Fish Eating Birds and Salmonids in Scotland

 

2. The abundance of sawbills

M Marquiss

SUMMARY

A literature review summarises the breeding distributions, movements and numbers of goosanders and red-breasted mergansers in Europe. Populations increased up to the mid-1980s but have recently declined. The most recent estimates of the breeding populations on Scottish rivers are 2600 pairs of goosanders and 800 pairs of red-breasted mergansers.

The present study monitored numbers and breeding production of goosanders and red-breasted mergansers on two salmon rivers in northeast Scotland, to provide density estimates and to indicate the main factors influencing their abundance.

The relative abundance of goosanders on the Rivers Dee (1987-94) and North Esk (1987-93) was predictable in that each year the birds were most abundant on the lower river in winter, on the upper mainstem and lower tributaries in spring, and on the mainstem duckling nursery areas in mid-summer.

Annual variation was low for wintering numbers of goosanders (averaging 64 on the Dee and 24 on the North Esk) but otherwise high. Spring numbers declined from 74 to 28 pairs on the Dee and 12 to 6 pairs on the North Esk within the study period and summer duckling numbers varied more than twofold (75 to 183 on the Dee and 0 to 41 on the North Esk).

Annual production was mainly the consequence of variation in the survival of ducklings during the week after hatching. Summer abundance differed between rivers apparently determined by local factors, whereas trends in spring numbers were correlated between rivers and could be determined by food availability in wintering areas elsewhere.

The relative abundance of red-breasted mergansers on the North Esk in spring (78 to 43 pairs) and summer (10 to 59 ducklings) varied in parallel with those of goosanders, suggesting similar influences were at work.

Sawbills on these rivers lived at densities averaging 0.15 birds/ha, rarely exceeding 0.33 bird/ha. Goosander density was highest on lower sections of both rivers in midwinter, and on the mainstem of the Dee in summers of high duckling production. Red-breasted merganser density was highest on the North Esk mainstem in May.

INTRODUCTION

Estimates of the densities of sawbills on rivers are required to calculate the numbers of salmonid fish consumed relative to fish density (Chapter 4). Previous studies (Carter & Evans 1986, Marquiss & Duncan 1993, 1994a, Gregory et al. 1997) had already established that abundance varies between Scottish rivers, different parts of rivers, and between seasons so one aim of the present study was to extend the data to examine variation between years. This would allow estimates of the fish consumed with respect to time and place.

This chapter reviews the literature on sawbill population ecology then details our studies on the Rivers Dee and North Esk. An analysis of the variation in our counts shows the levels of differences between bird species, rivers, seasons and years. The populations of goosanders on the Rivers Dee and North Esk, and of red-breasted mergansers on the North Esk are then presented. The data are compared between duck species, and within and between years in search of patterns that could indicate the main factors influencing abundance. Finally, we give the average densities of sawbills (per hectare of river) for specific times and places.

Population ecology of goosanders and red-breasted mergansers: a literature review

Goosanders and red-breasted mergansers have circumpolar breeding distributions that overlap considerably in boreal and low arctic zones, lying between latitudes of 40o and 80o N (Cramp & Simmons 1977). There is much overlap in the habitats they use, but goosanders are mainly found in fresh waters and red-breasted mergansers in shallow inshore marine waters. Goosanders are larger (males weigh about 1.6 kg, and females 1.2 kg) than red-breasted mergansers (males 1.1 kg, and females 0.9 kg) and have stouter beaks. Both species feed on small fishes but red-breasted mergansers forage in more open, sometimes deeper waters, and more often take crustacea (Appendix 3). Individual goosanders and mergansers range over large areas and are difficult to catch and mark, so there are few studies of movements and migration. Population studies have therefore concentrated on estimating breeding production and wintering numbers.

GOOSANDER POPULATIONS

Habitat, breeding and movements

The goosander is a predominantly freshwater duck rearing young on rivers and sometimes on lakes. Recoveries of ringed birds show northern breeding populations may move south for the winter, though most of those breeding in the southern and western part of the range, particularly in the more oceanic climes, are resident (Boyd 1959, Meek & Little 1977a, Hofer & Marti 1988). Though 'resident', goosanders can move some distance to winter on the lower reaches of rivers, estuaries and freshwater lakes (Owen et al.1986), and even further if freshwaters freeze (Chandler 1981). Changes in the numbers of birds on the River Dee, in North East Scotland, suggested many had left the catchment to winter elsewhere (Marquiss & Duncan 1994a). In Europe there are some very large winter aggregations in shallow brackish waters and large lakes (Ruger et al. 1986, Aspinall & Dennis 1988). Outside the breeding season, goosanders gather to roost communally on large standing waters, or on islands and backwaters on rivers. Most birds leave the roost before sunrise and return in late afternoon, near sunset on short midwinter days (Marquiss & Duncan 1994b).

By spring, mature goosanders (birds 2 years or older) have paired and move to the upper, faster flowing parts of rivers to breed. Egg laying varies with latitude; birds at the southern end of the range starting to lay in March (North East Scotland, Marquiss et al. 1991; Northumberland, B. Little & S.J.Petty pers comm), but those in the far north, not until May (Lid & Schandy 1984, Hilden 1964, Dement'ev & Gladkov 1952). Goosanders lay 7-14 eggs in cavities in trees, cliffs or among boulders and, led by the female, the young leave the nest within 2 days of hatching to feed themselves.

Occasionally, broods of ducklings will congregate into a creche, but this seems mainly to occur in areas of open water (e.g. Stutz 1965) where brood density is high or where broods have been harassed (A.J.Erskine, 1971 & pers comm). On two rivers in North East Scotland, where brood amalgamation was rare, average breeding production varied over 5 years from 3.6 to 0.0 ducklings fledged per spring adult female (Marquiss & Duncan 1994a, Carss & Marquiss 1992), and average brood size from 6 to 7 ducklings per brood. In different years and 10-km sections of river, brood density varied from 0.0 to 0.4 broods per km. A similar range in average brood sizes and brood densities has been found on the River Tweed (Murray 1988, & 1992) and on rivers in Wales (Tyler 1986, Griffin 1990).

Once females have commenced incubation, males aggregate once more and migrate north in June to moult in July and August, on estuaries, firths and shallow fjords. In the south of the range, such as in Wales, males can leave the breeding areas as early as April (Tyler 1987) whereas in the north of Scotland they do not leave until late May or early June (Marquiss & Duncan 1994a). Adult goosanders become flightless as they moult their wing feathers and most congregate on expansive areas of shallow open water where they can dive to evade predation but where there is also an abundance of prey. The majority of European males appear to moult in North Norway (Little & Furness 1985) although a few remain in Scotland to moult on estuaries alongside females (Marquiss & Duncan 1994a). By late October, fledged juveniles have dispersed and moulted adults return to their wintering areas. Marquiss & Duncan (1994b) argued that diurnal and seasonal movement, including switching between habitats, is associated with food availability, but also with security from predation and the needs of pairing, nesting and brood production.

Trends in the breeding populations of Europe and Great Britain

Goosanders breed at relatively low densities so the monitoring of breeding populations is difficult and most statements regarding widespread changes in its breeding abundance are vague. For example, goosanders have apparently declined in southern Sweden in association with the acidification of lakes, but the decline has not been quantified (Hansen 1980). Most well documented trends in breeding populations have been restricted to small study areas (e.g. in Finland, Niittyla 1980) or have recorded changes in the breeding range. Thus the colonisation and increase of goosander populations in the southern part of the range are quite detailed. The species was a rare breeding bird on Lake Geneva in the early decades of this century but increased dramatically to about 700 pairs in 1984 (Geroudet 1985).

The colonisation of Great Britain is well described (e.g. Berry 1939, Mills 1962b, Meek & Little 1977b & 1980, Tyler et al. 1988, Sitters 1988, Griffin 1990, Carter 1993) alongside some of the local increases that have characterised the colonisation process. In contrast, apart from counts in 3 years on the River Tweed (Murray 1992), 3 years on the Dee and 4 years on the North Esk (Marquiss et al. 1991, Carss & Marquiss 1992), long-term changes in breeding abundance within the established range have not been documented.

Trends in the winter populations of Europe and Great Britain

An analysis of Monval & Pirot's (1989) count indices by Kirby et al. (1995), suggested long term increases in the European population from the mid 1960s to the mid 1980s, particularly in Central Europe, where there has been a parallel increase in breeding birds (see above). The numbers wintering in Britain have been monitored using National Waterfowl Counts and Wetland Bird Survey (WeBS) counts (Starling et al. 1992, Kirby et al. 1995). An analysis of these count data also showed a sustained increase in the wintering population of Britain from the mid-1960s to the mid-1980s, in parallel with the range expansion of breeding birds. The increase in counts occurred as a result of more birds on more sites covering the full spectrum of habitats, though particularly gravel pits. The most important wintering site was the Beauly Firth (inner Moray Firth) where there was well over a thousand birds throughout the winter from the late 1970s (Aspinall & Dennis 1988).

Since the mid-1980s however, the numbers on the Beauly have declined (Evans 1998) to one or two hundred which leave before midwinter. This change was thought to be associated with a decline in clupeid shoals (and in other fish-eating birds and mammals) and no similar concentration of goosanders has arisen elsewhere in Britain. The birds had presumably died because there is no evidence that they dispersed to other locations. Counts have decreased at many other sites and the overall population index fell in the winter of 1989/90, to a level equivalent to that in the late 1970s, and has not increased since.

In both the European and British data sets there is evidence of a 3 to 4 year population cycle superimposed on the overall trends in abundance (Kirby et al. 1995). It seems unlikely that factors operating on the wintering grounds are responsible for such a cycle; for example there is no correspondence with severe winter weather. If factors affecting breeding production are responsible for cycles, it implies some periodicity in fish availability on the breeding grounds.

Current estimates of breeding populations in Britain and Europe

Breeding distribution maps have been produced from fieldwork done by BTO members in 1968-72 (Sharrock 1976) and in 1988-91 (Gibbons et al. 1993). In the earlier survey, breeding goosanders were widespread in northern Britain, but concentrated in southern Scotland and northern England. The second survey showed more goosanders breeding further south, in the Pennines, Wales and southwest England. The second survey also incorporated indices of abundance to provide a quantitative baseline estimate of the breeding population. Additionally, the BTO organised a national sawbill survey in 1987 to estimate the density of birds in various parts of the breeding range and the total number in Britain. Carter (in Gibbons et al. 1993) estimated average breeding densities of 0.18, 0.21 and 0.08 pairs per kilometre of river in Scotland, England and Wales, respectively.

The total breeding population of Western Europe is unknown, but could be in the region of 30,000 to 40,000 pairs (estimated from wintering numbers cited in Ruger et al. 1986). The population in Britain was thought to be between 1000 and 2000 pairs in 1968-72 (Sharrock 1976), between 915 and 1,246 pairs in 1975 (Meek & Little 1977b) and about 2,600 pairs in 1987 (Gregory et al. 1997).

Current estimates of winter populations in Britain and Europe

A total population of about 8,000 goosanders was estimated to be wintering in Great Britain and Ireland in 1981-84 (Chandler, in Lack 1986), of which about 5,000 were in northern England and Scotland. From National Waterfowl Count data, Owen et al. (1986) estimated about 5,000 birds and most recently Starling et al. (1992) estimated 5,500. The discrepancy between estimates could be due to few counts of birds wintering on rivers because the National Waterfowl Counts mainly census standing waters. A recent reassessment (Kirby 1995), which included birds on estuaries estimated from WeBS counts, suggested a British wintering population of about 8,900 birds (approximately 6% of the numbers estimated to winter in NW Europe). This figure, when compared with the estimate of British breeding birds, suggests that goosanders are largely resident in Britain with few continental winter visitors (Owen et al. 1986).

The winter population of western Europe in the early 1980s was put at 100,000 - 150,000 birds (Ruger et al. 1986) and most recently, Rose & Scott (1994) give an estimate of 150,000 birds wintering in NW Europe. In these studies the figures varied considerably, associated with uneven coverage between years. Ruger et al. (1986) suggested coordinated aerial counts of goosanders in the Baltic would produce a more accurate estimate of the total wintering population.

RED-BREASTED MERGANSER POPULATIONS

Habitat, breeding and movements

The red-breasted merganser is a predominantly marine species living in shallow coastal waters. Most of the British population breeds in sheltered sea lochs and estuaries, and a small proportion on the slower-flowing, wider, lower reaches of rivers (Marquiss & Duncan 1993) and on fresh water lakes, which is the commoner breeding habitat elsewhere in the range (Cramp & Simmons 1977). Almost all red-breasted mergansers winter on the coast, entering fresh waters in late April and May, laying in May and June and producing young in July and August (Marquiss & Duncan 1993).

Egg-laying is probably slightly later at higher latitudes (Dement'ev & Gladkov 1952, Rehfeldt 1986, Bengtson 1972, Hilden 1964 and Sjoberg 1989). Seven to 12 eggs (sometimes many more in 'multiple clutches', Young & Titman 1988) are laid in a scrape on the ground amongst vegetation, and the young leave the nest within a day or so of hatch to feed themselves, though led and protected by the female. One female can guide many ducklings that sometimes congregate into a single large group (crèche), particularly where several females nest close by one another (Martin 1988). The habit of crèching is far more common in red-breasted mergansers than in goosander broods (Bergman 1956) and the biggest of crèches have been found on shallow lakes and estuaries (Hilden 1964, Gardarsson 1979, Macdonald 1987b and Marquiss & Duncan 1993). Average production in individual years has varied from 1-3 ducklings per spring female, on the rich shallow lake Myvatn, Iceland (Bengtson 1972) and 1-2 in the shallow brackish Gulf of Bothnia (Hilden 1964), and on a river in North East Scotland (Marquiss & Duncan 1993). In this latter study, low duckling production was associated with high river flows at the time of hatching and in the subsequent 10 days.

Adult red-breasted mergansers moult their wing feathers in August and September, at this time congregating in open areas of shallow water. In Britain, current moulting sites are on the coast around Scotland, Northumberland and North Wales (Owen et al. 1986, Marquiss & Duncan 1993, Dickson 1987, Andrews 1987, Starling et al. 1992, R.Broad, L.Hatton, D.C.Jardine, B.Little and M.Marquiss unpublished). After moult, the birds become more widely distributed around the coast, with fewer large aggregations (Owen et al. 1986, Aspinall & Dennis 1988, Starling et al. 1992).

Population trends in NW Europe.

Red-breasted mergansers are relatively easy to count on the sea close inshore, and on estuaries and lakes, but more difficult in the open sea (without aerial surveys) or on upstream sections of rivers (where individuals can move over several kilometres of river, Marquiss & Duncan 1993). The numbers at specific count sites can fluctuate erratically from year to year so it is thought that birds readily shift foraging and moulting areas, perhaps associated with shifting food availability. Moreover, there are relatively few ringed bird recoveries so their movements are poorly documented (Cramp & Simmons 1977, Owen et al. 1986, Ruger et al. 1986). It is therefore not surprising that the wintering population of North West Europe is roughly estimated at about 100,000 (Rose & Scott 1994) and the total breeding numbers are unknown.

Trends in the British breeding population

As with goosander, the best documented population change is in Britain where numbers have increased alongside an expansion in breeding range. Red-breasted mergansers have long bred in Scotland but the population increased dramatically from about 1885 to 1920, spreading into northern England in 1950 and North Wales in 1953 (Parslow 1973, Cramp & Simmons 1977, Owen et al. 1986). Although the breeding population of Britain continued to be concentrated in western Scotland, by the time of the 1968-72 Breeding Bird Atlas (Sharrock 1976) there were well established populations in the Lake District, Anglesey and North Wales. The 1988-91 breeding atlas (Gibbons et al. 1993) documents further colonisation in the Peak District and southwards in Wales, and gives the best current estimate of population distribution.

There has been no census of the British and Irish breeding population. The 1987 national sawbill survey was designed to estimate goosander numbers, so it ignored most of the red-breasted merganser breeding habitat on the coast, and conducted spring river counts in April (many red-breasted mergansers do not arrive until May, Marquiss & Duncan 1993). The first breeding bird atlas of 1968-72 estimated the population at 2,000 to 3,000 pairs in Britain and Ireland (Sharrock 1976) by multiplying the area occupied by birds by a guess at breeding density (3 to 5 pairs per 100 km2). The abundance indices accompanying the most recent breeding distribution map (Gibbons et al. 1993) give a current (1988-91) estimate of 4,300 in Britain and 1,400 in Ireland, suggesting minimum populations of 2,150 and 700 pairs respectively. It was estimated that there were a minimum of 800 pairs on British rivers in 1987 (Gregory et al. 1997).

Trends in the British wintering population

The wintering numbers of red-breasted mergansers have been monitored using NWC and WeBS Counts. Trends have broadly followed those for goosander, with an increase from the mid-1960s to the mid-1980s, and a subsequent decline, though not as pronounced as that of goosanders (Starling et al. 1992, Kirby et al. 1995). Owen et al. (1986) estimated the winter population at 6,000 to 10,000 birds and Rose & Scott (1994) give an estimate of 10,000, about 10% of the overall estimate for NW Europe (Kirby 1995).