Developing embryos of the inland silverside fish, Menidia beryllina, were exposed to conidiospores of the entomopathogenic fungus, Metarhizium anisopliae. Several adverse effects were observed in both embryos and newly hatched larvae. These included transitory effects on the heart resulting in decreased cardiac output or circulation velocity, rupture of the chorion, fungal growth on the mandibles of larvae, focal vertebral abnormalities in larvae and teratogenic expressions in embryos and larvae. An ordinal ranking system was used to enumerate responses to conidiospores. This ranking system allowed significance to be determined by nonparametric analysis of variance. Responses were highly variable with significant (p0.05) adverse effects observed in five of the six experiments conducted. Heat-killed spores failed to cause significant adverse effects, indicating that viable spores were required for adverse effects to occur.

Key Words: Metarhizium anisopliae, Menidia beryllina, biological control/nontarget effects

Metarhizium anisopliae is a pathogen of many insect species (Vey et al., 1982). This fungi imperfecti has received much attention for its potential use in the control of a variety of insect pests. It is extensively used as a biocontrol agent in Brazil where, produced under the name Metaquino®⁽¹⁾, it is sprayed on sugar cane fields to control the spittle bug, Mahanarva posticata (Moscardi, 1989). In Australia, Metarhizium anisopliae may soon be commercialized for control of the subterranean pasture scarab, Adoryphorus couloni. In tests with a granulated formulation of Metarhizium anisopliae, Bio-Green®, one application proved effective against the pasture scarab for over four years (Goettel, 1994). In May, 1993, Metarhizium anisopliae was registered by the U.S. Environmental Protection Agency for control of nuisance flies and cockroaches. For this application, Metarhizium anisopliae is introduced in a patented bait-station, Bio-Path®, where exposed insects can spread the fungus through direct contact.

With the exception of insects, little is known about the effects of this organism on poikilothermic organisms. Donovan-Peluso et al. (1980) reported that exposing frogs to Metarhizium anisopliae spores by gastric intubation produced no pathologic lesions or infection of viscera. Recently, we adapted a chemical toxicity and teratogenicity test that utilized developing embryos of the inland silverside fish, Menidia beryllina, to evaluate potential toxicity and pathogenicity of a related insect pathogenic fungus, Beauveria bassiana (Genthner and Middaugh, 1992; Middaugh and Genthner, 1994). Various adverse effects were observed in the embryos including rupture of the chorion, developmental defects (vertebral abnormalities) and death; in the hatched larvae fungal infections on the mandibles and vertebral abnormalities were noted.

Because of the recent commercialization of Metarhizium anisopliae based products and the possibility that this fungus may possess a host-range different from Beauveria bassiana, potential adverse effects of Metarhizium anisopliae on developing Menidia beryllina embryos were investigated. We report significant adverse effects including one not observed with Beauveria bassiana. Exposure to this particular strain of Metarhizium anisopliae caused developmental effects that were less severe than those observed with Beauveria bassiana - grasshopper strain (GH).

Cultivation of fungi and recovery of spores. Metarhizium anisopliae 683 (USDA-ARS collection of entomopathogenic fungal cultures, Ithaca, NY) was obtained from Rosalind James, ManTech Environmental Technology, Corvallis, OR.

The fungus was cultured at 25^oC on glucose-yeast extract-basal salts agar medium (GYBS) (Boucias et al., 1988). At 5-12 days post-inoculation, progeny conidiospores were harvested by scraping confluent mycelial mats with a sterile spatula. Conidia were suspended in sterile, moderately hard water (MHW) (U.S. EPA 1991) by gentle aspiration in a hand-held tissue homogenizer. Spore density was adjusted to approximately 1 x 10⁸ spores ml^-1, using a hemocytometer. Spore suspensions were either used immediately or allowed to soak for 24 hrs (Table 1). Viable spore counts were performed by diluting spores in sterile distilled water containing 0.03% Triton X-100 (Union Carbide Chemicals and Plastics Co., Inc., Indianapolis, IN) and spreading the dilutions onto GYBS. Fungal colonies were counted after 5 days.

Corn earworm, Helicoverpa zea, were exposed to Metarhizium anisopliae by dipping 2nd instar worms into a conidiospore suspension (ca. 1 X 10⁷ ml^-1). Five exposed worms were placed in a petri dish at 25C containing approximately 3 g of a sterile diet consisting of ground raw pinto beans, 14 g; wheat germ, 10 g; torula yeast, 6.3 g; casein 5.0 g; agar, 2.3 g; and distilled water, 135 ml. Three days after exposure, dead worms displaying external conidiogenesis were stored frozen at -80^oC in 20% glycerol. Frozen worms were thawed at room temperature and homogenized in sterile water. A loopful of the homogenate was streaked onto a GYBS + nalidixic acid (Nx, 0.5 mg ml^-1) plate. Isolated colonies of Metarhizium anisopliae were selected to produce conidiospores on Nx-free GYBS for exposure experiments.

Metarhizium anisopliae was recovered from exposed embryos and larvae by washing them (twice in 100 ml sterile water) to eliminate the spores carried over in water from the exposure tubes and then homogenizing the tissue and spreading dilutions onto the surface of GYBS + Nx plates.

Embryo tests. Embryonic inland silversides, Menidia beryllina, were obtained from adults that spawned naturally in the laboratory at 5 salinity and 25C as described by Middaugh et al. (1987; 1988). The photoperiod was 16 hrs light and 8 hrs darkness. Light intensity at the water's surface in brood tanks was approximately 17 microeinsteins/m²/s.

Embryo exposures were performed as described previously (Genthner and Middaugh, 1992). Single embryos were placed in Leighton® culture tubes. Six ml of sterile MHW were added to each control tube, and 6 ml of 1/10 serial dilutions of the conidiospore suspensions were added to each exposure tube. A conidiospore-killed control was provided from the dilution containing the highest conidiospore density by sterilization in an autoclave [20 min, 15 lb/in² (=10⁵ Pa)]. This design yielded a "no exposure" control, a "heat-killed" control, and three 10-fold dilutions of conidiospores with 30 replicate embryos for each treatment. Exposure tubes were maintained in a horizontal position and incubated at 25C. The photoperiod was 14 hrs light and 10 hrs darkness. Light intensity was ca. 20 microeinsteins/m²/s (Genthner and Middaugh, 1992). The observed effects in embryonic and larval Menidia beryllina were scored daily on the basis of the responses described in Table 2. Observations were made with a Zeiss Axiovert® inverted microscope equipped for photomicrography.

At the end of each 7 to 9 day experiment, scores for each embryo were summed. This procedure resulted in 30 individual ranks for each of the four to five treatments in each experiment. The nonparametric Kruskal Wallis Analysis of Variance (KW-ANOVA) by ranks was conducted to determine if the rank scores for each treatment within an experiment differed significantly (p 0.05) (Statsoft Inc., 1991). This test assumes that the variables (respective embryo responses at differing conidiospore concentrations) under consideration were measured on at least an ordinal (rank order) scale as shown in Table 2. In experiments where statistically significant differences were detected with the KW-ANOVA, post hoc (pairwise comparisons) were conducted to determine if the rank scores for the control embryos and each of the conidiospore concentrations were significantly different (alpha 0.05) from each other (Siegel and Castellan, 1988).

In addition to the nonparametric KW-ANOVA, a parametric one way analysis of variance (ANOVA) and Tukey HSD post hoc test were conducted to determine if the number of "hatched-normal" larvae in control treatments and each of the conidiospore density treatments were significantly different (alpha 0.05), Statsoft Inc. (1991). In this analysis, the 30 embryos in each treatment were divided into three replicate groups of 10 each to provide three replicates per treatment. The first 10 embryos in each treatment were assigned to the first replicate, the next 10 to the second replicate and the final 10 to the third replicate. All "hatched-normal" larvae in a treatment were assigned a value of one (1) and all "not-hatched" or "hatched-not-normal" larvae a value of zero (0). The values for the 10 embryos in each replicate were summed prior to conducting the one way ANOVA.

Developing embryos of Menidia beryllina were exposed to conidiospores of Metarhizium anisopliae in 6 separate experiments. The date each experiment was performed and history of conidiospore culture is presented in Table 1.

There was substantial variability in the observed effects of Metarhizium anisopliae within and among experiments (Table 3). In the first Experiment (A) spores were tested after passage through corn earworm, Helicoverpa zea (Table 1). The KW-ANOVA and post hoc analyses revealed statistically significant differences from the control at all three conidiospore densities (Table 3). These differences were largely the result of a reduction in cardiac output (approximately 50%) during the second day of conidiospore exposure (Table 4). This transitory effect was not observed on subsequent days and there was no apparent adverse effect on the course of normal development and successful hatching of embryos. The parametric one way ANOVA and post hoc test for hatch data revealed no significant differences (NS) between the numbers of "hatched normal" larvae for the control and each of the conidiospore treatments (Table 4).

In the second experiment (B) conidiospores from Experiment A were produced after 2 additional saprophytic passages of Metarhizium anisopliae on GYBS plates (Table 1). Statistically significant differences from the control response were obtained at the intermediate and high conidiospore density treatments but not in the low spore treatment of 1.0 x 10⁴ ml^-1 (Table 3).

At the intermediate spore density several embryo responses were observed. A slight to moderate reduction in the size and structure of the craniofacial complex was often associated with a failure of the heart to develop normally. The severity of these cardiac manifestations ranged from a lack of a discernible heart, to a "tube" heart lacking chambers incapable of normal cardiac output, to a structually normal looking heart that displayed a transient reduction in cardiac output. Skeletal abnormalities included a slight-to-moderate stunting of the skeletal axis, and in some cases the presence of gross vertebral abnormalities resulting in the lack of any organized structures or vertebrae. Embryo rupture was also occasionally observed. Germinating conidia on the surface of a ruptured embryo suggested that penetration of the chorion by the fungus caused the rupture. Spreading dilutions of homogenized embryos onto the surface of GYBS+Nx plates revealed that dead embryos exposed at the intermediate spore density possessed approximately 4 x 10³ colony forming units of Metarhizium anisopliae per embryo while living embryos possessed only about 75 colony forming units of Metarhizium anisopliae per embryo.

In the second experiment, (B) only seven (24%) of the 30 embryos exposed at the intermediate spore density (1.0 x 10⁵ ml^-1) hatched. The hatched larvae showed deformities similar to the ones described above for developing embryos. While statistically significant responses were obtained at the highest spore density (Table 3), they were less severe at the intermediate density. All embryos exposed at the high spore density hatched, however many exhibited a persistent reduction in cardiac output from day two through hatching that appeared to be very similar, if not identical, to the transitory cardiac response observed in Experiment A. Although this reduction in cardiac output persisted until hatch, there was no apparent detrimental effect on the developing embryos or hatched larvae nor was hatching delayed. Moreover, the ANOVA comparison revealed that differences in the number of control versus conidiospore exposed individuals that "hatched-normal" was significantly different only for the intermediate exposure density (1.0 x 10⁵ spores ml^-1, Table 4).

Although conidiospores used in Experiments B and C were cultured from the same source (Table 1), no significant responses were obtained in Experiment C (Table 3). A single "no spore" control embryo, however, had a teratogenic response characterized by a moderate reduction in size of the craniofacial complex, an amorphous heart and a stunted skeletal axis. This individual died prior to hatching. A single embryo exposed at the low spore density (9.7 x 10³ ml^-1) also showed slight stunting of the skeletal axis at hatching. The ANOVA revealed no significant differences in numbers of "hatched normal" larvae for control, killed-control, and conidiospore exposed groups (Table 4).

Metarhizium anisopliae, isolated from an infected corn earworm stored frozen in 20% glycerol, was used to produce conidiospores for Experiment D (Table 1). Statistically significant larval responses occurred at the high (1.0 X 10⁶ ml^-1) conidiospore density (Table 3) where 13 out of 30 (43%) larvae had skeletal abnormalities. While not statistically significant, at the intermediate spore density, fungal growth was observed on the mandibles of 2 larvae and vertebral abnormalities with slight stunting of the skeletal axes was apparent in 4 larvae on the day of hatching. The ANOVA revealed significant differences in the number of "hatched normal" control larvae and larvae from embryos exposed at the intermediate and high conidiospore densities (Table 4).

For Experiment E, conidiospores were cultured from the same source as in Experiment D. In this experiment, however, no teratogenic effects were observed in the larvae. Instead, fungal growth was observed on the mandibles of newly hatched larvae exposed at all 3 conidiospore densities. The KW-ANOVA and post hoc tests revealed that all conidiospore densities resulted in responses that were statistically different from the controls (Table 3). The ANOVA and post hoc test also revealed significant differences between the number of "hatched-normal" control larvae and larvae from embryos exposed at all three conidiospore densities (Table 4).

For the final experiment (F) spores were cultured from the same source as in Experiments D and E (Table 1). Responses in Experiment F resembled those observed in Experiment C. A single control embryo displayed teratogenicity that included a slight reduction in the size of the craniofacial complex and reduced cardiac output. Conidiospore exposed embryos showed a nearly complete and normal survival and hatch with only a few of the larvae from the intermediate and high conidiospore treatment groups with fungal growth on their mandibles. The KW-ANOVA indicated that none of these observed effects were statistically different from control responses (Table 3). The ANOVA also indicated that there were no significant differences between the number of "hatched-normal" control larvae and larvae from embryos exposed at the three conidiospore densities (Table 4).

In previous studies, we exposed fish embryos to conidiospores of Beauveria bassiana, an insect pathogenic fungus related to Metarhizium anisopliae, and observed several adverse effects (Genthner and Middaugh, 1992; Middaugh and Genthner, 1994). In the present study, the adverse effects we observed in Metarhizium anisopliae exposed embryos were similar, although less severe, to those obtained with Beauveria bassiana. An exception was a transitory effect on the embryo heart that resulted in a decrease in cardiac output or circulation velocity. Observed in Experiments A and B, this response was not unique to Metarhizium anisopliae - exposed embryos. Reductions in cardiac output, typically between 20-60%, were also observed in embryos exposed to fractions of partially degraded crude oil (Middaugh, unpublished). In these oil tests, however, delayed hatch and other adverse effects often followed reductions in cardiac output.

As with partially degraded crude oil, the cardiac response with Metarhizium anisopliae suggested some form of toxin involvement. Metarhizium anisopliae produces several toxic secondary metabolites whose precise role in pathogenesis and virulence remains unclear (Gillespie and Claydon, 1989). It is possible that mycotoxins were produced by conidiospores germinating on the surface of the embryos. These mycotoxins may have elicited a narcotic effect on the embryos with no lasting adverse effects on embryogenesis or larval hatching. Transport of the mycotoxin through the chorion may not be necessary to produce this effect since at least one mycotoxin, destruxin E, was shown to possess "contact" activity (Poprawski et al., 1994). The other more severe and permanent effects such as vertebral abnormalities (focal and stunting) could also have resulted from the action of mycotoxins. It is also possible that a cardiac response was absent in embryos exposed to Beauveria bassiana (Middaugh and Genthner, 1994) because mycotoxins produced by this fungus are different than those produced by Metarhizium anisopliae.

Highly variable responses were common in tests with Beauveria bassiana and Metarhizium anisopliae exposed embryos (Genthner and Middaugh, 1992; Middaugh and Genthner; 1994, this study). In our study, two procedures were tested to attain an invariable, virulent response between tests. First, conidiospores were consistently produced from Metarhizium anisopliae recently isolated from corn earworm (Table 1). Thus, the culture had not been repeatedly subcultured on solid media, a practice that may decrease virulence. Second, Metarhizium anisopliae conidiospore suspensions were incubated in MHW at 25^oC for 24 hrs before testing (Table 1). Dillon and Charnley (1985) reported that soaking conidiospores of Metarhizium anisopliae in distilled water accelerated and synchronized germination. Soaked conidiospores of Metarhizium anisopliae were also more virulent towards the tobacco hornworm, Manduca sexta, than were non-soaked conidiospores (Hassan and Charnley, 1983). In our study, soaking of conidiospores did not produce consistent results (compare Experiments B and C). Thus, we can only conclude that unexpected changes in virulence, as reported by Weiser (1982), may have contributed to response variability.

Statistically significant responses were detected by the

KW-ANOVA and ANOVA tests (Tables 3 and 4). These analyses yielded differences in significance in 6 out of 18 viable conidiospore treatments (Table 4). In Experiments A and B, significant differences from control responses were detected by the KW-ANOVA post hoc test for five of six conidiospore densities (Table 4). In contrast, the ANOVA and post hoc test revealed only one conidiospore density that was significantly different from controls (Table 4). While reductions in cardiac output in these two experiments were important as an observed effect of exposure to Metarhizium anisopliae conidiospores, there appeared to be no biological significance for embryo hatching. The embryos continued to develop and hatched at the normal 7 to 8 days post fertilization, and hatched larvae appeared normal in all respects (Table 3). Thus, while the KW-ANOVA assesses significance in observed effects, the parametric ANOVA evaluates the apparent biological significance.

In Experiment D, skeletal stunting and other abnormalities went undetected until hatch. Because these differences included skeletal abnormalities and skeletal stunting they also involved an important biological impact. Thus, both the KW-ANOVA and ANOVA for "hatched normal" larvae indicated that these skeletal abnormalities were statistically significant at the high conidiospore density (Tables 3 and 4).

In Experiment E fungal growth was observed on mandibles of hatched larvae from all 3 conidiospore density exposures. While this growth was apparent on the day of hatching, the following day the hyphae had sloughed off of the mandibles. As in previous studies with Beauveria bassiana (Middaugh and Genthner, 1994), the biological significance of fungal growth on the mandibles of fish larvae remains to be interpreted.

Boucias, D.G., Pendland, J.C., Latge, J.P. (1988). Nonspecific factors involved in attachment of entomopathogenic deuteromycetes in host insect cuticle. Appl. Environ. Microbiol. 54: 1795-1805

Dillon, R.J., Charnley, A.K. (1985). A technique for accelerating and synchronising germination of conidia of the entomopathogenic fungus Metarhizium anisopliae. Arch. Microbiol. 142: 204-206

Donovan-Peluso, M. Wasti, S.S., Hartmann, G.C. (1980). Safety of entomogenous fungi to vertebrate hosts. Appl. Entomol. Zool. 15:498

Genthner, F.J., Middaugh, D.P. (1992). Effects of Beauveria bassiana on embryos of the inland silverside fish (Menidia beryllina). Appl. Environ. Microbiol. 58: 2840-2845

Gillespie, A.T. and Claydon, N. (1989). The use of entomogenous fungi for pest control and the role of toxins in pathogenesis. Pestic. Sci. 27: 203-215

Goettel M.S. (1994). Microbial Control News from Australia. Society for Invertebrate Pathology Newsletter Vol 26, No. 2

Hassan A.E.M., Charnley, A.K. (1983). Combined effects of diflubenzuron and the entomopathogenic fungus Metarhizium anisopliae on the tobacco hornworm, Manduca sexta. 10th Int Congr. Plant Prot. BCPC Publications, UK, p 790

Middaugh, D.P., Genthner, F.J. (1994). Infectivity and teratogenicity of Beauveria bassiana in Menidia beryllina embryos. Arch. Environ. Contam. Toxicol. 27: 95-102

Middaugh, D.P., Hemmer, M.J., Goodman, L.R. (1987). Methods for spawning, culturing and conducting toxicity tests with early life stages of four Atherinid fishes: the inland silverside, Menidia beryllina, Atlantic silverside, Menidia menidia, tidewater silverside, Menidia peninsulae and the California grunion, Leuresthes tenuis. Report EPA-600/8-87/004. U.S. Environmental Protection Agency, Gulf Breeze, FL

Middaugh, D.P., Hemmer, M.J., Lores, E.M. (1988). Teratological effects of 2,4-dinitrophenol, "produced water" and naphthalene on embryos of the inland silverside Menidia beryllina. Dis. Aquat. Org. 4:53-65

Moscardi, F. (1989). Production and use of entomopathogens in Brazil. In: Roberts, D.W., Grandos, R.R. (eds.) Biotechnology, Biological Pesticides and Novel Plant-Pest Resistance for Insect Pest Management, Insect Pathology Resource Center, Boyce Thompson Institute for Plant Research, Cornell University, Ithaca, NY.

Poprawski, T.J., Maniania, N.K., Robert, P.H. (1994). Contact toxicity of the mycotoxin destruxin E to Empoasca vitis (Gothe) (Hom., Cicadellidae). J. Appl. Entomol. 117: 135-143

Siegel, S., Castellan, N.J. Jr. (1988). Nonparametric statistics for the behavioral sciences. McGraw-Hill Book Co., Inc., New York

StatSoft, Inc. (1991). Complete statistical system: statistica, version 3.0, p. 237-270. StatSoft, Inc., Tulsa, OK

U.S. Environmental Protection Agency (1991). Methods for measuring the acute toxicity of effluents and receiving waters to freshwater and marine organisms, p. 293. Report EPA/600/4-90/027. U.S. Environmental Protection Agency, Washington, D.C.

Vey, A., Farques, J., Robert, P. (1982). Histological and ultrastructural studies of factors determining the specificity of pathotypes of the fungus Metarhizium anisopliae for scarabaeid larvae. Entomophaga 24: 387-397

Weiser, J. (1982). Persistence of fungal insecticides: influence of environmental factors and present and future applications. In: Kurstak E. (ed.) Microbial and viral pesticides. Marcel Dekker, New York, p. 531-557

Genthner Table 1

Genthner Table 2

Genthner Table 3

Genthner Table 4

1. Mention of tradenames or commercial products does not imply endorsement by the U.S. Environmental Protection Agency.