The development from kinetic coefficients of a predictive model for the growth of Eichhornia crassipes in the field . IV . Application of the model to the Vernon Hooper Dam — a eutrophied South African impoundment

A m odel developed for Eichhornia crassipes (M art.) Solms was used to identify the lim iting nutrient in the V ernon H ooper D am and to predict population sizes, yields, growth ra tes and frequencies and am ounts o f harvest u n d e r varying conditions o f nutrient loading and clim ate. Predicted d a ta w ere used to evaluate the effectiveness o f harvesting m easures curren tly being em ployed fo r contro lling b o th nutrient inputs and the population size in this im poundm ent. Predictions o f th e population size, before harvesting was initiated, generally com pared favourably w ith th a t estim ated visually. Predictions o f th e quan tities o f P th a t could b e rem oved daily by a 20 ha population indicate that such a population in the im poundm ent could reduce P concentrations in th e epilim nion during sum ­ m er stratification to levels limiting for algae. T his may explain th e observed reduction in algal concentrations since th e in troduction o f harvesting. E stim ates o f th e am ounts and frequencies o f harvest required to contain the p re ­ dicted po ten tial yields o f a 20 ha population indicate that 100 m etric tonnes o f fresh w ater hyacinths harvested daily from th e im poundm ent, under present conditions o f reduced nutrient loading, are adequate during w in ter, but not during sum m er.

Harvesting Eichhornia crassipes (M art.)Solms (water hyacinth) growing in eutrophied aquatic sys tems may constitute an effective means of removing nutrients and controlling excessive growth of plants (Boyd, 1970;Yount & Crossman, 1970).However, to achieve maximum nutrient removal efficiency by E. crassipes in a nutrient removal scheme, it is necessary to establish the size of the population re quired to m aintain desirable nutrient concentrations in the w ater, under varying conditions o f nutrient loading and climate, and the amounts and frequen cies o f harvest required to control the population size.Musil & Breen (1985a, 1985b, 1985c) developed, tested and refined a model, incorporating culture and field derived kinetic coefficients, for predicting growth o f water hyacinth in eutrophied aquatic sys tems. in this paper we demonstrate how this model can be used to identify the limiting nutrient and pre dict population sizes, yields, growth rates and fre quencies and am ounts o f harvest, under varying con ditions of nutrient loading and climate, in the Ver non H ooper Dam.Predicted data are used to evalu ate the effectiveness o f harvesting m easures currently being'em ployed for controlling both nutri ent inputs and the population size in this im pound ment.

ST U D Y SIT E
The Vernon H ooper Dam at Ntshongweni in Na tal is located in a deep, rocky gorge below the origi nal confluence of the Mlazi, Sterkspruit and Weke weke (Ugede) Rivers, A description o f the catch ment and the physicochemical and hydrological characteristics o f the impoundment are presented in Archibald & Warwick (1980).For many years this impoundment has served as an important water sup ply for the city o f Durban, but its continuation as a water resource is jeopardized by poor water quality.Effluent discharges from domestic sewage treatm ent works and industrial complexes in the Mlazi and Sterkspruit River catchments have resulted in a con siderably enriched impoundment with little opportu nity for effluent diversion.A history o f the develop ment in the catchment area, which includes a rccord of changes in water quality and subsequent action taken by the Durban City Engineers D epartm ent, is given by Howes (1976).
Before 1979, water hyacinth was observed occa sionally in the impoundment.T he populations, how ever, were generally small and confined to the major inlets with aggregations of plants occasionally being wind-blown across the reservoir (C.G .M. Archi bald, pers.com m .t).Water quality at this time was described by Howes (1983) as consisting o f three types, each requiring different chemical treatments.These were: (i) 'Normal', (ii) 'Algal laden' and (iii) 'Manganese' water.The latter results from anaer obic conditions during summer when severe stratifi cation occurs in the impoundment (Archibald & Warwick, 1980).
During 1979, water hyacinth spread extensively and covered 70 to 80% of the reservoir by March, 1980(Everitt, 1980).Light wind action compressed this population to ca 50% of the water surface area.A marked improvement in the quality of the 'Algal laden' water, noted by the Chemical Branch of the Durban City Engineers D epartm ent, was attributed to P removal by the water hyacinth population (A.M. Howes, pers.comm.*).In view o f this, the **Durban City Engineers Departm ent decided against eradicating the water hyacinth population in the impoundment by chemical control measures.Instead, ca 20 ha of water hyacinths were retained behind a floating boom in the Mlazi leg of the reser voir (Fig. 1) at the end o f 1981 (Howes, 1983), and the populations harvested regularly with the aid of a mobile crane.U p to 100 metric tonnes o f fresh water hyacinths are harvested daily from the impoundment (P. A. Larkan, pers.com m .1)which has resulted in reduced water treatm ent costs (Howes, 1983).

Physicochemical measurements Flow
M ean daily inflow rates (m3 s-1) for the impound ment were derived from flow-measuring weirs lo cated on the Mlazi, Sterkspruit and Wckeweke Riv ers, whereas outflow rates were derived from the spillway, scour and dam outlet (draw).These rates, m easured fortnightly, were supplied by the National Institute for W ater Research (NIW R), Council for Scientific and Industrial Research (CSIR), D urban, and converted to monthly flows using the following formula; Monthly flow (M€) = mean daily flow rate per month (m3 s-1) x num ber o f days in m onth x (*,0864 x 103.
Total monthly inflow was calculated as the sum of monthly flows o f the Mlazi, Sterkspruit and Wekeweke Rivers, whereas total monthly outflow was cal culated as the sum o f monthly flows of the spillway, scour and dam outlet.

Chemical analysis o f wafer
Chemical analyses of water samples, collected fortnightly from the Mlazi, Sterkspruit and Wekeweke Rivers and from the outflowing w ater, were carried out by the NIW R, CSIR, Durban.These analyses were autom ated and derived from pub lished methods (Environmental Protection Agency, 1974;American Public Health Ass.: Standard Meth ods, 1975).
T he following N and P fractions were analysed in the water samples: a) in filtered samples, nitrate-nitrogen (N 0 3-N ) by colorimetry after reduction to nitrite and soluble reactive phosphorus (SRP) (Twinch & Breen, 1980) by colorimetry using the molybdenum blue method.b) in unfiltered samples, Kjeldahl nitrogen as am monium (NH4-N ) after digestion of the samples with conc.H2S 0 4 in the presence of a mercury catalyst and total phosphorus (total P) as SRP after digestion of the samples with H2S 0 4 and persulphate.Total nitrogen (total N) was calculated as the sum of Kjel dahl nitrogen and nitrate plus nitrite (NH4-N + N 0 3-N + NOr N).

Nutrient loading and release
Monthly flows of the Mlazi, Sterkspruit and Wekeweke Rivers were multiplied by the average monthly total N and total P concentrations in the water o f each river and summed to give monthly total N and total P inflow loads (inputs) for the im poundment.Total monthly outflows via the spill way, scour and dam outlet were multiplied by the average monthly total N and total P concentrations in the outflowing water to give monthly total N and total P outflow loads (release).

Environment
Mean daily air tem perature data for the Vernon Hooper Dam, derived from monthly averages over the period 1932to 1946(Climate of South Africa, 1954), were supplied by the W eather Bureau, De partment of Transport.No measurements of radiant flux density or relative humidity were available for the impoundment.

Chemical and biological transformations
Estimates of the proportions o f N and P inflow loads retained or lost by chemical and biological transformations in the impoundment, i.e. through sediment adsorption and denitrification, were based on monthly inflow and outflow data and N and P bal ances for the impoundment for the period January to December, 1976 (Table 1).W ater hyacinth was either absent, or present as small populations in the reservoir during this period (C .G .M .Archibald, pers. comm,).The differences between total N and total P inflow and outflow loads, expressed as per centages of inflow loads, estimated the percentages of N and P inflow loads retained o r lost by chemical and biological transformations in the impoundment monthly and, although potentially available, were assumed not to be readily available to water hya cinths for growth.A net export of N from the system was evident during September, 1976.However, this could not be attributed to a net loss of water from the reservoir during this month since the recorded inflow was larger than the recorded outflow (Table 1).

Chemical and biological transformations and water hyacinth uptake
Monthly inflow and outflow data and N and P bal ances for the impoundment for the period August, 1979to July, 1980, when an extensive cover o f water hyacinth was present (Evcritt, 1980), are given in Table 2.The differences between total N and total P inflow and outflow loads, expressed as percentages of inflow loads, estimated the percentages o f N and P inflow loads removed by water hyacinth plus those retained o r lost by chemical and biological trans formations in the impoundment monthly.Net ex ports of P from the system were evident during Janu ary and March, 1980, Again, these could not be attributed to net losses of water from the reservoir during these two months since recorded inflows were larger than recorded outflows (Table 2),

Water hyacinth uptake
Subtracting the percentages of N and P inflow loads estimated to have been retained, lost or re moved monthly in Table I from those in Table 2 and multiplying these by the recorded monthly N and P inflow loads for the period August, 1979to July, 1980, gave estimates of the quantities o f N and P re moved monthly by the water hyacinth population in the reservoir, The results (Table 3) suggest that dur ing the periods March to April, 1980and November, 1979to June, 1980, N and P respectively were ex ported from the system.The estimated percentages of N and P inflow loads removed by water hyacinth plus those retained or lost by chemical and biological transformations in the reservoir during these periods being considerably lower than those estimated to have been retained o r lost by chemical and biological transformations only.These apparent net exports of N and P from the system, however, could not be attributed to any large net losses of water from the reservoir during these periods since recorded in flows, with the exception o f May and June, 1980, were generally considerably higher than recorded outflows (Table 2), The apparent net exports of P from the system may be explained by a release of sediment-bound P 0 4-P during anoxic conditions produced in the im poundment by the extensive water hyacinth cover (Everitt, 1980) and accentuated by summer stratifi cation in the reservoir (Archibald & Warwick, 1980), In fact, Everitt (1980) reported a dissolved oxygen concentration of less than 0,5 mg €-'(ppm ) in the water at the dam wall during March, 1980 which indicated anaerobiosis below the thermocline.This suggestion is supported by the large (up to 1 100%) increase in the total P concentration in the water during the period November, 1979 to June, 1980, even though P inflow loads during this period were 25 to 50% lower than those during the prece ding period, i.e, August to October, 1979 (Table 3).An increased P release from sediments under anoxic conditions is well documented (M ortim er, 1941) and Vollenweider (1972) has shown that oxygen deple tion is accompanied by a breakdown in the ability of sediments to adsorb P 0 4-P so that sediments act as a source rather than a sink for P.
The apparent net exports of N from the system could not be readily explained, since total N concen trations in the water relative to N inflow loads did not show any significant differences during March and April, 1980 compared with other months (Table 3).These apparent net exports of N from the system, however, could have been due to increased rates of N O ,-N loss via denitrification (Keeny, 1973;Chon & Knowles, 1979) The possibility of P inputs from internal sources (sediments) and possible increases in rates o f N loss via denitrification, during those months when net ex ports of N and P from the system were not apparent, introduce constraints to modelling the growth of water hyacinth in the impoundment.This is because it is not feasible to accurately estimate the propor tions of N and P inflow loads removed by the water hyacinth population.However, during those months where net exports of N and P from the system were not apparent, the model was applied using estimates made o f the proportions of N and P inflow loads re moved by the water hyacinth population.

Limiting nutrient
The nutrient (N o r P) 'limiting water hyacinth growth rate in the impoundment was predicted from total N and total P concentrations in the water (monthly averages) using the half saturation (Ks) concentrations of 976 fig N €-■ and 94,1 fig P derived for E. crassipes in culture (Musil & Breen, 1985a), in the Monod model.!t can be assumed (Musil & Breen, 1985b) that the specific growth rate of E. crassipes is limited not in a multiplicative or additive manner, but in a threshold mode by the single nutrient (N o r P) in shorter supply.Estimates were based on total N and total P concentrations, rather than on soluble N and P fractions, in the water since specific growth rates of E. crassipes in the field are more accurately predicted from these concentrations (Musil & Breen, 1985b).For ex ample, the average total N and total P concentra tions in the water of the impoundment during Au gust, 1979 (Table 4)  The results show that during August, 1979, E. crassipes would achieve a lower percentage of the Umax at the average total P than at the average total N concentrations in the water, indicating that P was the limiting nutrient.
Estimates of the limiting nutrient in the impound ment for the period August, 1979to July, 1980, are given in Table 4.The results show that during July, 1980 and for the period August to O ctober, 1979, P was limiting for water hyacinth, whereas for the period November, 1979 to June, 1980, N was limit ing, The change from P to N limitation after Oc tober, 1979 was reflected in the considerable in crease in the total P concentration in the water (Table 4 ).

Specific growth rate
Specific growth rates of water hyacinth occurring in loosely crowded populations in the impoundment were predicted from mean daily air temperatures (monthly averages) and limiting total N or total P concentrations in the water (monthly averages) us ing the following mathematical expression (Musil & Breen, 1985b): U = 5,2631 x 10se *6540/T x Ks + S where U = specific growth rate g fresh mass g~' d_1; T = absolute mean daily air temperature °K; S = limiting nutrient concentration pg Ks = half saturation concentration jug €-*.
This expression, essentially a combination of the Arrhenius and Monod equations, incorporates the Ks concentrations of 976 fig N £_l and 94,1 fig P t-K Specific growth rates of water hyacinth occur ring in densely crowded populations in the impound ment were estimated using a correction factor of 0,2236.This correction factor (Table 5) was derived from a mean value of ratios calculated between specific growth rates (estimated maximum specific growth rates) measured for water hyacinths growing in loosely and densely crowded field populations at a nearby sewage maturation pond (Musil, 1982).For example, during August, 1979 the mean daily air temperature at the impoundment and the limiting total P concentration in the water were 16,5°C and 45 jug P €_1 respectively (Table 4).Specific growth rates (U ) of water hyacinth occurring in (i) loosely and (ii) densely crowded populations in the im poundment during this month were predicted as fol lows: * N utricni present a t concentrations below that re q u ited for m ax im um plant grow th and hence restricting the grow th rate.
The Yc expresses the relationship between mass of plant material produced and mass of limiting nutri ent absorbed (Musil & Breen, 1985a).For example, during August, 1979 the proportion of the limiting P inflow load estimated to have been removed by the water hyacinth population was 265 kg F (Table 3) or 265/31 kg P d_I.The potential yield (Xpy) of water hyacinth during this month was predicted as follows: Xpy = x 17 248 31 = 147,4 metric tonnes of fresh water hyacinths d -1 Predictions of the potential yields of water hya cinth in the impoundment for the period August, 1979to July, 1980, are given in Table 4.These ranged from 86,7 to 886,7 metric tonnes o f fresh water hyacinths d-1.

Population size
Population sizes (loosely and densely crowded) re quired to produce the predicted potential yields of water hyacinth in the impoundment were estimated using the following form o f the general growth equa tion (Malek & Fencl, 1966, Radford, 1967): Xo + Xpy = Xoeut where Xo = population size metric tonnes; Xpy = potential yield metric tonnes d -1; u = specific growth rate g fresh mass g~! d-1; t = time interval between initial biomass (Xo) and final biomass (Xo + Xpy) days.
For example, population sizes (loosely and densely crowded) required to produce the predicted potential yield of water hyacinth during August, 1979 of 147,4 metric tonnes of fresh water hyacinths d -' at a predicted specific growth rate of (i) 0,0267 and (ii) 0,0060 g fresh mass g-1 d -' for loosely and densely crowded populations respectively, were esti mated as follows:  4.These ranged from 1,0 x 103 to 8,2 x 103 and from 4,6 x 103 to 38,3 x 103 metric tonnes of fresh water hya cinths for loosely and densely crowded populations respectively.

Population area
Assuming stand densities (dry mass basis) of 2,21 and 21,3 metric tonnes ha 1 for loosely and densely crowded populations respectively (Boyd & Scarsbrook, 1975) and a mean water content of water hya cinth of 94,75% (Penfound & Earle, 1948;West lake, 1963;Bock, 1969), it is possible to calculate the areas occupied by the loosely and densely crowded populations and express these as percentages of the surface area of the impoundment, which at full sup ply level is 84 ha (Everitt, 1980).For example, the estimated population sizes in the impoundment dur ing August, 1979 were 5,4 x 103 and 24,5 x 103 met ric tonnes of fresh water hyacinths for loosely and densely crowded populations respectively.The areas occupied by the populations in (i) loosely and (ii) densely crowded situations were, therefore, calcu lated as follows:  4.These ranged from 23,7 to 194,8 ha (28,2 to 231,9% o f the full surface area o f the impoundment) and from 11,3 to 94,4 ha (13,4 to 112,4% of the full surface area o f the impoundment) for loosely and densely crowded populations respectively.Everitt (Í980) visually estimated a 50% coverage o f the impoundment by water hyacinth during M arch, 1980, when the population was compressed by light wind action.However, the reservoir was only ca 65% full at the time with the water surface area covering ca 75% of that at full supply level (A.M. Howes, pers.comm.).Consequently, the water hyacinth population in the reservoir under crowded conditions in actual fact covered only ca 31,5 ha or 37,5% of the full surface area of the impoundment.This visual estimate of cover compares favourably with the areas estimated to have been occupied by densely crowded populations in the reservoir, when expressed as percentages o f the full surface area of the impoundment (Table 4), during May, 1980 (36,9% ), December, 1979 (39,0% ), O ctober, 1979 (41,7% ) and even during September, 1979 (54,3%) and June, 1980 (53,3% ).It, however, does not com pare favourably with those during Novem ber, 1979 (13,4% ), January, 1980 (112,4%) and February, 1980 (17,0% ).Inaccurate estimates of the propor tions o f N inflow loads removed by the w ater hya cinth population during these months may partly ex plain this.

P redictions based on a 20 h a population
Presently, ca 20 ha of water hyacinths are retained in the impoundment behind a floating boom (Howes, 1983).The population is maintained in moderately crowded situations with up to 100 metric tonnes of fresh water hyacinths being harvested daily (P. A. Larkan, pers.com m .).The potential yields, am ounts and frequencies o f harvest required and quantities of N and P, i.e. assuming no luxury up take o f these nutrients by the water hyacinths, that could be removed by a 20 ha moderately crowded population in the impoundment were predicted as follows.Estimates were based on the period August, 1979to July, 1980.since chemical and hydrological data for the impoundment after this period was in complete.It was assumed that (i) the 20 ha moder ately crowded population had an average stand den sity of 11,7 metric tonnes ha-' (dry mass basis), i.e. a mean value of 2,21 and 21,3 metric tonnes ha-i (dry mass basis) for loosely and densely crowded popula tions respectively, (ii) the water hyacinths had a mean water content o f 94,75% and (iii) specific growth rates were mean values o f those predicted tor loosely and densely crowded populations in the im poundment in Table 4.

Harvesting interval
Assuming that 100 metric tonnes of fresh water hyacinths are harvested daily from the impound ment, the time required for a 20 ha moderately crowded population to produce an additional 100 metric tonnes of fresh water hyacinths was estimated using the general growth equation (Malek & Fenci, 1966;Radford, 1967) and is referred to as the har vesting interval, viz: t = €n X t-fri Xo U where: Xt = final biomass (X o + 100) metric tonnes; Xo = initial biomass of 20 ha population (moderately crowded) metric tonnes; U = specific growth rate (moderately crowded population) g fresh mass g-1 d_1; t = harvesting interval, i.e. time interval between Xo and Xt days; in = loge (natu r a l logarithm).
t For example, it was predicted that a 20 ha moder ately crowded population, or 4 457,1 metric tonnes of fresh water hyacinths, would produce 73,2 metric tonnes of fresh plant m aterial daily during August, 1979 at an estimated specific growth rate of 0,0163 g fresh mass g-1 d-i (Table 6).The time (t) required for this population to produce an additional 100 metric tonnes o f fresh water hyacinths was: ( [ = In (4 457,1 + 100) -In (4 457,1) 0,0163 = 1,4 days The harvesting interval in this example was 1,4 days.
After this period, one day's removal would need to be initiated and this would have to be repeated after a further 1,4 days grow'th.Harvesting intervals estimated for a 20 ha moder ately crowded population in the impoundment for the period August, 1979 to July, 1980 are given in Table 6.The results indicate that during winter (May to August) 100 metric tonnes of fresh water hya cinths harvested daily would more or less be ade quate to account for the predicted potential yields of the population.The estimated harvesting intervals ranging from ca 0,8 to 1,4 days.During summer (Septem ber to April), however, this quantity of fresh water hyacinths harvested daily would gener ally be insufficient.Approximately two to three times as much fresh plant material (ca 161,5 to 287,9 metric tonnes) would generally need to be harvested daily from the impoundment during summer to con tain the predicted potential yields of the population.
With respect to the amounts of harvest, it should be pointed out that the 100 metric tonnes o f fresh water hyacinths apparently harvested daily from the impoundment (P. A. Larkan, pers, comm.) are high when com pared with those reported in the literature for various mechanical water hyacinth removal oper ations.Van Dyke (1971), for example, reported that a stationary, land-based mechanical harvester proto type (Sarasota Weed and Feed Incorporation) was only capable of removing an average of ca 5,9 metric tonnes of fresh water hyacinths per hour o f operat ing tim e, i.e. ca 47 metric tonnes of fresh water hya cinths per day assuming an 8 hour working day, when time required for general maintenance and re pairs and that lost due to unfavourable weather was taken into consideration.Similar results were ob tained by Phillippy & Perryman (1972b) using an Aquam arine S-650 Shore Conveyor (Linder Indus trial Machine Company, Florida) where an average of ca 52 metric tonnes of fresh water hyacinths were removed per 8 hour working day of operating time.Somewhat higher values, however, have been ob tained by Touzcau (1972), using an Aquamarine H-650 Harvester combined with an S-650 Shore Conveyor system (Linder Machine Company, Flor ida), where an average of ca 74 metric tonnes of fresh water hyacinths were removed per 8 hour working day of operating time and by Phillippy & Perryman (1972a), using a modified, stationary, land-based mechanical harvester prototype (Sara sota W eed and Feed Incorporation), where an aver age o f ca 96 metric tonnes of fresh water hyacinths were removed per 8 hour working day of operating time.The latter was the highest value that could be traced in the literature.

Nitrogen and phosphorus removal
The quantities of N and P that could be removed daily by a 20 ha moderately crowded population in the impoundment were estimated from the predicted potential yields of the population (Table 6) using the yield coefficient (Yc) values (fresh mass basis) of 1 768,5 for N and 17 248 for P.These were ex pressed as percentages of those proportions of N and P inflow loads estimated not to have been retained o r lost by chemical and biological transformations in the system, i.e. those proportions of N and P inflow loads assumed to be readily available to the water hyacinths for growth.For example, the predicted potential yield of a 20 ha moderately crowded popu lation in the impoundment during August, 1979 was 73,2 metric tonnes of fresh water hyacinths d 1 (Tabic 6).The quantities of (i) N and (ii) P that could be removed daily by the population during this month were estimated as follows: 73.2 X 1 000 1 768,6 73.2 X 1 000 During August, 1979, 20 218 kg N and 1 598 kg P entered the impoundment (Table 2 ), o r daily inflow loads of 652,2 kg N d -1 and 51,5 kg P d-t, of which 26,3% with respect to N and 78,9% with respect to P were estimated to be retained o r lost by chemical and biological transformations in the system (Table 1).Consequently, the proportions of daily (i) N and (ii) P inflow loads estimated to have been readily avail able to the water hyacinths for growth during Au gust, 1979 were: Predictions o f the quantities of N and P that could be removed by a 20 ha moderately crowded popula tion in the impoundment for the period August, 1979to July, 1980 are given in Table 6.The results indicate that such a population could, at least during most of the above-mentioned period, remove larger quantities of P daily than those entering the system that were readily available for water hyacinth growth.The predicted quantities of P that could be removed daily by the population, expressed as per centages of the estimated available P inflow loads, ranging from 158,1 to 3 875,0% except during Au gust, September and O ctober, 1979 when these ranged from only 38,5 to 59,1% (Table 6).In con trast to P, the results indicate that the population would generally remove smaller quantities of N daily than those entering the impoundment that were readily available for water hyacinth growth.The pre dicted quantities o f N that could be removed daily by the population, expressed as percentages of the esti mated available N inflow loads, ranging from 8,6 to 82,4% except during March and April, 1980 when these ranged from 513,5 to 1 348,2% (Table 6).
The above estimates were based on the minimum quantities of N and P that could be removed daily by

Water quality
A comparison of chemical treatm ent costs re ported by Howes (1983) for the different 'water quality types' in the Vernon Hooper Dam, prior to and after retention of ca 20 ha o f water hyacinths behind a floating boom and the introduction of har vesting are presented in Table 7.A reduction of 61% in chemical treatm ent costs was achieved initially through the introduction o f harvesting.The cost re duction dropped to 45% during the second six month period of harvesting due primarily to in creased nutrient loading and poor rainfall.Cost of harvesting and disposal of water hyacinths varied be tween R600 and R1 000 per day which was initially justified by savings in chemical treatm ent costs when treating in excess of 37 M/ d-i.Justification no longer exists financially.However, the resultant reduction in algal concentrations, i.e. improvement in quality of the 'Aleal laden' water, is highly beneficial (Howes. 1983).
W hether the ca 20 ha of water hyacinths presently confincd in the impoundment could reduce nutrient concentrations in the water to levels limiting for al gae and account for the observed reduction in algal concentrations is difficult to ascertain.The average N : P ratio in the water is ca 25.5 (Archibald & War wick, 1980) suggesting that P may be the nutrient most frequently limiting for algae in the impound ment.Furtherm ore, it was predicted that a 20 ha moderately crowded population in the impound ment during the period August, 1979 to July, 1980 could generally remove larger quantities of P daily than those entering the system that were readily available for plant growth (Table 6).However, this does not necessarily m ean that the present popula tion could reduce P concentrations in the water of the reservoir to levels limiting for algae.This would be dependent on a num ber of factors, viz: (i) rate and efficiency o f P uptake by water hyacinths, (ii) magnitude of P inflow loads, (iii) residence time of inflowing water beneath the water hyacinth mat, (iv) extent of mixing between inflowing water and reser voir water and (v) influence of the water hyacinth population on chemical and biological transforma tions in the impoundment.It is clear that a large pro portion of the P entering the impoundment is re tained by adsorption onto sediments (H epher, 1958;Hayes & Phillips, 1968).This source of P is poten tially available to plants for growth, since sediment P and dissolved P exist in equilibrium (H epher. 1958;Pomeroy et al., 1965).The equilibrium concentra tion increases with increased P content in the sedi ment (Pomeroy et al., 1965).Removal of P from the water by hyacinths during growth could, therefore, displace the P equilibrium allowing additional P to be released from sediments into the overlying water.In addition, anoxic conditions that might be pro duced beneath the water hyacinth mat could also provide conditions conducive for the release of sedi ment P (M ortimer. 1941;Vollenweidcr, 1972).
Available chemical and hydrological data for the impoundment after July, 1980 are incomplete.How ever, they do indicate that since the end of 1981, when ca 20 ha of water hyacinths were retained be hind a floating boom in the impoundment and har vesting was initiated (Howes, 1983), the magnitude of the monthly P inflow loads have generally not been very much different from those during 1979 and 1980 (Table 8), Consequently, if one extrapo lates from the predicted quantities of P that could be removed by a 20 ha population, relative to the esti mated available inflow loads, for the period August.!979 to July, 1980, it would appear that the 20 ha of water hyacinths confined in the impoundment since the end o f 1981 could have removed those propor tions of P inflow loads not removed by processes other than water hyacinth uptake in the system.Fur thermore, during summer the reservoir is stratified and a well defined thermocline develops at a depth of 6 to 8 m (Archibald & Warwick, 1980).There fore.one may speculate that the development of this thermocline and consequent density gradient in the impoundment could allow the water hyacinth popu lation to reduce P concentrations in the epilimnion to levels that could be limiting for algae, at least dur ing summer when maximum algal growth rate and production would be expected.Any P released from sediments into the hypolimnion would theoretically be restricted from diffusing into the epilimnion by the thermocline.This may partly explain the ob served reduction in algal concentrations in the reser voir since the introduction o f harvesting.---data incom plete o r unavailable U nder conditions of increased inflow and P load ing, evident from monthly inflow and P loading data for the impoundment for the period January to D e cem ber, 1976 (Table 8), a larger population would, however, generally be needed in the reservoir to re move those proportions of P inflow loads not re moved by processes other than water hyacinth up take in the system.This is evident from the predicted specific growth rates, potential yields and N and P removal potentials of a 20 ha moderately crowded population in the impoundment for this period (Table 9).The results indicate that such a population would, at least during 8 months o f the above-men tioned period, remove smaller quantities o f P daily than those entering the impoundment that were readily available for plant growth.The predicted quantities of P that could be removed daily by the population, expressed as percentages of the esti mated available P inflow loads, ranging from 21,4 to 97,8% , except during March, June, September and Decem ber, 1976 when they ranged from 104,3 to 115,2%.Using the model, the population sizes that would be required in the impoundment to remove the estimated available P inflow loads were pre dicted.These ranged from 20,6 to 93,4 ha, except during March, June, September and December, 1976 when they ranged from 17,3 to 19,2 ha only (Table 9).An example of the derivation is as fol lows: during January, 1976 the daily P inflow load was 77,7 kg P d-> of which 39,7 kg P d-' was esti mated to have been readily available to plants for growth (Table 9).The potential yield (Xpy) of water hyacinth during this month would be: Xpy = 39,7 X 17 248 = 684,7 metric tonnes of fresh water hya cinths d_1 The population size required to produce this poten tial yield at a predicted specific growth rate o f 0,0449 g fresh mass g-1 d_! for a moderately crowded popula tion (Table 9) would be: Xo + 684,7 = Xoe * i = 14 910,7 metric tonnes of fresh water hyacinths Assuming an average stand density (dry mass basis) of 11,7 metric tonnes ha-1 for a moderately crowded population and a mean water content o f water hya cinth of 94,75%, the area occupied by the popula tion would be: 14 910,7 x 5,25 11,7 x 100 = 66,9 ha CO N C L U SIO N S Harvesting water hyacinth growing in eutrophied aquatic systems directly addresses the problem of nutrient enrichment of water and not only the ex cessive aquatic plant growth which is a manifestation of the problem.In designing an effective harvesting strategy for water hyacinth, the model serves as a useful aid for identifying the limiting nutrient and predicting population sizes, yields, growth rates and frequencies and amounts of harvest, under varying conditions o f nutrient loading and climate, to control both nutrient inputs and excessive growth in eutro phied aquatic systems.However, accurate predictive estimates using the model will require the incorpor ation of mathematical expressions from which those proportions of N and P inflow loads retained or lost by chemical and biological transformations in such systems can be predicted.Such mathematical ex pressions will also need to integrate the influence of the water hyacinth population on these transforma tions, Furtherm ore, the relationship between maxi mum specific growth rate of water hyacinth and den sity of the population will need to be mathematically form ulated, since this presents a potential constraint to the model's application.It is clear that the nutri ent removal capacity of water hyacinth is a function of the population size, its density and growth rate.An inverse relationship exists between the two latter    August, 1983, the ca 20 ha of water hyacinths confined in the impoundment has been adequate to remove those proportions of P inflow loads that are readily available for plant growth and account for the observed reduction in algal concentrations.However, the 100 metric tonnes of fresh water hya cinths harvested daily from the impoundment, al though adequate during winter, would appear to be insufficient during summer.It is estimated that about two to three times as much fresh plant m a terial (ca 161 to 288 metric tonnes) would need to be harvested daily from the impoundment during sum m er, under reduced nutrient loadings, to contain the predicted potential yields of the population.U nder conditions o f increased inflow and nutrient loading, such as those prior to 1979, the population size and the daily am ounts of harvest would have to be in creased accordingly.These can be predicted from the nutrient loading data using the model.
rient omass o f 2 0 h a p op u latio n m ctric tonnes o f fresh water ir tem perature °C cific grow th rate fresh m a s s g 'l d 'l (X 100 = % d 'l ) itential yield o f the pop u latio n m etric tonnes o f fresh inths d" I arve sting interval d a y s (assum ing 100 m etric tonnes o f hyacinths harvested daily) lantities o f N rem oved by th e p o p u latio n kg N d 'l )w load kg N d"' i o f N inflow load retained o r lost by chem ical and ransfonnations roportion o f N inflow load readily available for plant M d' i antitics o f N rem oved as % o f estim ated available N antities o f P rem oved by the population kg P d 'l w load kg P d 'l i o f P inflow load retained o r lost A C K N O W L E D G E M E N T S We wish to thank the National Institute for W ater Research of the Council for Scientific and Industrial Research, in particular Mr C. G. M, Archibald, for providing chemical and hydrological data for the Vernon H ooper Dam; Prof, D, F. Toerien and Drs P, J. Ashton and M. C, Rutherford for their valu able comments and criticisms; Mr A. M. Howes of the Chemical Branch of the City Engineers D epart m ent, Durban for permission to publish his results; Messrs P. A .Larkan and B. N. T. Smith o f the Umgeni W ater Board for information supplied and Mrs S. S. Brink for typing the manuscript.
1975.Standard methods fo r the examination o f water and waste water, 14th e d n , p p .1193.New Y ork.A R C H IB A L D , C .G .M , & W A R W IC K , R .J .. 1980.V ernon H ooper D am .In R. D .W alm sley & M .T .B atty.Limnology o f some selected South African impoundments, 187-194.P re to ria: C SIR Press.B O C K .J. H .. 1969.Productivity o f th e w a te r hyacinth (Eichhor nia crassipes).Ecology 50: 460-464.B O Y D , C .E .. 1970.V ascular aquatic plan ts for mineral nutrient rem oval from polluted w aters.Econ.Bot.24: 95-103.B O Y D , C .E .& S C A R S B R O O K , E ., 1975.Influence o f nutrient additions and initial density o f plan ts o n production o f w ater hyacinth.Eichhornia crassipes.Aquat.Bot.1: 253-261.C H O N , Y -K .& K N O W LE S. R ., 1979.M easurem ent o f denitrifi cation in tw o freshw ater sedim ents by a n in situ acetylene in h ib itio n m eth o d .A ppl. Environ.Microbiol.37,6: 1067-1072.D E B U S K .T .A ., R Y T H E R , J. H ., H A N 1SA K , M .D .& W IL L IA M S , L. D ., 1981.E ffects o f seasonality and plant density o n th e productivity o f som e fresh-w ater m acrophytes.Aquat.Bot.10,2: 133-142.E N V IR O N M E N T A L P R O T E C T IO N A G E N C Y , 1974.Manual o f methods fo r chemical analysis o f water a n d wastes.R eport No. E PA -625/6-74-003.W ashington, D .C .: U .S .E nviron m ental Protection Agency O ffice o fT echnological T ransfer.E V E R IT T .I.. 1980.W ater hyacinth -Shongw eni D am .D u r ban: C ity E ngineer's D epartm ent.H A Y E S, F. R .& PH ILLIPS.J. E ., 1968.Lake w ater and sedi m ent.IV.R adio-phosphorus equilibrium w ith m ud, plants and bacteria un d er oxidized a n d reduced conditions.Limnol.Oceanogr.3: 459-475.H E P H E R .B ., 1958.O n the dynam ics o f pho sp h o ru s a d d ed to fish pon d s in Israel.Limnol.Oceanogr.3: 84-100.H O W E S , A .M .. 1976.R eport o n the Shongw eni catchm ent: w ater quality a n d trea tm e n t.D urban: C ity E n g in eer's D e partm ent.H O W E S , A .M ., 1983.Shongw eni w a te r tre a tm e n t and hyacinth control.D u rb an : City E n g in eer's D ep artm en t.K E E N Y , D , R ., 1973, T he nitrogen cycle in sedim ent w a te r sys tem s.J. Environ.Qual.2,1: 15-29.M A L E K .I. & F E N C L .Z .. 1966.Theoretical and methodological basis o f continuous culture o f micro-organisms.Prague: C ze choslovak A cadem y o f Sciences.M O R T IM E R , C .H ., 1941.T he exchange o f dissolved substances betw een m ud a n d w ater in lakes.J. Ecol.29: 280-329.M U SIL , C .F ., 1982.The use o f growth kinetics in the develop m ent o f a predictive model fo r the growth o /E ic h h o rn ia cras sipes iM art.)Solms in the field.P h .D thesis.U niversity o f N atal.M U S IL , C .F .& B R E E N , C. M ., 1985a.T he developm ent from kinetic coefficients o f a predictive m ode! for th e grow th o f Eichhornia crassipes in th e field.I. G e n e ra tin g kinetic coef ficients fo r the m odel in greenhouse c u ltu re.Bothalia 15.C .F .& B R E E N .C .M .. 1985b.T he developm ent from kinctic coefficients o f a predictive m odel for the grow th o f Eichhornia crassipes in the field.II.T esting a n d refining the m odel un d er field conditions.Bothalia 15: 705-724.M U SIL , C .F .& B R E L 'S , C \ M .. 1985c.T he developm ent from kinetic coefficients o f a predictive m odel for the grow th of Eichhornia crassipes in the field.111.T esting a m odel fo r predicting grow th ra te s from plant n u trien t concentrations.Hothalia 15: 725-731.P E N F O U N D , W . T .& E A R L E .T .T .. 1948.T he biology o f w ater hyacinth.Ecol.Motiogr.18: 448-472.PH IL L IP P Y , C. L. & P E R R Y M A N .J. M " 1972a.M echanical harvesting o f w ater hyacinth I Eichhornia crassipes) in G ant L ake C an al. Sum ter C ounty.F lorida, p p .2 l.Florida G am e and Fresh W ater Fish C om m ission.UnpuM .P H IL L IP P Y .C .L .& P E R R Y M A N , J. M ., 1972b.M echanical harvesting o f w ater hyacinth (Eichhornia crassipes) in T ro u t L ake.Lake C o u n ty .F lorida, p p .24.Florida G am e and Fresh W ater Fish C om m ission.U npubl.P O M E R O Y .L. R ., SM IT H . E .E .& G R A N T .C .M .. 1965.The exchange o f phosphate betw een estuarine w ater and sedi m ents.Limnol.Oceanogr.10: 167-172.R A D F O R D .P. J .. 1967.G ro w th analysis form ulae -their use a n d abuse.Crop Sci.7: 171-175, 'T O U Z E A U , L. F., 1972.M echanical w ater hyacinth rem oval o p e ratio n s.A quam arine C orp o ratio n , B lufton.Florida, pp.16.Florida G am e a n d Fresh W ater Fish C om m ission.U npubl.T W IN C H , A. J .& B R E E N .C .M .. 1980.Advances in under standing phosphorus cycling in inland waterstheir signifi cance for South African limnology.South A frican N ational Scicntific Program m es, R eport No. 42.V A N D Y K E .J .M .. 1971.M echanical harvesting o f w ater hya cinth (Eichhornia crassipes) in Shell C reek R eservoir, C h ar lotte C o u n ty .F lorida, pp.25.Florida G am e and Fresh W ater Fish C om m ission.U npubl.V O L L E N W E ID E R .R .A ., 1972.Input-output models with special reference to the phosphorus loading concept in limno logy.P aper given at W orkshop C onference o n C hem icalecological considerations for defining th e goal o f w ater pol lution control.K austanienbaum .Sw itzerland, A p ril 19-21.1972.W E A T H E R B U R E A U .R SA .1954.Climate o f South Africa.C lim ate Statistics, Part 1. W B19. P reto ria: T h e G overnm ent Printer.W E S T L A K E , E. F ., 1963.C om parisons o f plant productivity.Biol.Rev. 38: 385-425.Y O U N T .J .L. & C R O S SM A N .R .A .. J r .1970.E utrophication co n tro l by p lan t harvesting.J. Wat.Po/lut.Control Fed.42: 173-183.
TA BLE 1. -Inflow and o u tflo w d a ta and to ta l N and to ta l P balances for the V ernon H ooper Dam (m o n th ly averages: January to Decem ber, 1976) • A negative value indicates e x p o rt.TABLK 2. -Inflow and o u tflo w data and to ta l N and to ta l P balances fo r the V ernon H ooper Dam (m onthly averages: August, 1979 to Ju ly , 1980) TA B L E 4 .-Predicted g row th ra te s, p o ten tial yields and po p u latio n sizes o f w ater h y a cin th in th e V ernon H ooper Dam(A ugust, 1979(A ugust,   t o Ju ly , 1980)   ) a te d specific g row th rate a s % o f m axim um specific g row th rate .mated specific g row th rate a s % o f m axim um specific g row th rate A B U -7 ,-A com parison o f chem ical tre a tm e n t costs fo r the differen t w ater quality ty p es in the V ernon H ooper Dam , p rio r to and follow ing the re tention o f ca. 2 0 ha o f w ater h yacinths behind a floating boom and the in tro d u c tio n o f harvesting, according toHowes (1983) T

TABLE 8 .
-Inflow and P loading d a ta for th e V ernon H ooper Dam (m o n th ly averages) IF. 9 .-Predicted yields, grow th ra te s and n u trien t removal potentials o f a 2 0 h a po p u latio n o f w ater h yacinths confined in m oderately crow ded situations in the V ernon H ooper Dam (Jan u ary to D ecem ber, 1976)