Cybernetic mechanisms in lake plankton systems: how to control undersirable algae

(Reprinted from Nature, Vol. 273, No. 5659. pp. 228-230, May 18,1978) @
Macmillan Journals Ltd., 1978
Cybernetic mechanisms in lake plankton
systems: how to control undersirable algae
.,
TOXIC algal blooms represent a serious nuisance in lakes and
ponds. Not only do they form floating scums and impart odours and
tastes to the water, but they are also associated with the occurrence
of dermatitis in our own species and fatal poisoning in livestock. To
reduce algal levels, man resorts to various practices that range from
chemical 'control' to artificial mixing of the lake'. Most of these
have deleterious, if not immediately toxic, effects on the other
compartments of the system as well. In this report safer alternatives
are presented.
As toxicity is primarily a defence mechanism evolved against
herbivores, the problem of controlIing toxic algae can be transposed
to the simpler one of control1ing inedible algae. A simple loop
model of a lake plankton community containing an inedible
compartment is then sufficient to reveal how the system wil\
respond to perturbations introduced at different trophic levels. This,
in turn, suggests various ways to achieve biological control of the
inedible (undesirable) species.
The model considered (Fig. la) is a simplified planktonic food
web reduced to six variables: nutrients (N), inedible algae (AI),
edible algae (A2), herbivores (H), carnivores (C) and planktivorous
predators (P). The distinction between edible and inedible algae is
based on the overwhelming evidence2-s that phytoplankton celIs are
selected on the basis of size, taste and morphology.
Here the term 'inedible' is used realistically: we recognise the
likelihood of some members of group Al being occasionalIy
ingested by grazers but consider these instances truly exceptional in
a community in which 'edible' algae are also present. Accordingly,
throughout this report Al refers to an heterogeneous group of species
including alI blue-green algae, irrespective of size, plus all other
phytoplankton species larger than 50 /.Lm, but excluding the
common diatom Asterionella formosa. As the latter is easily broken
down and ingested"'s it was integrated in A2. Blue-greens excepted,
alI algae smalIer than 50 /.Lm were considered edible. The few taxa
possessing an ambiguous trophic status were excluded from either
group. This applied to the gelatinous greens Sphaerocystis
schroeteri and Elakatothrix gelatinosa that may be enhanced by
grazers', the colonial diatoms Fragilaria and Tabellaria that can be
ingested by cope pods when fragmented (K. Porte.r, personal
communication) and the colonial chrysomonad Dinobryon that may
be a major food source for daphnids when it is present in high
density'..
The dynamics of this planktonic model can be readily explored
by loop analysis. This technique recently developed by Levins"'" is
particularly applicable to ecological systems in which most
interactions among the biological variables can only be specified
qualitatively. Loop analysis is based on the equivalence between
differential equations near equilibrium and matrices and their loop
diagrams. Thus, in our particular system the elements aij of the
matrix and the loop diagram (Fig. 1) represent the effect of variable
j on the growth of variable i when the equation dXi/dt= f; (Xl, . . .
Xj, . . . , Xn) is solved at equilibrium. As all stability require
ments are satisfied for our model, one can assess mathematicalIy the
consequences on the entire system of changes in each variable. The
reader interested in the details of the technique is referred to the
formal treatment of loop analysis theory".
As shown in Table I, the predictions derived from our model are
often far from obvious. For example, increasing predator density
(case 6) wil\ leave the herbivore level
_oNA)
-ONAz
0
0
°A,N
0
0
0
0
0
°AzN
0
0
_0
AzH
0
0
0
0
°H
Az
0
0
0
0
°CH
0
0
0
°pH
-ONN
0
.
_oH
_oHP
C
0 _ocP
0
ope
Fig. 1 a, Loop diagram of the six-variable system. Positive effect of
variable i on variable j is indicated by an arrow going from i to j.
Negative effect is indicated by a circle. Nutrient resources are not selfreproducing and are thus self-damped. b, Matrix of the system, taking
all the au as positive numbers and representing
the directions of their effects by the sign in front.
unaltered yet increase the inedible algae. Also the addition of
herbivores to the system (case 4) has no effect on the standing stock
of alI animal consumers. Thus, one cannot infer the response of one
variable to changes in another from the simple knowledge of the
direct link between the
Table 1 Predicted response of plankton systems to perturbations in
each variable
PerturbaCase
tion*
Effect on the level of
N
+N 0
Predictions
A, A.
H
C P
+
0
0 0
0
supported by
Fertilisation
experievidenc
ments and
e
from 'cultural' eutrophication (refs 16-18)
- 0
2
3
+A,
+A.
0
0
0
0
4
+H
0
5
+c
0
+
- 0
- + 0
6
+p
0
+
-
+
0
+
-
+ (Currently being tes ted
0 O
in long-range experi-
0 + J ments in situ)
-
o Lake manipulations
(ref. 21)
two. Clearly the structure of the entire network needs to be taken
into account, as its effects may regularly overpower simple-link
effects. This is where the power of loop analysis lies and why it
often yields unexpected results"''', especially when there is no direct
link between the two variables considered. For instance we can
predict in this manner that adding inedible algae to our system (case
2) will increase the carnivore level but will reduce the standing
stock of the other animal consumers.
As the evolution of inedibility in algae leads to the development
of toxicity that is considered undesirable by man, we can treat AI as
the variable to be reduced. According to our model, this can be
achieved in four ways (see column AI in Table 1): that is, either by
reducing the rate of nutrient supply, by increasing carnivore levels
or by reducing predator or herbivore stocks.
The first method is particularly applicable when there is a point
discharge of nutrients into the water body: this was demonstrated
spectacularly by the diversion of sewage from Lake Washington".
Unfortunately nutrient inputs are not controllable in most cases.
Biological manipulation of lake communities may then well
represent the only means for selective algal control. Although this
approach has been convincingly advocated by Shapiro et aI." it has
remained surprisingly neglected. The results of our work suggest
simple ways to achieve biological control of undesirable algae.
Among these, the introduction of herbivore-specific disease or the
addition of primary carnivores would seem particularly worth
trying.
It is clear that the table of predictions generated by our model
would take years to be tested in its entirety. Its general validity,
however, is supported by long-term field experiments and indirect
evidence from previously published work. A particularly well
documented case, because of the great concern for the consequences
of artificial eutrophication, is that of nutrient addition. We predict
here (case 1), in agreement with previous models of nutrient enrichment".'..17, that the increment in nutrients is picked up entirely by
the inedible algae. Evidence from 'cultural' eutrophication".l8 and
fertilisation experiments" supports our contention. The common
response of lakes to nutrient enrichment is an immediate increase in
the density of the algae considered inedible and a change in the
qualitative composition, but not in the standing stock, of the
consumer levels. Further, as N itself remains unaltered, it can be
understood why nutrient levels are poor indicators of eutrophication".
Conversely, we predict that a decrease in the rate of nutrient
input (case 7) will reduce only the level of inedible algae without
altering those of N, A2, H, P or C. This would indicate, as already
pointed out by Phillips'. for marine systems, that inedible algae
serve as a buffer against variations in the nutrient supply. This
hypothesis was tested by following from June to November 1976
the effects on lake plankton of cutting the nutrient input from the
sediments. The communities studied were enclosed in a series of
polyethylene tubes, each 1 m in diameter and 10 m deep, suspended
in Heney Lake, a lake with low levels of nutrients 60 miles north of
Ottawa. Community structure was monitored by weekly sampling at
0, 3, 6 and 9 m, followed by microscopic enumeration using the
sedimentation technique of Utermohl21.
Our predictions (case 7) were all confirmed. Comparing tubes
open at both ends and sunk in the sediments with tubes sealed at the
bottom revealed that their population stocks were similar at each
trophic level throughout the 5-month study. (Wilcoxon matchedpairs signed-ranks test) except for the inedible algae that were
significantly reduced (P<0..o25) in the latter case. In light of this
'enclosure effect' and of the increasing popularity of this technique
among aquatic ecologists, the use of tubes in
contact with the sediments is recommended as the closest
simulation of lake conditions.
Support for our predictions concerning the response of planktonic
communities to changes in predation levels (cases 6 and 9) is found
in the classic investigations by Hrbacek".23. He documented a shift
from nannoplankton to netplankton and a change in the species
composition but not in the standing crop of the herbivores after increases in stocks of planktivorous fish. Then, by eliminating fish
stocks altogether he was further able to induce a shift from
netplankton to nannoplankton. More recently Smyly" detected no
noticeable change in herbivore levels following the reduction in fish
predation in experimental enclosures.
The response of plankton systems to reduction in both herbivore
and carnivore levels in polyethylene tubes at Heney Lake was also
investigated. All components of the planktonic community were
represented in our experimental enclosures, including, in addition to
99 phytoplankton species, herbivores such as Bosmina, Chydorus,
Daphnia and Diaptomus, the primary carnivores Epischura and
Mesocyc!ops, and the predators Chaoborus and Leptodora.
Throughout the 5-month study Hand C levels were reduced each
week by passing a net of 0.75 m in diameter and 135 JIm mesh
aperture through the tubes.
Examination of tl1e loop diagram (Fig. 1) shows that the net
outcome of decreasing jointly Hand C levels is to reduce the
carnivore population alone, because of the negative CH loop. The
consequent predictions (case 8) were
generally verified by our experiments. Using unperturbed
enclosures for comparison, we found that the manipulations resulted
in a significant increase (P<O.025) in the biomass of inedible algae
and in no significant changes in herbivore and carnivore levels
(Wilcoxon test). The only discrepancy was that no significant
decrease in edible algae was recorded.
We thank R. Trucco and D. Wall for field assistapce, and R.
Levins, J. R. Strickler and P. Yodzis for discussions. This work was
supported by the National Research Council of Canada in the form
of a grant to F. B. and a predoctoral scholarship to E. M.
FREDERIC BRIAND
EDWARD MCCAULEY
Department of Biology,
University of Ottawa, Ottawa,
Canada KIN 6N5
Received 14 February; accepted 20 March 1978.
1. Dunst, R. C. e/ al. Survey of Lake Rehabilitation Techniques and Experiences
(Tech. Bull. No. 75, Dept Nat. Resources. Madison. Wisconsin, 1974).
2. Burns. C. Limnol. Oceanogr. 13, 675-678 (1968); Int. Rev. Ges. Hydrobiol. 53,
83-100 (1968).
3. Gliwicz. M. Bull. A cad. Polon. Sci.. CI. II, 17, 159-165 (1969); Verh.Int. Verein.
Limnol. 19, 1490-1497 (1975).
4. Arnold. D. E. Limnol. Oceanogr. 16,906-920 (1971).
5. Weers, Eo T. & Zaret. T. M. Verh. Int. Verein. Limnol. 19, 1480-1483 (1975). 6.
Porter, K. G. Nature 244,179-180 (1973).
7. Porter, K. G. Am. Sci. 65, 159-170 (1977).
8. Nadin-Hurley. C. M. & Duncan. A. Freshwater BioI. 6, 109-123 (1976).
9. Porter. K. G. Verh. Int. Verein. Limnol. 19, 2840-2850 (1975); Science 192,
1332-1334 (1976).
10. Tappa. D. W. Ecol. Monogr. 35, 395-423 (1965).
11. Levins, R. Ann. N. Y. Acad. Sci. 231, 123-138 (1974).
12. Levins, R. in Ecology and E,'olution of Communities (eds Cody, M. &
Diamond, J.) 16-50 (Harvard University Press, Cambridge, Massachusetts, 1975).
13. Lane, P. & Levins, R. Limnol. Oceanogr. 22, 454-471 (1977).
14. Edmonson, W. T. in Nutrients and Eutrophication (ed. Likens, G. E.) 172-193
(Am. Soc. Limnol. Oceanogr. Special Symposium, 1972).
15. Shapiro. J., Lamarra. V. & Lynch, M. in Water Quality Management Through
Biological Control (eds Brezonic, P. L. & Fox, J. L.) 85-96 (Report No.
ENV.-07-75-1, University of Florida, Gainesville, 1975).
16. Phillips. O. M. Arch. Hydrobiol. 73, 310-333 (1974). 17. Smith, F. E. in
Eutrophication: Causes, CQnsequences,'''Correctil'es 631-645
(National Academy of Sciences. Washington, DC, 1969).
18. Larkin, P. A. & Northcote, T. G. in Eutrophication: Causes, Consequences,
Corredives 256-273 (National Academy of Sciences, Washington, DC, 1969).
19. Schindler, D. W., Armstrong, J., Holmgren, S. K. & Brunskill, G.J. J. Fish.
Res. Bd. Can. 28,1763-1782 (1971).
20. Hooper, F. F. in Eutrophication: Causes, Contequences, Correctives 225-235
(National Academy of Sciences, Washington. DC, 1969).
21. Utermohl, H. Arch. Hydrobiol. Plankt. 22, 643-645 (1931).
22. Hrbacek, J. Rozpravy Ceskosl. Akad. Ved, Rada Matem. 72, 1-114 (1962). 23.
Hrbacek. J. & Novatna-Dvorakova Rozpravy Ceskosl. Akad. Ved, Rada Matem.
75, 1-64 (1965).
24. Smyly, W. J. P. Freshwater Bioi. 6, 241-251 (1976).
Printed in Great Britain by Henry ling ltd., at the Dorset Press, Dorchester, Dorset
`