paynter, C. Liggett, M. Proctor, S. Richardson, D. Rigby, D. Salmon, P. Shimmings, J. Smith, A. Temple, P. Tovey, P. Turner and A. Watts. Thanks also to L. Cranna, T. Keatinge and K. Scott of the Nature Conservancy Council for help, support and advice with the project and to S. Bell for supplying information on water quality from the previous NCC freshwater survey of 1987. D. Stroud supplied previous survey data and he, M. Owen, I. Winfield and an anonymous referee all improved earlier drafts.
References Andren, H., P. Angelstam, E. Lindstrom & P. Widen, 1985. Differences in predation pressure in relation to habitat fragmentation: an experiment. Oikos 45: 273-277. Avery, M. I., 1989. Effects of upland afforestation on some birds of adjacent moorlands. J. appl. Ecol. 26: 957-966. Barko, J. W. & R. M. Smart, 1983. Effects of organic matter additions to sediment on the growth of aquatic plants. J. Ecol. 71: 161-176. Batterbee, T. W., T. G. Appleby, K. Odell & R. J. Flower, 1985. 2 0Pb dating of Scottish lake sediments, afforestation and accelerated soil erosion. Earth Surf. Process. and Landf. 10: 137-142. Burt, T. P., M. A. Donahoe &A. R. Vann, 1983. The effect of
forestry drainage operations on upland sediment yields: the results of storm-based study. Earth Surf. Process. and Landf. 8: 339-346. Campbell, L. H. &T. R. Talbot, 1987. The breeding status of the Black-throated Diver (Gavia arctica) in Scotland. British Birds 80: 1-80. Danell, K. & K. Sjoberg, 1978. Habitat selection by breeding ducks in boreal lakes in northern Sweden. Viltrevy 10: 161190. DesGranges, J.-L. & M. Darveau, 1985. Effect of lake acidity and morphometry on the distribution of aquatic birds in southern Quebec. Holarct. Ecol. 8: 181-190. Forsyth, R., 1987, PC/Beagle user guide. Pathway Research Ltd., Nottingham. Fox, A. D., N. Jarrett, H. Gitay & D. Paynter, 1989. Late summer habitat selection by breeding waterfowl in northern Scotland. Wildfowl 40: 106-114. Fox, A. D., D. A. Stroud & I. S. Francis, 1990. Up-rooted Common Cotton-grass Eriophorum angustifolium as evidence of goose feeding in Britain and Ireland. Bird Study 37: 210-212. Francis, I. S. & J. A. Taylor, 1989. The effect of forestry drainage operations on upland sediment yields: a study of two peat-covered catchments. Earth Surf. Process. Landf. 14: 73-83. Genstat 5 Committee, 1989. Genstat 5 Reference Manual. Clarendon Press, Oxford. Harriman, R., 1978. Nutrient leaching from fertilised forest catchments in Scotland. J. appl. Ecol. 15: 933-942. Hill, M. O., 1979. TWINSPAN. A Fortran Program for arranging multivariate data in an ordered two-way table by classification of the individuals and attributes. Cornell University, Ithaca. Hill, M. O., R. G. H. Bunce & M. W. Shaw, 1975. Indicator species analysis, a divisive polythetic method of classification, and its application to a survey of native pinewoods in Scotland. J. Ecol. 63: 597-613. Holopainen, I. J. & L. Paasivirta, 1977. Abundance and biomass of the meiozoobenthos in the oligotrophic and mesotrophic Lake Paajarvi, southern Finland. Ann. Zool. Fenn. 14: 124-134. Hornung, M. & M. D. Newson, 1986. Upland afforestation: influences on stream hydrology and chemistry. Soil Use Mgmt 2: 61-65. Lindsay, R. A., D. J. Charman, F. Everingham, R. M. O'Reilly, M. A. Palmer, T. A. Rowell & D. A. Stroud, 1988. The Flow Country: the peatlands of Caithness and Sutherland. Nature Conservancy Council, Peterborough, 174 pp. Milner, N. J., J. Scullion, P. A. Carling & D. T. Crisp, 1981. The effects of discharge upon sediment dynamics and consequent effects on invertebrates and salmonids in upland rivers. Appl. Biol. 6: 153-220. Nature Conservancy Council, 1989. Guidelines for selection of biological Sites of Special Scientific Interest. Nature Conservancy Council, Peterborough, 118 pp.
307 Ormerod, S.J., N. S. Weatherley, P. V. Varallo & P. G. Whitehead, 1988. Preliminary empirical models of the historical and future impact of acidification on the ecology of Welsh streams. Freshwat. Biol. 20: 127-140. Reed, T. M., D. R. Langslow & F. L. Symonds, 1983. The breeding waders of the Caithness Flows. Scottish Birds 12: 180-186. Robinson, M. & K. Blythe, 1982. The effects of forestry drainage operations on upland sediment yields: a case study. Earth Surf. Process. Landf. 7: 85-90. Roff, J. C. & R. E. Kwiatkowski, 1977. Zooplankton and zoobenthos communities of selected northern Ontario lakes of different acidities. Can. J. Zool. 55: 899-911. Stoner, J. H. & A. S. Gee, 1985. Effects of forestry on water
quality and fish in Welsh rivers and lakes. Envir. Pollut. 35A: 125-157. Stroud, D. A. & T. M. Reed, 1986. The effects of plantation proximity on moorland breeding waders. Wader Study Group Bulletin 46: 25-28. Stroud, D. A., T. M. Reed, M. W. Pienkowski & R. A. Lindsay, 1987. Birds, Bogs and Forestry: the peatlands of Caithness and Sutherland. Nature Conservancy Council, Peterborough, 121 pp. Stroud, D. A., T. M. Reed & N. J. Hardy, 1990. Do moorland breeding waders avoid plantation edges? Bird Study 37: 177-187. Webster, R. & M. A. Oliver, 1990. Statistical methods in soil and land resource survey. University Press, Oxford, 316 pp.