Fish Eating Birds and Salmonids in Scotland

DISCUSSION

The first part of this discussion makes some broad comparisons between the diet of goosanders, red-breasted mergansers and cormorants on Scottish waters with that elsewhere. The second part considers the dietary patterns revealed by the present study in relation to current knowledge of the foraging behaviour of the birds and the ecology of their prey. Wherever appropriate, reference is made to predation on larger salmon and trout because these fishes are of particular concern to fisheries managers. The discussion closes with guidelines for future studies on bird diet.

COMPARISONS WITH OTHER DIETARY STUDIES

Goosander

Almost all of the published studies of goosander diet we found relied upon samples of shot birds, except for that by Heard & Curd (1959) which investigated the stomachs of birds killed incidentally by gill nets. Methods of stomach analysis were not always detailed, so fish in the diet have been ranked (Appendix 3.1.) according to their importance as reported. The various studies are not strictly comparable because some authors merely give counts of various fishes in stomachs whereas others have estimated their size and we can thus assess diet composition in terms of fish biomass. Some authors have ignored all but fairly intact fish, which gives bias towards larger specimens whereas others have used the presence of key bones which produces less bias. Irrespective of the methods used there are some broad similarities with our results.

The diet of goosanders is diverse, mainly small fish of various species between 5 and 11 cm taken from rivers, although slightly larger fish are taken from lakes. In spring, other items include carrion (Rae & Duncan 1989), and frogs from ponds, 'oxbows' and backwaters can be important then. In summer, small ducklings feed on large insects as well as small fish, and at times adults may take freshwater crayfish where they occur. As in our study, there is an obvious bias in the published literature towards estimates of diet from places where goosanders were thought to be eating commercial or sporting fishes but it is clear that in many places this was not the case.

In the Old World (Palaearctic) the main species of fish taken on rivers include salmonids, eel, lamprey, bullhead, grayling, minnow, stickleback and perch. Goosander ducklings consume minnows, salmon and trout. On lakes outside the breeding season, the diet is mainly cyprinid and percid fish and sticklebacks. In winter and spring, in shallow coastal waters, eel, eelpout, stickleback, gobies, sandsmelt, butterfish, cottids and roach are taken. Goosanders fed on salmonids on only a few rivers where they took trout, salmon and Arctic charr in spring and summer.

On rivers in North America, the main elements of goosander diet are suckers, sculpins, salmonids, shiners and eels, with the inclusion of fallfish and alewife in autumn. Ducklings feed mainly on shiner, with some small salmonids and cottids. The diet on lakes includes a similar array of fish types but also includes 'chub', smelt, yellow perch and killifish in the north. In the warmer waters of the southern lakes and reservoirs, where large congregations of birds spend the winter, the diet is mainly 'forage fishes' - gizzard shad, carp, perch, freshwater drum and white crappie. The birds were feeding on fish of direct commercial interest in only a few instances. In winter, they took stocked rainbow trout from Washington lakes and 'trout' (including stocked brook trout) from Michigan rivers. The other instances were in summer and autumn, on several rivers in New Brunswick and Nova Scotia, where they fed on juvenile Atlantic salmon.

Red-breasted merganser

Much of the information on red-breasted merganser diet comes from samples of shot birds from areas where it was thought the birds could conflict with fishery interests. Despite this emphasis, most studies (Appendix 3.2.) have shown that mergansers eat small fish, (2-10 cm, Cramp & Simmons 1977) of a variety of species (mainly sticklebacks, shiners, minnows and gobies) and also shrimp and large insects. Amongst the studies reviewed, salmon were an important element in the diet only on one river in Nova Scotia (White 1957) and on the more northern rivers of Scotland in spring (Feltham 1990, 1995b). White (1957) found salmon parr prevalent in samples from far upstream but other fish species on the lower river and estuary. He concluded that while feeding in the river during the smolt run, red-breasted mergansers were not selecting smolts. Similarly, Feltham found that the salmon consumed in Scotland during spring were small (mainly 5-12 cm), predominantly parr (mean 7 cm) but including some small smolts (mean 11.5 cm) from the beginning of the smolt migration.

Some studies have deduced diet from a combination of visual observations of foraging birds and an association with certain habitats that supported mainly one or two species of small fish. From this it is inferred that the diet consisted of sandeels, young clupeids, shrimp, flatfish and sticklebacks on the coasts of Scotland in autumn and winter (Berry 1936, Aspinall & Dennis 1988, M. Marquiss pers obs); predominantly insects and sticklebacks in spring and summer in the north of the breeding range (Hilden 1964, Bengtson 1971, Rad 1980); but including some other species such as eels (Berry 1936), minnow (Rehfeldt 1986) or perch fry (Atkinson & Hewitt 1978) in the south.

Cormorants

Cormorants are perceived as damaging to fisheries over a very wide geographical range and as a result there has been a large number of dietary investigations. We summarise the results of 37 European studies in Appendix 3.3. At least 77 species of fish are recorded as cormorant prey in Europe but only about a third of these feature regularly and, within habitats, different studies have shown similar prey spectra despite differing methods. In the sea, cormorants mainly feed on bottom-dwelling fishes, wrasse and gadoids over rocky and weed-covered substrates, flatfish over soft substrates and eel and eelpout in a variety of areas. On occasions small, shoaling, midwater fishes such as clupeids, capelin and sandeels, are taken. In estuaries, flounder, trout, eel and saithe are most frequent prey, and sandsmelt, mullet and sea bass are important in the south.

On rivers, diet varies according to stream characteristics. Salmonids are the main prey in fast-flowing streams, cyprinids in slower deeper ones and flatfish in the lower reaches. In studies at freshwater lakes, by far the commonest recorded prey are roach, perch and eel. Other cyprinid prey in 'rich' fresh waters include bream, rudd and tench, and other percids, notably ruffe and zander. In more 'acid' and/or species-poor waters, cormorants feed mainly on brown trout or perch. Finally, cormorants frequently use waters artificially stocked for recreational angling (brown and rainbow trout) as well as carp farm ponds. Even where farmed fish are confined within suspended cages, escaped rainbow trout congregate outside to feed from faeces and waste food and are then sometimes taken by cormorants.

THE PRESENT STUDY

Our results have enabled a reasonably comprehensive description of patterns in the diet of sawbills and cormorants in Scotland. Although diet was diverse, it showed some predictability. The greatest variation was between bird species and locations; less (though still statistically significant) between times of year and between years. The most consistent patterns were that larger bird species took larger fish (Table 3.2) and that the diet of all bird species was less diverse, with a greater salmon component, in the north (Figure 3.1).

Brown trout, salmon, eel and minnow were taken frequently by all three bird species throughout Scotland. Trout was the most widespread and important food, other elements predominating for particular bird species at specific times and places. Concerning fish of commercial importance, juvenile salmon greater than 90 mm long were consumed mainly by adult sawbills on more northerly rivers such as the Spey, Findhorn, Beauly, Deveron and North Esk but were notably scarce in the diet of cormorants throughout Scotland. Trout of 300 mm or more were eaten mainly by cormorants on rivers such as the Borders Esk, Laggan and Deveron, and at stillwaters such as Cobbinshaw Loch (Table 3.12).

These dietary patterns were consistent with the hypothesis that, as with predators in general (Begon et al. 1996), fish-eating birds take the most nutritious (usually the largest) prey that are available to them. Prey availability is difficult to measure but is probably influenced by several factors - a combination of the relative abundance of various species and size classes of fish, together with the ease with which they can be located, caught and eaten.

Goosander foraging behaviour and diet

Goosanders consumed some large items, eels up to 47 cm and trout up to 34 cm, but most of their prey was substantially smaller. Their ability to consume prey is probably less of a constraint on diet than is their ability to locate and catch it. Goosanders locate prey either by surface swimming with their bill and eyes submerged scanning for fish, or by diving (Lindroth & Bergstom 1959, Sjoberg 1988). They have a stout bill with serrated edges and a down-curved (almost hooked) tip to the upper mandible, suitable for seizing and handling larger prey and for probing crevices in stoney substrates. Once located, potential prey is chased either above or below the water surface, the duck propelling itself with legs and feet. Prey may be captured and swallowed, lost, or 'run to ground'. In the latter case, prey can sometimes be caught if, by probing with its bill, the goosander can dislodge it from its refuge and continue the chase. When no fish are visible, goosanders often probe with their bill under stones, flipping over smaller ones (Sjoberg 1988). Goosanders also sometimes probe in soft substrates and in vegetation, particularly in the fringes of lochs and along banksides, where they take eels and frogs.

Considering these foraging techniques we would expect that the prey most vulnerable to goosander predation would be those that cannot outswim a bird or cannot find secure refuge. Thus, away from adequate cover, large but slow swimming prey such as frogs, eels and lampreys might be vulnerable. Whenever or wherever such prey are scarce, we might expect goosanders to take faster swimming fish such as salmon or trout, but mainly small individuals because they have the lowest burst speed and least stamina (Peake & McKinley 1998). On present evidence this is sufficient explanation for the dietary patterns we have found.

Seasonal patterns in diet would thus be determined by the availability of the most sluggish prey. Frogs are taken in March and April when they are abundant in aquatic habitats as they gather to spawn (Heyer et al. 1994). Lampreys, which are usually hidden in the substrate, are taken in April and May as they migrate or spawn (Maitland & Campbell 1992). Eels are taken more in summer when they are most active and less likely to be buried in the substrate (Tesch 1977).

The same explanation could apply to the latitudinal trends in diet, i.e. more diverse, including fewer salmon, on rivers in the south. Generally, more fish species occur in the rivers of southern Scotland; e.g. 18 in the Annan and 17 in the Tweed, compared with only 11 in the Findhorn, 10 in the Spey and even fewer further north and west (Maitland & Campbell 1992). Moreover with fewer species in the north, trout and salmon increasingly dominate the fish communities there. The rivers where goosanders consistently consumed fewer fish species and more salmon were mainly fast-flowing Highland rivers such as the Dee, Deveron, Spey and Beauly. In these rivers, the fish community comprised relatively few species and salmonids predominated.

However, the greater diversity of fish species in the diet of birds from the south was not purely a function of the greater number of species present there. The main elements of diet in the south were trout, eel, minnow and lamprey, all of which also occur in northern rivers. If goosanders preferentially take slower swimming species, they might have taken more salmon in the north because the more easily caught prey were less abundant there. To test this idea, we need estimates of both bird diet and the relative abundance of their various prey in both northern and southern river systems.

Red-breasted merganser foraging behaviour and diet

Red-breasted mergansers are smaller than goosanders and have a thinner, proportionately longer, and slightly upturned serrated bill, features suited to taking small agile prey, and benthic prey from soft substrates. They forage in similar ways to goosanders except for a tendency to use their wings underwater (Dementev & Gladkov 1952) and for more diving, and less surface scanning, than goosanders when searching for prey (Sjoberg 1988). These morphological and behavioural differences between duck species are associated with habitat use. Mergansers tend to use deeper, slower moving waters, which can be turbid and often have soft or sandy substrates (Cramp & Simmons 1977). On the North Esk, we found mergansers to be more abundant in the wider, deeper, river sections where the current was less and the substrates finer-grained whereas goosanders were more abundant in upstream sections with cobbled substrates (Marquiss & Duncan 1993, 1994a, Chapter 2).

In the present study, mergansers more frequently ate small prey species than did goosanders, and had a greater proportion of salmon in the diet. Both the trout and salmon consumed by mergansers were smaller than those taken by goosanders. This is unlikely to have been because smaller salmonids are more abundant in red-breasted merganser habitat because there is no evidence that salmonids in pools on the lower reaches of rivers are any smaller than those on riffles further upstream.

The simplest explanation is perhaps that goosanders can swim faster than red-breasted mergansers and the sizes of salmonids taken reflect the size range that cannot outstrip the predator. Where studies have compared the sizes of smolts taken by mergansers and goosanders with the sizes of migrating smolts, ducks took mainly the smaller individuals (Feltham 1990, Kalas et al. 1993). Feltham & McLean (1996) showed that where smolts were Carlin-tagged on the North Esk, ducks did take some larger ones, but mainly those that had their adipose fin removed, a marking process which presumably reduced their capacity to evade capture.

On rivers and in estuaries there is some overlap in habitat use between the sawbill species (as in diet) but the observed differences emphasise that, on a broader scale, these ducks occupy different ecological niches. Red-breasted merganser is primarily a coastal species (almost exclusively so in winter) and breeds late (almost a month later than goosander), consistent with the seasonal cycle of the main prey, small fish such as minnows and sticklebacks (Maitland & Campbell 1992). Their use of small salmon and trout on northern Scottish rivers in April probably reflects the lack of alternative prey at that time because such alternative prey (which are the staple foods elsewhere in Europe) increase in the diet thereafter.

Cormorant foraging behaviour and diet

Cormorants are much larger than sawbill ducks enabling them to dive for longer, and possibly deeper, than the smaller species (see Cooper 1986). They have a stout bill with a pronounced sharp hook at the tip, allowing them to seize, maim, and kill very large fish (Takashima & Niima 1957, Carss 1990) prior to their being manipulated and swallowed head first (Reimchen 1991). The prolonged handling time involved for large fish means such activity is conspicuous and the literature is full of anecdotes of 'cormorants with big fish' (e.g. St John 1882, Snook 1992) even though most of the fish they consume are much smaller. The contents of the stomachs we examined showed cormorants took larger fish than did sawbills, including some very large individuals, but they also took very small fish such as minnows and 3-spined sticklebacks. The great diversity in our samples probably reflected the mobility of cormorants and their ability to switch between sites and prey species to exploit local fish abundance.

Diet varied according to the habitat cormorants used. Grayling and trout were the main prey in fast-flowing rivers, cyprinids (roach) in slower, deeper parts of rivers and flounder in the lowest reaches and estuaries. Our small samples from lochs suggested that the diet there was mainly brown and rainbow trout, eel, roach and perch. Much of the diversity in diet seemed to stem from seasonal switches between prey species. The cormorants at Loch Fad switched from feeding on rainbow trout, brown trout and perch in winter, to eel the following spring. On the Tweed cormorants switched from grayling in February and March, to grayling or trout in April, and eels in May. Large grayling were also taken in late winter and early spring on other southern Scottish rivers, as elsewhere in Europe (Suter 1995, Keller 1997).

Within habitats and locations, seasonal shifts in the importance of particular species in the diet of cormorants are common. For instance, in species-rich, standing waters in Europe there is a tendency for some cyprinids (mainly roach or bream) to predominate in the diet in early spring, followed by ruffe, then often eel in summer, and switching to perch or zander later in the year (van Dobben 1952, Dirksen et al. 1995, Veldkamp 1995, Suter 1991). Such 'prey switching' has been attributed to changes in the abundance of prey. For example, at a site in the Netherlands, the eutrophication of lakes was associated with increased abundance of cyprinids, and the diet of cormorants at a nearby colony changed as eel and ruffe became less important and roach, bream, silver bream and rudd became more so (Veldkamp 1995).

This phenomenon is particularly important for fisheries that artificially increase the abundance of one fish species above others, because by doing so they can attract the attention of cormorants towards stocked fish. At Loch Leven in Scotland, the diet of cormorants switched from predominantly perch, to predominantly trout following a decline in the perch, coincident with the release of substantial numbers of hatchery-reared trout (Carss et al. 1997b). On the River Bush in Northern Ireland, Kennedy and Greer (1988) found cormorants fed exclusively on hatchery-reared salmon smolts downstream from their release point, compared with medium-sized trout and some small salmon upstream of it. Only four cormorant stomachs were examined from above the hatchery, and four from below, but despite the small samples the difference seemed clear.

So far the only well-documented instances of cormorants taking large numbers of salmon smolt have been where their numbers were artificially enhanced. Nevertheless, cormorants are also often seen in numbers on the lower reaches (mainly estuaries) of Scottish rivers in spring and it is widely perceived that they are taking large numbers of salmon smolts. However, despite good samples of stomachs from the Rivers Nith, Annan, Tweed and Deveron, juvenile salmon were a minor component there (< 9% by mass in all samples > 10); only stomachs from the Beauly contained a high proportion of young salmon. Adult salmon (kelts) were found in the stomachs of one cormorant from the Beauly and in two from the Deveron. Although the daily food intake of cormorants is twice that of goosanders, their abundance on rivers is less than one sixth (Chapter 4) so, at least on the present evidence, it seems that cormorants are not the major source of loss to salmon fisheries in Scotland.

This could not be said for trout fisheries. Cormorants on Scottish rivers and lochs take brown trout in many places and at various times of year. Most of those consumed on rivers were small fish but in samples from 3 locations, the Borders Esk and Deveron in winter, and the Laggan in spring, many of the trout were large enough to have been taken by anglers. Also the brown and rainbow trout taken by cormorants from stocked lochs (e.g. Butterstone and Loch Fad) were mainly large fish and it may be that cormorants are in direct competition with anglers on such waters. To calculate the potential loss to the fishery we need to measure the abundance of fish of different sizes in these places. However, cormorants feeding on stocked lochs were not always taking fish of commercial concern. On Loch Fad, cormorants took mainly stocked trout in one winter, but eels and perch the following spring.

FUTURE STUDIES

The dietary patterns we observed suggest the numbers of commercially important fish (i.e. salmonids) taken by birds might be primarily influenced by the availability of non-commercial fish. There is thus a clear need for information on the relative abundance of fish (and their size) in fresh waters used by these birds. At present there is information on which species occur in various lochs and rivers, but data on abundance is largely restricted to a few species (mainly salmon and trout) in places where they are easily sampled. There are few estimates of the standing stock of fish in lochs and insufficient data on abundance for fish communities in the deeper and wider stretches of rivers, where most sawbills and cormorants feed.

Sawbills and cormorants took slightly larger salmon and trout in winter than in spring, perhaps because these fish become less agile as water temperatures fall in winter, so that larger individuals can be more easily caught. Alternatively, larger fish could simply be more abundant in the lower reaches of the river, where these birds forage in winter. To examine this further again requires data on the abundance of various sizes of fish in the places where birds have been sampled.

Finally, the suggestion that swimming speed is an important component in fish availability needs to be explored. A literature review together with laboratory studies quantifying the burst speeds of fish of different species and sizes, and the effects of water temperature, Carlin-tagging and adipose fin clipping, would be useful.

There is sufficient evidence from the contents of bird stomachs to suggest that birds might select their prey. Future work on diet should investigate selection, concentrating on factors that might predispose commercially important fish to predation.