Impact of the Biocontrol Agent Eccritotarsus Catarinensis, a Sap-Feeding Mirid, on the Competitive Performance of Waterhyacinth, Eichhornia Crassipes

Biological Control 32 (2005) 90–96 www.elsevier.com/locate/ybcon

Impact of the biocontrol agent Eccritotarsus catarinensis, a sap-feeding mirid, on the competitive performance of waterhyacinth, Eichhornia crassipes

Julie A. Coetzeea,*, Ted D. Centerb, Marcus J. Byrnea, Martin P. Hillc

a Ecophysiological Studies Research Group, School of Animal, Plant and Environmental Sciences, University of the Witwatersrand, Private Bag X3, Johannesburg 2050, South Africa b US Department of Agriculture, Agricultural Research Service, Invasive Plant Research Laboratory, 3205 College Ave., Fort Lauderdale, FL 33314, USA c Department of Zoology and Entomology, Rhodes University, P.O. Box 94, Grahamstown 6140, South Africa

Received 4 December 2003; accepted 6 August 2004 Available online 25 November 2004

Abstract

The mirid, Eccritotarsus catarinensis, was released in South Africa to aid in the biological control of waterhyacinth (Eichhornia crassipes). Post-release evaluations are needed to quantify the miridÕs impact on the weed in South Africa. The subtle feeding damage that it causes is not easily measured, but studies have shown that interactions with other plant stresses, e.g., plant competition, can often magnify impacts of plant-feeding insects. The impact of the mirid was therefore evaluated using an additive series analysis of competition between waterhyacinth and waterlettuce (Pistia stratiotes), as influenced by mirid herbivory. Competitive abilities of waterlettuce and waterhyacinth were determined using an inverse linear model with plant weight as the yield variable. In the absence of herbivory, waterhyacinth was 23 times more competitive than waterlettuce, but only 10 times more competitive when exposed to mirid feeding. Waterlettuce was only 0.9 times as aggressive as waterhyacinth that was free of herbivory, but 1.5 times as competitive when mirids were impacting waterhyacinth. Most importantly, in the presence of herbivory on waterhyacinth, interspecific competition coefficients from waterhyacinth on waterlettuce were no longer statistically significant. These results show that the mirid desta- bilizes waterhyacinthÕs competitive interactions between these two floating plant species, although impacts were subtle. This insect is unlikely to be an effective agent by itself, but it will be a useful complement to the existing biological control agents in South Africa. Ó 2004 Elsevier Inc. All rights reserved.

Keywords: Eccritotarsus catarinensis; Eichhornia crassipes; Pistia stratiotes; Biological control; Competition; Herbivory; Risk assessment

1. Introduction worldwide (Julien and Griﬃths, 1998). Some have suc- cessfully controlled waterhyacinth in tropical regions, Waterhyacinth (Eichhornia crassipes (Mart.) Solms- such as Papua New Guinea (Julien and Orapa, 1999) Laub (Pontederiaceae)) has achieved international noto- and Lake Victoria in Uganda (Cock et al., 2000). How- riety as the worldÕs worst aquatic weed (Holm et al., ever, additional agents are being considered for release 1977), and has, as a result, been targeted for biological in more temperate regions, such as South Africa and control in several countries (Harley, 1990). At least se- North America, where results have been more variable ven natural enemies have been used for this purpose (Cordo, 1999; Stanley and Julien, 1999). Five arthropod natural enemies of waterhyacinth * Corresponding author. Fax: +27 11 403 1429. have been released in South Africa, and more are being E-mail address: [email protected] (J.A. Coetzee). considered (Hill and Cilliers, 1999). The mirid Eccrito-

1049-9644/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.biocontrol.2004.08.001 J.A. Coetzee et al. / Biological Control 32 (2005) 90–96 91 tarsus catarinensis (Carvalho) (Heteroptera: Miridae) et al., 1999). Because no biocontrol agent comes without was introduced in 1996 (Hill et al., 1999) and, although risks, and because of doubts as to potential benefits of populations have established, its performance has yet to the mirid, we employed competition studies to magnify, be assessed. Inasmuch as each new biocontrol agent and thereby quantify, subtle sublethal effects of its feed- might present additional risk to the native flora (Sim- ing. The mirid does not feed on waterlettuce (Hill et al., berloff and Stiling, 1996), assessments of the perfor- 1999), another floating species that commonly coexists mance of existing agents are needed to determine with waterhyacinth, so it was chosen as the competing whether further introductions are warranted. species. Even though it is fundamentally important to under- This study follows the protocol employed by Pantone stand the effects of biological control agents on native et al. (1989), who suggested competition experiments flora, typically, their effectiveness is almost always as- using an additive series, and the application of the in- sessed by measuring their direct effects on the target verse linear model as effective methods for evaluating weeds (Callaway et al., 1999). However, impacts are of- the efficacy of potential control agents. These methods ten sublethal and experiments fail to detect subtle effects provide biologically meaningful competition coeffi- that accrue over time and provide significant long-term cients, which will allow for direct comparison with other control. Furthermore, feeding damage by sap-feeding in- similar competition studies. sects is often inconspicuous and without overt symp- toms. Attention has therefore recently focused on exploiting the interaction between herbivory and plant 2. Materials and methods competition, and their combined effects on plant performance, as a means of assessing the value of new agents Waterhyacinth and waterlettuce were grown out- (e.g., Ang et al., 1994; Story et al., 2000; Van et al., 1998). doors in plastic tubs at the University of the Witwaters- The effects of biological control agents that do not in- rand, Johannesburg, to determine competitive flict direct mortality on the target plant can be difficult interactions between the two plant species. Both plants to measure. It has been suggested that the impact of were obtained from stock cultures maintained at the invertebrate herbivory will become more noticeable university. The experimental design employed an addi- when grazed plants are competing with other plants tive series (Spitters, 1983) of factorial combinations of for resources, due to a gradual reduction of the host different densities of the two species in a randomized plantÕs vigor (Center et al., 2001). Therefore, any reduc- block design. The waterhyacinth:waterlettuce densities tion in the competitive ability of grazed plants may be were 0:3, 0:9, 3:0, 3:3, 3:9, 9:0, 9:3, and 9:9 plants per sufficient to increase their susceptibility to competition container. The waterhyacinth:waterlettuce density from other plants (Cottam, 1986; Whittaker, 1979). matrices were repeated twice—one series as a control Thus, a fundamental justification for using biological without insects and the other with the mirid E. catarin- control agents to suppress invasive plants is that by ensis in a randomized block design. The two matrices weakening the invader, indigenous species may gain a were repeated three times, resulting in 48 tubs of differ- competitive advantage (Callaway et al., 1999). Further- ent plant combinations. more, Crawley (1989) noted that the principal effect of The plastic tubs (51 cm by 40 cm and 28 cm deep) herbivores is not caused by them eating the plants to were filled with 23 L water. Nitrogen and phosphorus extinction, but more often from their feeding modifying were added as potassium nitrate (KNO3) and potassium the relative competitive abilities of the plants with one dihydrogen orthophosphate (KH2PO4), respectively, at another. concentrations of 50.5 mg N/L and 2.56 mg P/L, the Biological control studies conducted to determine the concentrations at which maximum N storage occurs in combined effect of herbivory and plant competition on waterhyacinth (Reddy et al., 1989, 1990). A commercial plant performance have shown that by combining iron chelate (13% Fe) was also added to the water at a strong plant competitors with biological control agents, concentration of 2 g/23 L of water. Water and nutrients the control obtained exceeds that exerted by either were replaced weekly. Each plastic tub was enclosed by mechanism alone (e.g., Ang et al., 1994; Story et al., a net canopy. 2000; Van et al., 1998). When grown together in the ab- Two weeks later, all daughter ramets were removed sence of natural enemies, waterhyacinth dominates wat- from the plants to reestablish the prescribed stocking erlettuce in nutrient enriched waters (Agami and Reddy, densities. Thereafter, 15 insects per waterhyacinth plant, 1990). Therefore, if this interaction changes in the pres- collected randomly from the Plant Protection Research ence of herbivory, we can assume that the biological Institute in Pretoria, South Africa, were released into control agent has reduced the vigor, and therefore the each cage. The male:female ratio was not controlled as competitive ability, of waterhyacinth. the mirid exhibits a 50:50 sex ratio. Feeding damage to Eccritotarsus catarinensis, a sap-sucking mirid, pro- the waterhyacinth plants and mirid numbers were mon- duces damage that is not usually lethal to the plant (Hill itored visually for the first 4 weeks to ensure that the 92 J.A. Coetzee et al. / Biological Control 32 (2005) 90–96 mirids had established and were feeding on waterhya- culated for both waterhyacinth and waterlettuce, in the cinth. It was not viable to monitor insect density during presence or absence of herbivory on waterhyacinth. the course of the experiment because the disturbance Mann–Whitney U tests were used to analyze whether would have been too disruptive to the plants and the the mean end-weights of waterhyacinth and waterlet- insects. tuce, were significantly different in the presence and ab- After 16 weeks, the insects were removed, the number sence of mirid herbivory. of adults was counted, the presence of nymphs was noted, and the plants were harvested. In each tub, the two plant species were separated and the total biomass 3. Results (fresh weight), including daughter plants, of each species was measured. These wet weight values were then di- After 4 weeks, a visual assessment of each tub in vided by the original plant stocking density to obtain which mirids had been released confirmed the presence mean wet-weight. Dead plant material was not removed of feeding damage to the waterhyacinth leaves. At the prior to weighing. Fresh weight was measured instead of end of 16 weeks, the mean number of adult mirids col- dry weight because both measures are highly correlated lected per plant, for the original planting densities, was (T.D. Center, unpublished data), and fresh weight was 37.25 ± 10.96, and nymphs were present in every tub the more expedient measure. in which adults had been released, indicating that the Data were analyzed using the inverse linear model as mirid populations were well established. described by Spitters (1983) and Pantone et al. (1989). This model involves multiple linear regressions of the in- 3.1. Waterhyacinth verse of the mean weight-yield of each species as the dependent variable, on the planting densities of water- Waterhyacinth remained the dominant species after hyacinth and waterlettuce, used as the independent vari- 16 weeks of selective feeding by the mirid, but its com- ables. The regression equation is of the form: petitive advantage over waterlettuce decreased as re- flected by the ratios of respective competition 1=wh ¼ ah0 þ ahhdh þ ahldl; coefficients (ahh/ahl). Waterhyacinth without herbivory was at least 23 times (P < 0.05) more competitive than 1=wl ¼ al0 þ alldl þ alhdh: waterlettuce, but only 10-fold greater with sustained Here 1/wh and 1/wl are the inverse biomass yields of damage from mirid feeding (P < 0.05), indicating a individual waterhyacinth and waterlettuce plants, 56% decrease in competitive ability (Table 1). Tests of respectively, and dh and dl represent their respective the pooled data indicate that the main effects of the planting densities. The coefficients ahh and all estimate regression equations were significantly different due to intraspecific competition, while the coefficients ahl and herbivory (F1, 32 = 7.76, P = 0.009). alh estimate interspecific competition, in terms of their These data were also graphically analyzed as three-di- effects on reciprocal waterhyacinth or waterlettuce yield. mensional, surface–response planes where the slope in The intercepts (ah0 and al0) measure the reciprocal of the one direction represents the effect of waterhyacinthÕs maximum weight of isolated plants. The ratios of the own density upon its yield, and the slope in the other coefficients ahh/ahl and all/alh measure the effects of direction represents the effects of waterlettuce density intraspecific competition by waterhyacinth and waterlet- on waterhyacinth yield (Fig. 1). The flat slope of the tuce, respectively, on their own yield relative to the waterlettuce density regression planes in both the her- effects of interspecific competition by one species on bivory treatment and the control indicated that interspe- the yield of the other (Pantone et al., 1989). cific competition from waterlettuce was negligible. The regressions were weighted using biomass as the Waterlettuce had little effect on waterhyacinth yield weighting variable because the wet weight (wh and wl) either with (waterlettuce density b = 0.05) or without variance decreased as plant biomass increased. There- herbivory (waterlettuce density b = 0.17). However, fore, greater wet weights values influenced the regression the steep slope of the waterhyacinth regression planes more than smaller ones (Spitters, 1983). Regressions for showed that intraspecific competition significantly af- the control series and insect-treated series were com- fected waterhyacinth yield with (waterhyacinth density pared using multiple regression analysis (Zar, 1996). b = 0.917) or without (waterhyacinth density An indicator variable was used to distinguish between b = 0.870) herbivory. the presence and absence of the control agent, and F tests of regressions of the pooled data determined 3.2. Waterlettuce whether the competition coefficients were significantly different as a result of herbivory by the mirid. Waterlettuce was obviously the weaker competitor For each experimental unit, the mean end-weights per inasmuch as adding one waterlettuce plant (all/alh original plant (i.e., total yield/planting density) were cal- = 0.974, P < 0.005) had the same impact on waterlettuce J.A. Coetzee et al. / Biological Control 32 (2005) 90–96 93

Table 1 Multiple regression analysis of the eﬀects of Eccritotarsus catarinensis herbivory and plant density on the reciprocal of waterhyacinth yielda and waterlettuce yield b (wet weight (g)) Treatment Regression coeﬃcients Intercept R2 F value (P)

Waterhyacinth ahh ahl ahh/ahl ah0 E. catarinensis absent 0.243 0.010 23.631 0.5214 0.755 23.11 (<0.05) E. catarinensis present 0.182 0.0181 10.007 0.6086 0.856 44.61 (<0.05)

Waterlettuce all alh all/alh al0 E. catarinensis absent 8.650 8.886 0.974 46.853 0.690 16.67 (<0.05) E. catarinensis present 3.294 2.069 1.592 6.392 0.460 6.40 (<0.05) a The intercepts ah0 estimate the reciprocal of the maximum weight of isolated waterhyacinth plants. The regression coefficients ahh and ahl measure intraspecific and interspecific competition respectively, for waterhyacinth. The ratio ahh/ahl measures the effects of intraspecific competition by waterhyacinth on its own weight relative to the effects of interspecific competition by waterlettuce. b The intercepts al0 estimate the reciprocal of the maximum weight of isolated waterlettuce plants. The regression coefficients all and alh measure intraspecific and interspecific competition, respectively, for waterlettuce. The ratio all/alh measures the effects of intraspecific competition by waterlettuce on its own weight relative to the effects of interspecific competition by waterhyacinth.

Fig. 1. Multiple regression planes indicating the combined effects of waterlettuce and waterhyacinth densities on the reciprocal of the mean wet weight (1/kg) per waterhyacinth plant (i.e., higher values represent lower yield). (A and B) Compare relative competitive abilities of waterhyacinth in the absence and presence of E. catarinensis feeding damage, respectively. Points indicate observations (n = 18) and the vertical lines between data points represent the residuals. Values on X and Y axes represent waterlettuce and waterhyacinth planting densities at the outset of the experiment. yield as adding one waterhyacinth plant without herbiv- itive ability had increased; while waterhyacinthÕs com- ory, but in the presence of the mirid, this ratio increased petitive ability had decreased. to 1.6 (P < 0.05) (Table 1). Tests of the pooled data indi- There was no significant difference between the mean cated that the main effects of the regression equations end-weights per original waterhyacinth plant, in either were significantly different due to herbivory the insect-free control or the herbivory treatment (origi- (F1, 32 = 6.50, P = 0.016). nal planting density of three waterhyacinth plants: The three-dimensional surface response planes were U9;9 = 32.5, P > 0.05; original planting density of nine again used to determine the effects of intra- and interspe- waterhyacinth plants: U9;9 = 18.5, P > 0.05) (Fig. 2A). cific competition on waterlettuce yield (Fig. 2). In the in- Similarly, there was no significant difference between sect-free control, both waterhyacinth and waterlettuce the mean end-weights per original waterlettuce plant planting densities significantly affected the wet weight for either treatment (original planting density of three of waterlettuce, indicated by the steep gradients in both waterlettuce plants: U9;9 = 33.5, P > 0.05; original directions (waterhyacinth density b = 0.568, waterlet- planting density of nine waterlettuce plants: tuce density b = 0.436) (Fig. 2A). Therefore, both inter- U9;9 = 39.0, P > 0.05) (Fig. 2B). and intraspecific competition were operative when there was no influence of herbivory. However, this relationship changed in the presence of mirids and only intra- 4. Discussion specific competition significantly affected waterlettuce yield (waterhyacinth density b = 0.21; waterlettuce Previous studies have demonstrated the superior density b = 0.625) (Fig. 2B), indicating that its compet- competitive nature of waterhyacinth when grown in cul- 94 J.A. Coetzee et al. / Biological Control 32 (2005) 90–96

Fig. 2. Multiple regression planes indicating the combined effects of waterhyacinth and waterlettuce densities on the reciprocal of the mean wet weight (1/kg) per waterlettuce plant. (A and B) Compare relative competitive abilities of waterlettuce in the absence and presence of E. catarinensis feeding damage to waterhyacinth, respectively. Points indicate observations (n = 18) and the vertical lines between data points represent the residuals. Values on X and Y axes represent waterhyacinth and waterlettuce planting densities at the outset of the experiment. ture with waterlettuce (Agami and Reddy, 1990; Sutton, employed the same nutrient levels in the growth media 1983; Tag El Seed, 1978). Waterhyacinth shades and have shown that under high mirid feeding intensity, the stresses waterlettuce plants through its high productivity number of waterhyacinth leaves, and shoot offsets is and morphological plasticity (Agami and Reddy, 1990). reduced (Coetzee, 2003). Furthermore, compensation However, in this study, herbivory by E. catarinensis re- for herbivory by waterhyacinth may have been a duced the vigor of waterhyacinth growing in culture short-term phenomenon. Had the study been allowed with waterlettuce and correspondingly increased the rel- to continue longer so that the mirid population could ative competitive ability of waterlettuce. Therefore, the increase, reduction in waterhyacinth yield might have most significant finding of our study is that mirid herbiv- been more evident. However, if we had allowed the ory reduced waterhyacinthÕs competitive ability in two experiment to go on for much longer, we might have ways. First, the effects of interspecific competition from missed seeing the effect of competition. Therefore, waterhyacinth on waterlettuce yield became negligible determining the proper length of time for these exper- when the mirid was present, as indicated by the flat iments is difficult. regression plane (Fig. 2); and second, there was a 56% In addition, compensation might have been possible reduction in intraspecific competition by waterhyacinth because of the high nitrogen (N) concentration in the (Table 1). nutrient media. Sap feeders, such as the mirid, remove We had expected waterhyacinth wet weight to be N from the plant, but waterhyacinth exhibits high con- significantly less, and waterlettuce wet weight to be sumption of N in high nutrient conditions (Gossett greater, in the presence of herbivory, but no difference and Norris, 1971; Reddy et al., 1989, 1990), so there was evident. Compensation for herbivory might explain may have been N to spare for the mirid to consume. Un- why there was no reduction in waterhyacinth yield. der low nutrient conditions, we expect the effects of mir- Generally, plant–herbivore interactions are considered id herbivory on waterhyacinth yield to be more severe. antagonistic because of the negative direct effects herbi- The concentrations of N and P used in this study are vores have on plants, through biomass consumption very high, but are realistic values in that they are the (De Mazancourt and Loreau, 2000). However, the threshold concentrations for N storage of waterhyacinth grazing optimization hypothesis has challenged this (Reddy et al., 1989, 1990). According to South African theory by suggesting that herbivores can enhance plant Water Quality standards, the N and P concentrations primary production under certain conditions, where used are hypertrophic (Walmsley, 2000), and impound- low grazing intensity will increase primary production ments exist in South Africa with N and P concentrations up to an optimum at medium grazing intensity, after similar to those used in this study (e.g., Naicker et al., which production will decrease as grazing increases 2003). A study conducted at Hammarsdale Dam (Dyer et al., 1986; Hilbert et al., 1981). The initial in- (KwaZulu Natal Province, South Africa) concluded that sect inoculation was small and the extent of feeding a site close to a wastewater inlet was hypertrophic damage observed in this experiment was not extreme, (1.561 mg/N/L and 0.731 mg/P/L) (Oberholzer, 2002). and might be classified as low to intermediate intensity, Both Neochetina eichhorniae Warner (Coleoptera: Cur- which could explain why waterhyacinth production culionidae), another waterhyacinth control agent, estab- was not affected by herbivory. Other experiments that lished at this site since 1989, and E. catarinensis, J.A. Coetzee et al. / Biological Control 32 (2005) 90–96 95 established here since 1996, have reached very high pop- References ulation densities but appear to have had little impact on the waterhyacinth infestation (Hill and Olckers, 2000). It Agami, M., Reddy, K.R., 1990. Competition for space between is predicted that the hypertrophic conditions have ne- Eichhornia crassipes (Mart.) Solms and Pistia stratiotes L. cultured in nutrient-enriched water. Aquat. Bot. 38, 195–208. gated the effects of herbivory by these two control Ang, B.N., Kok, L.T., Holtzman, G.I., Wolf, D.D., 1994. Competitive agents. Thus, efficacy of an agent is difficult, if not growth of Canada thistle, tall fescue and crownvetch in the impossible, to predict prior to its release. Perhaps the presence of a thistle defoliator, Cassida rubignosa Mu¨ller (Coleop- mirid will be more effective when nutrients are less avail- tera: Chrysomelidae). Biol. Control 4, 277–284. able to the plant. Callaway, R.M., DeLuca, T.H., Belliveau, W.M., 1999. Biological control herbivores may increase competitive ability of the noxious Using competition experiments such as this to evalu- weed Centaurea maculosa. Ecology 80, 1196–1201. ate the efficacy of a new control agent of waterhyacinth Center, T.D., Van, T.K., Hill, M.P., 2001. Can competition exper- has many advantages. Individuals of both competing iments be used to evaluate the potential efficacy of new waterhy- species are easily identifiable. Also, the nutrient status acinth biological control agents? In: Julien, M.H., Hill, M.P. of the water is easily controlled, and the critical crite- Center T.D., Jianqing, D. (Eds.), Biological and Integrated Control of Waterhyacinth, Eichhornia crassipes. Proceedings of the 2nd rion, wet weight, is easily measured. Because the effect Meeting of the Global Working Group for the Biological and of feeding damage by E. catarinensis on waterhyacinth Integrated Control of Waterhyacinth, Beijing, China, 9–12 October vigor is subtle, it is difficult to measure the direct effects 2000. ACIAR Proceedings No. 102, pp. 77–81. of the mirid. Waterhyacinth has a very plastic pheno- Cock, M., Day, R., Herren, H., Hill, M.P., Julien, M.H., Neuensch- type in leaf and petiole shape and size, which can be wander, P., Ogwang, J., 2000. Harvesters get that sinking feeling. Biocontrol News Inf. 21, 1–8. influenced by many factors. Comparative experiments Coetzee, J.A., 2003. Suitability of Eccritotarsus catarinensis as a with other biocontrol agents of waterhyacinth, Neoche- Biological Control Agent of Water Hyacinth. PhD Thesis, tina eichhorniae and N. bruchi Hustache (Coleoptera: University of the Witwatersrand, Johannesburg, South Africa. Curculionidae), have shown that without herbivory, 41 Cordo, H.A., 1999. New agents for biological control of waterhya- waterlettuce plants produce a competitive effect equiva- cinth. In: Hill, M.P., Julien, M.H., Center, T.D. (Eds.), Proceedings of the First IOBC Global Working Group Meeting for the lent to one waterhyacinth plant, while weevil feeding Biological and Integrated Control of Waterhyacinth, 16–19 damage reduced this ratio to 0.7 for N. bruchi alone, November 1998, Zimbabwe. Plant protection Research Institute, 1.4 for N. eichhorniae alone, and 0.6 for the two species Pretoria, South Africa, pp. 68–74. together (T.D. Center, unpublished data). By this yard- Cottam, D.A., 1986. The effects of slug-grazing on Trifolium repens stick, the weevils are more effective control agents, but and Dactylis glomerata in monoculture and mixed sward. Oikos 47, 275–279. nonetheless, E. catarinensis does reduce waterhyacinth Crawley, M.J., 1989. Insect herbivores and plant population dynamics. vigor and it should be a valuable adjunct to the existing Annu. Rev. Entomol. 34, 531–564. biological control program. De Mazancourt, C., Loreau, M., 2000. Grazing optimization, nutrient These data indicate that even though the competitive cycling, and spatial heterogeneity of plant–herbivore interactions: ability of waterhyacinth is reduced by mirid herbivory, should a palatable plant evolve? Evolution 54, 81–92. Dyer, M.I., De Angelis, D.L., Post, W.M., 1986. A model of herbivore the mirid will not be very effective in high nutrient situ- feedback on plant productivity. Math. Biosci. 79, 171–184. ations, although competitors might better be able to Gossett, D.R., Norris Jr., W.E., 1971. Relationship between nutrient withstand encroachment by waterhyacinth. In biological availability and content of nitrogen and phosphorous in tissues of control programs, it would be desirable to evaluate the the aquatic macrophyte, Eichhornia crassipes (Mart.) Solms. efficacy of a potential control agent prior to its release. Hydrobiologia 38, 15–28. Harley, K.L.S., 1990. The role of biological control in the management Experiments such as those reported here could evaluate of waterhyacinth, Eichhornia crassipes. Biocontrol News Inf. 11, the potential value of the new control agent in quaran- 11–22. tine, thereby preventing the introduction of questionable Hilbert, D.W., Swift, D.M., Detling, J.K., Dyer, M.I., 1981. Relative agents that might provide little control. While the meth- growth rates and the grazing optimization hypothesis. Oecologia od has shortcomings (see Center et al., 2001), it is worth 51, 14–18. Hill, M.P., Cilliers, C.J., Neser, S., 1999. Life history and laboratory considering. host range of Eccritotarsus catarinensis (Carvalho) (Heteroptera: Miridae), a new potential natural enemy released on waterhyacinth (Eichhornia crassipes (Mart.) Solms-Laub.) (Pontederiaceae) in South Africa. Biol. Control 14, 127–133. Acknowledgments Hill, M.P., Cilliers, C.J., 1999. A review of the arthropod natural enemies, and factors that influence their efficacy, in the biological The Working for Water Programme of the Depart- control of waterhyacinth, Eichhornia crassipes (Mart.) Solms- ment of Water Affairs and Forestry, South Africa, are Laubach (Pontederiaceae), in South Africa. In: Olckers, T., Hill, acknowledged for their financial assistance with this M.P. (Eds.), Biological Control of Weeds in South Africa (1990– 1998). Afr. Entomol. Mem. No. 1, pp. 103–112. project. The authors also thank Kerith Nel and Jennifer Hill, M.P., Olckers, T., 2000. Biological control initiatives against Lancaster (University of the Witwatersrand) for techni- water hyacinth in South Africa: constraining factors, success and cal assistance during the trials. new courses of action. In: Julien, M.H., Center, T.D., Hill, M.P. 96 J.A. Coetzee et al. / Biological Control 32 (2005) 90–96

(Eds.), Proceedings of the Second Global Working Group Meeting Simberloff, D., Stiling, P., 1996. How risky is biological control? for the Biological and Integrated Control Water Hyacinth. Beijing, Ecology 77, 1965–1974. China, 9–12 October 2000. Spitters, C.J.T., 1983. An alternative approach to the analysis of mixed Holm, L.G., Plucknett, D.L., Pancho, J.V., Herberger, J.P., 1977. The cropping experiments. 1. Estimation of competition effects. Neth. J. WorldÕs Worst Weeds. Distribution and Biology. University Press Agric. Sci. 31, 1–11. of Hawaii, Honolulu, Hawaii. Stanley, J.N., Julien, M.H., 1999. The host range of Eccritotarsus Julien, M.H., Orapa, W., 1999. Structure and management of a catarinensis (Heteroptera: Miridae), a potential agent for the successful biological control project for waterhyacinth. In: Hill, biological control of water hyacinth (Eichhornia crassipes). Biol. M.P., Julien, M.H., Center, T.D. (Eds.), Proceedings of the First Control 14, 134–140. IOBC Global Working Group Meeting for the Biological and Story, J.M., Good, W.R., White, L.J., Smith, L., 2000. Effects of the Integrated Control of Waterhyacinth, 16–19 November 1998, interaction of the biocontrol agent Agapeta zoegana L. (Lepidop- Zimbabwe. Plant protection Research Institute, Pretoria, South tera: Cochylidae) and grass competition on spotted knapweed. Africa, pp. 123–134. Biol. Control 17, 182–190. Julien, M.H., Griffiths, M.W., 1998. Biological Control of Weeds: A Sutton, D.L., 1983. Aquatic plant competition. Aquatics 5, 10–14. World Catalogue of Agents and their Target Weeds, fourth ed. Tag El Seed, M., 1978. Effect of pH on the nature of competition C.A.B.I. Publishing, CAB International, Oxon, UK. between Eichhornia crassipes and Pistia stratiotes. J. Aquat. Plant Naicker, K., Cukrowska, E., McCarthy, T.S., 2003. Acid mine Manag. 16, 53–57. drainage arising from gold mining activity in Johannesburg, South Van, T.K., Wheeler, G.S., Center, T.D., 1998. Competitive interac- Africa and environs. Environ. Pollut. 122, 29–40. tions between hydrilla (Hydrilla verticillata) and vallisneria (Val- Oberholzer, I.G., 2002. The Effect of Nutrient Rich Water on lisneria americana) as influenced by insect herbivory. Biol. Control Biological Control of Water Hyacinth. M.Sc. Thesis, University 11, 185–192. of Pretoria, Pretoria, South Africa. Walmsley, R.D., 2000. Perspectives on eutrophication of surface Pantone, D.J., William, W.A., Maggenti, A.R., 1989. An alternative waters: Policy/research needs in South Africa. WRC Report approach for evaluating the efficacy of potential biocontrol agents KV129/00. Water Research Commission, Pretoria, South Africa. of weeds. 1. Inverse linear model. Weed Sci. 37, 771–777. Whittaker, J.B., 1979. Invertebrate grazing, competition and plant Reddy, K.R., Agami, M., Tucker, J.C., 1989. Influence of nitrogen dynamics. In: Anderson, R.M., Turner, B.D., Taylor, L.R. (Eds.), supply rates on growth and nutrient storage by water hyacinth Population Dynamics. The 20th Symposium of the British (Eichhornia crassipes) plants. Aquat. Bot. 36, 33–43. Ecological Society, London, 1978. Blackwell Scientific Publica- Reddy, K.R., Agami, M., Tucker, J.C., 1990. Influence of phosphorus tions, Oxford. supply on growth and nutrient storage by water hyacinth (Eich- Zar, J.H., 1996. Biostatistical Analysis, third ed. Prentice Hall, hornia crassipes) plants. Aquat. Bot. 37, 355–365. London.