Figure 1. A ruffe (from Minnesota Sea Grant).

The original goal of this review was to provide background information about ruffe to North American researchers and managers who were contemplating the "best strategry" for controlling ruffe that had been accidentally introduced into the St. Louis River Harbor (SLRH) of the Lake Superior drainage (Pratt 1988; Simon and Vondruska 1991; Pratt et al. 1992). A thorough review of the existing literature became more important as ruffe were found in other new locations (Maitland et al. 1983; Chiara 1986; Winfield 1992; Kalas 1995; Rosch and Schmid 1996). In fact, in 1996, Rosch et al. suggested that "a synopsis of this species is desirable and should be attempted." This review and this web page serves as an on-going and interactive attempt to meet this important suggestion.

Many people have contributed to this work. J. Bruner, A. Byla, J. Janssen, D. Jensen, R. M. Newman, J. H. Selgeby, and I. J. Winfield have kindly opened their personal libraries or provided partial ruffe literature reviews to me. J. H. Selgeby and A. Byla provided translated materials. A. Specziar provided the Hungarian name and A. Kangur corrected the Estonian name for ruffe.

Geographic Distribution
Originally found throughout Europe and Asia, G. cernuus is the most widespread species of the genus. The native range of ruffe included northeastern France, eastern rivers of England, the rivers entering the Baltic Sea, the rivers entering the White Sea, most of Siberia (Figure 2; Wheeler 1974; Collette and Banarescu 1977; Lelek 1987), and the brackish archipelagos of the Baltic Sea (Hansson 1984; 1985; 1987). Popova et al. (1998) updated the distribution of ruffe in the former USSR.

Figure 2. European distribution of ruffe (from Lelek 1987).

Ruffe have recently been found in several European water bodies outside of their original distribution. These water bodies include Loch Lomond, Scotland (Maitland et al. 1983), Italy (Chiara 1986), Llyn Tegid, North Wales (Winfield 1992), the Lake District, England (Winfield 1992), Lake Mildevatn, Norway (Kalas 1995), Lake Constance, Germany (Rosch and Schmid 1996), the Camargue region, France (E. Rosecchi, Station Biologique de la Tour du Valat, pers. comm), a Zurich, Switzerland lake (H. Persat, pers. comm.), and two new reservoirs on the Drava River, Croatia (H. Persat, pers. comm.). The expansion of ruffe has probably been aided by the construction of canals (Wheeler 1974) and the use of ruffe as bait for northern pike (Esox lucius; Winfield, personal communications).

In addition, in the early to mid 1980s, ruffe were accidentally introduced, apparently through ship ballast, to the St. Louis River (Pratt 1988; Simon and Vondruska 1991; Pratt et al. 1992; Selgeby and Ogle 1991, 1992). Ruffe established a poplation there and by the early 1990s had included several other river mouths of Lake Superior in their range (Pratt et al. 1992; Slade and Kindt 1992). Ruffe are currently also found in Lake Huron (Ogle 1998).

Literature Synthesis

The general taxonomic nomenclature of ruffe is generally well agreed upon, with the exception of subfamily or tribal membership. Collette (1963) and Collette and Banarescu (1977) placed Gymnocephalus in the tribe Percini within the subfamily Percinae (includes Perca and Percarina). Etheostomatini was the other tribe within Percinae. Page (1985) removed Etheostomatini from Percinae, leaving Percinae with only Perca and Gymnocephalus (Figure 3). He felt that egg-stranding, thought to be common in Perca and Gymnocephalus, was vastly different from egg-burying, common in darters. This change is tenuous because I have found no evidence that Gymnocephalus produce egg-strands (see Reproduction and Early Life History section). Wiley (1992) proposed that Gymnocephalus should be in Etheostomatinae, a subfamily that included all Percid genera except Perca. In this organization, Gymnocephalus is the basal taxon for all Percids except Perca (Figure 3). Coburn and Gaglione (1992) rejected Wiley's proposal because an equally parsimonious relationship with Perca and Gymnocephalus in the same subfamily could be found by simply including Page's egg-stranding information. Again, the inclusion of egg-stranding is tenuous, but Coburn and Gaglione's analysis is similar to Collette's, in that Perca and Gymnocephalus are closely related and probably sister groups. A general nomenclature for ruffe is shown in Table 2.

Figure 3. Three cladograms depicting alternative hypotheses of relationships among percids (from Coburn and Gaglione 1992): (A) Collette (1963) and Collette and Banarescu (1977); (B) Page (1985); (C) Wiley (1992). Abbreviations: Et = Etheostomatinae; Lu = Luciopercinae; Pe = Percinae; E = Etheostomatinae; L = Luciopercini; P = Percini; R = Romanichthyini; Gym = Gymnocephalus; Per = Perca; Rom = Romanichthys; Stz = Stizostedion; Zin = Zingel.

Table 2. General Taxonomic nomenclature of ruffe. Note that there is some controversy.

Morphological Description
Several morphological characters led to the taxonomies described above. Percinae are separated from Luciopercinae by an enlarged anteriormost interhaemal bone, large and well developed anal spines, and a lateral line that usually does not extend onto the caudal fin (Collette 1963). Percini are separated from Etheostomatini by a well developed swimbladder, a compressed body, an auxiliary interneural bone that supports the first dorsal fin, a well developed supraoccipital crest, a strongly serrated preoperculum, small or absent genital papilla, and lack of breeding tubercles (Collette 1963). Characteristics synapomorphic for Gymnocephalus are enlarged preopercular spines, a distinctly shorter anterior ramus compared to the ventral ramus, a club-like distal end of the premaxilla, a retrarticular that is distinctly higher than long when viewed medially (Wiley 1992), and hypertrophy of the cephalic lateral-line canals (Collette and Banarescu 1977). In addition, two characteristics found only in Gymnocephalus and Zingel were a distinctly enlarged foramina of the lower jaw and a reduced or absent articular process. Scales of Gymnocephalus, Percina, and Stizostedion are identical with respect to eight characters observed by Coburn and Gaglione (1992). Gymnocephalus appear to have fewer vertebrae than other percids with the exception of Percarina and some Etheostoma (Collette 1963). Within Gymnocephalus, G. Baloni and G. cernuus have a similar number of vertebrae, which are significantly fewer than G. acerina and G. schraetser (Holcik and Hensel 1974). Based on vertebrate counts (and some coloration differences), Holcik and Hensel (1974) separated Gymnocephalus into two subgenera, Gymnocephalus and Acerina (Figure 3). Elshoud-Odenhave and Osse (1976) thoroughly described the cephalic skeletal and muscular systems of ruffe. A compilation of meristics and morphological characters of ruffe is shown in Table 3. Matkovskiy (1987) determined the relationship between total length and size of several body parts.

Figure 4. Presumed phylogeny of the genus Gymnocephalus (from Holcik and Hensel (1974)).

Different morphotypes of ruffe may exist. "Shallow-bodied" and "deep-bodied" forms of ruffe existed in the same stretches of the Dneiper (Aleksandrova 1974). Differences between littoral- and pool-dwelling populations of ruffe in Syamozero were identified (Kuderskiy 1966). Some of the morphological and meristic characters that differed between the two groups were the number of dorsal rays and spines, scales above and below the lateral-line, opercular spines, vertebrae, and the relative size of the head, snout, eyes, and fins (Aleksandrova 1974).

In addition to different morphotypes, geographical variation in meristic and morphological characters exist (Biliy 1967; Witkowski and Kolacz 1990). There is a general gradient (longer to shorter) from east (Europe) to west (Siberia) in the following characters: maximum and minimum body depth, predorsal and preanal distance, depth of first dorsal fin, length and depth of anal fin, and length of pectoral fin (Witkowski and Kolacz 1990). Differences may occur over a smaller geographic distance. Ruffe from the Nadym River had a more fusiform body, shallower head, longer snout, and pelvic, pectoral, and dorsal fins positioned more posteriorly than ruffe from the nearby Ob' (Kolomin 1977). Ruffe from the lower Ob' were deeper bodied than the fusiform ruffe from the upper Ob' (Petlina 1967). Morphological differences may be caused by environmental differences among locations (Witkowski and Kolacz 1990). For example, differences between ruffe from the Nadym and Ob' Rivers may be an adaptation to long spawning and overwintering migrations (Kolomin 1977).

Sexual dimorphism in ruffe is present in some characters. The morphological characters that differ might differ between populations (Petlina 1967). Opalatenko (1967) found that least body depth, head length, pelvic fin length, and upper lobe of the caudal fin length were greater in males than in females (Opalatenko 1967). Petlina (1967) found differences between the sexes in pelvic fin length, eye diameter, head length, body thickness, head depth through the middle of the eye, and preventral, prepectoral, and anal-caudal distance. Sexual dimorphism in 30 morphological and meristic characters was not observed in either "deep-bodied" or "shallow-bodied" ruffe from the Dneiper (Aleksandrova 1974). Opalatenko (1967) found no sexual differences in meristic characters. Furthermore, Shmidtov and Varfolomeyev (1952), Shilenkova (1962), and Biliy (1967) did not observe sexual dimorphism in ruffe.

Larval ruffe are distinguished from all native North American percids by (1) "a slightly concave head becoming attenuate at larger length intervals", (2) "a pointed snout with teeth on the maxillary and premaxillary" by 11 mm, (3) "a large dorsally pigmented swim bladder", (4) "a serrated preopercle", and (5) "few postanal myomeres (usually 18-22)" (Simon and Vondruska 1991).

Ruffe have 2n=48 chromosomes (Lieder 1964; Nygren et al. 1968; Bozhko et al. 1978; Rab et al. 1987; Klinkhardt 1990). Bozhko et al. (1978) found 2 metacentric, 11 submetacentric, 8 subelocentric, and 3 acrocentric chromosome pairs. Rab et al. (1987) discovered 1 metacentric, 16 submetacentric, 4 subelocentric, and 3 acrocentric chromosome pairs. Mayr et al. (1987) found only one acrocentric chromosome pair (no. 14). Logvinenko et al. (1983) described a diallelic codominant system of inheritance at two independent loci and proposed that these alleles may serve as genetic markers. Nyman (1969, 1975) demonstrated that ruffe have a simple two-allele serum esterase polymorphism and suggested that this information could be used in population investigations. The karyotypes of the four Gymnocephalus species are unique at the species level (Rab et al. 1987).

All known Gymnocephalus fossils are G. cernuus from interglacial deposits in Denmark, Germany, Russia, England, and Poland (Holcik and Hensel 1974). G. cernuus are apparently of Paleo-Danube origin (Holcik and Hensel 1974; Rab et al. 1987) and may have arose from Perca (Collette and Banarescu 1977; Rab et al. 1987). Collette et al. (1977) hypothesized that the lack of centrarchids in Europe may have allowed for the evolution of the four moderate-sized Gymnocephalus species. G. cernuus and G. baloni appear to be more primitive than G. acerina and G. schraestser, with G. baloni apparently derived from G. cernuus (Holcik and Hensel 1974). Thus, G. cernuus may be the basal Gymnocephalus species with most speciation due to geographic isolation (Holcik and Hensel 1974).

Figure 5. The position of the cephalic lateral-line canals on ruffe (from Kuiper 1967).

The ultrastructure and electrical physiology of the lateral-line system has been actively studied (Jielof et al. 1952; Flock 1967). Ruffe are sensitive to frequencies of 50-150 cycles per second (cps) and are able to distinguish between the low and high end of this range (Kuiper 1967). Van Netten (1991) found that the "hydrodynamic excitation of the cupulae ... can satisfactorily be described with a frequency-dependent combination of viscous and inertial fluid forces, which result from the boundary layer around the cupula." Gray and Best (1989) and Denton and Gray (1989) described the relative stimulation of each cephalic neuromast as a ruffe passed over a prey source (Figure 6). The stimulus pattern changes as the fish moves over the source. This change in nerve activity should allow ruffe to accurately detect prey. Wubbels (1991) described a phase reversal of the responses of adjacent neuromasts that may allow ruffe to detect and locate a moving object.

Figure 6. Diagram from Gray and Best (1989) to illustrate changes of stimulus pattern to the head neuromasts. Within each of the three parts (A, B, & C) are three diagrams arranged in a column representing from above down the supraorbital, infraorbital, and mandibular canals. Within each diagram the posterior neuromast is the left and the anterior to the right. Each vertical line is proportional to the velocity of liquid at that neuromast, relative to that at neuromast 16, for a transient in one direction; the direction of the line above or below the base indicates the relative direction of the velocity along the canal. A full line outlines the envelope of the vertical lines; a similar dashed line indicates whet the envelope would be like for an identical transient of the opposite polarity. The diagram has been calculated for a fish moving forward with its mouth at a constant distance of 10 mm above a 'mud surface' and its body held at 60° to that surface (see inset). The source is taken as being at the surface with its axis of vibration along the surface in the direction of the fish's axis. A, is for a fish 5 mm before reaching the surface; B, is when the mouth is exactly above the surface; C, is for a position 5 mm beyond the source.

Ruffe possess a retinal tapetum lucidum similar to that of Luciopercini (Ahlbert 1970). The tapetum lucidum is a reflective material found in the dorsal two-thirds of the pigmented epithelial layer of the ruffe retina (Wunder 1930; Harder 1975; Craig 1987). In the dark-adapted state (low light), light is reflected back and forth between the tapetal processes with additional absorption by the rods after each reflection (Harder 1975; Zyznar and Ali 1975; Ali et al. 1977). Ali and Anctil (1968) found a relationship between the developmental state of the tapetum lucidum and habitat use by walleye (Stizostedion vitreum) and sauger (Stizostedion canadense). Sauger, which have a more developed tapetum lucidum than walleye, prefer more turbid waters. Thus, the presence of a tapetum lucidum aids vision in low-light conditions.

The photoreceptors of ruffe are arranged in an irregular twin cone pattern that varies between row and square configurations (Ahlbert 1970; Collette et al. 1977). The existence of a square pattern is positively correlated with poor movement perception (Ahlbert 1970). The greatest cone density in ruffe is distributed in the temporal region (Sroczynski 1981), suggesting that ruffe are positively adapted for vision in the posterior field and that vision in the anterior field is of lesser importance. Further specific aspects of the ruffe optical system were described in Sroczynski (1981).

Ruffe are found in all habitats of the SLRH (Selgeby and Ogle 1992). Ruffe are found in the deepest channels (8 - 10 m) at ice-out, move into the shallows to spawn, remain in 1-3 m water throughout the summer, and then return to the deeper channels in September and October. Adult ruffe are generally found in deep, dark waters during the day, move to shallower waters to feed at night, and return to deeper waters at dawn (Ogle 1992). Large numbers of age-0 ruffe were captured in both deep and shallow waters at night, but only in shallow water during the day, although substantially fewer were caught there than during the night (Ogle 1992).

The abundance of ruffe generally increases with eutrophication until hypereutrophy is reached (Entz 1977; Leach et al. 1977; Hansson 1985; Johansson and Persson 1986; Bergman 1991; Persson et al. 1991; Figure 7). In contrast to ruffe, the abundance of most other percids is greatest in mesotrophic waters (Leach et al. 1977; Ryder and Kerr 1978). Ruffe abundance increased with anthropogenic additions of nutrients in several situations (Heinonen and Falck 1971; Anttila 1973; Hansson 1987; Neuman and Karas 1988), but Peirson et al. (1986) found that ruffe abundance increased after severe nutrient additions were eliminated. Leopold et al. (1986) found no correlation between level of eutrophication and ruffe catch because year-to-year ruffe catch was highly variable. Biro (1977) reported that ruffe abundance declined with increasing eutrophication, but did not state where on the trophic continuum the increase occurred.

Figure 7. Schematic representation of the relationship between productivity and coregonid, cyprinid, and percid abundance (top panel; from Bergman 1990a) and abundance of perch and ruffe in eight Swedish lakes ordered by increasing chlorophyll a values (bottom panel; from Bergman 1991).

There are four possible hypotheses to explain the correlation between increased ruffe abundance and increased eutrophication. First, ruffe forage more efficiently under the reduced light conditions associated with increased algal production (Johansson and Persson 1986; Bergman 1988, 1991). Second, the benthos may increase in abundance and diversity and shift to smaller species in response to the storage of increased energy in the sediment due to eutrophication (Leach et al. 1977). Ruffe are primarily benthic feeders and would presumably be favored by the increased benthic production and shift to smaller species. Third, increased productivity may release predation pressure on ruffe (Bergman 1991). Fourth, ruffe may simply be physiologically more tolerant of eutrophic conditions than other percids.

Ruffe appear to be tolerant of a wide range of temperatures (Bergman 1987). Ruffe have been found to be abundant in water as cold as 0-2ºC (Neuman 1979). The upper lethal temperature for juvenile ruffe appears to be 30.4ºC, whereas the critical thermal maximum is 34.5ºC (Alabaster and Downing 1966 and Horoszewicz 1973, as summarized in Hokanson 1977). The upper lethal temperature for ruffe was near the middle of 9 species tested (Alabaster and downing 1966). Evidence of stress did not occur below 29.8ºC (Horoszewicz 1973). Optimal temperature for larval growth is 25-30ºC (Kammerer 1907), whereas optimal growth for age-0 ruffe is 21ºC (Edsall et al. 1993). The foraging ability (capture rate, handling times routine swimming performance, and reaction distance) is affected by temperature, but not nearly as drastically as that of perch (Bergman 1987). Nyman (1975) concluded that ruffe may be able to genetically adapt to a wide range of temperatures. In contrast, Hokanson (1977) states that the temperature requirements of G. cernuus, Perca fluviatilis, and Perca flavescens can not be distinguished with the available data.

Optimal temperature for growth of age-0 ruffe was at 21°C, with a fundamental thermal niche of 18-22°C (Edsall et al. 1993). However, ruffe grew at all temperatures in the test range (7.0-24.8°C). The fundamental thermal niche for age-0 ruffe corresponds closely to that of walleye, sauger, and yellow perch. Thus, using a relationship developed for walleye, the authors estimated that 58% of Lake Erie, 21% of Lake Huron, 12% of Lake Michigan, 7% of Lake Ontario, and 2% of Lake Superior was optimal growth habitat for ruffe.

Ruffe mortalities occur at low levels of oxygen and high levels of toxic chemicals. Ruffe, which prefer oxygen concentrations of 5-6 mg/l (Holcik et al. 1989), died in 1-3 h at oxygen concentrations of 0.69-0.97 mg/l (Holcik 1986). Severe mortalities of ruffe may occur in the anoxic hypolimnion of some lakes (Lelek 1987). Ruffe prolarvae were deformed and died when exposed to 1 mg/l triethyl-stannic chloride solutions (Danil'chenko 1982). Development ends at the prolarvae stage when ruffe are exposed to 10 mg/l silicylanilide, 0.1 mg/l triethyl lead chloride, and 0.1 mg/l pentachlorphenolate of sodium (Danil'chenko and Stroganov 1975). Ruffe eggs are resistant to some toxins that embryos are sensitive to (Danil'chenko 1977).

Age at maturity has increased since ruffe were first discovered in the SLRH. During the initial years of their invasion of the SLRH (1988 and 1989), nearly 100% of ruffe were mature after one year of life (Selgeby and Ogle 1991, 1992). By 1990 and 1991, only about 85% (Selgeby and Ogle 1992) and in 1992, only about 50% (Selgeby 1993) of age-1 ruffe were mature. Mean absolute fecundity of a 15 cm female ruffe is approximately 45000 eggs (Selgeby and Ogle 1992).

Ruffe are categorized as non-guarding, open substrate, phytolithophil spawners, laying eggs on available submerged plants in clean water habitats, or on other submerged items such as logs, branches, gravel, or rocks (Balon et al. 1977). The summaries of Maitland (Maitland 1977; Maitland et al. 1983) agree with Balon's generalizations, but the review of Collette et al. (1977) did not agree with Balon, as they concluded that ruffe spawn on hard bottoms of sand, clay, or gravel. Available field evidence supports both conclusions. Nadym River ruffe lay eggs on submerged vegetation or silted snags at approximately 2 m (Kolomin 1977). In Denmark, ruffe have been found to move from deep areas to shallow areas of stone-sand bottoms, sometimes near vegetation, or into rivers to spawn (Johnsen 1965). In Lake Ilmen, ruffe spawn on firm sand, sand-stone, or clay bottoms in less than 3 m of water (Kovalev 1973; Fedorova and Vetkasov 1974) or, on occasion, ruffe will spawn on plant remains and moss in abandoned smelt spawning grounds (Fedorova and Vetkasov 1974). In the Danube, ruffe spawn along shore line in "rock-fills" (Specziar and Vida 1995). In summary, ruffe spawn on a variety of substrates at depths of approximately 3 m or less.

Ruffe spawn between mid-April and July (Collette et al. 1977; Hokanson 1977; Neja 1988); beginning when the water warms to 6.0°C and continuing to water temperatures between 12 and 18°C. Nadym River ruffe begin spawning in June when the river is ice-free and temperatures are between 6 and 8°C (Kolomin 1977). Lake Ilmen ruffe spawn from mid-April to June when water temperatures are between 6.5 and 13°C (Kovalev 1973; Fedorova and Vetkasov 1974). Ruffe in other Russian waters spawn at temperatures between 6 and 10°C (Berka 1990). Lake IJssel ruffe spawn in May and June at water temperatures of 12 to 18°C (Willemsen 1977). Ruffe in Lake Aydat spawned at temperatures between 15.0°C and 18.0°C. Knowles (1974) reported that ruffe were "ready to spawn at unusual times of the year", presumably late summer and fall. The lowest recorded spawning temperature of 11.6°C for ruffe, as reported by Hokanson (1977), appears to be erroneous.

Very little other physical or chemical parameters concerning ruffe spawning are reported in the literature. Kiyashko and Volodin (1978) did, however, determine that ruffe eggs will develop normally at pH values between 6.5 and 10.5, one of the widest ranges from a broad set of fish tested.

Available evidence indicates that ruffe eggs are extruded from the female without sticking together, but will, upon contact with the water, become adhesive and stick to the substrate (Johnsen 1965; Kovalev 1973). Page (1985), however, categorized ruffe as "stranders" with "the unique habit of encasing their eggs in long gelatinous stands" and cited Seeley (1886) and Wheeler (1969) as sources. Page did note, though, that other authors did not mention egg-stranding in their ruffe spawning summaries. No evidence for egg-stranding appears in the literature located for this review.

Ruffe may spawn intermittently, laying eggs in two or more batches (Koshelev 1963; Travkina 1971; Kovalev 1973; Fedorova and Vetkasov 1974; Kolomin 1977; Craig 1987; Neja 1988; Jamet and Desmolles 1994). Koshelev (1963) extensively studied the physiology of intermittent spawning ruffe. Rapid and asynchronous oocyte growth and the lack of a resting stage II oocyte led to batch spawning and longer spawning periods in ruffe. The first batch of oocytes mature in 165 days during winter and spring, whereas the second batch matures in 30 days during summer. Resorption of the first batch of ova does not interfere with growth of the second batch because resorption and growth occur simultaneously. The number and size of eggs is reduced in the second batch (Kolomin 1977). Neja (1988) notes that at one point in time the ruffe contains three types of eggs: (1) small, hyaline, and colorless, (2) larger, opaque, white or pale yellow to yellow and orange in color, and (3) large, partly hyaline, and yellow-orange and orange in color. Only type (2) and (3) eggs will be released in an upcoming spawning event.

The number of eggs laid depends on the size of the female (Neja 1988, Kolomin 1977) and the batch of eggs (Kolomin 1977). Individual absolute fecundity (total eggs per female) was 13388-82233 in Lake Dabie (Neja 1988), 6900-64665 in Ob' River (Petlina 1967), and 4220-29600 in the first batch and 352-6012 in the second batch in the Nadym River (Kolomin 1977). Maximum absolute fecundity appears to be about 200000 eggs (Collette et al. 1977; Berka 1990). In contrast to absolute fecundity, the relative fecundity of Lake Dabie ruffe was uncorrelated to body size, gonad weight, or age (Neja 1988). Relative fecundity was between 305-1540 eggs per g of fish (Bastl 1983; Neja 1988; Jamet and Desmolles 1994). Average gonosomatic index for spawning female ruffe was between 7.11 and 15.6 (Kolomin 1977; Neja 1988; Jamet and Desmolles 1994), with an individual maximum of 27.78 (Bastl 1983). The gonosomatic index for spawning males was between 7.0 and 9.96 (Bastl 1983; Neja 1988; Jamet and Desmolles 1994).

Egg diameter also depends on the size of the female and the batch of eggs. Egg diameters ranged from 0.34 to 1.3 mm (Johnsen 1965; Kovalev 1973; Fedorova and Vetkasov 1974; Disler and Smirnov 1977; Kolomin 1977; Bastl 1983). Bastl (1983) reported that mean egg weight was 0.45 (± 0.05) g and mean egg volume was 0.59 (± 0.48) mm3. Eggs from a first batch are larger (0.90-1.21 mm) than eggs from a second batch (0.36-0.47 mm) and yellow compared to whitish in color (Kolomin 1977).

Gonadal activity in males occurs during spring and autumn waves of spermatogenesis (Butskaya 1980, 1985), with the autumn wave being more important (Butskaya 1985). In the laboratory, spermatogenesis and the extrusion of ripe sexual products in ruffe are thermo-dependent processes that do not depend on a fixed daylight regime. There is a continuum of effects of temperature on gonadal activity. At extreme temperature (< 2°C and > 18°C), spring spermatogenesis is limited. From 5 to 15°C, the rate of spermatogenesis increases with increasing temperature. From 15 to 17°C individual sensitivity is evident, as shown by a negative effect on spermatogenesis. The number of degree days required for the development of spermatogenesis in overwintering ruffe, prior to the formation of new spermatozoids, was greater under experimental conditions than during spring spermatogenesis in nature. This observation suggests that the gonadotrophic system of males in nature is influenced by the synergic action of additional factors not detected by his experiments, such as photoperiod or temperature (Butskaya 1980). Furthermore, in nature, temperature may interact with other forces to influence the presence of females on the spawning grounds and to initiate spawning conditions in males (Butskaya 1980).

Instances of hermaphroditism in ruffe have been recorded (Knowles 1974; Butskaya 1976). In the Gulf of Finland, approximately 25% of all ruffe gonads had both testicular and ovarian cells (Butskaya 1976). In 85% of these fish, the sexual organs are of a "testicular type." About half of these fish function as "normal" males. In the remaining half, fecundity is varyingly decreased from the "normal" levels down to none. In only 2% of the fish can a changeover to functional hermaphroditism occur. The presence of large numbers of these "inter-sexual" fish did not have a negative influence on overall ruffe reproductive levels.

Ruffe eggs hatch in 5 to 12 days at temperature between 10 and 15°C (Johnsen 1965; Maitland 1977; Craig 1987; Berka 1990). The newly hatched embryo of 3.5-4.4 mm (Fedorova and Vetkasov 1974; Disler and Smirnov 1977) are in a relatively underdeveloped stage, compared to yellow perch (Perca flavescens; Disler and Smirnov 1977). The embryo remains sedentary on the bottom for 3 to 7 days until it reaches a size of 4.5 to 5.0 mm, which is the same stage of development as a newly hatched yellow perch embryo (Collette et al. 1977; Disler and Smirnov 1977). Disler (1960) and Disler and Smirnov (1977) described the morphological development of ruffe larvae.

French and Edsall (1992) described the hatching of artificially fertilized eggs and the development of ruffe protolarvae from the SLRH. Detailed developmental descriptions will not be given here. Fertilized eggs are 0.9-1.2 mm diameter. Newly hatched protolarvae are 2.5-3.2 mm TL. Feeding and swimming begin when the yolk-sac is fully absorbed about 1 week after hatching. Ruffe protolarvae are distinguishable from other Lake Superior percids by having fewer preanal myomeres, the head deflected over the yolk sac, continuous finfold of even width, and a total length of less than 4.0 mm.

Ruffe have no, or only a brief, pelagic larval stage (Johnsen 1965; Fedorova and Vetkasov 1974; Disler and Smirnov 1977). The onset of active feeding and the later transition to exogenous feeding (diet described in the Diet and Foraging Behavior section) takes place in the benthopelagic layer (Disler and Smirnov 1977), within about one week of hatching (French and Edsall 1992). Larval ruffe are positively phototactic (Disler and Smirnov 1977). At the larval stage, ruffe are secretive and solitary, not forming schools (Disler and Smirnov 1977). For these reasons, ruffe larvae are not highly vulnerable to plankton sampling gear (Disler and Smirnov 1977; French and Edsall 1992).

The temperature requirements of young ruffe have been determined. The lower TL50 for embryos is 10°C and the upper TL50 is 21.5°C (Hokanson 1977). Survival of larval ruffe is poor below 10°C (Hokanson 1977). Optimal growth of larval ruffe occurs between 25 and 30°C (Hokanson 1977), whereas optimal growth of age-0 ruffe was 21°C (Edsall et al. 1993). However, age-0 ruffe grew at temperatures between 7.0 and 24.8°C (Edsall et al. 1993).

Most authors used scales to age ruffe (Aleksandrova 1974; Fedorova and Vetkasov 1974; Holker and Hammer 1994). Jamet and Desmolles (1994) found that ruffe produced only one annulus per year and that the annulus was formed in the winter. Holker and Hammer (1994) provided a relationship between scale radius and ruffe total length.. In some instances, scales have been found to be inadequate for assigning age (Mills and Eloranta 1985). Thus, dorsal spine or otolith sections have been used to assign age (Bast et al. 1983). Aleksandrova (19974) used otoliths as "controls" for scale-assigned ages, but did not mention any discrepancies between the two methods. Doornbos (1979) and Matkovskiy (1987) provided a relationship between otolith length and ruffe total length.

Ruffe are typically less than 20 cm, rarely attain a size greater than 25 cm (Lind 1977; Craig 1987; Lelek 1987; Berka 1990), but may be as large as 29 cm (Moller et al. 1988). Ruffe mature at a young age (see Reproduction and Early Life History section), so most of the overall length is attained in the first or second year (Table 4; Kolomin 1977).

Growth of ruffe is affected by sex, morphotype, water type, intraspecific density, and food supply. Females typically grow faster than males (Table 4; Bast et al. 1983; Berg 1949; Fedorova and Vetkasov 1974; Neja 1989; Holker and Hammer 1994), although, in some instances no difference was observed (Neja 1989). Aleksandrova (1974) determined that growth was different between shallow- and deep-bodied forms of ruffe. Hansson (1985) found that ruffe growth was reduced in areas where ruffe densities were high. Ruffe captured in clear and brackish waters tend to grow faster (Neuhaus 1934; Lind 1977; Bast et al. 1983). Bergman and Greenberg (1994) showed a clear decline in growth of ruffe in enclosures with increasing densities of ruffe. Poor ruffe growth may also result if the benthos is impoverished (Boikova 1986; Bakanov et al. 1987) or, in one case, largely inaccessible due to oxygen deficiencies (Boikova 1986).

The principal prey of juvenile and adult ruffe are chironomids or macrocrustaceans (Table 5). The principal genera of chironomids consumed are Chironomus (especially plumosus) and Procladius spp. (Fedorova and Vetkasov 1974; Boron and Kuklinska 1987; Nagy 1988; Kangur and Kangur (1996); Kangur et al. 1999, 2000). The prevalence of chironomids in the diet may decrease with age (Leszczynski 1963; Fedorova and Vetkasov 1974). Other macrobenthos that are prevalent in the diet are Ephemeroptera, Trichoptera, and Hirudinea (Table 5). Ruffe collected from brackish or very deep waters, also feed heavily on macrocrustaceans such as Pallasea quadrispinosa, Pontoporeia affinis, Mysis relicta, Neomysis integer, Diaporeia affinis and Gammarus spp. (Sandlund et al. 1985; Van Densen 1985; Selgeby 1998; Table 5). For example, Hansson (1984, 1986) found the primary prey of ruffe in brackish waters to be Gammarus spp. and Pontoporeia affinis, with chironomids important at some locations. Larger ruffe eat some fish (Fedorova and Vetkasov 1974; Kozlova and Panasenko 1977; Bagge and Hakkari 1985). Ruffe of all sizes may eat fish eggs (see Egg Predation section).

The diet of ruffe differs little between lakes with different trophic status or ruffe densities or spatially within lakes. Ruffe fed mainly on chironomids and ephemeropterans in both lakes of moderate and high productivity, although diet breadth was greater in the more productive lake (Bergman 1991). Bergman and Greenberg (1994) determined that ruffe diet consisted of mostly macrobenthos at all levels of ruffe density, with only the contribution of trichopterans and Pisidium affected by ruffe density (Bergman and Greenberg 1994). Hansson (1987) and Nilsson (1979) found ruffe diet to differ little between sampling locations, but Hansson (1984) found differences between locations in a brackish archipelago.

After switching to a diet composed of largely macrobenthos early in life, ruffe diet changes little with size. Bergman (1988) found the diet of three size-classes of ruffe to be similar. Boron and Kuklinska (1987) also did not find any diet differences due to size of age-1 and older ruffe and Jamet and Lair (1991) came to the same conclusion for age-2 and older ruffe. Boikova (1986) identified a slight shift in ruffe diet at 8-10 cm, although the shift is not very evident.

Ruffe appear to continue to feed in late autumn and winter, although not to the extent as in summer months (Kangur and Kangur 1996; Kangur et al. 1999, 2000).

Ruffe in the SLRH are primarily benthophagous. Age-0 ruffe fed mostly on cladocerans and copepods in early summer and chironomids in later summer and fall. Adult ruffe <12 cm fed mostly on chironomids and other macrobenthos, but also consumed large numbers of microcrustaceans. Adult ruffe >12 cm fed mostly on chironomids, Hexagenia spp., and caddisflies. Ruffe consumed very few fish eggs. Patterns in catch rates and stomach contents weight suggest that all age-0 and adult ruffe (in deeper waters) fed throughout the day. Adult ruffe, though, moved to shallower waters at night to feed most heavily. Local weather systems and spawning activities may have disrupted these patterns. Ogle (1992) also calculated daily rations for adult ruffe (0.004-0.046 g per g of fish) and made the first gastric evacuation calculations for ruffe.

Juvenile and adult ruffe may select chironomids (Leszczynski 1963; Nagy 1986), ephemeropterans (Nagy 1986), and Sialis spp. (Bergman 1990b; Bergman and Greenberg 1994), but select against oligochaets (Leszczynski 1963; Nagy 1986) and Hirudinea (Nagy 1986). On a species level, some species of chironomids may be selected against (Nagy 1986). In addition, age-0 ruffe may select larger Daphnia and copepods (Van Densen 1985). Johansson and Persson (1986), Kangur and Kangur (1996), Kangur et al. (1999) summarized that ruffe are not selective and consume organisms in proportion to the organisms abundance.

Prey items found in the diet can give some indication of ruffe habitat use and feeding behavior. Adult ruffe probably feed in the littoral or sublittoral zones (Leszczynski 1963; Holcik and Mihalik 1968; Boron and Kuklinska 1987; Jamet and Lair 1991). A high percentage of Chaoborus in large ruffe led Shamardina (1967) to conclude that large ruffe are in deeper water during the summer than smaller ruffe, a conclusion also made by Bagge and Hakkari (1985). Holcik and Mihalik (1968) felt that ruffe moved to shallow waters to feed in the evening. Ruffe probably feed in soft bottom areas because they penetrate the bottom substrate to capture some prey (Boron and Kuklinska 1987), as evidenced by predation on burrowing chironomids such as Chironomus plumosus (Hilsenhoff 1966; Coffman 1978) and Procladius spp. (Ford 1962; Coffman 1978). However, high levels of habitat complexity may lead to a reduction in the ruffe predation rate (Mattila 1992). Ruffe in very deep waters typically consume Mysis relicta, Pontoporeia affinis, and Pallasea quadrispinosa (Sandlund et al. 1985).

The feeding behavior and functional morphology of the feeding system in ruffe was described by Elshoud-Oldenhave and Osse (1976). In this paper, the movements and interactions of the cephalic skeletal and muscular systems during a feeding event were described in great detail. The author's developed a general model of teleost feeding from these observations. In addition to the development of this model, interesting observations on ruffe feeding behavior were described.

Ruffe appear to have two types of feeding behavior (Figure 8; Elshoud-Oldenhave and Osse 1976). "Horizontal" feeding occurs when ruffe visually detect the prey in the water column, approach the prey, and suck the prey into it's mouth. "Back-lifting" feeding occurs when a ruffe detects the prey on the bottom, lifts and curves it's body so that the ventral portion of it's head is at a 20-30° angle to the prey, and then sucks the prey into it's mouth. If the ruffe detect the prey visually then it will assume the "back-lifted" position as it approaches the prey. In contrast, if the ruffe detects the prey with the cephalic lateral-line system then it must swim backwards to get into the "back-lifted" position. Debris captured in the sucking action is expelled through the opercular cavity or spat out. In most cases, a last spitting action follows the final swallowing of the prey.

Figure 8. Two types of ruffe feeding -- A = "Horizontal", B = "Back-lifting" -- adapted from Elshoud-Odenhave and Osse (1976).

Ruffe can feed under a variety of conditions but seem to exhibit crepuscular patterns of activity or stomach fullness. Westin and Aneer (1987) found ruffe held in aquariums exposed to natural light levels were generally active during twilight periods with a tendency toward diurnal activity patterns in the winter. Ogle (1992) found adult ruffe to be capable of feeding throughout the 24-h period, but did not feed during the day, began feeding at dusk, and continued to feed at night on some sample dates. Jamet and Lair (1991) provided limited evidence that ruffe fed mostly at night. Adams and Tippett (1991) claimed that no diel feeding periodicity was observed for ruffe, but offered no evidence for this conclusion.

The primary sensory system used to detect prey by ruffe appears to be the lateral-line system. The physiology of the extremely sensitive cephalic lateral-line system is discussed in the Sensory Physiology section. Blinded ruffe could localize immobile prey that were made mobile by "involuntary trembling of the hand" (Kuiper 1967). Furthermore, electromagnetic stimulation of nerves attached to the cupulae cause ruffe to "snap for food" (Kuiper 1967). Gray and Best (1989) concluded that the ruffe lateral line is sensitive enough and has sufficient discriminatory power to detect and locate a source 2-5 cm from it's snout. Denton and Gray (1989) also concluded that ruffe could detect and locate chironomid larvae in the top layers of the bottom substrate using the head canals.

Vision may also be used to detect prey by ruffe. The organization of the cone cells in the retinae and presence of the tapetum lucidum (see the Sensory Physiology section) is consistent with the bottom-feeding behavior of ruffe (Ahlbert 1970). Thus, vision may be used to locate prey, even in low-light situations. Ruffe had relatively high levels of choleacetyltransferase and acetycholine leves in the brain, which is typical of fish with well-developed visual systems (Szabo et al. 1991). However, Ahlbert (1970) suggested that vision might not be that important in ruffe because the ruffe's square cone pattern is related to poor movement perception. Furthermore, Bergman (1987) showed that the reaction distance of ruffe was only 4 cm, compared to 21 cm for perch.

Bergman (1988) provided both field and experimental data to support the conclusion that ruffe are adapted for feeding in low-light conditions. Ruffe were found in all zones of the lake but were most abundant in low-light benthic areas. In the laboratory, the feeding ability (i.e., attack success, capture rate, handling time, and swimming speed) of ruffe was affected little by reduced light levels. The attack success and capture rate by ruffe decreased little with decreasing light intensity.

Only one calculation of annual per capita food ration of ruffe appears in the literature. Kozlova and Panasenko (1977) determined that an "average" ruffe consumed approximately 130 g of food per year and 870 g of food per lifetime.

Figure 9. Percent occurrence and weight of prey in Loch Lomond ruffe (from Adams and Tippett 1991).

Ruffe have been implicated in the decline of several Coregonus stocks. Pokrovskii (1961) described several cases where Coregonus stocks have declined, presumably due to egg predation by ruffe. Unfortunately, Pokrovskii (1961) provided little supportive data and cited documents that are exceedingly obscure. In some instances, 80-90% of spawning C. albula eggs were eaten by ruffe which may result in a two or three fold decline in C. albula catch. Good catches of C. albula occurred in Svaytozero Lake, where the Tetracotyle parasite caused a mass mortality of ruffe (Pokrovskii 1961). In other Russian lakes, there was an inverse relationship between the catch of ruffe and C. lavaretus in subsequent years (Balagurova 1963; Titova 1973). In Loch Lomond, Scotland, ruffe were found to prey substantially on the eggs of C. lavaretus during and immediately after the C. lavaretus spawning period (Adams and Tippett 1991). Adams and Tippett (1991) thought that the rate of predation on the eggs of this stock could be great enough to negatively affect the abundance of C. lavaretus. However, ruffe abundance, ruffe consumption rates, and number of C. lavaretus eggs laid were not estimated (Winfield 1992). Ruffe have also been found to prey on smelt eggs (Fedorova and Vetkasov 1974).

Competition between ruffe and perch is likely high because they consume the same food items and ruffe are generalists in respect with several parameters that dictate habitat use. Perch undergo three ontogenetic diet shift in their lifetime: first they feed on zooplankton, then on benthic macroinvertebrates, and finally fish (Collette et al. 1977; Hartman and Numann 1977; Johansson and Persson 1986). During the benthivorous stage, when perch eat mostly chironomidae and ephemeroptera, diet overlap with ruffe will likely be high. For example, diet overlap between ruffe and small (< 17.5 cm) and large (> 17.5 cm) perch was significant in a Baltic archipelago (Hansson 1984). Furthermore, because the foraging ability of ruffe is more independent of light (Bergman 1988) and temperature (Bergman 1987) than that of perch, ruffe should have a competitive advantage over perch (Bergman and Greenberg 1994). Previous studies showed that the presence of roach forced perch to begin feeding on macroinvertebrates at a smaller size (Persson 1986; Persson and Greenberg 1990a, b). However, when ruffe were present, perch did not increase it's consumption of macroinvertebrates (Bergman 1990b), suggesting an affect of ruffe on perch. Bergman and Greenberg (1994) showed that, with increasing ruffe density and constant roach density, that perch included more zooplankton in their diet and their growth correspondingly declined. This is evidence that ruffe and benthivorous-feeding perch compete for food resources.

Intraspecific competition has also been indirectly shown for ruffe. Both Hansson (1985) and Bergman and Greenberg (1994) found ruffe growth to decline with increasing ruffe density.

One of the prerequisites for competition is that a fish can reduce the biomass of available prey. Mattila and Bonsdorff (1989) suggested, based on feeding rates determined in the laboratory, that ruffe should be able to structure the benthic community through predation. However, in a weakly designed experiment, they found no effect of ruffe on the abundance or composition of the bottom fauna in the Baltic Sea. Bergman (1990b) and Bergman and Greenberg (1994) found that Sialis spp., a preferred food item of ruffe, were reduced significantly in enclosures containing ruffe. Nagiec (1977) suggested that a depauperate benthos may have been caused by ruffe, bream, or eel.

Ruffe was one of the species used to identify fish "assemblages" in Finland. Salojarvi and Ekholm (1990) classified 33 Finnish lakes based on the percentage catch of 14 common fish, including ruffe. Lakes with a high percentage of ruffe were generally in a cluster labeled "trash fish." The "trash fish" cluster also included a high percentage of roach and a low percentage of pike. The success of stocking Coregonus lavaretus was poor in lakes in the "trash fish" cluster. Ruffe formed a small percentage of the catch in lakes in 4 other clusters. Tonn et al. (1990) found that the presence/absence or relative abundance of ruffe was unimportant in discriminating fish assemblages for 113 Finnish lakes.

Ogle et al. (1996) investigated the feeding behavior and diet of predators (1989-1991) likely to consume ruffe in the SLRH. Laboratory studies showed that walleye, northern pike, and burbot will eat ruffe, but walleye and northern pike preferred soft-rayed fish to ruffe. Predation on ruffe in the SLRH by most predators remained low, but overall predation increased slightly from 1989 to 1991. Most ruffe eaten were age-0 (1989 and 1990) or were small age-1 fish (1991). The primary predators of ruffe were bullheads Ictalurus spp. and northern pike, but yellow perch, smallmouth bass Micropterus dolomieu, black crappie Pomoxis nigromaculatus, and burbot all ate ruffe. No ruffe were found in the nearly 1000 walleye stomachs examined. Additional work reported in Selgeby (1993), indicated that ruffe were 5-7% of the diet of walleye in 1992 and 1993.

The rate of predation on ruffe is affected by the abundance of ruffe or the availability of alternative prey. Pikeperch prefer soft-rayed fish, especially smelt, to ruffe (Deedler and Willemsen 1964; Collette et al. 1977). In some cases, ruffe are eaten in proportion to their abundance, except that larger pikeperch (> 40 cm) may increase their consumption of ruffe (Fedorova and Vetkasov 1974; Bonar 1977). Pikeperch generally consume more ruffe in years of low smelt abundance (Pihu and Pihu 1974; Popova and Sytina 1977; Samokhvalova 1982; Willemsen 1977). For example, Pihu and Pihu (1974) found that ruffe were 10-15% of the annual ration of pikeperch in years of high smelt abundance but 80-85% of the annual ration in years of low smelt abundance. In the absence of smelt, ruffe and cyprinids are the preferred food of pikeperch (Popova and Sytina 1974; Willemsen 1977). Perch also consume more ruffe in years of low smelt abundance (Popova and Sytina 1977). The diet of northern pike shifted from Coregonus lavaretus to ruffe after the introduction and population explosion of ruffe in Loch Lomond (Adams 1991).

Ruffe have a wide array of morphological, physiological, and behavioral adaptations to avoid predation. The most conspicuous of these adaptations are the large dorsal, anal, pelvic, and preopercular spines (see Figure 1). Spines have multiple anti-predation adaptations. Spines make a small forage fish appear larger than it actually is, especially to a predator that generally attacks the center of a prey mass (e.g., pike; Webb and Skadsen 1980; Eklov and Hamrin 1989). They also require that a predator swallow the fish head first, therefore, limiting the focus of the attack to the head, increasing the handling time required to maneuver the prey for ingestion, or decreasing retention. Spines may also puncture the throat or stomach linings of the predator (Eklov and Hamrin 1989; Lammens et al. 1990). In addition, ruffe are equipped with a retinal tapetum lucidum, that allows ruffe to exploit low-light twilight conditions (Ahlbert 1970), and numerous lateral-line sensors that are sensitive to large wavelength disturbances from predators (Collette et al. 1977). Finally, cryptic coloration and utilization of benthic habitats may reduce predation (Swenson 1977) as most predators forage in littoral or pelagic areas (e.g. Esocidae, Centrarchidae, some Percidae) and strike prey from beneath ("Twilight theory"; Pitcher and Turner 1986).

Selgeby (personal communications) documented the parasites of St. Louis River ruffe. A total of 22 species were found; 7 were endemic to North America, 5 were endemic to Eurasia, and 10 were of common origin or uncertain taxonomic classification. At least three of the parasites found were new to North America. Cone et al. (1994) also showed that Dactylogyrus amphibothrium was transported to the United States on the gills of ruffe.

Abnormalities of the liver and fins of ruffe have been documented. Kranz and Peters (1985) recorded the occurrence of excessive fat deposition in the hepatocytes, localized discoloration, and pale neoplastic nodules in ruffe livers and Peters et al. (1987) observed shrinkage of liver cells, blood clots, tissue necroses, and neoplastic liver nodules in ruffe. In both instances, they related the occurrence of these liver abnormalities to anthropogenic pollutants, specifically PCBs in the latter case. Lindesjoo and Thulin (1983) and Thulin et al. (1988) found instances of fin erosion and curvature in ruffe captured in bleached kraft effluents in the Gulf of Bothnia. Weissenberg (1965) noted that lymphocystis, a viral disease whose symptoms include wart-like proturbences on the fins, jaws, and opercula, was first recognized in ruffe.

Figure 10. Change in proportional contribution of ruffe and perch to the catch of fish impinged on a water intake screen in Loch Lomond, Scotland (from Maitland 1990).

Several characteristics of ruffe make them successful invaders of new communities. Taylor et al. (1984) list 12 attributes of exotic species that preadapt them for successful colonization and population growth in novel environments (Table 7). Within "broad physiological tolerances," ruffe can withstand temperature extremes and turbidity and pollution. In addition, they can withstand relatively low oxygen levels and a relatively broad range of salinities. Within "feeding habits and diet," ruffe can feed under a variety of light and temperature conditions and are well-adapted to discourage or avoid predation. In addition, they appear to be strong competitor for benthos. Finally, within "reproductive behavior," ruffe show rapid growth, early maturation, and multiple clutches. From this analysis it is apparent that ruffe are very good colonizers of new environments.

The establishment of ruffe in new areas may also be enhanced by instabilities in the invaded community. Lelek (1987) stated that "the ruffe is dominant in unbalanced populations of fishes." Instabilities in a community may predispose a community for invasion by an exotic species (Christie et al. 1987).

There is great concern as to what is the best managerial response when an exotic species invades a body of water. This concern is heightened in the case of ruffe in the Laurentian Great Lakes for three reasons. First, in Europe and Asia, ruffe are of very little economic importance but are of great ecological concern. Commercial and recreational harvest is limited due to the ruffe's small size. Furthermore, ruffe are thought to compete with perch (Perca fluviatilis) and prey on the eggs of native whitefish (Coregonus lavaretus and Coregonus albula; Pokrovskii 1961; Mikkola et al. 1979; Sterligova and Pavlovskiy 1984; Pavlovskiy and Sterligova 1986; Adams and Tippett 1991). Second, the Great Lakes are of great economic importance. Talhem (1988) reported that the total (all species) Great Lakes commercial fishery.generated $270 million in total reginal economic activity. Third, the Great Lakes management community is still in the midst of understanding and managing other harmful exotic fishes. Most notable in this group is the sea lamprey (Petromyzon marinus), which had invaded all of the Great Lakes by 1946 and has been partially blamed for the collapse of some lake trout stocks (Fry 1953; Lawrie 1970; Coble et al. 1990). The United States and Canadian governments have spent millions of dollars controlling lamprey with chemicals since the 19??s.

To determine the best management response to ruffe in the Great Lakes, a Ruffe Task Force was appointed by the Great Lakes Fishery Commission in 1991. The Task Force immediately concluded that ruffe were "a threat to North American fisheries" and felt that there was a "window of opportunity" to "prevent or delay the spread of ruffe in the Great Lakes and inland waters by containing the species to its current range in western Lake Superior" (Ruffe Task Force 1992; Busiahn 1993). However, opinions about control of ruffe in the Great Lakes fish management community ranged from deployment of all physical, chemical, and biological methods to contain ruffe to beliefs that ruffe are established and will spread regardless of control efforts (Busiahn 1993).

Some management efforts, however, have not resulted in declines of ruffe stocks. In Lake Tjeukemeer, the termination of the gillnet fishery resulted in significantly more and larger pikeperch, but catches of ruffe were not affected (Lammens et al. 1990). Predation on ruffe in Tjeukemeer did not increase, even though pikeperch abundance did increase, because the abundance of smelt, the preferred prey of pikeperch, was uncoupled with pikeperch abundance due to migration from an adjoining lake. Thus, pikeperch did not have to switch to the less desirable ruffe. In Lake Vortsjarv, intensive bottom trawling did not result in decreased ruffe numbers (Pihu 1982; Pihu and Maemets 1982).

Legislation against the use of ruffe as bait for northern pike has been used to attempt to control the spread of ruffe to other areas of England (I. J. Winfield, personal communications). This measure has been largely ineffective at restricting the spread of ruffe in England (I. J. Winfield, personal communications).

Ruffe are not specifically mentioned in management plans for several lakes where ruffe are present (Steinmetz 1990; Steinmetz et al. 1990; Van Densen et al. 1990).

It is difficult to reduce the numbers of ruffe with traditional methods (i.e., removal with traditional gear or a top-down approach) because ruffe have several adaptations for compensating for high mortality rates (Lind 1977). As mentioned in previous sections, ruffe may grow quickly, mature early, and spawn more than once in a season. For example, Lelek (1987) noted that ruffe abundance will rebound quickly if water quality improves after severe ruffe mortalities due to low oxygen.

Very few fisheries are prosecuted with the idea of catching ruffe. In the Elbe River, ruffe were an important commercial fishery at the beginning of the century with a catch of 350,000 kg in 1990 (Sterner 1916). By 1965 - 1975, ruffe catch was only 2,500 kg (Holker and Hammer 1994). In Lake Illmen, total mortality was 65% and natural mortality was 42%, leaving 23% of the ruffe being removed by the fishery (Fedorova and Vetkasov 1974). In Tjeukemeer, ruffe are captured in the fishery for baiting long-lines for eel (Goldspink and Banks 1975). Ruffe are caught incidentally is several fisheries.

Some chemicals are selectively toxic to ruffe. Bills et al. (1992) determined that ruffe could be selectively killed by application of the lampricide 3-trifluoromethyl-4-nitrophenol (TFM). Tests in the Brule River, Lake Superior suggested that an average application of 3.2 mg/l selectively killed up to 97% of ruffe and lamprey (Petromyzon marinus), but killed very few non-target finfish. Lethal concentrations of TFM were determined for ruffe under a variety of water quality conditions in laboratory experiments.

In some instances, the abundance of ruffe may be limited by benthos production. Ruffe were present in low numbers in Lake Vastra Kyrksundet when it was meromictic (Bonsdorff and Storberg 1990). When the lake became limnic after isolation from the Baltic Sea, benthos production and ruffe biomass increased.