Hydrobiologia 190: 193-214, 1990. © 1990 Kluwer Academic Publishers. Printed in Belgium.

193

Impact of intensive cage fish farming on the phytoplankton and periphyton of a Scottish freshwater loch Hadrian P. Stirling & Trideep Dey Institute of Aquaculture, University of Stirling, Stirling FK9 4LA, Scotland, U.K. Received 18 May 1988; accepted 20 January 1989

Key words: Microcystis aeruginosa,phytoplankton, periphyton, cage fish culture, environmental impact Abstract Nutrients, phytoplankton and periphyton were monitored in a 71 ha shallow, unstratified lake used for intensive cage culture of rainbow trout. Inorganic nitrogen, ortho-phosphate and suspended solids were significantly higher near the cages and the bottom and, although declining during summer, nutrients did not reach levels which limit phytoplankton growth. Microcystis aeruginosadominated the phytoplankton, with surface chlorophyll a reaching 189 g 1- in August, but with no subsequent bloom collapse or deoxygenation. A sub-dominant community of 'vernal' diatoms and Pediastrum spp. persisted. Periphyton was dominated by Melosira italica-subarctica.Algal species and water quality showed the lake to be highly eutrophic. Chlorophyll values predicted from a phosphorus-dependent eutrophication model agreed with observations but light limitation by self-shading and suspended farm wastes, aided by wind-induced turbulence, is believed to control algal growth rates and biomass. Implications for environmental management of intensive freshwater cage farms are discussed. Introduction In recent years there has been a rapid increase in the number of Scottish fish farms which employ floating net cages to rear salmonids, principally rainbow trout, Salmo gairdneri Richardson, in natural freshwater lakes. Most of these farms are in the Highlands of Scotland because of the large number of available sites, with 16 such cage farms and an annual production exceeding 900 tons in 1984. Most farms are situated on the larger, deeper lochs (e.g. L. Awe & L. Earn) where the environmental impact on the water body as a whole is likely to be slow to develop because of the large physical 'buffering' (Bailey-Watts & Duncan, 1981). Nevertheless, there is concern about eutrophication where cage farming in lochs

used for supplying drinking water, such as Loch Lomond, has been proposed (Beveridge & Muir, 1982). The main impact of a cage farm, as with any fish farm, is to increase the load of phosphorus, nitrogen and organic matter which enrich both the water and underlying sediment (Alabaster, 1982). The effect depends primarily on the annual fish production and lake morphometry (area, depth and water residence time). Beveridge (1984) produced a model of the impact of cage culture of salmonids, based on previous work on the relation between phosphorus loading of lakes and resulting chlorophyll levels; the model predicts the capacity of a lake to produce fish while keeping water quality within acceptable limits. Phillips (1984) monitored the effect on dissolved

194 nutrients and chlorophyll of introducing fish cages to a 3.4ha Highland loch, for which Phillips, Beveridge and Muir (1985) provided a nutrient budget, while Merican and Phillips (1985) quantified the nutrient loads to the sediments of several Scottish farms. Similar nutrient budgets have been reported for cage farms in Poland (Penczak et al., 1982) and Sweden (Enell & Lof, 1983). These works describe the overall impact of cage culture in terms of nutrient and chlorophyll concentrations, but not other ecological effects. To date there has not been a detailed study of the phytoplankton and periphyton in a Scottish loch with a cage farm. The only similar study is that by Kilambi et al. (1976) of a warm monomictic lake in Arkansas, U.S.A. in which experimental cages were placed. Loch Fad was chosen for the present study because it is the site of one of the largest commercial freshwater cage farms in Scotland. The farm was established in 1976 and produced c. 200 tonnes of rainbow trout in 1981, increasing to c. 300 tonnes in 1986. Loch Fad (Fig. 1) is fairly small and shallow with an area of 71 ha, a mean depth of 5 m and a water residence time of 0.48 y (estimated by Bostock, 1987) so such a large fish farm can be expected to have had a considerable influence on water quality and plankton ecology. Unfortunately, biological information on the lake prior to 1976 is lacking, but it now shows many signs of eutrophication, including frequent dens blooms of cyanobacteria. Concern by the farm's management as to whether present production levels could continue prompted a number of investigations (see Bostock, 1987 for review).

Study area

Fig. 1. Loch Fad, showing bathymetry in metres (according to Murray & Pullar, 1910) and location of fish cages (stippled area) and sampling stations (F: cage farm; S: south).

wooded slopes on both sides. The lake was deepened in the 18th century by construction of a low dam at the northern end with a sluice outlet giving a maximum depth of 12 m. In 1981 the farm consisted of 75 floating cages, each 6 x 5 x 4 m (120 m3), moored together in 6 rafts of approxi-

Loch Fad (Fig. 1) on the Isle of Bute (55 ° 48 4 N: 5° 0.38 W) lies immediately to the south of the Highland Boundary fault which bisects the Island, with predominantly metamorphic schists and quartz rocks to the northwest and sedimentary sandstones to the southeast. It is fed by a few small streams draining arable land and steeply

mately 12 cages, each raft having a single point anchorage. The cages were stocked with rainbow trout at an average density of 15 kg m3 , giving a mean total biomass of 135 tonnes. The fish were fed on a commercial pelleted diet, occasionally supplemented with frozen blocks of shellfish processing waste.

195 Material and methods

temperature, secchi disc transparency and pH were measured in situ at around mid-day, dissolved oxygen by a modified Winkler titration (Strickland & Parsons, 1972) and nutrient ions by the methods described in Golterman, Clymo & Ohnstad (1978). Samples for chlorophyll-a determination were treated with 1% MgCO 3, stored in the dark for a maximum of 4 h before filtration through 0.45 Mm Sartorius membrane filters and

Water and plankton were sampled at monthly intervals from November 1980 to December 1981 at two stations, one among the fish cages and the other 1.61 km to the south west, henceforth referred to as the south station. Sampling was by means of a bucket at the surface and a transparent 5 litre Friedinger sampler at 4 and 9 m. Water

Table 1. Physico-chemical parameters (annual mean of 12 monthly values + standard error) at cage and south stations at Loch Fad, 1981, with significant results oft-test for paired comparisons between stations, averaged over all depths (* and ***indicate significance at P levels of ~<0.05 and 0.001, resp.).

Temperature

C

Depth

South station

Cage station

t

Om

10.48 + 1.43 10.16 + 1.43

Cage station Cage station

NS

9m Transparency m

1.99 + 0.13

1.83 + 0.12

5.08 ***

6.50 + 1.65 6.77 + 1.62

8.83 + 1.75 9.94 + 1.77

8.77 ***

Suspended solids mgl I

Om

Chlorophyll a #lgl L

Om 9m

Dissolved oxygen mgl-

Om 9m

9.94 + 0.29 9.55 + 0.29

9.77 + 0.32 9.20 + 0.31

2.18*

pH

Om 9m

7.38 + 0.085 7.28 + 0.092

7.34 + 0.080 7.19 + 0.075

2.35 *

a

b

Conductivity

9m

5m

28.4 20.6

175

+ 1.4 + 1.4

+ 4.2

30.8 18.3

175

+ 1.4 + 1.5

+ 4.8

NS

NS

#S cm Hardness

5m

1.55 + 0.064

1.09 + 0.045

NS

5m

0.604 + 0.024

0.572 + 0.027

NS

Orthophosphate mg P1-'

Om

0.043 + 0.0058 0.046 + 0.0053

0.048 + 0.0053 0.056 + 0.0057

6.02 ***

9m

Nitrate mg N1-'

Om 9m

5.73 + 0.65 6.45 + 0.68

6.11 + 0.88 7.57 + 0.94

2.65 *

Nitrite mg NI

Om 9m

0.030 + 0.075 0.033 + 0.0082

0.039 + 0.0098 0.046 + 0.011

NS

Ammonia mg N 1-

Om 9m

0.346 + 0.053 0.415 + 0.046

0.438 + 0.054 0.496 + 0.049

7.88***

Dissolved silica

Om 9m

3.39 + 0.42 3.70 + 0.41

3.42 + 0.41 3.67 + 0.43

NS

b

mmol 1b

Alkalinity

m equiv. 1-

mg SiO2 1 a b

Geometric mean; otherwise arithmetic. Data for April 1980 - March 1981 provided by Central Scotland Water Authority.

196 .

.

.

.

.

.

.

.

.

.

P

E

S

E

i

r

I

Fig. 2. Seasonal changes in physico-chemical parameters in Loch Fad. Temperature at 0 m (continuous line) and 9 m (dashed line); otherwise means of water column at south station (dashed line) and cage farm station (continuous line).

197 pigment extraction in 90 % acetone; chlorophyll-a was determined according to Tailing and Driver (1963). 500 ml water samples for phytoplankton analysis were fixed immediately in acid Lugol's iodine and left to concentrate 100-fold by sedimentation; cells of all species were enumerated in a Sedgwick-Rafter cell after sub-sampling the concentrate. To study the periphyton, artificial substrate consisting of 4 x 4 cm blocks of softwood were tied to horizontal wooden bars which were suspended at the cage and south stations just below the surface and at 4 m and 9 m depths by means of ropes held vertically between a surface float and a sinker. On each monthly sampling occasion (from January to December 1981), two wooden blocks were removed at each depth and preserved in 10% formalin and replaced by 2 fresh blocks. Subsequently the algae were scraped and washed off the block surface, dispersed uniformly and sub-sampled for identification and cell counts, the results being expressed as cells per cm 2 of substratum.

Results Physico-chemical characteristics Fig. 2 presents the variation in the major nonconservative water quality parameters form Nov. 1980 to Dec. 1981. It shows that the water column remains isothermal, with no persistent temperature stratification in summer. The maritime climate ensures that Loch Fad is virtually free of ice during winter. Thus wind-induced mixing by convection and turbulence extends right to the bottom throughout the year. This greatly reduces the extent of chemical stratification as shown by the annual mean values at 0 and 9 m in Table 1. Even dissolved oxygen in bottom water never fell below 7.2 mg 1- (70 % saturation), the minimum value being at the cage station in September. Apart from chlorophyll which was always much higher in surface water (with a maximum of 200 /g 1- ' ) and a tendency for a greater increment in suspended solids and nutrients in bottom water at the

cage station, differences between surface and bottom values, though consistent, were relatively small. Figure 2 shows that there was a clear annual cycle in Secchi disc transparency, total suspended solids and chlorophyll a with a distinct peak in the latter parameters in August which coincided with a bloom of Microcystis aeruginosa. All major nutrients showed a similar annual cycle, with high levels during winter and low levels during May-August. The nitrite maximum in September coincided with the oxygen minimum, at a time of collapse of the algal bloom and regeneration of other nutrients. The results in Table 1 of t-tests, using sampling dates as the basis for paired comparisons between stations, show that the most significant increments at the cage station over the south station (P < 0.001) occurred with suspended solids, orthophosphate and ammonia. These parameters represent the most immediate impact of the fish farm on water quality, representing the original waste materials (faeces and uneaten food, and metabolic wastes from the fish). The less significant and non-significant increments displayed by nitrate and nitrite, respectively, are consistent with their being derived from ammonia by bacterial nitrification, and becoming dispersed away from the cage site during this process. The less significant reduction in dissolved oxygen and pH at the cage station (P < 0.05) probably results from the counteracting effects of fish respiration and high rate of photosynthesis at mid-day. Site differences would undoubtedly be greater at night, since a 24-hour study at the cage station in April revealed a decline in D.O. of 0.8mgl -1 and in pH of 0.5 unit between 13.00 and 04.00. Not surprisingly, there are no station differences in dissolved silica which is not directly influenced by fish farming.

Phytoplankton biomass The contribution of phytoplankton to total suspended solids was estimated by converting chlorophyll values into dry organic biomass, assuming a chlorophyll content of 2.5% of ash-free dry

198 year at the cage station and 2.3 mg 1-' at the south station. Suspended particles limit light for phytoplankton and periphyton, especially at the cage site, but the effect is probably too localised to depress the phytoplankton biomass there.

Phytoplankton species E E

........ -- r---n

N

D

J

F

M

A

M

J

J

A

S

N

D

1981

Fig. 3. Dry organic biomass of phytoplankton in Loch Fad estimated from chlorophyll concentration. Means of cage and south stations at surface (continuous line) 4 m (dotted line), and 9 m (dashed line).

weight for the diatom-dominated community of winter 1980, and 1.2% for the Microcystis-dominated community of 1981 (Reynolds, 1984). The monthly means of the cage and south stations at each depth are presented in Fig. 3 and the paired t-test gave significant differences between all depths (P < 0.05). This concentration of biomass in surface and mid-water, especially during summer, is of significance to the phytoplankton in overcoming the light-limiting effects of deep mixing. On an annual basis, phytoplankton organic biomass accounted for a higher proportion of total suspended solids at the surface, being also higher at the south station (65 and 48 % at 0 and 9 m, respectively) than at the cage station (53 and 32%, respectively). During the August bloom,

these percentages were all c. 20% greater (range 50-87 %). This supports the finding of significant station differences in suspended solids and transparency and yet similar chlorophyll levels, and reflects the large contribution of non-living suspended matter from the cage farm. At the surface this amounted to c. 4.1 mg 1- throughout the

Table 2 shows the algal species recorded in the phytoplankton and their abundance throughout the year, as indicated by the geometric mean of cell counts at 0, 4 and 9 m (ascribing a value of 1 cell 1- to zero counts). The geometric mean gives particular emphasis to persistent species occurring throughout the year. In order of abundance in each group these are: the diatoms: Melosira italica-subarctica,Asterionella formosa, Navicula viridis, Cyclotella comta and Stephanodiscus astraea; the green algae: Pediastrum boryanum, P. duplex and Staurodesmus cuspidatus; and the blue green algae Microcystis aeruginosa, Coelosphaerium naegelianum and Anabaena flos-aquae. Seasonal species reaching significant peak abundances of over 60000 cells 1-1 include the green algae Botryoccoccus braunii,Sphaerocystis schroeteri,Pandorinamorum and Eudorina elegans and the blue greens Oscillatoria limosa and 0. agardhii; the seasonal cycles of these species are illustrated in Fig. 4. The major seasonal diatoms, not illustrated in Fig. 4 because of their low mean abundance, had the following maximum abundances (cells 1- ' x 10- 3 and month in parentheses): Amphora ovalis (20.4 August), Synedra ulna (18.6 May, Meridion circulare (18.2 April) Gomphonema constrictum (16.6 August), and Nitzschia linearis (15.5 May). The abundance of most species, including most green algae and the diatoms Melosira and Navicula, was significantly lower at 9 m than at 0 or 4 m, where counts were similar. Cyclotella comta was exceptional in being more abundant at 9 m. The blue-green algae were much more abundant at the surface that at either 4 or 9 m, and for much of July and August the lake was characterised by a distinct surface bloom consisting predominantly of Microcystis aeruginosa,with a peak

199 Table 2. Species of algae recorded from the phytoplankton and periphyton in Loch Fad, with their annual geometric mean abundance. Separate means at south (S) and cage farm (F) stations are shown where paired t-test gave significant differences (* and **indicate P levels of <0.05 and 0.01, resp). Phytoplankton a

Periphyton b

cells 1-

cells cm - 2

x 10- 3

Bacillariophyceae Amphora ovalis Kutz. Asterionellaformosa Hass. Cyclotella comta (Ehr.) Kutz. Cymbella ventricosa Ag. Gomphonema constrictum Ehr. Melosira italica (Ehr.) Kutz subarctica 0. MOll. Meridion circulare (Grev.) Ag. Navicula viridis Kutz. Nitzschia linearis (Ag.) W. Sm. Stephanodiscus astraea(Ehr.) Grunow. Synedra ulna (Nitz.) Ehr. Tabellariafenestrata (Lyngb.) Kutz.

0.18 75.1 21.0 0.03 S 0.76, F 0.80* S 131.7, F 107.4**

230 69 168 32 278 S 835,

0.49 S 61.1, F 91.9** 0.03 17.0 0.11 S 0.64, F 0.98**

39 470 27 158 9 57

0.03 1.70 0.03 S 0.73, F 1.36* 0.03 0.18 S 2.11 S 87.4, F 130.7* S 59.6, F 79.3* 0.02 0.15 0.71 0.16 0.46

24

Chlorophyceae Ankistrodesmusfalcatus (Corde) Ralfs Botryococcus braunii Kutz. Closterium kutzingii Breb. Eudorina elegans Ehr. Gleocystis gigas (Kutz.) Legarh. Oocystis elliptica West Pandorinamorum (Mull.) Bory. Pediastrum boryanum (Turp.) Menegh. Pediastrum duplex Meyen Scenedesmus abundans (Kirchn.) Chod. Scenedesmus quadricauda(Turp.) Breb. Sphaerocystis schroeteri Chod. Spirogyra gracilis(Hass.) Kutz. Staurodesmus cuspidatus (Breb.) Teiling

6

89 33 3 21 90

Dinophyceae Ceratium hirudinella (0. Mull.) Schrank Peridinium willei Huntf.-Kass.

0.03 0.19

Cyanophyceae Anabaenaflos-aquae(Lyngb.) Breb. Coelosphaerium naegelianum Unger Microcystis aeruginosa Kutz. Oscillatoriaagardhii Gomond Oscillatorialimosa Ag. a b

0.83 0.99 9.570 2.29 2.43

Geometric mean of (monthly mean of counts at 0,4 and 8 m) + 1. Geometric mean of (monthly mean plaque count at 0 and 4 m) + 1.

564

257 339

F 1117**

200 I

A

I

I

0

2

IJ I

Microcystis aeruginoe O0'

t

I I

I~~~~~~~ " - W

I

I b

Oscillatoril agardhii

b

p.

t-

o'

Oitori

limo

Cyciotllc mtm N D J

F

.

M AM. J J

.

.

.

.

A S ON

.

D

N D

F MAM

J

J

A S O

N D

Fig. 4. Seasonal changes in abundance of major species of phytoplankton in Loch Fad. Means of counts at 0.4 and 9 m at cage farm station (continuous line) and south station (dashed line).

201

Total Bacillariophyceae

Melosira italicasubarctica

Asterionella formosa

Navicula viridis

Cyclotella comta

--

50--

Total Chlorophyceae

40-

20-

- A"

'\ Pediastrum

duplex

20-

I-1

I

P. boryanum

I

Botryococcus

0-

braunii

Sub-dominant

20:

Cyanophyceae _

Oscillatoria LO limosa

I,

'f

20-

U.

10

surface count of 4.9 x 108 cells 1-'. The most striking feature of the phytoplankton community was the continuous predominance of this species throughout 1981, during which its contribution to total abundance never fell below 94% and remained above 99% from June to September, although it was not recorded in November and December 1980. There was a significantly higher abundance of many of the commoner species at the cage farm station, using logs of monthly water column means as the basis for paired comparisons with the south station (Table 2). This difference was not apparent among blue-green algae, including Microcystis, possibly due to the horizontal dispersal of the surface bloom. The diatoms show a more confusing picture; of the three major species only Navicula was more abundant at the cage station, while Melosira was more abundant at the south station and Asterionella showed no difference. Figure 5 illustrates seasonal changes in the relative abundance of the major components of the phytoplankton community apart from Microcystis aeruginosa, including the percentage contribution of each algal group. The inclusion of Microcystis would have masked changes in the community of subdominant phytoplankton species, which exceeded a mean total count of 1 x 106 cells 1-' from April to September. Diatoms, chiefly Melosira and Asterionella comprise well over 50% of this sub-dominant community from November to March, with Navicula and Cyclotella also contributing during November-December. Blue-green Oscillatoria spp, notably 0. agardhii,contribute over 20% from March to May while green algae, mainly Pediastrum boryanum, comprise about 50% of the community from June to October. The absolute and relative abundance of these diatoms and green algae during November and December was not markedly

'/ .

0. agardhii .

.

.

.

.

N D J FMAMi

..

.

.

.

J ASOND 1981

Fig. 5. Seasonal changes in percentage composition (weighted mean% total cells at 0 and 9 m) of major components of subdominant phytoplankton, excluding Microcystis aeruginosa at the cage station (continuous line) and south station (dashed line).

202

4-

E to

TotalPeriphyton

O3 -

3-

% Total

70 -, 50-

' /

,.'

I Bacillario'- phyceae 30

- -

Melosira _

-/

italica-

affected by the presence of Microcystis in 1981, compared with 1980 when it was absent. Differences between stations in the relative abundance of the major species are similar to those in absolute abundance described above, but paired t-tests on monthly percentages (using the arcsin transformation) revealed that total% Chlorophyceae was significantly higher at the cage station (P < 0.001) while total% Bacillariophyceae was significantly higher at the south station (P < 0.05).

1 subarctica

20I --

I

Navicula

I

O

-// 20 Amphora

0IL ,-

10

-\

ovalis

'

Cymbella ventricosa

\ ~10

0~~~~

[1

s > __~~

20

Meridion circulate

-

Cyclotella 0

-

-

comta

%Total chlorophyceae

.,--C_. ~/

'

- -

' I ok'-. \

t10 Spirogyra L 0 gracilis

%Total Cyanophyceae Oscillatoria agardhii

O.limosa

Anabaena floe- aquae J

F M A MJ

J

1981

Periphyton

viridis

A S O N D

The geometric mean annual abundance of periphytic algae recorded on wooden plaques suspended at 0 and 4 m is shown in Table 2. Very small populations were observed at 9 m. Apart from the absence of Microcystis in the periphyton, many species occurred in both plankton and periphyton. The presence in the plankton of many tychoplanktonic or typically periphytic algae was probably due to the large surface area of suspended netting and wood constituting the fish cages, coupled with turbulence which maintained detached, heavier algae in suspension. Melosira italica-subarcticawas by far the most abundant periphytic species overall, and was the only species with a significant population difference between the cage and south stations but in this case higher at the cage station. Navicula viridis and Cyclotella comta were also abundant, as they were in the phytoplankton, but many typically periphytic diatoms were much more significant in the periphyton, including Amphora ovalis, Gomphonema constrictum, Tabellariafenestrata, Meridion circulare, Cymbella ventricosa and Nitzschia linearis. Green algae were less significant in the periphyton with the exception of Spirogyragracilis, while among blue-green algae Anabaena flosaquae was more significant in the periphyton, its Fig. 6. Total cell numbers of periphyton and seasonal changes in percentage composition of major components at the cage station (continuous line) and south station (dashed line).

203 annual mean abundance exceeding that of the Oscillatoriaspp. Figure 6 illustrates seasonal changes in total cell numbers of periphyton and in % composition of algal groups and principal species, expressed as a weighted mean % of total cells at 0 and 4 m. Compared with the phytoplankton, periphytic diatoms dominate throughout the year, their total contribution not falling much below 50 % even in summer. Total periphytic green algae show a smaller mid-summer peak, while Cyanophyceae contribute most at the beginning and end of the year. Differences between stations in group percentages are not significant but in September and October only, diatoms (notably Amphora) contributed much more at the south station, while blue-green algae (notably Oscillatorialimosa) and Spirogyragracilis made higher contribution at the cage station. Differences in certain individual species between stations were more significant in terms of relative abundance, when tested by paired comparisons after arcsin transformation. Not only did Melsoira contribute significantly more at the cage station (P < 0.01) but Navicula less (P < 0.05), a situation opposite to that found in the phytoplankton. Oscillatoria agardhiimade significantly higher contribution at the south station (P < 0.05).

these nutrients from the sediment as a result of mineralisation of organic matter. Certainly the increment in nutrients (except silica) at the bottom was greater at the cage station which received the higher organic input. Bottom water was fairly well oxygenated even under the cages, agreeing with studies of redox potential which showed that bottom sediments were aerobic to a depth of 4.5 cm. This should accelerate mineralisation by aerobic micro-organisms and the remobilization of phosphorus. Drake and Heaney (1987) observed significant phosphorus release by desorption from aerobic sediments of eutrophic Esthwaite Water, especially at elevated pH associated with algal blooms. Another explanation of lower nutrients in surface water is uptake by phytoplankton which was significantly more abundant at the surface, with correspondingly higher chlorophyll concentrations (Table 1). The uptake of nutrients by phytoplankton is supported by the strong inverse relation between chlorophyll and NO3-N, NH 3-N and P0 4 seen in Fig. 2 (correlation coefficients all significant at P < 0.001). This vertical gradient in nutrients parallels the station differences described in the results and reflects the complex dynamic balance between rates of supply from fish cages and sediments and uptake by algae, as modified by mixing, dilution with low-nutrient inflows and losses by sediment/adsorption and via the outflow.

Discussion Nutrient distribution

Comparativefeatures of phytoplankton community

The observed lack of temperature stratification in Loch Fad is not unexpected because the mean depth of 5 m is less than the summer mixing depth (Zm) of 6.3 m predicted by Ragotskie's (1978) for-

The dominance of Microcystis throughout an entire year is unusual in British lakes, although it frequently dominates the late summer/autumn phytoplankton of shallow, productive lakes, such as Rostherne Mere (Reynolds, 1980), as well as experimentally fertilised Blelham Tarn enclosures (Lund & Reynolds, 1982). In these situation it overwinters as vegetative colonies on the bottom, where the necessary stimulus for subsequent development appears to be hypolimnetic oxygen depletion coupled with low light intensity (Reynolds & Walsby, 1975). Reducing conditions, however, did not preceed its appearance in

mula (Zm = 4 x Fetch of 2.5 km). Vertical mixing

will be enhanced by the lake's exposure to prevailing westerly winds, magnified by the windfunelling effect of hills to the NW and SE. A small but consistent increase with depth, however, was observed in nutrient concentrations; for example, paired t-tests on NO 3 and P0 4 increments between 0 and 4 m and 4 and 9 m were significant at P < 0.001. This may well reflect a source of

204 Loch Fad in Jan. 1981, and it remained in suspension in considerable numbers during the following winter. It is quite possible that propagules of Microcystis were present in suspension during Nov.-Dec. 1980, but were not detected by the counting method, whose sensitivity was limited to about 200 cells 1- '. Reynolds (1973) did report the persistence of Microcystis throughout the winter in eutrophic Crose Mere, where the annual cycle was remarkably similar to that in Loch Fad, but with a lower peak abundance of 22-69 x 106 cells 1- 1'in July - August. Microcystis continuously dominates warmer, shallow, eutrophic waters such as Hartbeespoort Dam, S.A. (Scott etal., 1980). Lake George, Uganda (Ganf, 1974) and temple ponds in India (George, 1962). The vast numerical dominance of Microcystis is due in part to its small size. This dominance is smaller on a biomass basis. Employing the mean dry cell weights given in Reynolds (1984), the annual mean biomass of Microcystis is c. 0.3 mg 1-' compared with c. 0.05 mg 1- ' each for Melosira and Stephanodiscus. Differences between peak biomasses are large, with maximum surface values of c. 15 mg l-' for Microcystis, 0.2 for Melosira and 0.14 for Asterionella, while Ceratium briefly attains 0.16 mg 1-'. The coexistence of a variety of species with Microcystis (up to 33 during the summer) contrasts with the exclusion or active suppression of algal competitors observed by Reynolds (1980, 1984) in Rostherne Mere and Blelham enclosures. Also total diatom numbers in November December, 1980, when Microcystis was not recorded, were not significantly greater than in the same period of 1981 when it continued to dominate (overall column means for 1980 and 1981 were 2.4 x 105 and 2.2 x 105 cells 1- , respectively). The apparent absence of growth inhibition indicates that Microcystis dominated as a result of successful competition with other species for such factors as light (through attenuation) and possibly organic carbon sources, as discussed in detail below. Thus Microcystis appears to have dictated the physical environment, primarily light availability, to which other algae were exposed.

The subdominant algal community, underneath this 'parasol' created by Microcystis, reached a surface maximum of 2.6 x 106 cells 1-' in July, which compares quite favourably with neighbouring Scottish lochs. The diatom community and species succession in Loch Fad is similar to that in the south basin of Loch Lomond (Maulood & Boney, 1980; Bailey-Watts & Duncan, 1981). In both lakes Melosira italicasubarctica dominates during winter and peaks in April, with Asterionella formosa also peaking in April and again in August; Tabellariafenestrata and Cyclotella comta are important in autumn in Loch Lomond, but not Stephanodiscus. The same four prominent diatoms also characterise the spring community of Lake of Menteith, (Maulood & Boney, 1981). As in Loch Fad, blue-green algae dominate the phytoplankton, but in this case with frequent blooms of Anabaena spp., while Microcystis was common from July to November. Dominance by Melosira and Asterionella in spring also occurs in oligotrophic Windermere and Grasmere (Hutchinson, 1967). There is a lack of information on the phytoplankton of poorly stratifying waters in Britain (Bailey-Watts, 1978), but parallels can be drawn between the phytoplankton community of Loch Fad and those of the shallow, well mixed and eutrophic Loch Leven and Lough Neagh. These are also characterised by persistent blue-green blooms, but in this case chiefly of Oscillatoriaspp. instead of Microcystis. Both lakes display a continuous growing season with considerable year to year variation in species composition and biomass development. Loch Fad, on the other hand, appears to have been dominated by Microcystis almost continually since 1980 (at least during the summer) but biomass has varied, with peak chlorophyll a levels ranging from 80 to 360tg 1-' (Bostock, 1987). The diatoms of Lough Neagh are similar to those of Loch Fad with 9 species in common, of which Melosira and Stephanodiscus frequently dominate for long periods in spring, later giving way to Oscillatoria redekei or 0. agardhii(Gibson et al., 1971). Loch Leven also shows prolonged dominance by diatoms (mainly minute species of Stephanodiscus

205 and Cyclotella) early in the year, followed either by varied populations of green algae or by long-lived blooms of nannoplanktonic Oscillatoria redekei, Synechococcus sp.n. and latterly by the larger 0. agardhiior Anabaenaflos-aquae (Bailey-Watts, 1974, 1978). Cryptophyceae, Chrysophyceae and Xanothophyceae were apparently absent from Loch Fad and were also unimportant in Loch Leven and Lough Neagh. Recently, however, a species of Cryptomonas dominated the Loch Fad phytoplankton in April/May of 1987 (Bostock, 1987).

Seasonal changes in phytoplankton The absence of ice cover or sustained stratification in Loch Fad prevents the seasonal succession of distinct species assemblages described for stratifying lakes by Reynolds (1980) and is probably responsible for the persistence in the plankton of heavy diatoms such as Melosira and the periphytic species Navicula and Gomphonema. This situation is similar to Loch Leven, where the growing season of cold and low light-adapted species also extends throughout the winter. (Bailey-Watts, 1978). A gradual shift in the major sub-dominant species did occur over the year in Loch Fad, starting with diatoms in winter and spring, then a brief burst of Oscillatoriaagardhii followed in mid-summer by green algae, with diatoms becoming established by the following winter. Seasonal species in Loch Fad were rarely important; further details are given in Dey (1984).

Trophic status of Loch Fad Loch Fad is highly eutrophic or even hypertrophic as judged by a number of biological and chemical criteria, and there is little doubt that the principal cause of eutrophication is the fish farm. The clearest biotic indicator of eutrophy is dominance by Microcystis aeruginosa. Rawson (1956) considered that dominance by this species definitely indicated eutrophy in Western Canada, while Hutchinson's (1967) review of earlier work on

phytoplankton associations stressed the link between blue-green blooms and eutrophy. Most of the subdominant species in Loch Fad are also strongly associated with eutrophic conditions, including the major diatoms Melosira, Stephanodiscus and Asterionella and the major green genus Pediastrum (Hutchinson, 1967). With only 2 species of desmids and a preponderance of centric over pennate diatoms, the phytoplankton indices of Thunmark (1945) and Nygaard (1949), with the latter's 'compound index' = 10.5, clearly classify Loch Fad as productive. Rawson (1956) however, cautioned against too much reliance on such indicators because of the dependence of species counts on the sensitivity of the sampling method, the 'swamping' effect of dominant species and the difficulty of separating truly euplanktonic from littoral and periphytic forms in small shallow lakes. Geographical differences are also important. Many diatoms common in Loch Fad, such as Asterionellaformosa and Tabellaria fenestrata, are typical of oligotrophic lakes in Canada while Cyclotella comta dominates oligotrophic lakes in Finland; similarly, both species of Pediastrum are considered by Rawson (1956) as mesotrophic indicators in Canada. Algal genera and species may indicate the degree of organic pollution, apart from nutrient enrichment. Most of the common algae in Loch Fad are listed by Sladecek (1973) as indicative of P-mesosaprobic conditions, in which primary production is largely supported by a moderate input of allochthonous organic material and is not nutrient limited. The mean saprobic index of 24 species covered by Sladecek is 1.7 + 0.08 S.E., i.e. between 2 for ,-mesosaprobic and 1 for oligosaprobic, confirming that organic loading is only moderate. Indeed Sphaerocystis schroeteri and Closterium kutzingii are claimed to be good indicators of oligosaproby. A similar conclusion is reached by inspecting Palmer's (1969) lists of algae tolerating organic pollution; Loch Fad possesses 5 out of the 10 most pollution-tolerant species, but of these only Oscillatoria limosa and Pandorina morum attain the necessary threshold peak abundance of 50 cells ml- . Applying Palmer's algal

206 genus index gives a value of 11 and his species index is 5, both well below the critical value of 15 for 'probable evidence of high organic pollution'. Caution is needed, however, in applying these indices to Loch Fad as neither emphasises British algae, while Palmer's index is largely based on small, hypertrophic sewage stabilization ponds which have relatively short residence times favouring small, fast-growing green algae at the expense of blue-greens. Palmer (1969) fails even to list Oscillatoria agardhii or Microcystis. While local organic loading underneath fish cages is extremely high (estimated by Phillips et al. (1985) to be 150-300 kg of waste food and 250-300 kg dry weight of faeces per tonne of production), the loading over the entire Loch is likely to be more moderate in its effect on the plankton, but the indices do appear to underestimate the organic pollution. Dominance and number of species can be related by the concept of diversity, which declines with increasing nutrient enrichment. A simple diversity index is that of Margalef (1958) ds = (S - )/logeN where s is the number of species and N the total number of individuals per litre.ds ranges in Loch Fad from 1.3 in January to a maximum of 2.8 in May, when s reaches 32 species while abundance is still only moderate. This range in d is intermediate between oligotrophic, nutrient limited Windermere (range 1.78-6.44) and highly enriched Blelham enclosures (range 0.24-1.75) (Reynolds, 1984). This supports the above conclusion that the phytoplankton is only moderately stressed by enrichment, but d will tend to be higher and have a smaller range in permanently mixed lakes compared with the seasonally stratified examples quoted by Reynolds (1984). Water quality parameters can also be used to assess trophic state, as in the fixed and open boundary classifications of OECD (1982) which are based on a comparison of over 100 north temperate lakes. According to the fixed boundaries, the mean and minimum Secchi transparency in Loch Fad (1.9 and 1.0 m, resp.) indicate eutrophic conditions, while the annual mean and maximum chlorophyll a (46.4 and 160 ig 1- 1,

resp.) indicate hypertrophy. The open boundary classification takes account of the statistical uncertainty of any predicted trophic category. Thus the mean Secchi transparency gives probabilities of 50% for eutrophic, 40% for hypertrophic and 10% for mesotrophic, the mean chlorophyll 80% for hypertrophic and 20% for eutrophic and the maximum chlorophyll 70% for hypertrophic and 30% for eutrophic.

Prediction of phosphorus load and resulting chlorophyll levels in Loch Fadfrom eutrophicationmodels Phosphorus is the most important nutrient determining the trophic state of most fresh water bodies (OECD, 1982) and there are many publications on the prediction of eutrophication from total phosphorus loading (see Beveridge, 1984). As total phosphorus was not measured in Loch Fad, the first step is to estimate this from the total phosphorus loading to the lake from the fish farm plus the natural phosphorus load from the watershed. From an experimental study of trout culture in cages under conditions similar to those in Loch Fad, Phillips (1984) estimated the total phosphorus load to the environment at 18.8 kg per tonne of fish produced. Therefore, P loading to Loch Fad from the fish farm (200 tons production from 71 ha) = 5.3 g m- 2 y - . This loading can be substituted into models which predict the average annual in-lake total phosphorus concentration, [P], allowing for loss to sediments and/or outwash. A simple empirical model which corrects for flushing is that of Vollenweider (1976), as modified by OECD (1982) for shallow lakes and reservoirs: [P] = 154mg m - 3 of/g 1- (95 % confidence limits c. 70-400) An alternative model is that of Dillon & Rigler (1974), with phosphorus retention estimated from flushing rate (Larsen & Mercier 1976) on which Beveridge (1984) based his carrying capacity model. According to Beveridge (personal communication) this model underestimates the

207 proportion of P lost permanently to sediments from fish farm wastes, which contain a high proportion of insoluble apatite - P, so applying his correction:

Similarly, the maximum chlorophyll - concentration can be predicted from max [Chl] = 0.64 [] 1.0 5 = 132 /g I- 1 (95 % c.l. 40 - 400)

[P]= 131gl--1.

The agreement between these estimates is acceptable in view of the wide confidence limits of the inherent models. To these estimates of phosphorus originating from the fish farm must be added an estimate of in-lake total P derived from other sources. Bostock (1987) estimated the later to be 12 ug 1- ' according to the land use of the catchment, rainfall and evaporation. An alternative estimate can be derived from the standard ratio of silica-silicon to total phosphorus = 110: 1 in unenriched watersheds (Golterman & Kouwe, 1980). Thus maximum winter silicate = 5.5mgSiO 2 1 = 2.6mgSi l-, so naturally occurring phosphorus should be 2.6/110 = 24 g l-'. Both these estimates fall within the range of total phosphorus in mesotrophic lakes (OECD, 1982) and probably represent the condition of the lake prior to establishment of the fish farm. Assuming a mean value of natural phosphorus of 18 g 1-, and a mean of the above estimates for P derived from the fish farm of 142 ug 1- , the total in-lake P concentration =

160 Mg 1- '. Compared with a mean overall orthophosphate-P value of 48 ug 1- , this is higher than might be expected on the basis of a P0 4 -P to total phosphorus ratio in eutrophic lakes of about 0.5 (OECD, 1982), but fish wastes contain a lower fraction of biologically available P0 4 -P to total P than sewage or agricultural wastes. Bostock (1987) observed that this fraction in Loch Fad in Spring 1987 was 0.3 - 0.5. According to the open boundary classification of OECD (1982), 160 1g P 1- gives Loch Fad an equal probability of being eutrophic or hypertrophic. Finally the mean annual chlorophyll a concentration ([Chl]) can be predicted from the in-lake total P level according to the empirical model derived by OECD (1982): [Chl] = 0.28 [p]0 .96

= 36.6 ugl

-

(95% c.l. 12 - 140 gl - 1 )

Considering the very wide confidence limits in making individual predictions from these models, including those initially used to estimate [P], and the uncertainty in assuming a fixed P loading, these predicted chlorophyll levels are surprisingly close to the observed mean and maximum values. The OECD models were derived from data which were thoroughly screened to exclude lakes in which there was internal P-loading, light or nitrogen limitation or artificial aeration. The former two factors cannot be excluded from Loch Fad, and yet the development of phytoplankton biomass in 1981 is consistent with its being dependent on external phosphorus loading. This is not to conclude that phosphorus was necessarily limiting phytoplankton in 1981, although the large increase in P-loading since the establishment of the fish farm has undoubtedly been the principal cause of eutrophication (Schindler, 1978). Even after screening of OECD lakes, the coefficients of determination (r2 ) of the above equations were at best 0.77-0.81, implying that 19-23 % of the variation in predicted chlorophyll levels was unexplained by phosphorus. Various attempts have been made to improve chlorophyll prediction from phosphorus loads, including the incorporation of total nitrogen into a multiple regression model by Smith (1982) and the development of a separate regression for mixed lakes by Riley and Prepas (1985). These models reduce the amount of unexplained variation, but substantial error still remains. Bostock's (1987) analysis of historical trends in annual mean P0 4 -P and chlorophyll a levels in Loch Fad has revealed a very variable relationship between these parameters, with particularly high chlorophyll in 1980 and 1983 while P04 levels were slowly increasing, since when [PO4 ] has more than doubled (to 137 Mg 1- 1 in 1986) while chlorophyll has halved to 22 Mg 1- . Thus phosphorus does not appear to limit chlorophyll in recent years,

208 and phosphorus models must be applied with caution. The other factors limiting phytoplankton development in Loch Fad are considered in detail below.

Factors limiting phytoplankton biomass The factors limiting the development of phytoplankton in Loch Fad should be related to the morphological and physiological adaptations of Microcystis because of its dominance throughout 1981. Seasonal cycles of water temperature and availability of incident light control the rate of biomass development (Moss, 1980) and are primarily responsible for the unimodal cycle in Fig. 6. At 55 ° N, daily photoperiod increases from 7 h in December to 17 h in June, but peak biomass coincided with maximum water temperature in July-August, as has occurred in subsequent years (Bostock, 1987). The greater influence of temperature may be explained by significant self-shading by algae which prevented the underwater light climate from keeping pace with photoperiod. Microcystis started to grow exponentially in February when the water temperature was 5 C, much lower than 15 C at which it started to grow rapidly in Crose Mere (Reynolds, 1973). Apart from temperature and light, growth rate and ultimate standing crop can be limited by the availability of nutrients, while the observed net rate of increase depends upon the balance between rates of cell division and losses through the outflow and by sinking, grazing, parasitization etc. Under optimal conditions, Microcystis aeruginosais capable of a growth rate of 0.25 d (Reynolds, 1973). In Loch Fad this species maintained an average net growth rate from February to July of 0.023 d- 1, with the highest rate between March and April of only 0.04 d- . Although observations at monthly intervals may have obscured higher short-term growth rates, some factor was clearly limiting its growth, yet did not prevent its dominance. Inorganic carbon supply is unlikely to be limiting because the moderately high alkalinity kept the pH from rising above 7.8 during the summer

maximum, ensuring enough bicarbonate availability which blue-green algae are better able to utilize than other algal groups (Hutchinson, 1967). The possibility of exhaustion of free CO 2 during the summer bloom, which can limit green algae (Shapiro, 1973), is reduced by respiratory CO 2 production from the high fish biomass and the large bacterial and benthic community which is generated by farming activities. Dissolved organic carbon compounds are undoubtedly abundant in Loch Fad and may favourMicrocystis, although evidence of heterotrophy in cyanobacteria is equivocal. Hutchinson (1967) and Van Liere & Mur (1980) consider cyanobacteria to be photoautotrophs, but Reynolds & Walsby (1975) consider most species to be photoheterotrophic, being able to assimilate simple organic compounds at low irradiance. This could be a considerable advantage to Microcystis, enabling growth to continue while below the euphotic zone, which is much less than the mixing depth in Loch Fad. Silica, while not required by cyanobacteria or green algae, is a major nutrient for diatoms and its limitation is the commonest cause of bloom collapse (e.g. Youngman, Johnson & Farley, 1976 in hypertrophic Farmoor Reservoir; Gibson et al., 1971 in Lough Neagh). In Loch Fad there was a peak in total diatoms of c. 8 x 105 cells 1- in April, before dissolved SiO2 declined to a minimum of 1.8 mg 1- in May and stayed below 2.2 mg 1-1 into August, when a second diatom peak of similar size occurred. This periodicity is clearly demonstrated by Asterionella and Melosira in Fig. 3. In spite of a significant negative correlation between total diatom numbers and silica (P < 0.001), silica did not appear to limit diatoms because the minimum concentration was above the critical minimum SiO2 level of 0.5 mg 1- 1 for Asterionella (Lund, 1950), 0.2 for Tabellaria and 0.8 for Melosira (Maulood & Boney, 1980). The maintenance of high numbers of periphytic diatoms (c. 6 x 103 cells cm - 2 ) from April to August may have kept down SiO 2 levels over the summer. The mid-summer decline in planktonic diatoms may have been due to competition for light with green algae, or possibly to grazing by

209 cladoceran zooplankton. It has been demonstrated above that the estimated phosphorus loading could have accounted for the observed phytoplankton biomass. Phosphorus limitation is also indicated by a marked departure from the ratio of inorganic nitrogen to reactive phosphorus of between 7: 1 and 16: 1 by weight, required for optimum algal growth (Golterman & Kouwe, 1980; OECD, 1982). Maximum nutrient levels in February were 9.62 mg 1-' for inorganic N and 0.07 mg-1 for P 4 giving a N:P ratio of 135; this ratio increased to 153 during the August bloom. The decline in available N and P as biomass increased to its maximum was noticeable, but much less severe than during blooms of Oscillatoria spp. in Lough Neagh (Gibson & Stevens, 1979), Lake Wolderwijd (Zevenboom, de Vaate & Mur, 1982) and Windermere (Tett, Heaney & Droop, 1985). Table 3 compares the minimum inorganic N and P0 4 -P levels in Loch Fad with estimates of concentrations below which these nutrients limit algal growth. Nitrogen (mainly as NO3 ) is vastly in excess of any limiting level and even phosphate remains above these limits. Even the severe cases of nutrient depletion referred to above were not generally associated with limiting internal cell quotas of N or P, partly because of 'luxury' uptake and storage of P which occurs in cyanobacteria (Reynolds & Walsby, 1975). The water column in March contained enough reactive phosphorus to account for the maximum mean dry biomass of Microcystis of 13.3 mg 1- in August, giving a calculated P-content of 0.45 % which is close to the uppermost value recorded by Gerloff and Skoog

(1957). The supply of inorganic nitrogen was certainly more than adequate to account for the subsequent biomass development, allowing for a maximum N-content of 5%. Although these rough calculations ignore competition for nutrients from periphyton and bacteria, the probability of nutrient limitation is further reduced by continuous high nutrient fluxes from the fish farm as well as the sediments. As stated by Zevenboom, de Vaate & Mur (1982), external nutrient levels do not necessarily reflect steady state values but rather the balance between rates of supply and uptake by algae, the latter depending on their physiological state. Thus low concentrations of external nutrients or extreme N:P ratios do not necessarily reflect the nutrient status of algal populations. High loss rates of Microcystis cannot account for its slow net growth in Loch Fad. Microcystis forms large mucilaginous colonies which resist ingestion by filter-feeding zooplankton, which in any case were not common in Loch Fad. Total Cladocera (mainly Daphnia longispina Mull.) reached a maximum density of only c. 30 adults l- in May-June, while Diaptomus gracilis Sars, the only herbivorous copepod, reached c. 10 1in June-July. Grazing by amoeboid and ciliate protozoa and fungal parasitization are possible but unlikely to be significant (Reynolds et al., 1981). Loss by sedimentation is prevented because Microcystis has a well-developed buoyancy control mechanism, aided by large volumes of low-density mucilage and large colony diameter (Reynolds, 1973; Reynolds & Walsby, 1975). The sharp decline after August was probably caused

Table 3. Limiting concentrations of inorganic nitrogen and orthophosphate for growth of phytoplankton, compared with summer minima in Loch Fad. N (g I- )

P (/g l- )

Reference

Loch Fada General Phytoplankton

2000 28

13 4.7

Wofsy (1983)

Microcystis aeruginosa Oscillatoria agardhii

240 168

8 10

Reynolds & Walsby (1 9 7 5 )b Zevenboom et al. (1982)C

Surface values at south station, August 1981. Estimated from loading per m2 need to maintain light-limited population/mean depth. c Estimated as 10 x K,, the half-saturation constant for nutrient-limited growth. a

b

210 by declining light and temperature leading to loss of physiological condition, reduction in relative volume of gas vacuoles and eventual loss of buoyancy. Part of the population may have been removed by displacement of surface colonies onto leeward shores, as observed by Reynolds (1973) in Crose Mere, but there were no prolonged calm spells or noticeable surface scums during 1981. Light appears to be the principal factor limiting Microcystis growth and maximum standing crop, mediated by self-shading and turbulent mixing which restrict the light received per cell. The maximum chlorophyll per unit area of the euphotic zone depends on its absorption characteristics which, in the absence of irradiance measurements, can be predicted from the observed relationship between chlorophyll concentration, [Chl] and Secchi disc transparency, Zs. The linear regression equation relating inverse of transparency to surface chlorophyll a, calculated separately for each station and then pooled as the coefficients and intercepts did not differ significantly, is 1/z, = 0.00258 [Chl] + 0.418. This is highly significant, with r2 = 0.90 and S.E.'s of coefficient and intercept = 1.70 x 10- 4 and 0.0125, respectively. Assuming that the euphotic zone is limited to the depth (Zeu) receiving 1 % of the photosynthetically active irradiance which penetrates the surface and that the Secchi transparency similarly corresponds to the 15% light level (Westlake, 1980; Van Liere & Mur, 1980), according to Beer's Law: keu = 4.6/Zeu = 1.9/z, where keu is the effective extinction coefficient for photosynthetically active radiation. It also follows that Zeu = 2.4zs. Therefore multiplying both sides of the above empirical equation by 1.9: keu = 0.00484 [Chl] + 0.794

or keu = ks [Chl] + kq

k, is the specific attenuation coefficient per m per mg Chl a m - 3 and kq is the average attenuation coefficient due to non-algal components (water colour, organic and inorganic particles). Applying

the same coefficient of variation as in the original regression on Secchi depth, the 95% confidence limits of ks are 0.0043 - 0.0055. This is lower than published values of between 0.006 and 0.020 m (Westlake, 1980) and may be due to the small cell size of Microcystis. Bindloss (1976) also found a low ks value in Loch Leven dominated by nannoplankton as well as a similar value for k of 0.69 m- . Since the euphotic zone is defined as 4.6/keu (equivalent to 3.7/kmin found by Tailing,

1971) the theoretical upper limit of chlorophyll per m2 of euphotic zone, assuming kq = 0, Bma = 4.6/ks = 946 mg Chl a m- 2. As kq is sig-

nificant, this estimate must be reduced by the fraction kq/keu so the fraction of total attenuation caused

by the algae

will

be

(1 - kq/ke).

This = 0.53 during the August 1981 bloom, when mean Secchi transparency was 1.1 m, i.e. maximum keu = 1.7 m- . Hence the predicted maximum euphotic chlorophyll = 0.53 Bmax or c.

500 mg m -2, which compares with an observed maximum of 490 mg m- 2 (from product of mean surface [Chl] = 189 mg m - 3 and estimated Zeu = 1.1 x 2.4 = 2.6 m). Thus suspended waste par-

ticles significantly reduce the light available for phytoplankton and limit biomass to levels well below those caused by self-shading alone. Turbulent mixing also has a significant effect. Light limitation of algal biomass sets in when the depth of mixing (Zm) sufficiently exceeds the

euphotic depth. Horn & Paul (1984) stated that this critical ratio of

Zm:

zeu = 2.5, which is not

reached in Loch Fad if Zm is assumed to equal the mean depth of 5 m and minimum Zeu in August is 2.6 m. Wofsy (1983) showed that, in eutrophic lakes with non-limiting nutrients, algal biomass will build up to a steady state limit dictated by the product of its own extinction (due to self-shading) and the mixing depth. For sediment-rich waters the product keu x Zm, known as the optical depth, the has an average value of 8. Since keu = 4.6/, ratio Zm : Zeu should be restricted by light limi-

tation to 8/4.6 = 1.7, which is considerably less than the limit suggested by Horn & Paul (1984) and indicative of light limitation in Loch Fad if Zm

= 5 m. Alternatively the effective mixing depth

of the algae might be less than that indicated by

211 the isothermal temperature structure, as supported by the concentration of biomass towards the surface (Fig. 3). During the August 1981 bloom this depth can be estimated as 1.7 Zeu = 4.4 m. Horn & Paul (1984) concluded that restricted vertical mixing in S aidenbach reservoir during isothermal conditions in spring allowed phytoplankton to overcome light limitation. Variations in effective mixing depth according to weather, as well as changes in non-algal extinction (kq) can account for the considerable interannual variation in peak algal biomass in Loch Fad. Thus surface [Chl] reached a record peak of 360/Mg 1- in August 1983 with z, = 0.3 m, i.e. ke = 6.3 m-'. This coincided with fine settled

weather and a surface temperature of 19 °C which favoured temporary stratification and caused the predicted Zm to decline to 8/6.3 = 1.3 m. Chlorophyll might have exceeded 1100 g I - according to the above empirical equation relating [Chl] to keu, but presumably kq increased greatly to a predicted 4.6 m-

in order

to restrict [Chl] to 360 pg 1 - . A 50% increase in fish production in 1983 to c. 300 tonnes might have raised kq. Similarly, unsettled conditions in July-August 1985 and 1986, with maximum surface temperatures of only 16.2 C, were associated with peak [Chl] of only 80-85 pg 1-1 and mean z, = 1.4 m, i.e. ku = 1.4 m- . This is con-

sistent with light limitation if effective mixing depth increased to 5.7 m and the potential maximum [Chl] of c. 125 pg 1- was restricted by a kq of c. 1.0. These hypothetical calculations rely on a number of assumptions including an upper optical depth limit of 8 and a constant k, which cannot be verified independently without more detailed data from Loch Fad, but they illustrate the high dependence of peak algal biomass on water turbulence and non-algal attenuation. To conclude, the absence of nutrient limitation in Loch Fad enables Microcystis to increase up to the limit imposed by the underwater light climate, as postulated by Reynolds etal. (1981). Continuous mixing and non-algal attenuation help to maintain an optically deep water column which is presumably responsible for the slow net growth

rate, while low loss rates and physiological adaptations to shade appear to maintain its dominance over other algae.

Practicalimplicationsfor cage culture The overall impact of cage fish culture on lake phytoplankton does not differ from other forms of cultural eutrophication, such as from sewage or agricultural run-off. The experience with Loch Fad shows that a well-mixed lake can continue to provide an adequate environment for intensive cage culture, even under conditions of advanced eutrophication, if consequences such as blooms of cyanobacteria can be tolerated. Limitation of algal biomass by nutrients soon evolves towards light limitation. This process is accelerated during summer because the feeding rates of farmed salmonids at temperatures above 12 C are more than double those at 4 C; this leads not only to a significant increase in nutrient loads but also to an increase in suspended wastes which absorb light, making light limitation more likely. The suitability of Loch Fad, and other lakes with a similar ratio of maximum length to mean depth, for heavily stocked systems relies on continuous turbulence to maintain both oxygenated conditions throughout the column (especially for fish and nitrifying bacteria) and a high optical depth which limits algal development. The greatest environmental threat lies in prolonged settled weather, particularly during late summer, which creates stratification. Reduced optical depth and higher surface temperatures will promote the rapid development of massive blooms of cyanobacteria, with possible breakdown of buoyancy control resulting in serious surface accumulations, mass death and catastrophic deoxygenation as described by Reynolds & Walsby (1975). Other problems may arise from decomposing blooms, such as potential toxicity to fish (Phillips et al., 1985) and earthy tainting of the flesh with geosmin. To date, models for predicting cage farm carrying capacity have been based on annual nutrient loads derived from overall food conversion ratios

212 and rely on controlling phosphorus to keep water quality within trophic state limits dictated by a variety of multiple uses. Because they take no account of seasonal changes in loading and fish growth efficiency, nor of hydrographic parameters such as column stability or possible light limitation, these models overlook the possibility of short-term deleterious phytoplankton blooms in summer in response to higher temperatures, increased nutrient flux and stratification. Future models should incorporate such short-term dynamics. Also, existing models cannot predict the ultimate limit of sustained fish production in a given lake when the principal water quality criterion is what the fish can tolerate. Often this is the most important question for smaller, privatelyowned lakes where cage culture is practised or contemplated. The answer will depend largely on physical characteristics such as flushing rate, morphometry and degree of mixing. Shallow, well-mixed lakes would appear to have an upper limit of annual fish production of 3-4 tonnes per ha of lake area, but the higher the production, the greater the risk from freak weather of algal blooms, poor water quality or disease. The environmental management of cage systems in lakes at high production levels is no different from the management of intensive fish ponds. Practical advice includes: monitoring temperature structure for incipient stratification and water colour for algal blooms; harvesting as many fish as possible early in summer or at least before any bloom collapses; taking precautions to minimise overfeeding and even withholding rations during critical conditions; and artificial mixing or aeration to maintain turbulence and D.O. during stratified periods. Summary 1. Nutrients, phytoplankton and periphyton were monitored over 14 months in Loch Fad, a shallow unstratified Lake used for intensive cage culture of rainbow trout. 2. Inorganic nitrogen, ortho-phosphate and suspended solids were significantly higher at the

cages and near the bottom, reflecting nutrient enrichment from soluble and solid cage wastes. These nutrients plus silicate declined during summer but did not reach levels which limit phytoplankton growth. 3. The phytoplankton community was dominated throughout by Microcystis aeruginosa, with surface chlorophyll a reaching 189 g 1- ' in August, but no drastic bloom collapse or deoxygenation occurred. Continuous turbulence and excess nutrients maintained a persistent sub-dominant community of 'vernal' diatoms and Pediastrum spp. Periphyton was dominated by diatoms, especially Melosira italica-subarctica.Algal species indicators and water quality confirmed the highly eutrophic state of Loch Fad. 4. Application of a phosphorus-dependent eutrophication model gave chlorophyll values consistent with observations, but light limitation by self-shading and suspended farm wastes, aided by turbulence, is believed to control algal growth rates and biomass. Changes in effective mixing depth, dependent on windinduced mixing and buoyancy of Microcystis, can account for inter-annual variation in maximum chlorophyll. 5. Implications for environmental management of intensive cage farms are discussed. Wellmixed lakes can yield 3-4 tonnes ha- yr-' but precautions are necessary to minimise the effects of excessive cyanobacterial blooms during stratified conditions.

Acknowledgements The authors are grateful to the management of Rothesay Sea Foods Ltd. for permission to undertake this study and assistance with boats, and to Mr. A. Stewart of the Institute of Aquaculture for post-1981 data on Loch Fad. We also acknowledge the unpublished information of Dr. M.C.M. Beveridge and undergraduate and postgraduate students of Stirling University who undertook projects in Loch Fad.

213 References Alabaster, J. S., 1982. Report of the EIFAC workshop on fish-farm effluents, Silkeborg, Denmark, 26-28 May, 1981. EIFAC Tech. Pap. 41, 166 pp. Bailey-Watts, A. E., 1974. The algal plankton of Loch Leven, Kinross, Scotland. Proc. R. Soc. Edinb. B74: 135-156. Bailey-Watts, A. E., 1978. A nine-year study of the phytoplankton of the eutrophic and non-stratifying Loch Leven (Kinross Scotland) J. Ecol. 66: 741-771. Bailey-Watts, A. E. & P. Duncan, 1981. The phytoplankton. In The Ecology of Scotland's Largest Lochs, Lomond, Awe Ness, Morar and Shiel, P. S. Maitland (ed.), W. Junk, The Hague: 91-118. Beveridge, M. C. M., 1984. Cage and pen fish farming. Carrying capacity models and environmental impact. FAO Fish. Tech. Pap. 225, 133 pp. Beveridge, M. C. M. & J. F. Muir, 1982. An evaluation of proposed cage fish culture in Loch Lomond, an important reservoir in Central Scotland. Can. Water Res. J. 7: 181-196. Bindloss, M. E., 1976. The light climate of Loch Leven, a shallow Scottish lake, in relation to primary production and phytoplankton Freshwat. Biol. 6: 501-518. Bostock, J. C., 1987. Carrying capacity models for freshwater cage culture: a test of their applicability in heavily exploited systems. M. Sc. thesis, University of Stirling. Dey, T., 1984. Study of plankton in relation to cage and pond culture of trout in Scotland. Ph. D. thesis University of Stirling 226 pp. Dillon, P. J. & F. H. Rigler, 1974. A test of a simple nutrient budget model predicting the phosphorus concentrations in lake water. J. Fish. Res. Bd Can. 31: 1771-1778. Drake, J. C. & S. I. Heaney, 1987. Occurrence of phosphorus and its potential remobilization in the littoral sediments of a productive English Lake. Freshwat. Biol. 17: 513-523. Enell, M. & J. Lof, 1983. Miljoeffekter av vattenbruksedimentation och narsaltbelastning from fishkasseodlingar. Vatten 39: 364-375. Ganf, G. G., 1974. Diurnal mixing and the vertical distribution of phytoplankton in a shallow equatorial lake (Lake George, Uganda). J. Ecol. 62: 611-629. George, M. G., 1962. Occurrence of a permanent algal bloom in a fish tank at Delhi with special reference to factors responsible for its production. Proc. Ind. Acad. Sci. 56: 354-362. Gerloff, G. C. & F. Skoog, 1957. Nitrogen as a limiting factor for the growth of Microcystis aeruginosa in southern Wisconsin lakes. Ecology 38: 556-561. Gibson, C. E. & R. J. Stevens, 1979. Changes in phytoplankton physiology and morphology in response to dissolved nutrients in Lough Neagh, N. Ireland. Freshwat. Biol. 9: 105-109. Gibson, C. E., Wood R. B., E. L. Dickson & D. H. Jewson, 1971. The succession of phytoplankton in Lough Neagh, 1968-70. Mitt. int. Ver. theor. angew. Limnol. 19: 146-160.

Golterman, H. L. & F. A. Kouwe, 1980. Chemical budgets and nutrient pathways. In The Functioning of Freshwater Ecosystems, E. D. Le Cren & R. H. Lowe-McConnell, (eds) Cambridge University Press, Cambridge: 85-140. Golterman, H. L., R. S. Clymo & M. A. M. Ohnstad, 1978. Methods for Physical and Chemical Analysis of Fresh Waters. Blackwell Scientific, Oxford, 213 pp. Horn, H. & L. Paul, 1984. Interactions between light situation depth of mixing and phytoplankton growth during the spring period of full circulation. Int. Rev. ges. Hydrobiol. 69: 507-519. Hutchinson, G. E., 1967. A Treatise on Limnology Vol. II: Introduction to Lake Biology and the Limnoplankton. Wiley, New York, , 115 pp. Kilambi, R. V., C. E. Hofman, A. K. Brown, J. C. Adams & W. A. Wickizer, 1976. Effects of cage culture fish production upon the biotic and abiotic environment of Crystal Lake, Arkansas Final Rep. Arkansas Game Fish Comm. U.S. Dept. Commerce. NOAA National Marine Fisheries Service, Project No. 2 - 166R Dept. of Zoology, University of Arkansans, 127 pp. Larsen D. P. & H. T. Mercier, 1976. Phosphorus retention capacity of lakes. J. Fish. Res. Bd Can. 33: 1742-1750. Lund, J. W. G., 1950. Studies on AsterionellaformosaHass. 2 Nutrient depletion and the spring maximum. J. Ecol. 38: 1-35. Lund, J. W. G. &C. S. Reynolds, 1982. The development and operation of large limnetic enclosures in Blelham Tarn, English Lake District, and their contribution to phytoplankton ecology. In Progress in Phycological Research, F. E. Round & D. J. Chapman (eds). Elsevier, Amsterdam, 1: 1-65. Margalef, R., 1958. Temporal succession and spatial heterogeneity in phytoplankton In Perspectives in Marine Biology A. A. Buzzati-Traverso, (ed). University of California Press, Berkeley: 323-349. Maulood B. K. & A. D. Boney, 1980. A seasonal and ecological study of the phytoplankton of Loch Lomond. Hydrobiologia 71: 239-359. Maulood, B. K. & A.D. Aboney, 1981. Phytoplankton ecology of the Lake of Menteith, Scotland. Hydrobiologia 79: 179-186. Merican, Z. O. & M. J. Phillips, 1985. Solid waste production from rainbow trout, Salmo gairdneri Richardson, cage culture. Aquacult. Fish. Mgmt 1: 55-69. Moss, B., 1980. Ecology of Freshwaters. Blackwell Scientific Publications, Oxford, 332 pp. Murray, J. & L. Pullar, 1910. Bathymetrical Survey of the Freshwater Lochs of Scotland, Vols. 1-6, Challenger, Edinburgh. Nygaard, G., 1949. Hydrobiological studies of some Danish ponds and lakes II. Biol. Skr. 7, 293 pp. OECD, 1982. Eutrophication ofWaters: Monitoring, Assessment and Control. Organisation for Economic Cooperation and Development, Paris, 154 pp. Palmer, C. M., 1969. A composite rating of algae tolerating organic pollution. J. Phycol. 5: 78-82.

214 Penczak, T., W. Galicka, M. Molinski, E. Kusto & M. Zalewski, 1982. The enrichment of a mesotrophic lake by carbon phosphorus and nitrogen from the cage aquaculture of rainbow trout, Salmo gairdneri. J. Appl. Ecol. 19: 371-393. Phillips, M. J., 1984. Environmental effects of fish farming. Implications for wild salmonids. Proceedings of the Institute of Fisheries Management 15th Annual Study Course, 10-13 Sept. 1984 A. Holden (ed.) Stirling University, Scotland: 138-155. Phillips, M. J., M. C. M. Beveridge & J. F. Muir, 1985. Waste output and environmental effects of rainbow trout cage culture. International Council for the Exploration of the Sea, Mariculture Committee Memorandum F: 21. Phillips, M. J., R. J. Roberts, J. A. Stewart & G. A. Codd, 1985. The toxicity of the cyanobacterium Microcystis aeruginosa to rainbow trout, Salmo gairdneriRichardson. J. Fish Dis. 8: 339-344. Ragotskie, R.A., 1978. Heat budgets of lakes. In Lake Chemistry Geology, Physics, A. Lerman, (ed.) Springer Verlag, New York: 1-19. Rawson, D. S., 1956. Algal indicators of trophic lake types. Limnol. Oceanogr. 1: 18-25. Reynolds, C. S., 1973. Growth and buoyancy of Microcystis aeruginosa Kutz. emend. Elenkin in a shallow eutrophic lake. Proc. R. Soc. B 184: 29-50. Reynolds, C. S., 1980. Phytoplankton assemblages and their periodicity in stratifying lake systems. Holarct. Ecol. 3, 141-159. Reynolds, C. S., 1984. The Ecology of Freshwater Phytoplankton. Cambridge University Press, Cambridge, 384 pp. Reynolds, C. S. & A. E. Walsby, 1975. Water Blooms. Biol. Rev. 50: 437-480. Reynolds, C. S., G. H. M. Jaworski, H. A. Cmiech & G. F. Leedale, 1981. On the annual cycle of Microcystis aeruginosa Kutz emend: Elenkin. Phil. Trans. R. Soc. B 293: 419-477. Riley, E. T. & E. E. Prepas, 1985. Comparison of phosphorus-chlorophyll relationships in mixed and stratified lakes. Can. J. Fish. Aquat. Sci. 42: 831-835. Schindler, D. W., 1978. Factors regulating phytoplankton production and standing crop in the world's fresh waters. Limnol. Oceanogr. 23: 478-486. Scott, W. E., P. J. Ashton, R. D. Walmsley & M. T. Seaman, 1980. Hartbeespoort dam: a case study of a hypertrophic, warm, monomictic impoundment. In Hypertrophic Ecosystems J. Barica & L. R. Mur, (eds) W. Junk, The Hague: 317-322.

Shapiro, J., 1973. Blue-green algae: why they become dominant. Science 179: 382-384. Sladecek, V., 1973. System of water quality from the biological point of view. Arch. Hydrobiol. Ergebn. Limnol. 7, 218 pp. Smith, V. H., 1982. The nitrogen and phosphorus dependence of algal biomass in lakes: an empirical and theoretical analysis. Limnol. Oceanogr. 27: 1101-1112. Strickland, J. D. H. & T. R. Parsons, 1972. A practical handbook of sea water analysis Bull. Fish. Res. Bd. Can. 167, 310 pp. Tailing, J. F., 1971. The underwater light climate as a controlling factor in the production ecology of freshwater phytoplankton. Mitt. int. Ver. theor. angew. Limnol. 19: 214-243. Tailing, J. F. & D. Driver, 1963. Some problems in the estimation of chlorophyll a in phytoplankton. Proc. Conf. on Primary Productivity Measurement, Marine and Freshwater, Hawaii 1961 U.S. Atomic Energy Comm TID-7633: 142-146. Tett, P., S. I. Heaney & M. R. Droop, 1985. The Redfield ratio and phytoplankton growth. J. mar. biol. Ass. U.K. 65: 487-504. Thunmark, S., 1945. Zur Soziologie des Susswasser-planktons. Eine methodologish-okologische Studie. Folia Limnol. Scand. 3, 66 pp. Vollenweider, R. A., 1976. Advances in defining critical loading levels for phosphorus in lake eutrophication. Mem. Ist. Ital. Idrobiol. 33: 53-83. Westlake, D. F., 1980. Primary production. In The Functioning of Freshwater Ecosystems E. D. Le Cren & R. H. Lowe-McConnell (eds), Cambridge University Press, Cambridge: 141-246. Wofsy, S. C., 1983. A simple model to predict extinction coefficients and phytoplankton biomass in eutrophic waters. Limnol. Oceanogr. 28: 1144-1155. Youngman, R. E., D. Johnson & M. R. Farley, 1976. Factors influencing phytoplankton growth and succession in Farmoor Reservoir. Freshwat. Biol. 6: 253-263. Zevenboom, W. & L. R. Mur, 1980. N2 -fixing cyanobacteria: why they do not become dominant in Dutch hypertrophic lakes. In Hypertrophic Ecosystems J. Barica & L. R. Muir (eds) Dr W. Junk, The Hague: 123-130. Zevenboom, W., A. B. De Vaate & L. R. Mur, 1982. Assessment of factors limiting growth rate of Oscillatoriaagardhii in hypertrophic Lake Wolderwijd, 1978, by use of physiological indicators. Limnol. Oceanogr. 27: 39-52.