Biological Control of Water Hyacinth Under Conditions ... - Springer Link

7 downloads 78 Views 1MB Size Report
Fort Lauderdale, Florida 33314, USA. GREG P. JUBINSKY. Florida Department of Environmental Protection. 3917 Commonwealth Blvd. Tallahassee, Florida ...
Biological Control of Water Hyacinth Under Conditions of Maintenance Management: Can Herbicides and Insects Be Integrated? TED D. CENTER* US Department of Agriculture Agricultural Research Service Aquatic Plant Control Research 3205 College Ave. Fort Lauderdale, Florida 33314, USA F. ALLEN DRAY, JR. University of Florida Institute of Food & Agricultural Sciences 3205 College Ave. Fort Lauderdale, Florida 33314, USA GREG P. JUBINSKY Florida Department of Environmental Protection 3917 Commonwealth Blvd. Tallahassee, Florida 32399, USA MICHAEL J. GRODOWITZ US Army Engineer Waterways Experiment Station 3909 Halls Ferry Rd. Vicksburg, Mississippi 39180, USA ABSTRACT / We hypothesized that repeated herbicidal (maintenance) control of water hyacinth infestations in Florida suppressed biological control agent populations, especially the weevils Neochetina eichhorniae and N. bruchi. We therefore sampled water hyacinth and weevil populations at 54 sites distributed statewide. Half were under

Water hyacinth, Eichhornia crassipes Mart. (Solms) originated in lowland tropical South America, probably in the Amazon basin (Barrett and Forno 1982). It was first introduced into the United States in the late 1800s, then into many countries throughout tropical, subtropical, and some warm-temperate regions (Gopal 1987). Its erect free-floating habit and showy flowers made it attractive for use in ornamental ponds and garden

KEY WORDS: Eichhornia crassipes; Neochetina eichhorniae; Neochetina bruchi; Phytophagy; Integrated control; Aquatic weeds *Author to whom correspondence should be addressed.

Environmental Management Vol. 23, No. 2, pp. 241–256

maintenance control, half were not treated with herbicides. General site conditions were assessed, demographic data were collected on weevil and plant populations, the reproductive condition of the weevils was determined, and plant nutrient and proximate composition of water hyacinth leaves were analyzed. Water hyacinth infestations under maintenance control were minimal when compared to unmanaged sites. Likewise, on a population basis, all weevil cohorts were much lower due to the paucity of plants. Plants at unmanaged sites, where weevil intensities were much higher, suffered high levels of stress and showed low growth potential. Lower percentages of the female weevils were reproductive at unmanaged sites when compared to managed sites, so densities of reproductives and immatures were similar at both site types. Reproductive status of the weevils improved with increased plant quality. Plant quality, in turn, declined as stresses arising from weevil feeding increased. Plant quality was positively correlated with plant growth potential and flower production. Thus, maintenance control improved plant nutritive quality thereby inducing reproductive vigor of the weevils, but ensuring plant regrowth and the need for future control. This suggests that biological and herbicidal controls should be integrated, using herbicides to maintain water hyacinth infestations below management thresholds but in a manner that conserves biological control agent populations. This approach would lead to improved plant nutritional quality that would, in turn, stimulate reproduction in biological control agent populations.

pools, which inevitably led to anthropogenic spread. The growth form is sympodial and individual rosettes produce clones that form extensive floating mats supporting canopies that, in mature stands, extend a meter or more above the water surface (Center and Spencer 1981). Problems caused by water hyacinth are well documented (Center 1994), as are its negative impacts on aquatic environments (Schmitz and others 1993). Water hyacinth remains the world’s most troublesome aquatic weed as it continues to invade waterbodies and wetlands in new regions (Julien and others 1996). The state of Florida manages a legislatively mandated program of ‘‘maintenance’’ water hyacinth control (Schmitz and others 1993). Maintenance control is defined by statute as ‘‘A method for the control of

r 1999 Springer-Verlag New York Inc.

242

T. D. Center and others

non-indigenous aquatic plants in which control technologies are utilized in a coordinated manner on a continuous basis in order to maintain the plant population at the lowest feasible level as determined by the Florida Department of Environmental Protection’’ [Florida Statues 369.22(2)(d)]. This policy is usually implemented only in waterbodies under state control that have public access, so private lakes and waterbodies lacking public access often are unmanaged. Maintenance of infestations at the ‘‘lowest feasible level’’ is generally accomplished by repeated applications of herbicides, most often 2,4-D (Schardt and Schmitz 1991). The desired result is the extirpation of plants from infested waterbodies, but this remains an elusive goal. This maintenance approach is effective (although expensive; Schardt and Schmitz 1991), producing sparse infestations comprised of a few isolated plants or small patches of plants. Without persistent retreatment, however, regrowth occurs from untreated plants and via seed germination. Three insects have been released in Florida as biological controls of water hyacinth: two weevils, Neochetina eichhorniae and N. bruchi (released in 1972 and 1974, respectively), and a pyralid moth, Sameodes albiguttalis (released in 1977) (see Center 1994 for a review of bionomics and impacts). Questions regarding the effectiveness of these insects remain unsettled, presumably because of the subtle and long-term nature of the impacts. More acute effects probably result when stress induced by these agents interacts with that of other factors (i.e., plant competition, phytopathogens, droughts, freezes, nutrients, etc.). For example, water hyacinth acreage reductions reported in Louisiana apparently resulted from the inability of the plants to recover quickly after annual winter diebacks. These reductions were clearly due to reduced annual growth (i.e., spring acreages remained relatively stable; fall acreages declined) which is credited to the presence of biological control agents (Center and others 1990, Cofrancesco and others 1985). Thus, this probably illustrates an interactive effect between plant-feeding insects and winter freezes. Nonetheless, some declines cannot be ascribed to any cause other than the biological control agents themselves (e.g., Center and Durden 1986, Haag and Center 1988, Bodle 1988). These introduced insects have clearly contributed to water hyacinth control, but the degree and extent of these contributions are uncertain and are probably masked by other management activities. Whether direct or indirect, the contribution of these insects to water hyacinth control has been largely ignored in management programs. Maintenance control is certainly not utilized in ‘‘a coordinated manner’’

with biological control. As a result, herbicides frequently are applied even where plant growth suppression by biological control agents has produced water hyacinth populations that are static or declining (Center personal observation). Our observations suggest that elimination of weed infestations induces the loss of the biological control agents. Plant regrowth then, when it occurs, produces healthy, vigorous plant populations that grow to problem proportions very quickly due to this release from herbivory. In fact, static, formerly stressed populations seem to convert to actively growing infestations of healthy plants during this regrowth phase. In order to assess the accuracy of these impressions, we compared populations of water hyacinth and of biological control agents in areas routinely subjected to maintenance control (managed sites) with areas that had not been treated with herbicides (unmanaged sites). We reasoned that, if our assumptions were correct, sites subjected to maintenance control would harbor healthier plants with smaller populations of biological control agents than sites not subjected to herbicide treatments. Our objective was not to compare the relative merits of the two control methods. Instead, we wished to determine if biological and herbicidal control could potentially be used in a more effective, coordinated, and integrated manner. The results showed our initial impression to be true, that maintenance control does indeed reduce populations of biological control agents and produce healthier plants. However, it also showed that in some ways the insect populations benefit from herbicidal control.

Methods and Materials Regional biologists with the Bureau of Aquatic Plant Management (Florida Department of Environmental Protection) provided lists of sites containing water hyacinth within each of the six state regions. Each list included sites that had regularly been subjected to maintenance control (managed sites) and sites not subjected to control measures (unmanaged sites). The only criteria that we imposed were that water hyacinth was known to occur at each site and that five of each site type be provided from each region. Although the biologists were unable to provide the number of sites requested, they did provide equal numbers of each site type (27 managed and 27 unmanaged). Consequently, our study consisted of 54 sites ranging from Lake Seminole in northern Florida to Big Cypress Swamp in the south (Figure 1). These sites were all sampled within an eight-week period between 12 September and 4

Water Hyacinth Management in Florida

Figure 1. Locations of water hyacinth populations at 54 waterbodies in Florida that were sampled during this study.

November 1989, 80% (43 sites) within the three-week period between 12 September and 5 October. Sites were first surveyed by airboat to acquire general impressions about the waterbody and the attendant aquatic weed situation, particularly with regard to water hyacinth. We then selected a water hyacinth infestation that we judged to be typical for the site. Sample locations were selected according to the classification criterion for the site. In other words, if a site was classified as a managed site but contained some areas that had not been subjected to recent management, then these were excluded from consideration and only areas containing plants that represented regrowth from herbicide treatments were sampled. This was important at only one site. A study area of approximately 0.2 ha (100 ft 3 200 ft) was visually delimited at each site and the LORAN-C coordinates for this area were recorded. Photographs and video tapes were obtained showing both the general area and the specific site. Coverage of water hyacinth and other aquatic plants within the delimited area was visually estimated by at least two observers. The colonizing stage of the water hyacinth infestation (phenostage; Figure 2) was recorded and the proportion of the mat composed of each plant type (Figure 3) was independently estimated by two observers. Further, the degree of insect-induced stress to the plants was estimated (also by two observers), both on a 5-point scale (0 5 none, 1 5 light, 2 5 moderate, 3 5 heavy, 4 5 severe) and proportionately (based upon representation

243

of each category within the delimited area). Comparative studies in experimental tanks (Center and Van 1989, Van and Center 1994) coupled with observations at field sites (Center and Dray 1992, Center and Durden 1986, Center and Spencer 1981), have shown that plants stressed by biological control agents exhibit characteristic symptoms. First, weevil feeding scars are abundant on the leaf blades and petioles of the leaves. Petioles often are girdled by feeding scars at the narrow junction of the blade, which results in curled leaf margins (see Figure 2 in Deloach and Cordo 1983). Larval tunneling in the petioles interferes with water transport, thereby inducing wilting and drying out of leaves. These tunnels usually become necrotic from secondary microbial infections (Charudattan 1986), causing affected petiole bases to appear dark and waterlogged. The crowns of severely biological controlstressed plants typically are submerged. Dead and dying (,50% green tissue, but still erect) leaves frequently outnumber living (.50% green) leaves. Undamaged leaves often consist of brittle, spindly petioles and tough leathery blades. Regrowth leaves produced on plants chronically damaged by biological controls have a similar appearance, even though such plants may exhibit few signs of feeding. Mats comprised of these plants are poorly consolidated with a great deal of open water among plants, which reveals large amounts of attached or suspended litter near the water surface. Despite the open space among them, these plants are producing few offsets. A 0.25 m2 (0.5 m 3 0.5 m) frame was dropped on the mat at each site and the enclosed mat segment was photographed. Canopy height (distance from the water surface to the modal height of the uppermost leaves) was measured within the frame. The number of independent shoot systems (mother 1 daughter plants) was counted, as was the total number of ramets (stoloniferous offsets bearing roots or root initials) within the frame. All plants were removed from the frame and placed in a large plastic bag. The bag was placed on ice in a polystyrene cooler and transported to the Fort Lauderdale laboratory for processing. The ice was replenished as needed until the sample could be processed. We wanted to examine the relationship between insect-induced stress and water hyacinth nutritional status, so we collected 25–30 leaves at each site for chemical analysis. The leaves were placed in a zip-lock plastic bag, which was first compressed to squeeze out the air and then sealed. Earlier studies indicated that Neochetina weevils preferentially feed upon the youngest leaf tissue available (Center and Wright 1991). We therefore selected the youngest exposed leaves. These

244

T. D. Center and others

Figure 2. Water hyacinth phenostages: (A) incipient infestation—a few, small plants bearing inflated leaf petioles (phenostage 1); (B) scattered infestation—scattered, open-canopied patches of small plants, some bearing inflated leaf petioles and others bearing transitional attenuating leaf petioles (phenostage 2); (C) coalescing infestation—larger mats with mostly closed canopies (phenostage 4); (D) mature healthy infestation—mostly tall plants bearing fully attenuated leaf petioles and coalesced into a large mat with a completely closed canopy (phenostage 5). Phenostage 3 (not shown) is transitional between 2 and 4, and often presents a domed appearance. Also shown: mature infestations (phenostage 5) in early (E) and advanced (F) stages of decline. Note the spindly, attenuated leaf petioles on these insect-stressed plants.

usually consisted of emerging central leaves that were still wrapped around the petiole of the next older leaf. A gap of sufficient width for a weevil to crawl in had to be present between the leaf blade and the petiole it was wrapped around for the leaf to be included in the sample. If such leaves were not available, the youngest unfurled leaves were collected. All leaves were immediately frozen and stored on dry ice to prevent tissue deterioration and concomitant changes in leaf chemistry. The leaf samples from each site were lyophilized, ground, and analyzed for nitrogen (N), phosphorus (P), potassium (K), magnesium (Mg), calcium (Ca), and manganese (Mn) after extraction with a sulfuric acid–hydrogen peroxide digest (Center and Van 1989). Proximate constituents (i.e., ash, nonvolatile lipids, crude fiber, and total reducing sugars) were analyzed following Allen and others (1974). We also wanted to relate plant nutrition with weevil

reproductive state, so a sample of weevils (separate from those contained in the plant samples) was collected for this purpose at each site. We attempted to collect at least 100 weevils per site, but in some areas this was impossible. We therefore limited collecting time to one person-hour. Weevils were immediately frozen and stored on dry ice to prevent deterioration of reproductive structures. They were then transported to our laboratory in Fort Lauderdale where each sample was sorted to species (N. eichhorniae or N. bruchi) and sex. Female N. eichhorniae (up to 30 from each site) were dissected for age grading using methods described by Grodowitz and others (1997). Processing of the 0.25-m2 plant samples consisted of first counting the number of living and dead leaves. The biomass was then apportioned into live above-water, live below-water, and dead components. Leaves were peeled back and the condition of the subtending axillary

Water Hyacinth Management in Florida

245

Figure 3. Water hyacinth were classified as one of four plant types: (A) short, healthy plants bearing inflated leaf petioles with laminas wider than long; (B) moderate-sized, healthy plants with inflated to attenuating leaf petioles and laminas about as long as wide; (C) tall, healthy plants with attenuated leaf petioles and laminas longer than wide; (D) small to moderate plants with tough, spindly leaf petioles and curled laminas. Inset shows leaf girdling common on heavily damaged type D plants.

(lateral) buds was assessed. Expanding buds were noted, as was destruction by insect damage. Any developing stoloniferous offset that had not yet produced root initials was counted as an expanding lateral bud. The number of inflorescences were counted, as was the number of individual blossoms per inflorescence. We also recorded whether the apical bud had been destroyed by insects. Weevils were picked by hand from the plants during processing and the biomass components were then placed into separate Berlese funnels to extract additional weevil adults and larvae. These data were used to estimate bioagent population densities (number per unit area). The 27 unmanaged sites were compared to the 27 managed sites to determine the relationships between management history, plant phenology, plant morphometry, and demography of water hyacinth weevils. We again stress, however, that these comparisons were not made to assess the relative efficacy of biological vs

herbicidal control. Site selection required the presence of water hyacinth plants. This necessarily excluded sites where complete control had been achieved (regardless of the means). Hence, the sites we sampled might only represent examples of incomplete biological control or incomplete herbicidal control. This limitation of the data should be borne in mind when interpreting the results. Data were analyzed using SigmaStat and SigmaPlot software packages (v. 2.01 and 3.02, respectively) (Fox and others 1995). Managed and unmanaged sites were compared using Student’s t tests as appropriate. Numbers of observations for each site type were 27, unless otherwise noted. Data that violated normality or homoscedasticity assumptions (as determined by assumptionchecking subroutines integrated into the software) were analyzed using the Mann-Whitney rank-sum (T ) test. All summary data for site types are presented as means for ease of interpretation, regardless of which test was performed. Frequency distribution data were examined

246

T. D. Center and others

pletely covered with water hyacinth this late in the growing season, plant coverage exceeded 50% at only half of the unmanaged sites. The type of plants comprising these mats was also strongly influenced by management history (x2 5 110.4, P , 0.0001; Figure 4B). For instance, plants characterized by inflated petioles (type A; see Figure 3) predominated at 44% of managed sites, but were absent from unmanaged sites (Figure 4B). Conversely, declining plants (type D; see Figure 3) predominated at 60% of unmanaged sites, but were rare at managed sites (Figure 4B). Thus, water hyacinth at managed sites were generally quite healthy, but were often declining at unmanaged sites. Estimates of insectinduced plant stress were consistent with these observations, being much lower at managed than unmanaged sites (1.4 vs 2.7; t 5 7.45, P , 0.001). Stress also was greatest on mature plants and mature mats (Figure 4). Water Hyacinth Demographics

Figure 4. Management history affected both (A) which mat phenostage was present, and (B) which plant types predominated at 54 waterbodies in Florida. Insect-induced plant stress increased in intensity on more mature infestations (A) and plants (B) as well.

via chi-square contingency tables. Relationships between variables were assessed by correlation analysis and multiple linear regression.

Results Site Characteristics Water hyacinth were present at nearly every site we visited (see Figure 1), the exceptions being Ocean Pond (Baker County) and Lake Iamonia (Leon County). Likewise, Neochetina spp. were found at every site where water hyacinth was present. Water hyacinth populations were in earlier stages of mat development at managed than at unmanaged sites (x2 5 132.5, P , 0.0001; Figures 2 and 4A). Consequently, plants covered less of the study area at a typical managed site than at a typical unmanaged site (15% vs 60%; t 5 6.5, P 5 0.0001). Although we might have expected to find virtually all unmanaged sites com-

Dynamics of the existing water hyacinth populations proved to be linked to management history as well, plants at managed sites being much more dynamic than those at unmanaged sites. In order to understand these dynamics, however, it is necessary to understand patterns of water hyacinth growth and population expansion. Weber (1950), Richards (1982), Watson (1984), and Watson and Cook (1987) provide detailed information on this. In summary, plant growth and population increase is the result of the differentiation of two meristem populations, apical and axillary. A single apical meristem occurs at the tip of each stem. This apical meristem can be either vegetative, producing leaves that have axillary buds, or reproductive, producing axillary flowers. If an inflorescence develops, the apical meristem terminates and leaf production ceases. In this event, the axillary bud in the leaf immediately below the inflorescence differentiates into a continuation (or renewal) shoot. The continuation shoot produces a new apical meristem that allows leaf production and vertical growth of the shoot to proceed. During outgrowth of axillary buds, if the bud is not forming a continuation shoot, then it produces a stolon consisting of a single internode that elongates, thus moving the axillary bud apex away from the parent rosette. The axillary bud apex then produces short internodes that grow vertically, producing a new rosette. An axillary meristem exists at the intersection of the stem and petiole of each leaf, so the number of axillary buds is equal to the number of leaves. Axillary buds that do not become stolon offshoots (ramets) or continuation shoots, remain undifferentiated and mature and senesce as buds. The number of independent plants (single unat-

Water Hyacinth Management in Florida

Table 1.

247

Water hyacinth (Eichhornia crassipes) demographics in Florida during autumn 1989a Site

Parameters compared Plant density (N/m2 ) Rosette density (apices/m2 ) Rosette intensity (N/plant) Apical meristems (% of differentiated meristems) Damaged apices (% of total apices) Axillary meristem density (N/m2 ) Expanding axillary meristem density (N/m2 ) Expanding axillary meristem density at sites with expanding axillariesb Damaged axillaries (% of total axillaries)b Inflorescence density (N/m2 ) Inflorescence density (at sites with inflorescences)c Inflorescences (% of differentiated meristems) Flowers density (blossoms/m2 ) Flowers density (at sites with inflorescences)c Flowers/inflorescence aData

Managed

Unmanaged

Statistics

36.3 91.0 2.7 70 25 545 15.8 18.6 9 36.0 42.3 4.7 288 338 8.8

54.8 63.0 1.2 83 33 434 8.1 16.9 8 5.9 16.0 1.4 58 156 10.8

t 5 4.55, P , 0.0001 T 5 565.5, P 5 0.0008 T 5 431.0, P , 0.0001 t 5 2.42, P 5 0.019 t 5 1.68, P 5 0.0999 T 5 628.0, P 5 0.0486 T 5 576.5, P 5 0.004 T 5 209.5, P 5 0.315 T 5 247.5, P 5 0.830 T 5 521.0, P , 0.0001 T 5 124.0, P 5 0.075 T 5 536.0, P , 0.001 T 5 534.0, P , 0.001 T 5 137, P 5 0.203 t 5 1.58, P 5 0.124

are presented as means of 27 managed (under maintenance control) and 27 unmanaged waterbodies unless otherwise noted.

bSample

sizes were 23 and 13 for managed and unmanaged sites, respectively.

cSample

sizes were 23 and 10 for managed and unmanaged sites, respectively.

Data were analyzed using Student’s t-tests (t) or the Mann-Whitney rank sum (T) test, as appropriate.

Figure 5. Management history influenced both (A) the allocation, and (B) the occurrence of water hyacinth meristems.

tached rosettes or interconnected groups of rosettes) present at the sites was clearly affected by management history. Unmanaged sites contained about 50% more plants (per unit area of water hyacinth mat) than managed sites (Table 1). Conversely, managed sites had 44% more apical meristems per unit area, because plants were composed of more than twice as many rosettes (Table 1). These data indicate that maintenance control, by keeping density low, promoted expansion of axillary buds into stolon offshoots (Geber and

others 1992). Only 17% of the differentiated meristems were apportioned as either expanding axillary buds or inflorescences at unmanaged sites as compared to 30% at managed sites (Table 1, Figure 5A). Interestingly, similar proportions of stem apices were destroyed by insects (mostly weevil larvae) at unmanaged and managed sites (Table 1), with as many as 67% and 62% destroyed at some sites. The density of axillary buds (based upon the number of leaves) was 26% higher at managed sites than at

248

T. D. Center and others

unmanaged sites (Table 1, Figure 5A). Numbers of expanding axillary buds (stolon offshoots that had not yet produced root initials) were low at both site types, although on average nearly twice as many were present per unit area at managed sites (Table 1). This difference was attributable to reduced incidence of expanding axillary buds at unmanaged sites: expanding axillary buds were present at most managed sites but at only about half of the unmanaged sites (Table 1, Figure 5B). The densities per unit area were about the same, if only sites where plants with expanding axillary buds were present are included in the analysis (Table 1). Interestingly, about equal proportions (11%) of the available meristems consisted of expanding axillary buds at both site types. Insect destruction of expanding axillary buds was generally low at both managed and unmanaged sites (Table 1), although the maximum for an unmanaged site was 50% destroyed compared to only 36% at managed sites. We assume that seed abundance increases in proportion to increasing inflorescence abundance. If this is true, then seed production (as evidenced by floral production) was enhanced by maintenance control. Samples from 85% of managed sites contained flowering plants as compared to only about 37% of unmanaged sites (x2 5 11.22, P , 0.001; Figure 5B). About 1.4% of the differentiated meristems were allocated to inflorescences at unmanaged sites as compared to 4.7% at managed sites (Table 1). As a result, the average number of inflorescences per unit area at managed sites exceeded unmanaged sites by sixfold (Table 1, Figure 5A). This was attributable to not only reduced incidence of flowering at unmanaged sites (Figure 5B) but also lower flowering frequency within a site. If only sites where flowers were present are used in the comparison, the average density of inflorescences at managed sites was still twice as large as at unmanaged sites (Table 1). We discerned no difference in the number of individual flowers per inflorescence between site types. In fact, the inflorescences contained somewhat more flowers at unmanaged than at managed sites (Table 1). Flower density (flowers/inflorescence 3 inflorescences/ m2 ) was fivefold greater at managed than unmanaged sites (Table 1). If data for sites lacking inflorescences are excluded, flower density was twice as great at managed as at unmanaged sites (Table 1). Although this difference is still large, it was not statistically significant, so the primary management effect was manifested by the presence or absence of inflorescences, and not so much by differences in the flower densities. Managed sites were clearly much more dynamic than

unmanaged sites, showing a greater capacity for clonal growth via ramet production as well as greater reproductive potential via sexual means. All but one managed site (96%) contained plants that possessed either flowers and/or expanding axillary buds, as compared to only 56% of the unmanaged sites (x2 5 16.90, P , 0.001; Figure 5B). In fact, most (70%) managed sites contained both flowers and expanding axillary buds, whereas only 30% of unmanaged sites contained both (x2 5 7.41, P 5 0.006; Figure 5B). Biomass Allocation Standing crop (total harvestable biomass) differed greatly between site types, being nearly two times greater per unit area of water hyacinth mat at unmanaged sites than at managed sites (Figure 6). The amount apportioned above water and below water was nearly the same at both site types with 51.2% and 52.2% above water (exclusive of inflorescences) and 48.3% and 45.4% below water at the two respective site types. The quantity of biomass allocated to production of inflorescences was quite different, however. In contrast to vegetative biomass, inflorescence biomass was over twice as great at managed as at unmanaged sites and proportionately was nearly five times greater (2.4% vs 0.5% of standing crop). This difference was due primarily to the fact that flowers were absent from 16 of the 27 unmanaged sites as compared to only 4 of the 27 managed sites (x2 5 9.61, P 5 0.002; Figure 5B). When this comparison was made including only sites where flowers were contained in the samples, the difference between site types was negligible. Nonetheless, the proportion of biomass allocated to inflorescences remained twice as high (2.7% at managed sites as compared to 1.3% at unmanaged sites) when sites without inflorescences were excluded. Plant Nutrients and Proximate Composition Plant tissue was analyzed to assess the condition of the plants as well as the quality of the food available to the insect populations, especially the Neochetina spp. weevils. The weevils preferentially feed on the youngest leaves available (Center and Wright 1991), so the central, unfurling leaves were chosen for analysis. These were barely exposed to insect attack and so were generally undamaged. The resultant analyses were therefore not influenced directly by insect feeding. Water content, lipids, nitrogen (and thus crude protein), and potassium were higher at managed than at unmanaged sites (Table 2). Calcium content in the leaf tissue was about 18% higher at unmanaged than at managed sites, but this was the only nutritional parameter that was

Water Hyacinth Management in Florida

249

Figure 6. Allocation of plant biomass at 27 managed and 27 unmanaged waterbodies in Florida. The floral biomass comparison included only sites that had inflorescences.

Table 2.

Influence of management history on phytochemistry of water hyacinth leaves (mean 6 SEM)a Sites Managed

Unmanaged

Statisticsb

89.7 6 0.335 14.8 6 0.226 5.60 6 0.283 40.1 6 1.57 1.85 6 0.095 3.54 6 0.137 3.955 6 0.146 2.335 6 0.084 0.043 6 0.007 0.801 6 0.052 0.385 6 0.142

82.6 6 0.927 14.4 6 0.254 4.54 6 0.150 41.5 6 1.32 1.58 6 0.085 2.82 6 0.095 3.633 6 0.183 2.757 6 0.150 0.048 6 0.008 0.707 6 0.057 0.405 6 0.138

T 5 941.0,P , 0.0001 T 5 677.0, P 5 0.261 T 5 565.0, P 5 0.002 T 5 779.5, P 5 0.528 T 5 632.0, P 5 0.057 T 5 531.0,P , 0.0001 T 5 618.5, P 5 0.033 T 5 877.0, P 5 0.020 T 5 744.0, P 5 0.986 T 5 688.5, P 5 0.204 T 5 774.0, P 5 0.592

Constituents Water content (%) Ash (% dw)c Lipids (% dw) Fiber (% dw) Total reducing sugars (% dw) Nitrogen (% dw) Potassium (% dw) Calcium (% dw) Manganese (% dw) Magnesium (% dw) Phosphorus (% dw) aSamples bData c%

consisted of emerging central leaves that were still wrapped around the petiole of the next older leaf.

were analyzed using the Mann-Whitney rank sum (T ) test.

dw 5 percent of leaf tissue dry weight.

higher at unmanaged sites (Table 2). Thus, the newly forming leaves on plants at managed sites were more succulent (higher water content) and of higher nutritional quality than at their unmanaged counterparts. Water Hyacinth Weevil Demographics Adult weevils were found at all but two of the managed sites (Lake Jackson in Leon County and Lake Seminole in Gadsen County). We tried to collect at least 100 adult weevils from each site so as to compare characteristics of the resident weevil populations, but this was often difficult at managed sites, where the average number collected was only 69 per sample. The

average number collected at unmanaged sites was 129 per sample. Neochetina eichhorniae comprised 87% of the weevils we collected; N. bruchi comprised the remaining 13%. Species composition varied among site types (x2 5 135.75, P , 0.001), with N. bruchi constituting a larger percentage of the overall weevil populations at managed sites (20%) than at unmanaged sites (9%) (see also Center and Dray 1992). Generally, based on the totals collected across sites, sex ratios did not vary between species (x2 5 0.003, P 5 0.956) nor between site types (x2 5 0.006, P 5 0.937) (overall, 56% males: 44% females). Average adult weevil density per unit area of plant

250

Table 3.

T. D. Center and others

Water hyacinth weevil (Neochetina spp.) demographics in Florida during autumn 1989a Sites

Parameters compared

Managed

Unmanaged

Statistics

14 41 6.9 3.9 4.6 5.5 25.6 59 82 4.9 28 6.5 15.4 22

47 49 21.2 2.3 2.4 12.1 35.4 33 42 5.2 38 4.7 11.1 55

T 5 948.5, P , 0.0001 T 5 820.5, P 5 0.180 T 5 930.5, P 5 0.001 t 5 3.40, P 5 0.001 t 5 5.69, P , 0.0001 T 5 889.5, P 5 0.011 T 5 785.5, P 5 0.462 t 5 3.25, P 5 0.002 t 5 4.81, P , 0.001 T 5 725.0, P 5 0.769 x2 5 28.7, P , 0.001 t 5 4.64, P , 0.001 t 5 1.47, P 5 0.149 x2 5 131.4, P , 0.001

Adult weevils/m2 Larvae/m2 Females/m2 Eggs/female Eggs/female (only females with eggs) Egg-bearing females/m2 Eggs/m2 Reproductive females (% all females) Reproductive females (% of egg-bearers) Reproductive females/m2 Females without eggs (% all females) Unovulated follicles/ovariole Nulliparous females (% all females) Females with degenerated ovaries (% all females) aData

are presented as means of 27 managed (under maintenance control) and 27 unmanaged waterbodies.

Data were analyzed using Student’s t tests, the Mann-Whitney rank sum (T) test, or x2 contingency tables, as appropriate.

coverage was over three times greater at unmanaged sites than at managed sites (Table 3). Surprisingly, larval densities were not so different (Table 3), and larvae were found at all but three managed sites. Larvae were even present at the two sites which lacked adults. It was not feasible to census eggs embedded in plant tissue, but we were able to approximate the quantities held in ovaries of female N. eichhorniae. Overall, oviducts of females at unmanaged sites contained fewer ovulated eggs than at managed sites (Table 3). If individuals that lacked eggs are excluded, this difference becomes even more pronounced (Table 3). The density of females at unmanaged sites was three times the density at managed sites (Table 3), of which 62% and 67% contained eggs. Thus, unmanaged sites harbored twice as many egg-bearing females and higher egg densities (Table 3). Not all of the egg-bearing females were reproductive (parous with functional ovaries); however, these two variables were correlated (r 5 0.792, P , 0.0001). A greater proportion of the female population was reproductive at managed sites and a higher percentage of the egg-bearing females were also reproductive (Table 3). As a result, the density of reproductive females was similar at both site types (Table 3). These data were consistent with larval population data in that differences between site types were negligible, despite the difference in adult populations. Fewer females at unmanaged sites held ovulated eggs in their ovaries (Table 3), so adult abundances were not indicative of the numbers of eggs available for oviposition. In addition, the average number of unovulated follicles per ovariole was greater at managed than at unmanaged sites (Table 3). This suggests that the reproductive potential of the females was lower at unmanaged sites, assuming that

the number of eggs produced is related to the number of developing follicles. These data indicate that weevil populations at managed sites were younger, with relatively fewer adults and a larger proportion of immatures than at unmanaged sites. In addition, a higher proportion of the females were nulliparous (Table 3), indicating that they were not yet reproductive (preovipositional). Conversely, populations at unmanaged sites were older with proportionately higher numbers of postreproductive adults (ovaries had degenerated in twice as many adult females at unmanaged sites; Table 3) and relatively fewer immatures. Despite this difference in age structure, the level of reproduction evident in the populations was nearly the same per unit area at both site types. Nonetheless, these data pertain on a unit area basis and when the fourfold greater coverage at unmanaged sites is factored in, total weevil populations per site were obviously much greater than at managed sites. Moreover, postreproductive females can probably rejuvenate their ovaries (Grodowitz and Center 1997) when conditions become favorable, so the reproductive potential could quickly become much greater at unmanaged sites. Despite the relatively even densities of reproductive females, eggs, and larvae at the two site types, the numbers per rosette (intensity) were quite different. This higher intensity was attributable to lower rosette densities at unmanaged sites. Adult weevils were over four times as abundant per rosette at unmanaged as at managed sites, females were three times greater, the number of egg-bearing females was also about three times greater, the number of eggs contained within ovaries was over twice as great, and larvae were about

Water Hyacinth Management in Florida

Figure 7. Management history influenced the composition of weevil populations.

63% more abundant (Figure 7). Hence, the individual rosettes were much more stressed by insect feeding at unmanaged sites, as reflected by higher levels of insectinduced plant stress (Figure 4). Plant–Insect Interactions Insect-induced stress increased in conjunction with increasing weevil intensity (r 5 0.401, P 5 0.0026), with estimates for stress typically higher at unmanaged than managed sites. The finding that rosettes exhibited more feeding damage (t 5 7.453, P , 0.001) and harbored more weevils (T 5 946.0, P 5 0.001) at unmanaged than managed sites corroborates the stress estimates. Further, weevil populations had lower numbers of follicles in the ovarioles (r 5 0.615, P 5 0.000001), fewer eggs in their oviducts (r 5 0.478, P 5 0.00034), and proportionately fewer reproductively active females (r 5 0.554, P 5 0.00002) at sites where insect-induced stress was more intense (i.e., unmanaged sites). This suggests a feedback mechanism whereby increasing biological control pressure causes plant quality to decline (see also Center and Van 1989). Females consuming these lower quality (i.e., insect-induced stressed) plants reduce or cease reproduction, thereby limiting the imposition of additional stresses on the plants. Examinations of plant nutritional composition support this hypothesis. Nitrogen (r 5 20.488, P 5 0.00018), lipids (r 5 20.445, P 5 0.00074), and total reducing sugars (r 5 20.248, P 5 0.071) all decreased as insectinduced stress increased. Additionally, calcium was more prominent in highly stressed plants (r 5 0.302, P 5 0.0265). Associations between plant quality and weevil reproduction were also quite strong. For instance, 47% of the variation in the number of follicles/ovariole in female weevils was attributable to lower lipid (t 5 2.786,

251

P 5 0.008) and nitrogen (t 5 3.741, P , 0.001) levels in plants at unmanaged versus managed sites (Figure 8). Likewise, low lipids (t 5 3.603, P , 0.001) combined with elevated calcium (t 5 3.765, P , 0.001) to account for 39% of the variation in the proportions of reproductively active females found at unmanaged versus managed sites (Figure 9). Water hyacinth exhibiting high insect-induced stress allocated fewer resources to productivity than plants exhibiting less stress. This is illustrated by declines in the number of meristems (r 5 0.572, P , 0.001) and in the number of expanding axillary buds (r 5 0.272, P 5 0.0465) found on plants under increased insectinduced stress. Further, both plant density (r 5 0.403, P 5 0.003) and the number of rosettes comprising each individual (r 5 0.608, P , 0.001) were greatest at low stress sites and least at high stress sites. These indicators of stress could be construed to merely represent the effects of plant crowding, i.e., being essentially a shading response that develops as the mats mature ( J. Richards, personal communication). We examined this possibility by regressing insectinduced stress against the number of rosettes per plant after accounting for variability due to crowding (as measured by leaf biomass). The results of this analysis (Figure 10) indicate that both insect-induced stress and crowding strongly influenced ramet production, but that the effect of stress from plant-feeding insects was paramount (Table 4). Similar results (not shown) were obtained when other dependent variables (such as total meristems per plant, expanding axillary buds/plant, and rosettes per square meter) were used as indicators of productivity.

Discussion Joyce (1977) stated that as originally envisioned, maintenance control was to be an integrated pest management (IPM) approach to water hyacinth control. As currently practiced, however, it is largely an herbicide treatment plan. Biological control agents, because they are not effective at achieving immediate reductions of plant outbreaks (Harley 1990), have been largely ignored as possible contributors to maintenance control programs (but see Haag 1986, Haag and Habeck 1991). In fact, some have argued that biological control insects contribute little to water hyacinth control (c.f. Schardt and Schmitz 1991, p. 27). This perception has probably arisen because managed sites recover quickly and the resultant plants show few if any symptoms of stress attributable to the bioagents. It might also arise from the fact that infestations at unmanaged sites are typically larger that those at managed sites. This

252

T. D. Center and others

Figure 8. Influence of leaf tissue nitrogen and lipid contents on the number of follicles per ovary for female Neochetina eichhorniae. Dark circles represent managed sites, white circles represent unmanaged sites. Drop lines under circles are for orientation only. The plane represents the regression: follicles/ ovary 5 0.0192 (0.432 p lipids) 1 (1.057 p nitrogen) (r2 5 0.428, F 5 18.35, P , 0.001).

Figure 9. Influence of leaf tissue lipid and calcium contents on the proportion of female Neochetina eichhorniae that were reproductive. Dark circles represent managed sites, white circles represent unmanaged sites. Drop lines under circles are for orientation only. The plane represents the regression: reproductives 5 49.626 1 (9.951 p lipids) 1 (20.314 p calcium) (r2 5 0.389, F 5 15.63, P , 0.001).

perspective fails to consider, however, that plant-feeding insects often produce sublethal effects that can be important to area-wide management of this aquatic weed. Such effects are often unappreciated when only a local perspective is considered.

The managed sites we examined indeed had smaller, earlier phenostage water hyacinth infestations than unmanaged sites. However, they also harbored much smaller weevil populations than unmanaged sites. The higher herbivore load at unmanaged sites was associ-

Water Hyacinth Management in Florida

253

Figure 10. Water hyacinth productivity (as measured by the number of rosettes/individual) was strongly influenced by insectinduced stress and plant crowding (as measured by leaf biomass). Dark circles represent managed sites, white circles represent unmanaged sites. Drop lines under circles are for orientation only.

Table 4. Relative influence of insect-induced stress and crowding on plant productivitya Regression and source

df

Mean square

Model Leaf biomass Insect-induced stress Biomass 3 stress Error

3 1 1 1 50

9.85 6.48 11.00 3.29 0.58

F(P)

r2

17.11 (0.0001) 11.26 (0.0019) 19.11 (0.0001) 5.72 (0.0210)

0.506

aCrowding

was estimated by leaf biomass; plant productivity was estimated by the number of rosettes/individual.

ated with water hyacinth populations that had lower densities of apical meristems, fewer rosettes per plant, fewer expanding axillary buds, and fewer inflorescences when compared to managed sites. Although crowding could elicit similar responses, our data indicate that insect-induced stress was primarily responsible for the loss of vigor and reduced reproduction of plants at unmanaged sites. Prior to the establishment of biological control agents, water hyacinth typically exhibited reciprocal growth phases in terms of ramet production or plant size (see Center and Spencer 1981). Canopy height declined as plants became smaller during the winter, resulting in available open water surface. The plants responded to this newly available space by prolific ramet production in spring. Then, during the summer phase, mats became crowded, plants grew larger, the canopy

became more continuous, and ramet production lessened. Ramet production resumed during the fall, however, when plant size began to decline. Therefore, one could reasonably argue that the decreased ramet production observed at unmanaged sites was merely a manifestation of this phenomenon. However, as herbivoreinduced stress progresses at unmanaged sites, plants become smaller and the canopy opens up without reciprocal ramet production (see Figures 2E and 2F), and this was borne out in the analysis shown in Figure 10. This has strong implications for water hyacinth management programs. The long-term, chronic stress associated with weevil infestations clearly lessens plant vigor, and reduces vegetative propagation and sexual reproduction. This, in turn, suggests that water hyacinth colonies cannot recover from other control measures as quickly in the presence of biological controls as in their absence. Slower recovery times mean that retreatment could be required less often at such sites, resulting in a cost savings for resource managers. Further, less vigorous plants are more susceptible to other stress factors than are healthy plants. Thus, native pathogens (Charudattan 1986) and cold temperatures (Cilliers and Hill 1996) have a pronounced impact on highly stressed plant populations. This increased susceptibility could also be used to enhance herbicide effectiveness and may permit the use of lower herbicide concentrations during retreatments.

254

T. D. Center and others

Figure 11. Comparison of water hyacinth standing crop at 27 managed and 27 unmanaged Florida waterbodies in 1989 with historical records. Data from this study are presented as means 6 99% confidence limits. Historic data were adapted from: (1) Boyd and Scarsbrook (1975), (2 and 3) Knipling and others (1970), (4) Penfound and Earle (1948), (5) Wooten and Dodd (1976), and (6) Center and Spencer (1981).

Our estimates of insect-induced plant stress corresponded well with the actual herbivore loads we measured at the study sites. Water hyacinth under insectinduced stress typically grow more slowly and achieve smaller sizes than do unstressed plants (Center 1994, Grodowitz and others 1991). Further, weevil feeding accelerates leaf senescence and speeds leaf mortality (Center 1987, Center and Van 1989, Van and Center 1994). Several authors have pointed out that these biological control agent-induced stresses accumulate slowly (Center and Van 1989, Harley 1990, Van and Center 1994). This observation helps explain our finding that unmanaged sites, i.e., sites at which such stresses had accumulated, harbored less vigorous (later phenostage) water hyacinth populations than managed sites. Maintenance control eliminates plants before biological control stresses have accumulated and become apparent. Loss of the plants also results in severe reductions in biological control agent populations, so that few weevils are present to induce stress as the plant populations regrow. Consequently, the biological control agents appear ineffective at intensively managed sites. High levels of stress were also closely correlated with reduced plant nutritional quality at unmanaged sites. Center and Van (1989) showed that weevil feeding causes reductions in water hyacinth leaf nitrogen levels. Perhaps not coincidentally, plants at our managed sites showed nitrogen concentrations (Table 2) nearly identical to the ‘‘no-weevil’’ treatments (3.5%) of Center and Van (1989), whereas plants at unmanaged sites pre-

sented concentrations similar to the ‘‘with-weevil’’ treatments (2.8%). Low-quality plants cause the weevils to cease reproducing (Center 1994). This, in turn, promotes the establishment of persistent plant populations that are smaller and less robust than historic (prebiological control) levels, but not as small as at managed sites (Figure 11). Unlike at managed sites, however, a large (albeit reproductively inactive) weevil population persists at unmanaged sites. These insects can become reproductive again (Buckingham and Passoa 1985) should the plant population rebound and nutritional quality improve. This reflects the ‘‘lying in wait’’ strategy described by Murdoch and others (1985), in which the biological control agents prevent outbreak populations from occurring rather than eliminate outbreaks after they have occurred (the search-and-destroy strategy). This process is difficult to observe, which, perhaps, explains why biological control is often difficult to evaluate. Paradoxically, managed sites have better quality plants, so the weevils are more reproductive (although limited in numbers) at such sites. This factor suggests strongly that maintenance control practices might be used to enhance biological control efforts. Two approaches should be used to accomplish this integration. First, additional biological control agents must be developed for use against water hyacinth. These will add to the herbivore-induced stresses already experienced by water hyacinth at unmanaged sites and perhaps eliminate plants from areas where response time is not critical. Priority should be given to agents that multiply

Water Hyacinth Management in Florida

quickly and/or those that fare better on low-nutrient plants. Second, strategies should be developed for integrating biological controls into existing maintenance control protocols. For example, spray crews might be directed to leave pockets of water hyacinth untreated instead of spraying all accessible plants. These would serve as refugia for biological control agents. Biological control stresses would accumulate in such refugia, as they had at unmanaged sites in our study, and thereby slow or halt water hyacinth mat expansion. These refugia also would provide a reserve of insects that would readily disperse onto regrowing mats. Such a protocol could reduce the frequency at which waterbodies require retreatment, thus providing an economic savings to resource managers.

Conclusion This examination of water hyacinth populations that have historically been left undisturbed confirmed many of the benefits previously reported to result from biological control. Sites subjected to maintenance control typically did not share in these benefits, inasmuch as they harbored smaller weevil populations. Paradoxically, accumulated stress from biological control agents at unmanaged sites rendered individual plants less suitable for weevil populations, whereas routine maintenance control resulted in plant populations that enhanced weevil reproduction. These results argue strongly for investigation of a more integrated approach that effectively exploits benefits of both methods while minimizing the negative aspects of each. Whatever these integrated approaches finally entail, it is likely that our understanding of what constitutes the ‘‘lowest feasible levels of water hyacinth’’ will need to be adjusted to accommodate biological control agents if we are to achieve long-term, economical water hyacinth control.

Acknowledgments Willey Durden, Donna Niehaus, Vinda Maharajh, and Ann Jones assisted with various aspects of this research. Comments by Drs. Jennifer Richards, Maxine Watson, Nan-Yao Su, Tom Weisling, Robin Giblin-Davis, and two anonymous reviewers improved the manuscript. This research was supported, in part, through Agreement 58-43YK-9-0033 between the Florida Department of Environmental Protection and the United States Department of Agriculture, Agricultural Research Service.

255

Literature Cited Allen, S. E., H. M. Grimshaw, J. A. Parkinson, and C. Quarmby. 1974. Chemical analysis of ecological materials. Blackwell Publ., Oxford, UK. Barrett, S. C. H., and I. W. Forno. 1982. Style morph distribution in New World populations of Eichhornia crassipes (Mart.) Solms-Laubach (water hyacinth). Aquatic Botany 13:299–306. Bodle, M. 1988. Water hyacinth biocontrol: A case history. Aquatics 10(3):24, 26. Boyd, C. E., and E. Scarsbrook. 1975. Influence of nutrient additions and initial density of plants on production of water hyacinth Eichhornia crassipes. Aquatic Botany 1:253–261. Buckingham, G. R., and S. Passoa. 1985. Flight muscle and egg development in water hyacinth weevils. Pages 497–510 in E. S. Delfosse (ed.), Proceedings, VI International Symposium on Biological Control of Weeds, 19–25 August 1984, Vancouver, British Columbia. Agriculture Canada, Ottawa. Center, T. D. 1987. Insects, mites, and pathogens as agents of water hyacinth [Eichhornia crassipes (Mart.) Solms.] leaf and ramet mortality. Lake and Reservoir Management 3:285–293. Center, T. D. 1994. Biological control of weeds: Water hyacinth and waterlettuce. Pages 481–521 in D. Rosen, F. D. Bennett, and J. L. Capinera (eds.), Pest management in the subtropics: Biological control—a Florida perspective. Intercept Publishing Company, Andover, UK, 737 pp. Center, T. D., and F. A. Dray. 1992. Associations between water hyacinth weevils (Neochetina eichhorniae and N. bruchi) and phenological stages of Eichhornia crassipes in southern Florida. Florida Entomologist 75:196–211. Center, T. D., and W. C. Durden. 1986. Variation in water hyacinth/weevil interactions resulting from temporal differences in weed control efforts. Journal of Aquatic Plant Management 24:28–38. Center, T. D., and N. R. Spencer. 1981. The phenology and growth of water hyacinth [Eichhornia crassipes (Mart.) Solms] in a eutrophic north-central Florida lake. Aquatic Botany 10(1):1–32. Center, T. D., and T. K. Van. 1989. Alteration of water hyacinth [Eichhornia crassipes (Mart.) Solms] leaf dynamics and phytochemistry by insect damage and plant density. Aquatic Botany 35:181–195. Center, T. D., and A. D. Wright. 1991. Age and phytochemical composition of waterhyacinth (Pontederiaceae) leaves determine their acceptability to Neochetina eichhorniae (Coleoptera: Curculionidae). Environmental Entomology 20:323–334. Center, T. D., A. F. Cofrancesco, and J. K. Balciunas. 1990. Biological control of aquatic and wetland weeds in the southeastern United States. Pages 239–262 in E. S. Delfosse (ed.), Proceedings of the VII International Symposium on Biological Control of Weeds, 6–11 March 1988, Rome, Italy. Charudattan, R. 1986. Integrated control of water hyacinth (Eichhornia crassipes) with a pathogen, insects, and herbicides. Weed Science 34(Supplement 1):26–30. Cilliers, C., and M. Hill. 1996. Mortality of Eichhornia crassipes (water hyacinth) in winter following summer stress by biological control agents. Pages 507 in V. C. Moran and J. H.

256

T. D. Center and others

Hoffman (eds.), Proceedings of the IX International Symposium on Biological Control of Weeds, 21–26 January 1996, Stellenbosch, South Africa. University of Capetown, Capetown, South Africa. Cofrancesco, A. F., R. M. Stewart, and D. R. Sanders. 1985. The impact of Neochetina eichhorniae (Coleoptera: Curculionidae) on water hyacinth in Louisiana. Pages 525–536 in E. S. Delfosse (ed.), Proceedings of the VI International Symposium on Biological Control of Weeds, 19–25 August 1984, Vancouver, British Columbia. Agriculture Canada, Ottawa. DeLoach, C. J., and H. A. Cordo. 1983. Control of waterhyacinth by Neochetina bruchi (Coleoptera: Curculionidae) in Argentina. Environmental Entomology 12:19–23. Fox, E., K. Shotton, and C. Ulrich. 1995. SigmaStat statistical software for Windows user’s manual. Jandel Scientific Corporation, San Rafael, Calif. Geber, M. A., M. A. Watson, and R. Furnish. 1992. Genetic differences in clonal demography in Eichhornia crassipes. Journal of Ecology 80:329–341. Gopal, B. 1987. Water hyacinth. Elsevier, New York, 471 pp. Grodowitz, M. J., R. M. Stewart, and A. F. Cofrancesco. 1991. Population dynamics of water hyacinth and the biological control agent Neochetina eichhorniae (Coleoptera: Curculionidae) at a southeast Texas location. Environmental Entomology 20:652–660. Grodowitz, M. J., T. D. Center, and J. F. Freeman. 1997. A physiological age-grading system for Neochetina eichhorniae Warner (Coleoptera: Curculionidae), a biological control agent of water hyacinth, Eichhornia crassipes (Mart.) Solms. Biological Control 9:15–23. Haag, K. H. 1986. Effective control of water hyacinth using Neochetina and limited herbicide application. Journal of Aquatic Plant Management 24:70–75. Haag, K. H., and T. D. Center. 1988. Successful biocontrol of water hyacinth: A documented example. Aquatics 10:19–22. Haag, K. H., and D. H. Habeck. 1991. Enhanced biological control of water hyacinth following limited herbicide application. Journal of Aquatic Plant Management 29:24–28. Harley, K. L. S. 1990. The role of biological control in the management of water hyacinth, Eichhornia crassipes. Biocontrol News and Information 11(1):11–22. Joyce, J. 1977. Selective maintenance control plan St. Johns River, Florida. Pages 45–48 in Proceedings, Research Planning Conference on the Aquatic Plant Control Program, 19–22 October 1976, Atlantic Beach, Florida. Miscellaneous Paper A-77-3, US Army Engineers Waterways Experiment Station, Vicksburg, Mississippi.

Julien, M. H., K. L. S. Harley, A. D. Wright, C. Cilliers, M. Morris, T. Center, and A. Cofrancesco. 1996. International cooperation and linkages in the management of water hyacinth with emphasis on biological control. Pages 273– 282 in V. C. Moran and J. H. Hoffman (eds.). Proceedings of the IX International Symposium on Biological Control of Weeds, 21–26 January 1996, Stellenbosch, South Africa. University of Capetown, Capetown, South Africa. Knipling, E. B., S. H. West, and W. T. Haller. 1970. Growth characteristics, yield potential, and nutritive content of water hyacinths. Soil & Crop Science Society of Florida Proceedings 30:51–63. Murdoch, W. W., J. Chesson, and P. L. Chesson. 1985. Biological control in theory and practice. The American Naturalist 125:344–366. Penfound, W. T., and T. T. Earle. 1948. The biology of the water hyacinth. Ecological Monographs 18:448–472. Richards, J. H. 1982. Developmental potential of axillary buds of water hyacinth, Eichhornia crassipes Solms. (Pontederiaceae). American Journal of Botany 69:615–622. Schardt, J. D., and D. C. Schmitz. 1991. 1990 Florida aquatic plant survey. Technical report 91-CGA. Florida Department of Natural Resources, Tallahassee, Florida, 89 pp. Schmitz, D. C., J. D. Schardt, A. J. Leslie, F. A. Dray, J. A. Osborne, and B. V. Nelson. 1993. The ecological impact and management history of three invasive alien aquatic plant species in Florida. Pages 173–194 in B. N. McKnight (ed.), Biological pollution. The control and impact of invasive exotic species. Indiana Academy of Science, Indianapolis, Indiana, 261 pp. Van, T. K., and T. D. Center. 1994. Effect of paclobutrazol and water hyacinth weevil (Neochetina eichhorniae) on plant growth and leaf dynamics of water hyacinth (Eichhornia crassipes). Weed Science 42:665–6762. Watson, M. A. 1984. Developmental constraints: Effect on population growth and patterns of resource allocation in a clonal plant. The American Naturalist 123:411–426. Watson, M. A., and G. S. Cook. 1987. Demographic and developmental differences among clones of water hyacinth. Journal of Ecology 75:439–457. Weber, H. 1950. Morphologische und anatomische studien u¨ber Eichhornia crassipes (Mart.) Solms. Akademie der Wissenhaften und der Literature Mainz, Adhandllungen der Mathematisch—Naturwissenschaftlichen Klasse 6:135–161. Wooten, J. W., and J. D. Dodd. 1976. Growth of water hyacinths in treated sewage effluent. Economic Botany 30:29–37.