Watercress allelochemical defends high-nitrogen foliage against consumption: effects on freshwater invertebrate herbivores.
Although relatively few freshwater invertebrates appear to feed on live aquatic macrophytes, when these plants senesce or die, they are rapidly consumed and contribute substantially to the food base of aquatic systems (Newman 1991). Several authors have suggested that this low utilization of live plants may be due to their possession of chemical feeding deterrents which inhibit their consumption by aquatic invertebrates (e.g., Otto and Svensson 1981, Ostrofsky and Zettler 1986, Kerfoot 1989, Suren 1989; see reviews by Lodge 1991 and Newman 1991). However, there are few freshwater studies that link feeding response to specific chemicals.
One of the better studied examples of chemical defense in freshwater macrophytes is watercress (Nasturtium officinale = Rorippa nasturtium-aquaticum), an aquatic crucifer. Watercress possesses the glucosinolate-myrosinase system (Newman et al. 1992), which is regarded as a classic example of chemical defense in terrestrial crucifers (Feeny 1977, Louda and Mole 1991). The defensive agent, 2-phenylethyl isothiocyanate, is produced after enzymatic hydrolysis of 2-phenylethyl glucosinolate by myrosinase (Fenwick et al. 1989, Newman et al. 1992). The myrosinase is compartmentalized separately in myrosin cells (Thangstad et al. 1991), and is released upon tissue damage. The isothiocyanate is quite toxic to the aquatic amphipod Gammarus pseudolimnaeus (Newman et al. 1990).
Feeding trials showed that senescent watercress was greatly preferred to fresh green watercress (Newman et al. 1990, 1992). The four taxa of stream shredders investigated (two limnephilid caddisflies, a physid snail, and G. pseudolimnaeus) do not consume much fresh green watercress that is high in glucosinolate ([greater than]5 mg/g), but readily consume senescent yellowed watercress that is low in glucosinolate ([less than]1.5 mg/g), even though green watercress is higher in nitrogen than senescent watercress (5-7 vs. 3-4%). When the defensive system was blocked by simply heating leaf tissue to deactivate myrosinase and greatly reduce isothiocyanate release (Newman et al. 1992), we saw some striking preference reversals. Heating resulted in a shift in preference by the four taxa of shredders from the low-glucosinolate, but low-nitrogen, unheated senescent tissue to the high-nitrogen green tissue (Newman et al. 1992), suggesting that deactivation of the myrosinase eliminated the feeding deterrent (i.e., isothiocyanate production). Addition of myrosinase to the heated tissue resulted in reduced consumption of the high-glucosinolate green tissue for the two taxa tested (Newman et al. 1992). It thus appears that stream invertebrates are faced with the choice of a higher quality, but defended tissue (green leaves), vs. a lower quality but less defended tissue (senescent watercress). These results suggest a general explanation for the low consumption of live watercress but high consumption upon senescence and death (Newman et al. 1990, 1992, Newman 1991).
Although crucifers have been studied extensively for terrestrial plant-herbivore interactions (see Louda and Mole 1991), many of these studies have focused on adapted specialist herbivores rather than on generalist herbivores such as stream invertebrates. In addition, most studies of specialist herbivores have focused on responses to either glucosinolates (e.g., Erickson and Feeny 1974, Blau et al. 1978, Bodnaryk 1991) or isothiocyanates (e.g., Lichtenstein et al. 1964, Wadleigh and Yu 1988) and have not attempted to distinguish between these two components. Plant choice by herbivores may be restricted at two levels: preingestive palatability (deterrence) or postingestive physiological and toxicological responses (Louda and Mole 1991). Distinguishing these components is important to our understanding of mechanisms of deterrence across taxa and the factors that influence food choice by herbivores faced with "defended" plants.
The aim of this study was to determine the long-term consequences of watercress tissue consumption on the growth and survival of stream invertebrates. We chose taxa that commonly occur in North American watercress beds and that live and feed on senescent watercress and associated plant material: the amphipod Gammarus pseudolimnaeus, the limnephilid caddisflies Hesperophylax designatus, Limnephilus sp., and Pycnopsyche sp., and the physid snail Physella gyrina. Specifically, we first determined the no-choice consumption rates of fresh and senescent watercress for these invertebrates. We then determined the effects of heating (myrosinase deactivation) on no-choice consumption, and the effects of consumption of fresh and heated, green and senescent watercress on growth and survival of the amphipod G. pseudolimnaeus and the caddisflies H. designatus and Limnephilus sp. We related these responses to tissue nitrogen and glucosinolate levels and then compared the longer term consumption, growth, and survival responses to short-term preferences of these species.
Test protocol and consumption estimates
All tests were carried out with field-collected organisms acclimated to laboratory conditions. Watercress and invertebrates were collected primarily from two alkaline spring brooks: Stream 1 (Dakota County, Minnesota Township 27N, Range 23W, Section 18) and Valley Creek (Washington County, Minnesota T28N, R20W, S14). Invertebrates were maintained in an aerated tank at 12 [degrees] C with natural substrate and leaf litter. Experiments were carried out in a dark Precision Scientific 805 environmental chamber at 12 [degrees] C. Test invertebrates were placed into plastic petri dishes (one dish per replicate) that contained 40 mL filtered stream water (with [[O.sub.2]] at or near saturation) and starved for a specified period of time. Petri dishes (and organisms within them) were then assigned to treatments and organisms placed into the test dishes, using a stratified random design to block for order or position effects.
Leaf material was obtained either from fresh stream-collected plants or from stream-collected plants maintained under lighted conditions (natural spectrum grow lights on a 16L:8D photocycle) in a laboratory stock tank with stream sediment. Leaves for the experiments were harvested from the plants immediately prior to addition to an experiment, rinsed in well water, and blot dried; leaflets were then cut into [approximately equal to]100-[mm.sup.2] disks with a cork borer. Fully yellowed (senescent) leaves were harvested from the stock tank or developed by placing harvested green leaves in the dark at 12 [degrees] C for [approximately equal to]1 wk. Only firm and structurally intact senescent leaves were used. Yellow leaf disks were processed in the same manner as fresh green disks. Although we have not done a formal comparison of leaves that had senesced naturally vs. artificially, nor of field-collected vs. laboratory-reared plants, we noticed no consistent difference in chemical content of these beyond possible seasonal differences. Because chemical analyses were performed with a subset of the leaves used in each experiment and thus represent material to which the invertebrates were exposed, potential seasonal differences will not have affected our interpretation regarding our experimental responses. Heated leaf disks were produced by placing randomly selected leaf disks (green or yellowed) into filtered stream water at 70 [degrees] C for 2 min, and then transferring them to filtered stream water to cool at [approximately equal to]20 [degrees] C (Newman et al. 1992). Leaf disks were then randomly assigned to treatment dishes, where they were pinned (insect pins) to numbered silicon knobs to hold disks in place (Newman et al. 1990). Position and knob number were noted for each disk to ensure that disks that were completely consumed were recorded for the proper leaf type.
Consumption estimates were determined by measuring changes in leaf disk area. Disk areas were determined by digitizer and a set of control disks of each type was run with each experiment to estimate initial areas (see Newman et al. 1990). Because heating resulted in a shrinkage of tissues, we conducted an experiment to determine the relationship between area and mass and the effects of heating. This experiment indicated that heating did shrink the leaves a moderate amount (20% for green and 11% for yellow); the reduction in leaf area was highly correlated with reduction in leaf wet mass. Heating thus was shown to result in a loss in mass proportional to the loss in area. There were, however, significant differences in percent water among the four leaf types. Heated green leaves were 89% water, fresh green leaves were 93% water, and both fresh and heated yellowed leaves were 96% water. Soak time (e.g., 24 vs. 48 h), however, did not affect mass (wet or dry) or area. Thus, area provided an accurate measure of leaf wet mass across leaf types. Because dry matter content varied by leaf type, all three responses were analyzed.
Chemical analyses follow the methods presented in Newman et al. (1990, 1992). Pooled subsamples of leaves were collected from most experiments for glucosinolate and carbon: nitrogen (C : N) analyses. Samples of leaf parts ([approximately equal to]1 g wet mass) were either fixed in 70% methanol for later hydrolysis with excess myrosinase, or frozen and extracted with methylene chloride (see Newman et al. 1990). Glucosinolate content, as determined from isothiocyanate-yielding-glucosinolate (IYG), was analyzed by gas chromatography for phenylethyl isothiocyanate (Blua and Hanscom 1986). In this method, leaf glucosinolate content is determined from isothiocyanate content. Previous work showed that isothiocyanate yield from the frozen tissue (hydrolyzed with intact plant myrosinase) compared well with yield based on hydrolysis by the addition of excess myrosinase (Newman et al. 1990) and that 2-phenylethyl isothiocyanate is the primary ([greater than]90%) glucosinolate hydrolysis product in watercress (Cole 1976, Freeman and Mossadeghi 1973; Z. Hanscom, personal observation). The majority of our measures were made with frozen tissue. Previous experiments have also shown that heating deactivates the myrosinase and reduces isothiocyanate production (90-97% reduction) in heated green tissue and that addition of myrosinase to these heated tissues results in levels of isothiocyanate release similar to intact tissue (Newman et al. 1992), which indicates that the glucosinolates remain intact in the heated tissues.
Additional subsamples of leaf disks were dried and ground for nitrogen analyses. Nitrogen was determined for the dried subsamples with a LECO Model 800 CHN analyzer (LECO, St. Joseph, Michigan, USA).
Short-term consumption tests
No-choice consumption estimates for G. pseudolimnaeus, Limnephilus sp., H. designatus, and Physella gyrina were conducted over 96 h, with time intervals of 24, 48, 72, and 96 h. Each experiment consisted of five independent replicates of each leaf type (fresh green or senescent yellowed) and consumption interval (24, 48, 72, or 96 h). Two leaf disks were placed into each replicate container, which contained two caddisflies or five snails or amphipods. Disks were replaced when substantially consumed to ensure that food was always available. At the end of the specified time period, the disks were removed and measured and the test organisms were weighed. These experiments resulted in five independent estimates of consumption of green and yellow leaves over four time periods for each of the five species (20 observations per species). Consumption rates for each leaf type were determined by regression of leaf area lost (consumed) on time; the slope provides an estimate of consumption rate (in square millimetres per hour). A 24-h no-choice consumption test was conducted with the shredding caddisfly Pycnopsyche sp. collected from Hudson Mill Park, Michigan. Single caddisflies were randomly assigned to six individual petri dishes; three dishes contained fresh green disks and three contained senescent yellow disks. Consumption was estimated as leaf area lost over 24 h.
Long-term consumption tests
Long-term consumption estimates for G. pseudolimnaeus, Limnephilus sp., and H. designatus were conducted over periods of 30-85 d using fresh green, senescent yellow, heated green, and heated yellow watercress leaves. The experimental protocol was similar to the 96-h consumption tests with the following modifications. Generally, five replicate organisms were used for each leaf type. Each petri dish contained only one organism, which was weighed (wet mass) after 24 h starvation immediately prior to the beginning of the experiment. Disks were replaced when consumed, or after 3 d if not consumed, to ensure that food was always available and that it was in a relatively unaltered (fresh) state. At the end of [approximately equal to]30 d, or upon death, the organisms were starved for 24 h and reweighed. Survival was noted and length of life recorded. In several experiments, dead organisms were replaced with other organisms, providing an additional replicate albeit for a shorter time interval. Cumulative consumption of watercress was estimated as the sum of the lost tissue area, and daily consumption was determined by dividing cumulative consumption by the time alive (in days). Growth of invertebrates was determined by the change in mass over time alive (instantaneous growth = ln ([m.sub.t]/[m.sub.0])/t; m = mass in milligrams, t = time in days). In several of the experiments, the experiments were continued and the organisms were fed for another [approximately equal to]30 d. Growth and consumption were determined for both intervals and the total time period. At the end of each experiment, all organisms were dried and weighed after wet masses were obtained. These experiments resulted in [greater than or equal to]5 independent estimates of consumption and growth for each species on each leaf type over time and also provided estimates of survival.
To examine consumption, growth, and survival in a physiologically relevant manner for this experiment, we converted the leaf area estimates into wet and dry mass consumed, using the conversions mentioned earlier (see Methods: Test protocol . . .). Gross conversion efficiency was calculated as absolute growth divided by consumption (both as wet or dry mass, in milligrams). We used the nitrogen and glucosinolate determinations for each leaf type and experiment to estimate the masses of nitrogen and glucosinolate consumed by each animal, both on a daily basis (in milligrams per gram per day) and as a total for each individual over the entire experiment (in milligrams per gram mean body mass). It should be noted that with fresh (non-heated) leaf tissue, glucosinolates may be hydrolyzed to isothiocyanate during consumption and thus our measures of glucosinolate consumption include an unknown proportion of 2-phenylethyl isothiocyanate.
Statistical analyses were done with SYSTAT 5.1 (Wilkinson 1989), and unless stated otherwise, significance was set at [Alpha] = 0.05. Multiple comparisons of means were made using Tukey's hsd because we were interested in comparing [greater than]1 mean. Because we were not interested in an overall difference among all treatments we followed the suggestion of Wilkinson (1989) and did not conduct one-way ANOVAs. For all statistical tests, data were transformed when necessary; however, untransformed means and [+ or -]22 standard errors are presented for ease of interpretation. For mean survival times, transformations were not successful and these data were analyzed with a Kruskal-Wallis one-way ANOVA.
Short-term no-choice consumption tests without heat-treated tissues
In all cases, relative consumption of senescent tissue in 96-h tests was much greater than that of fresh green tissue [ILLUSTRATION FOR FIGURE 1 OMITTED]. Analysis of variance, with time as a covariate, showed significant time x tissue-type interactions for all taxa (all P [less than or equal to] 0.02). In the absence of choice, the amphipod Gammarus pseudolimnaeus, the caddisflies Hesperophylax designatus and Limnephilus sp., and the snail Physella gyrina all consumed 5-18x more senescent yellowed watercress than green watercress [ILLUSTRATION FOR FIGURE 1 OMITTED]. Consumption of green leaves was low [measured as leaf wet mass] (G. pseudolimnaeus: 0.4 mg[center dot][individual.sup.-1][center dot][d.sup.-1]; H. designatus: 0.7 mg[center dot][individual.sup.-1][center dot][d.sup.-1]; Limnephilus: 2.3 mg[center dot][individual.sup.-1][center dot][d.sup.-1]; P. gyrina: 0.1 mg[center dot][individual.sup.-1][center dot][d.sup.-1]) but consistent, suggesting a regular sampling of green leaves but minimal accumulative consumption. Consumption of senescent yellowed tissues by G. pseudolimnaeus and caddisflies was substantially higher [as wet mass] and nearly constant over time (G. pseudolimnaeus: 3.4 mg[center dot][individual.sup.-1][center dot][d.sup.-1]; H. designatus: 11.7 mg[center dot][individual.sup.-1][center dot][d.sup.-1]; Limnephilus: 8.9 mg[center dot][individual.sup.-1][center dot][d.sup.-1]). Consumption of senescent tissue by snails was initially low and similar to green-tissue consumption, then rose to more substantial rates (1.8 mg[center dot][individual.sup.-1][center dot][d.sup.-1]). The reasons for the initial lag were unclear, probably due to slower movements and food finding by snails relative to the other taxa; similar observations were made by Kerfoot with Physella gyrina in Michigan (Kerfoot and Newman, unpublished manuscript). The caddisfly Pycnopsyche also consumed significantly more wet tissue of senescent watercress (147.2 [mm.sup.2][center dot][individual.sup.-1][center dot][d.sup.-1] or 28 mg[center dot][individual.sup.-1][center dot][d.sup.-1]) than of green watercress (68.5 [mm.sup.2][center dot][individual.sup.-1][center dot][d.sup.-1], or 16 mg[center dot][individual.sup.-1][center dot][d.sup.-1]; independent t test, P [less than] 0.05) in 24-h no-choice tests.
Long-term performance on standard and deactivated tissue
Much as in the 96-h results, G. pseudolimnaeus, H. designatus, and Limnephilus sp. all consumed little high-glucosinolate tissue (fresh green watercress; Table 1), but consumed significantly more senescent and heated green watercress ([ILLUSTRATION FOR FIGURE 2 OMITTED]; Table 2). Repeated experiments at different times, with H. designatus and G. pseudolimnaeus, gave consistent results that further supported the generality of the observations (Table 2). Consumption by the two caddisflies was generally constant over time (i.e., linear) for all leaf types, whereas G. pseudolimnaeus exhibited a suggestion of decreasing (asymptotic) consumption of fresh and heated yellow tissue with time [ILLUSTRATION FOR FIGURE 2 OMITTED].
Wet-mass consumption of fresh green tissue by the caddisflies and G. pseudolimnaeus (Table 2) was less than that of heated green or of fresh or heated yellow (all P [less than] 0.05; Tukey's hsd). Dry-mass estimates showed relatively higher consumption rates of heated watercress for all taxa (Table 2), due to the lower water content of heated green leaves. For the caddisflies, dry-mass consumption was always highest for heated green, but for G. pseudolimnaeus consumption of heated green tissue was generally not different from heated or fresh senescent tissue (Table 2). Growth rates of all taxa were lowest on fresh green watercress and in many instances, invertebrates lost mass on a fresh green watercress diet (Table 3). In contrast, growth rates of the caddisflies were usually highest on heated green watercress, ranging from 0.4 to 0.7%/d (Table 3). Gammarus pseudolimnaeus, however, showed low or negative growth on both heated and fresh green watercress and had the highest growth, up to 0.25%/d, on senescent watercress.
Much as in the consumption and growth results, the caddisflies generally survived poorly on fresh green tissue and survived well on heated green tissue (Table 4). Survivorship on fresh green tissue decreased with time. Mortality was likely associated with a lack of feeding and starvation, because no alternative foods were provided in any experiments. Survival of the caddisflies on heated green tissue did not seem to be affected by time and ranged from 75% for H. designatus to 100% for Limnephilus sp. (Table 4). In contrast, G. pseudolimnaeus survived poorly on heated green tissue (0-24%; Table 4), and relatively better on fresh green tissue (40-80%). Survival on yellow tissue was similar to survival on fresh green tissue (60-80%). Because G. pseudolimnaeus consumed very little fresh green tissue, but survived relatively well, starvation did not seem to be a factor in their survival during these experiments.
For the caddisflies, daily relative nitrogen consumption was always highest for heated green watercress (all P [less than] 0.05, Tukey's hsd; Table 5). Daily nitrogen consumption was always lowest for fresh green tissue, but was only significantly lower than yellow tissue in one experiment (Table 5). Total relative nitrogen consumption per unit herbivore dry mass (milligrams per gram, over the experiment) followed similar trends, with consumption for heated green tissue ranging from 107 mg/g for H. designatus to 528 mg/g for Limnephilus sp. and consumption of nitrogen from fresh green tissue ranging from 11 mg/g for H. designatus to 186 mg/g for Limnephilus sp. Total nitrogen consumption from heated green tissue was higher than from the other three tissues in all three experiments (all P [less than] 0.05; Tukey's hsd).
Caddisfly daily glucosinolate consumption paralleled nitrogen consumption and was higher for heated green tissue than for the other tissue types (Table 5; all P [less than] 0.05). Consumption of glucosinolates from yellow tissue was significantly lower than that from fresh green tissue. Total glucosinolate consumption followed similar trends, with significantly higher consumption for heated green tissue (13-27 mg/g for H. designatus and 46 mg/g for Limnephilus sp.; all P [less than] 0.05). Glucosinolate consumption from senescent tissues was similar for heated and fresh yellow (P [greater than] 0.05) and ranged from 0.5 mg/g for H. designatus to [approximately equal to]3 mg/g for Limnephilus sp.
In contrast to the caddisflies, there was less of a [TABULAR DATA FOR TABLE 1 OMITTED] difference in nitrogen consumption among leaf types by G. pseudolimnaeus. Consumption of nitrogen from fresh green watercress was significantly lower than from heated green tissue in both experiments (Table 5), but there were few significant differences among the other types (Table 5). Due to higher mortality on heated green tissue, total cumulative nitrogen consumption was highest for fresh and heated senescent leaves; however, these were not significantly higher than heated green leaves. Daily relative glucosinolate consumption was significantly higher for heated green tissue than for the other three types in both experiments (Table 5), but total cumulative glucosinolate consumption was equal for heated and fresh green tissue (6-18 mg/g).
Conversion efficiencies were quite variable and ranged from -50% for H. designatus and G. pseudolimnaeus on fresh green tissue, to 11% for H. designatus on heated yellow and 13% for G. pseudolimnaeus on fresh yellow tissue. There were no differences among leaf types for any taxon (all P [greater than] 0.1) and even with various transformations the variances were quite heterogeneous.
To determine relations among consumption, growth, and survival, correlations and partial correlations among the responses were calculated. For both caddisfly species, instantaneous growth rates were positively correlated with daily consumption and daily nitrogen consumption (Table 6). Growth was also positively correlated with glucosinolate consumption, but not significantly. When the influence of nitrogen was accounted for, via partial correlation analysis, the partial correlation of growth with glucosinolate level was reduced or became negative. Survival was also positively related to daily consumption, but survival showed weaker relationships with instantaneous growth and nitrogen consumption and no significant relation to glucosinolate consumption. When nitrogen consumption was accounted for, the relationship of survival with glucosinolates became negative (Table 6). For G. pseudolimnaeus, growth was not significantly related to nitrogen or glucosinolate consumption as a proportion of leaf dry mass (Table 6). Survival was negatively related to nitrogen and glucosinolate consumption. When glucosinolate consumption was accounted for, the partial correlation of nitrogen consumption with growth and survival became positive. Thus when the effects of nitrogen and glucosinolate were separated, a general positive influence of nitrogen was found as opposed to the negative effects of glucosinolates.
North American stream communities contain abundant populations of facultative shredder-herbivore-detritivores that are apparently not adapted to overcome the watercress defensive system. These stream shredders do rapidly utilize senescent tissues that contain reduced nitrogen levels and reduced defensive compounds. Our findings hint that aquatic shredder-detritivores retain the ability to become shredder herbivores, if unprotected tissues are present. Stream invertebrates appear adapted to dealing with lower quality but less-defended senescent tissue and terrestrial leaf litter (Webster and Benfield 1986, Newman 1991). This is in contrast to terrestrial systems, in which most insects seem to be specialist phytophages that often use behavioral and physiological adaptations to overcome defended live tissue (e.g., Brattsten 1992, Slansky 1992). Utilization of senescent tissue in terrestrial systems seems poorly documented (but see Ghaout et al. 1991).
The glucosinolate-myrosinase system in watercress provides one of the first clear demonstrations of chemical defenses in a freshwater macrophyte. Glucosinolate levels in watercress are among the highest in the crucifers (e.g., Cole 1976, Feeny and Rosenberry 1982, and the Introduction above) and are high in both submersed and emergent leaves (R. M. Newman, Z. Hanscom, and W. C. Kerfoot, unpublished data). Although there appear to be no adapted aquatic herbivores in North America, emergent watercress is subject to attack by both aquatic shredders and terrestrial specialists such as pierids and chrysomelids (R. M. Newman, personal observation). The emergent plants tend to develop high densities of shaded submersed leaves, which senesce and harbor abundant shredders. Herbaceous plants that contain mobile defenses (sensu Coley et al. 1985, Bazzaz et al. 1987) have been shown to have both reduced defenses and reduced nitrogen levels in senescent leaves (e.g., Feeny and Rosenberry 1982, Newman et al. 1990, Ghaout et al. 1991, Stamp and Bowers 1994) presumably because compounds are being retranslocated for use in storage or in growing tissue or for maintaining an optimal allocation of resources to defense (Coley et al. 1985, Herms and Mattson 1992, Gershenzon 1994). Stream shredders are able to capitalize on the submersed senescent tissue. Living submersed plant tissues, however, still must maintain adequate levels of defense to deter feeding by the shredder community. Our experiments verify considerable green tissue consumption if green tissues are not protected by the glucosinolate-myrosinase system.
It is also possible that myrosinase, in addition to glucosinolate, may be affected by senescence. A reduction of myrosinase level or activity (which would influence the rate of isothiocyanate release) could enhance consumption of senescent tissue. More work on seasonal changes in glucosinolate levels along with the effects of variation in levels of myrosinase (Louda and Mole 1991) is needed. Such studies could show if watercress is strategically sequestering or resorbing defensive components, either in response to seasonal metabolic condition or in response to potential effects of herbivores.
We have now shown deterrence from feeding on fresh watercress and a preference for senescent watercress for five species of North American invertebrates from three classes. However, watercress is not indigenous to North America, but was introduced in the 1700's from Europe (Newman et al. 1990). There are several taxa in Europe which have been reported to eat watercress: Limnephilus lunatus (Gower 1967), a distinct but congeneric species with our Limnephilus, and Gammarus pulex, a European counterpart to our Gammarus pseudolimnaeus (H. B. N. Hynes, University of Waterloo, personal communication). These observations suggest that some congeners of North American taxa in Europe, long exposed to watercress, may be better adapted to feeding on watercress and may not be as deterred by the glucosinolate-myrosinase system. A comparative study of European herbivore responses to the watercress defense system would shed light on possible adaptations to freshwater macrophyte chemical defenses by stream invertebrates.
[TABULAR DATA FOR TABLE 2 OMITTED]
Responses to tissue glucosinolates and nitrogen
On a tissue level, the ranked results of our long-term feeding trials generally coincided with short-term preferences for tissues. Short-term preferences were fairly good indicators of long-term consumption. Common spring-brook shredders clearly avoid feeding on fresh green watercress (Table 7). When the deterrent system is deactivated, the caddisflies H. designatus and Limnephilus, and the amphipod Gammarus, will reverse preferences and consume large quantities of the high-nitrogen heated tissue for extended periods of time.
The positive relation of caddisfly growth to nitrogen consumption is not surprising; nitrogen, as an index of protein content, is often considered a good predictor of food quality (Slansky and Scriber 1985, Anderson and Cargill 1987). In fact, consumption may be stimulated [TABULAR DATA FOR TABLE 3 OMITTED] by higher levels of nitrogen or protein, so it is difficult in the current study to separate the effects of total consumption from nitrogen consumption for the caddisflies. However, our finding that areal and wet-mass estimates of consumption of heated green tissue were not higher than those for senescent tissue (in general they were lower, but not significantly) in the long-term growth studies contrasts with our previously reported short-term choice results where heated green watercress was strongly preferred (on an areal basis; Newman et al. 1992). This contrast could be due to complications from physiological toxic effects, but is more likely due to differences in short-term preferences vs. longer term nutritional requirements. Nitrogen intake may have been sufficient with the high-nitrogen tissue, but increased consumption of the lower nitrogen tissues may have been required to acquire adequate [TABULAR DATA FOR TABLE 4 OMITTED] nitrogen (e.g., Otto and Svensson 1981, Slansky 1992). For all three taxa and all experiments, daily wet-mass consumption was negatively correlated with nitrogen content (all r [less than or equal to] -0.5, all P [less than or equal to] 0.01), but there was no relationship of dry-mass consumption to nitrogen content. The correlation of wet-mass consumption and nitrogen was strongly influenced by the high nitrogen content of defended (unpalatable) tissues; if fresh green [TABULAR DATA FOR TABLE 5 OMITTED] tissue was deleted from the analysis, there was no significant relationship.
Our results do suggest that nitrogen may be the factor that influences preference for green tissues when the deterrent system is deactivated. Higher nitrogen may be associated with preference for heated green watercress (Newman et al. 1992); however, longer term effects on G. pseudolimnaeus appeared deleterious, probably [TABULAR DATA FOR TABLE 6 OMITTED] because nitrogen content is highly correlated with glucosinolate content. The partial correlation analysis showed that nitrogen was positively associated with growth and survival whereas glucosinolate content was negatively associated with growth and survival. Thus, the animals face a trade-off of low nitrogen vs. high glucosinolates. In order to definitively separate the deterrent, consumption, and growth effects from nitrogen, glucosinolate, isothiocyanate, and other components, experiments with manipulated diets (e.g., Bodnaryk 1991, McCloskey and Isman 1993) will be needed and are planned for future research.
As mentioned above, G. pseudolimnaeus and the caddisflies appeared to have different long-term responses to consumption of heated tissues. The caddisflies grew and survived best on the nitrogen-rich heated tissue, whereas G. pseudolimnaeus showed variable growth, mass loss, and reduced survival, which suggests that although the feeding deterrent (isothiocyanate) was eliminated, a component toxic to G. pseudolimnaeus remained. Although Sutcliffe et al. (1981) found growth rates of G. pulex on watercress leaves to be comparable to those on many other foods, they found a high mortality rate within 50 d.
There are several possible causes of higher mortality when Gammarus consumed heated green tissue. Starvation was clearly not the cause because survival was [TABULAR DATA FOR TABLE 7 OMITTED] generally much better on the lesser consumed fresh green tissue and we have been able to keep amphipods alive under similar conditions with no food for several weeks. The most likely explanation is that a toxic compound, or a compound that would be metabolized into a toxic compound, was present in the heated green tissue, and that the higher consumption rates of the heated tissue resulted in higher consumption rates of the compound. The most likely candidate in this scenario is 2-phenylethyl glucosinolate. It was present in high levels in heated green tissue (Table 1; see Newman et al. 1992 for heated-tissue levels) and was negatively related to survival.
However, the presence of other active compounds cannot be ruled out. One possibility is that the glucosinolate was converted to toxic isothiocyanate (Newman et al. 1990) in the gut, whereas another is that the glucosinolate alone is toxic to amphipods. For heated green tissue, total glucosinolate consumption by amphipods averaged 13-18 mg/g and resulted in high mortality, whereas total glucosinolate consumption by the caddisflies ranged from 13 to 46 mg/g with little or no mortality. The glucosinolate consumption rates we observed for heated green tissue by both caddisflies and Gammarus were comparable to the glucosinolate consumption rates that resulted in significant mortality of the umbellifer specialist Papilio polyxenes (0.25-3.3 mg[multiplied by][[g wet mass].sup.-1] [multiplied by] [d.sup.-1], calculated from Erickson and Feeny 1974). A crucifer specialist was not adversely affected by even higher levels of glucosinolate consumption (Erickson and Feeny 1974, Blau et al. 1978).
In contrast to G. pseudolimnaeus, mortality in the caddisflies appeared to be due to starvation. Survival rates were generally highest on the highly consumed heated green leaves and lowest on the fresh green leaves, which were consumed little. Insects with determinate growth may be less capable of surviving without food than the indeterminate-growth amphipods. The lack of an effect of ingested glucosinolate on caddisflies is somewhat surprising given the negative effects of glucosinolates demonstrated with generalist Lepidoptera (Blau et al. 1978); however, gut environments may differ substantially between Lepidoptera and Trichoptera even though they are distantly related. Differences in post-ingestive processes among taxa (e.g., see Nugon-Baudon et al. 1990, Pracros et al. 1992, Slansky 1992) need further examination.
The watercress glucosinolate-myrosinase system is an effective feeding deterrent to a wide range of stream shredders, who are apparently adapted to lower nitrogen senescent tissue but not to live defended tissue. Thus these organisms have taken an alternative approach to dealing with defended plants. Because freshwater systems share taxa with both marine and terrestrial systems that neither has in common, further study of generalist and specialist herbivore-plant relationships and plant semiochemicals in freshwater systems will help bridge the gap evident in marine-terrestrial comparisons (e.g., Hay and Steinberg 1992).
The assistance of B. Giese, E. Engnell, J. Muck, and M. Nemeth with the maintenance of the experiments, digitizing results, and entering data is greatly appreciated as is the assistance of C. Olow with the glucosinolate analyses and J. Ameel with the CHN analyses. We thank C. Bell, A. Glenn, and the Krischers for access to study sites. Discussion, and comments provided by D. A. Andow, A. S. Cargill, L. M. Maher, S. L. Solarz, A. E. Weis, and several anonymous reviewers were most helpful and greatly appreciated. This research was supported by grants from A. S. Cargill, II, and the Minnesota Agricultural Experiment Station and is published as Paper Number 21,409 of the contribution series of the Minnesota Agricultural Experiment Station based on research conducted under Project 74. W. C. Kerfoot thanks NSF DEB 82-07007 and support from the Waterways Experimental Station, Macrophyte Control Division, U.S. Army Corps of Engineers, during initial phases of the project, and Mary Breitsprecher for laboratory assistance.
Anderson, N. H., and A. S. Cargill. 1987. Nutritional ecology of aquatic detritivorous insects. Pages 903-925 in F. Slansky and J. G. Rodriguez, editors. Nutritional ecology of insects, mites, and spiders. John Wiley and Sons, New York, New York, USA.
Bazzaz, F. A., N. R. Chiariello, P. D. Coley, and L. F. Pitelka. 1987. Allocating resources to reproduction and defense: new assessments of the costs and benefits of allocation patterns in plants are relating ecological roles to resource use. BioScience 37:58-67.
Blau, P. A., P. Feeny, L. Contardo, and D. S. Robson. 1978. Allylglucosinolate and herbivorous caterpillars: a contrast in toxicity and tolerance. Science 200:1296-1298.
Blua, M. J., and I. Z. Hanscom. 1986. Isolation and characterization of glucocapparin in Isomeris arborea Nutt. Journal of Chemical Ecology 12:1449-1458.
Bodnaryk, R. P. 1991. Developmental profile of sinalbin (p-hydroxybenzyl glucosinolate) in mustard seedlings, Sinapis alba L., and its relationship to insect resistance. Journal of Chemical Ecology 17:1543-1556.
Brattsten, L. B. 1992. Metabolic defenses against plant allelochemicals. Pages 175-242 in G. A. Rosenthal and M. R. Berenbaum, editors. Herbivores, their interactions with secondary plant metabolites. Second edition. Volume II: evolutionary and ecological processes. Academic, San Diego, California, USA.
Cole, R. A. 1976. Isothiocyanates, nitriles and thiocyanates as products of autolysis of glucosinolates in Cruciferae. Phytochemistry 15:759-762.
Coley, P. D., J. P. Bryant, and I. F. S. Chapin. 1985. Resource availability and plant antiherbivore defence. Science 230: 895-899.
Erickson, J. M., and P. Feeny. 1974. Sinigrin: a chemical barrier to the black swallowtail butterfly, Papilio polyxenes. Ecology 55:103-111.
Feeny, P. 1977. Defensive ecology of the Cruciferae. Annals of the Missouri Botanical Garden 64:221-234.
Feeny, P., and L. Rosenberry. 1982. Seasonal variation in the glucosinolate content of North American Brassica nigra and Dentaria species. Biochemical Systematics and Ecology 10:23-32.
Fenwick, G. R., R. K. Heaney, and R. Mawson. 1989. Glucosinolates. Pages 1-41 in P. R. Cheeke, editor. Toxicants of plant origin. Volume II: Glycosides. CRC, Boca Raton, Florida, USA.
Freeman, G. G., and N. Mossadeghi. 1973. Studies on relationship between water regime and flavour strength in watercress (Rorippa nasturtium-aquaticum (L) Hayek), cabbage (Brassica oleracea capitata) and onion (Allium cepa). Journal of Horticultural Science 48:365-378.
Gershenzon, J. 1994. The cost of plant chemical defense against herbivory: a biochemical perspective. Pages 105-173 in E. A. Bernays, editor. Insect-plant interactions. Volume V. CRC, Boca Raton, Florida, USA.
Ghaout, S., A. Louveaux, A. M. Mainguet, M. Deschamps, and Y. Rahal. 1991. What defense does Schouwia purpurea (Cruciferae) have against the desert locust: secondary compounds and nutritive value. Journal of Chemical Ecology 17:1499-1515.
Gower, A. M. 1967. A study of Limnephilus lunatus Curtis (Trichoptera: Limnephilidae) with reference to its life cycle in watercress beds. Transactions of the Royal Entomological Society of London 119:283-302.
Hay, M. E., and P. D. Steinberg. 1992. The chemical ecology of plant-herbivore interactions in marine versus terrestrial communities. Pages 371-413 in G. A. Rosenthal and M. R. Berenbaum, editors. Herbivores, their interactions with secondary plant metabolites. Second edition. Volume II: evolutionary and ecological processes. Academic, San Diego, California, USA.
Herms, D. A., and W. J. Mattson. 1992. The dilemma of plants: to grow or defend. Quarterly Review of Biology 67:283-335.
Kerfoot, W.C. 1989. Glucosinolates and phenolics in aquatic macrophytes: implications for allelopathy studies and suggested practical uses for metabolic blocking agents. Pages 178-189 in Proceedings 23rd Annual Meeting of the Aquatic Plant Control Research Program, Miscellaneous Paper A-89-1, Waterways Experiment Station, Vicksburg, Mississippi, USA.
Lichtenstein, E. P., D. G. Morgan, and C. H. Mueller. 1964. Naturally occurring insecticides in cruciferous crops. Agricultural and Food Chemistry 12:158-161.
Lodge, D. M. 1991. Herbivory on freshwater macrophytes. Aquatic Botany 41:195-224.
Louda, S., and S. Mole. 1991. Glucosinolates: chemistry and ecology. Pages 123-164 in G. A. Rosenthal and M. R. Berenbaum, editors. Herbivores, their interactions with secondary plant metabolites. Second edition. Volume 1: the chemical participants. Academic, San Diego, California, USA.
McCloskey, C., and M. B. Isman. 1993. Influence of foliar glucosinolates in oilseed rape and mustard on feeding and growth of the bertha armyworm, Mamestra configurata Walker. Journal of Chemical Ecology 19:249-266.
Newman, R. M. 1991. Herbivory and detritivory on freshwater macrophytes by invertebrates: a review. Journal of the North American Benthological Society 10:89-114.
Newman, R. M., Z. Hanscom, and W. C. Kerfoot. 1992. The watercress glucosinolate-myrosinase system: a feeding deterrent to caddisflies, snails and amphipods. Oecologia 92: 1-7.
Newman, R. M., W. C. Kerfoot, and Z. Hanscom. 1990. Watercress and amphipods: potential chemical defense in a spring stream macrophyte. Journal of Chemical Ecology 16:245-259.
Nugon-Baudon, L., S. Rabot, J. M. Wal, and O. Szylit. 1990. Interactions of the intestinal microflora with glucosinolates in rapeseed meal toxicity: first evidence of an intestinal Lactobacillus possessing a myrosinase-like activity in vivo. Journal of the Science of Food and Agriculture 52:547-559.
Ostrofsky, M. L., and E. R. Zettler. 1986. Chemical defences in aquatic plants. Journal of Ecology 74:279-287.
Otto, C., and B. S. Svensson. 1981. How do macrophytes growing in or close to water reduce their consumption by aquatic herbivores? Hydrobiologia 78:107-112.
Pracros, P., C. Couranjou, and R. Moreau. 1992. Effects on growth and respiration due to the ingestion of the rapeseed meal glucosinolates in young larvae of Tenebrio molitor. Comparative Biochemistry and Physiology A Comparative Physiology 103:391-395.
Slansky, F. 1992. Allelochemical-nutrient interactions in herbivore nutritional ecology. Pages 135-174 in G. A. Rosenthal and M. R. Berenbaum, editors. Herbivores, their interactions with secondary plant metabolites. Second edition. Volume II: evolutionary and ecological processes. Academic, San Diego, California, USA.
Slansky, F., and J. M. Scriber. 1985. Food consumption and utilization. Pages 87-163 in G. A. Kerkut and L. I. Gilbert, editors. Comprehensive insect physiology, biochemistry and pharmacology. Volume four. Pergamon, Oxford, England.
Stamp, N. E., and M. D. Bowers. 1994. Effects of cages, plant age and mechanical clipping on plantain chemistry. Oecologia 99:66-71.
Suren, A. M. 1989. Histological changes in macrophyte tissue during decomposition. Aquatic Botany 33:27-40.
Sutcliffe, D. W., T. R. Carrick, and L. G. Willoughby. 1981. Effects of diet, body size, age and temperature on growth rates in the amphipod Gammarus pulex. Freshwater Biology 11:183-214.
Thangstad, O. P., K. Evjen, and A. Bones. 1991. Immunogold-EM localization of myrosinase in Brassicaceae. Protoplasma 161:85-93.
Wadleigh, R. W., and S. J. Yu. 1988. Detoxification of isothiocyanate allelochemicals by glutathione transferase in three lepidopterous species. Journal of Chemical Ecology 14:1279-1288.
Webster, J. R., and E. F. Benfield. 1986. Vascular plant breakdown in freshwater ecosystems. Annual Review of Ecology and Systematics 17:567-594.
Wilkinson, L. 1989. SYSTAT: the system for statistics. SYSTAT, Evanston, Illinois, USA.
|Printer friendly Cite/link Email Feedback|
|Author:||Newman, Raymond M.; Kerfoot, W. Charles; Hanscom, Zac, III|
|Date:||Dec 1, 1996|
|Previous Article:||Delayed inducible resistance in mountain birch in response to fertilization and shade.|
|Next Article:||Linking herbivory and pollination: defoliation and selective fruit abortion in Lindera benzoin.|