Fish Eating Birds and Salmonids in Scotland, page:17

Fish Eating Birds and Salmonids in Scotland

4. The numbers of salmon and brown trout consumed by fish-eating birds

D N Carss & M Marquiss

SUMMARY

The numbers of salmon and trout consumed per bird per day were calculated for sawbills and cormorants on 13 rivers and 2 lochs.

It was calculated that sawbills consumed 2.6 to 8.0 salmon per bird per day on southern rivers and 4.6 to 60.6 on northern ones. The numbers of larger salmon (>89mm) consumed by sawbills were fewer; 0 to 2.8 on southern rivers, 0.6 to 19.0 in the north.

Similar calculations for cormorants suggested they consumed 0.4 to 3.1 salmon per bird per day on southern rivers and 1.0 to 7.8 in the north. Corresponding figures for larger salmon (>89 mm) were 0.3 to 2.6 per cormorant per day on southern rivers and 1.0 to 4.2 in the north.

The numbers of trout calculated to be taken per bird per day varied from 0 to 23 for cormorants and 3 to 29 for sawbills. Mergansers took no trout over 199 mm length, goosanders took few (0.0 to 0.4 per bird per day) and cormorants most (0.2 to 2.6).

Further calculations incorporating both bird diet and abundance data, suggested that in April, goosanders consumed 1.9 to 2.8 salmon per hectare of river per day, on both the Dee and North Esk. Higher numbers were taken by goosanders in winter on the lower Dee (4.0), by ducklings on both rivers (4.2 to 8.9) and by mergansers at the mouth of the North Esk in April (11.6).

Comparing the numbers consumed by ducks with the best available data on fish densities suggested sawbills on these 2 rivers consumed 0.02 to 0.30% of the salmon standing stock per day on narrow (<30m) river sections and 0.40 to 2.4% on wider sections where birds were at their highest density.

The calculations are discussed in terms of the validity of the assumptions necessary. The highest figures could result from overestimated consumption, or underestimated salmon density. Future calculations might be improved by using estimates of diet, bird density, daily food intake and fish density appropriate to specific bird foraging sites.

INTRODUCTION

In this chapter we draw together data from previous ones to calculate:

the numbers of salmon and trout consumed each day by goosanders, red-breasted mergansers and cormorants on some Scottish fresh waters;
the consumption of salmon/ha by sawbill ducks on the Rivers Dee and North Esk in relation to stream width, and
the percent of salmon standing stock consumed by sawbills on those river sections where these ducks were at their highest density.

Two previous studies have used simple models to estimate the impact of sawbill ducks on salmon in the River North Esk. The first (Shearer et al. 1987) used available data from a wide range of sources to estimate numbers of smolts taken by sawbill ducks and the effect of sawbill removal on the number of adult salmon returning to the river. Simulations suggested that the maximum benefit would be a 35% increase in returning adult salmon but the authors believed that in practice the benefit was likely to be less than this. The second (Feltham 1995a) used estimates of the field metabolic rate (FMR) of captive goosanders and data from previous dietary studies to estimate the daily consumption of salmon smolts and parr by natural populations of these ducks on the same river. Annual predation of smolts by goosanders was estimated to be between 8000 and 15 000 during the period of the smolt run.

However, both these models include a large number of assumptions, some of which are valid but several of which are in retrospect demonstrably invalid or remain untested (Table 4.1). Much of the required information remains unknown so our aim here is to limit our modelling to calculations of the consumption of salmonids by birds at times and places where there are data available. This minimises the assumptions necessary because figures can then be compared directly with information on salmon densities where available.

Table 4.1. Details of some invalid/untested assumptions made in two simple models of impact by sawbill ducks on salmon in the River North Esk.

Shearer *et al.’s (1987)* simple model considering the effect of sawbill removal on the number of adult salmon returning to the river included the following assumptions:

1.	That any reduction in sawbill numbers was not accompanied by a concomitant local increase in the numbers of other predators -	untested.
2.	That the sawbill population was between 50 - 100 birds each month -	invalid.
3.	That during the smolt run, smolts were the sole component of sawbill diet -	invalid.
4.	That the salmon component of goosander and red-breasted merganser diet was 0.572 and 0.425, respectively (based on Mills 1962 from various Scottish rivers) -	invalid
	.
5.	That each smolt consumed weighed 21g -	invalid.

Feltham’s (1995a) model to estimate the daily consumption of average-sized smolts and parr during the period of the smolt run made the following assumptions:

1.	That the captive birds (used for energetics calculations) behaved naturally on release in the wild -	invalid.
2.	That birds were in ‘energy balance’ (i.e. not metabolising fat) -	untested.
3.	Calorific values for non-salmonid fishes -	untested.
4.	That smolts could be differentiated from parr in the stomach contents of birds on the basis of their length (i.e. smolts > 9 cm) -	invalid.
5.	That previous smolt population estimates (Shearer et al.1987) were accurate -	untested.

In this chapter we calculate two series of figures for the amount of fish taken by birds:

(i) The intake rate (nos. salmonid fish bird^-1 day^-1) of goosanders, mergansers and cormorants on a number of Scottish fresh waters, based on the dietary information presented in Chapter 3.

(ii) The consumption rate (nos. salmonid fish ha^-1 day^-1) of sawbill ducks on the Rivers Dee and North Esk in relation to river width, based on (i) above, and the duck density information presented, in Chapter 2.

Data on salmon abundance in relation to stream width were available for a number of Scottish rivers enabling a crude comparison of consumption estimates against fish standing stock for the Rivers Dee and North Esk.

METHODS

Intake rate

As in Chapter 3, we concentrated on those four categories of salmonids that are of most interest to fisheries managers: all salmon, all trout, salmon greater than 89mm in length and trout greater than 199mm.

Intake rate (nos. of these fish per bird per day) was calculated as the product of (i) the numbers of trout or salmon consumed per 100g of food intake (Chapter 3) and (ii) the estimated Daily Food Intake (DFI) of each bird species, based on their respective body masses. The numbers of fish per 100g of reconstituted stomach contents (Table 3.12) was used where ten or more stomachs had been sampled. Smaller samples were excluded because estimates of diet derived from them were likely to have had a large margin of error (Marquiss & Carss 1997).

The Daily Food Intake (DFI) of wild fish-eating birds is difficult to measure and this is reflected in the wide range of published estimates, derived from a number of techniques (reviews in Marquiss & Carss 1994, Russell et al. 1996, Carss et al. 1997a, Feltham & Davies 1997). These reviews concluded that some published DFI estimates, such as those derived from pellet analysis, were clearly underestimates. Others, such as those derived from samples of 'full' stomachs and some field measurements of energy expenditure, were overestimates. Amongst published estimates of DFI, the most likely values are in the order of 15-25% of lean body mass per day (e.g. Marquiss & Carss 1994, Russell et al. 1996) and for the purposes of the present calculations, the upper value of 25% was chosen.

The mean body mass for adult goosander (1496g), red-breasted merganser (1025g) and cormorant (3100g) was determined for a sample of mix-sex birds (30 of each sex) arbitrarily selected from the carcase samples provided for diet analysis (Chapter 3). The weight of goosander (469g) and red-breasted merganser (380g) ducklings was calculated from mean duckling weights from the Rivers Dee (17 goosander broods) and North Esk (9 merganser broods) in July (Alexander 1995). Goosander ducklings in August are full-grown ('juveniles') and were assumed to average the same as adults (i.e. 1496g). The DFI values used in our calculations were thus 25% of these values: 374g for adult/juvenile goosanders, 117 g for goosander ducklings, 256g for adult mergansers, 95g for merganser ducklings and 775g for cormorants.

Consumption rate

For the Rivers Dee and North Esk, the numbers of fish consumed by sawbills per hectare per day were calculated in relation to river width, as the product of intake rates and estimates of density (Chapter 2, Figures 2.5-2.6). In addition, we used the following density estimates for mergansers on the North Esk in April: 0.0365 (5m stream width), 0.1112 (15m), 0.092 (25m) and 0.5832 (35m) calculated from Marquiss & Duncan (1993). The consumption rate for large trout (>199mm) was ignored because they were seldom taken by sawbills on these rivers.

The percent of standing fish stock consumed by sawbills was only calculated for 'all salmon' because the appropriate data on fish abundance in relation to stream width were available for only this species. Densities of salmon likely to achieve a length of 90mm by the end of the growing season had been estimated on the North Esk (5 sites), Dee (3), Spey (5) and Tay (15) by electrofishing in summer in shallow areas (Cowx & Lamarque 1990) coupled, where necessary, with visual counting in deeper water while diving (Gardiner 1983).

RESULTS

Intake rates

Intake rates were calculated for goosanders on 12 river systems, mergansers on 7 and cormorants on 6 rivers and two stillwaters (Table 4.2). The highest salmon intake, 60 fish per bird per day, was calculated for juvenile goosanders on the Dee in August; the corresponding figure for large salmon was 12 fish per day. Other high values were calculated for the more northern rivers in spring (Beauly, Dee, Deveron, North Esk, Spey), winter (Dee) and summer (Dee ducklings). At all locations, salmon intake rates in spring were higher for mergansers than for goosanders, the highest being 27 fish per day on the River Laggan. For trout, the highest intakes were calculated for sawbills in spring on the Rivers Laggan, South Esk, Tay, Borders Esk and North Esk. Large trout were not consumed by mergansers and the intake of these fish was low for goosanders with highest values calculated for the Rivers Dee in spring 0.40 fish per day), Borders Esk (0.39) and Deveron (0.38) in winter.

Compared to sawbills, intake rates of salmon were low for cormorants, the highest (8 fish per bird per day) was calculated for the River Laggan in spring. The trout intake rates of cormorants were similar to those of sawbills, the highest (14-22 fish per day) being calculated for the Rivers Laggan and Annan in spring and the Deveron in winter. Corresponding figures for large trout on these three rivers were 2.6, 0.2 and 1.9 fish per day, respectively.