Hydrobiologia (2012) 693:169–181 DOI 10.1007/s10750-012-1106-0

PRIMARY RESEARCH PAPER

Feeding patterns and population structure of an invasive cyprinid, the rudd Scardinius erythrophthalmus (, ), in Buffalo Harbor (Lake Erie) and the upper Niagara River

Kevin L. Kapuscinski • John M. Farrell • Michael A. Wilkinson

Received: 16 July 2011 / Revised: 26 March 2012 / Accepted: 31 March 2012 / Published online: 19 April 2012 Ó Springer Science+Business Media B.V. 2012

Abstract Feeding patterns and population structure and supplemented their diet with algae and fish in of the non-native rudd Scardinius erythrophthalmus spring and fall. Feeding intensity was positively (Linnaeus) were examined to understand their ecology correlated with water temperature, but significantly in Buffalo Harbor and the Niagara River. We hypoth- reduced during spawning. Rudd condition and growth esized that (1) the diet of rudds would be omnivorous, were greater than estimates from other populations, but contain greater proportions of macrophytes in suggesting increases in abundance and range expan- summer months, (2) feeding intensity would increase sion are possible. Furthermore, reproduction was with water temperature, and (3) condition and growth successful at lotic sites but very poor at sites without would be similar to other populations. We collected measureable flow, contrary to the paradigm of optimal rudds with a variety of gears in 2009 to test these rudd habitat. Research is needed to understand how hypotheses, and used data from 2007 to 2010 seining herbivory by abundant rudd populations affects native surveys to determine if the relative abundance of aquatic communities. young-of-the-year rudd differed among sites with different flow conditions. Rudds were mostly herbiv- Keywords Rudd Feeding ecology Herbivory orous; they consumed aquatic macrophytes in summer Invasive Growth

Handling editor: John Havel Introduction

K. L. Kapuscinski (&) The rudd Scardinius erythrophthalmus (Linnaeus) is a College of Environmental Science and Forestry, State Eurasian cyprinid that was introduced to North University of New York, 104 Illick Hall, 1 Forestry Drive, Syracuse, NY 13210, USA America during the late nineteenth or early twentieth e-mail: [email protected] century (Burkhead & Williams, 1991; Mills et al., 1993). The popularity of rudd as a bait J. M. Farrell facilitated its spread to at least 21 states (Nico et al., College of Environmental Science and Forestry, State University of New York, 250 Illick Hall, 1 Forestry Drive, 2010), and although reproducing populations became Syracuse, NY 13210, USA established, they did not reach nuisance levels and therefore received little attention from fisheries biol- M. A. Wilkinson ogists (Burkhead & Williams, 1991). In New Zealand, New York State Department of Environmental Conservation, 270 Michigan Avenue, Buffalo, NY 14203, however, rudd populations became abundant enough USA in several bodies of water to receive attention by 123 170 Hydrobiologia (2012) 693:169–181 researchers (Hicks, 2003). The rudd is typically populations. However, abundant populations of rudd classified as omnivorous but undergoes ontogenetic have recently been documented in Buffalo Harbor shifts in diet. Young-of-the-year (YOY) rudds in New (eastern Lake Erie) and the upper Niagara River Zealand fed on planktonic cladocerans and chirono- (Kapuscinski et al., 2012). The rudd comprised 38% mid pupae, switched to benthic invertebrates and some of the total trap-net catch in 2007 (n = 2,066 or 9 per submerged aquatic macrophyte (hereafter macro- net night; Kapuscinski & Wilkinson, 2008) and 56% in phyte) material as juveniles, and fed mostly on 2008 (n = 4,821 or 29 per net night; Kapuscinski et al., macrophytes when [150 mm in length (Lake et al., 2009). The rudd was the most abundant species 2002; Hicks, 2003). Using a bioenergetics model, sampled during 2007–2008, and catches during 2008 Nurminen et al. (2003) estimated that an omnivorous were about four times greater than catches for the next population of rudd in a shallow eutrophic lake most abundant species, the brown bullhead Ameiurus consumed approximately its weight in macrophytes nebulosus (Lesueur) (n = 1,260). These large catches (wet weight) each year. The captive rudds studied by of rudd in Buffalo Harbor and the upper Niagara River Lake et al. (2002) consumed up to 20% of their weight were unexpected, and potential effects on native species in Egeria densa Planchon daily in spring and 22% of are unknown. However, it is unlikely that the popula- their weight in macroalgae (Nitella spp.) daily in tion dynamics of native species can be understood and summer. Rudd populations are thought to alter mac- properly managed without an understanding of the rophyte assemblages in New Zealand and aid the dynamics and ecology of the invasive rudd. For spread of exotic macrophytes by selectively feeding example, if rudds are consuming macrophytes, this on native species (Lake et al., 2002; Hicks, 2003). important habitat for native fishes might be impaired Rudds typically grow to about 350 mm in total and the potential benefits of ongoing habitat restoration length and 1,800 g in weight (Crossman et al., 1992); efforts may be nullified or reduced. Therefore, we the largest rudd on record was 617 mm and 3,623 g conducted a study of rudd that focused on (1) describing (Sˇprem et al., 2010). Although many neotropical fishes the diet, including an examination of differences among are herbivorous, such large, macrophyte-consuming seasons and between sexes, (2) quantifying several fish are uncommon in temperate North American population parameters, including condition and growth, freshwater ecosystems. Therefore, an abundant rudd and (3) documenting successful natural reproduction population could negatively affect aquatic communi- and comparing the relative abundance of YOY among ties by altering macrophyte assemblages and acceler- sites with different flow conditions. By comparing this ating internal nutrient loading (eutrophication) by information on condition, growth, and reproductive means of remobilizing nutrients stored in sediments success to that reported for other populations, we can and macrophytes (Hansson et al., 1987; van Donk & acquire a better understanding of the status of rudd in Gulati, 1995; Vanni, 2002; Hicks, 2003; Nurminen the Buffalo Harbor/upper Niagara River ecosystem and et al., 2003). In addition, the rudd may affect aquatic guide future research and management efforts. communities by (1) creating novel trophic links that serve as contaminant pathways (alaKwon et al., 2006), especially in an Area of Concern like the Niagara River Materials and methods (Environmental Protection Agency, 2009), (2) com- peting with native fishes for benthic invertebrate prey as We attempted to capture 50 rudds monthly from May juveniles (Crossman et al., 1992; Hicks, 2003), (3) to August and November of 2009. Two hoop-style spreading parasites, diseases, and viruses among native trap-nets (1.23 m diameter hoops, 30.48 m lead, fishes (Popovic´ et al., 2001; Schlueter, 2010), and (4) 6.10 m wings) were set once each month from May hybridizing with golden shiner Notemigonus crysoleu- to August (Fig. 1). A total of 50 rudds were captured in cas (Mitchill) (Burkhead & Williams, 1991), thereby May (nets were set at 16:02 and 16:31 on May 20, causing genomic extinction (i.e., the loss of local emptied at 10:10 and 10:30 on May 21, and emptied evolutionary lineages; Allendorf & Luikart, 2007). and pulled at 8:53 on May 22), 50 in June (nets were Concern regarding the potential negative effects of set at 10:30 and 10:55 on June 8 and emptied and invasion by rudd in North American waters has been pulled at 9:15 and 9:40 on June 9), 39 in July (nets minimal due to the relative scarcity of established were set at 16:14 and 16:45 on July 28 and emptied and 123 Hydrobiologia (2012) 693:169–181 171 pulled at 11:35 and 11:50 on July 29), and only 4 in where P is the number of rudds with food in their gut, August (nets were set at 13:30 and 13:51 on August 24, Q is the number of food types, and Wij is the weight of emptied at 8:35 and 8:40 on August 25, and emptied food type i in rudd j. Monthly differences in MWi were and pulled at 8:35 and 8:50 on August 26). Trap-nets determined via Analysis of Variance (ANOVA); were abandoned after August due to poor catches. A Tukey’s multiple comparisons test was used to large-mesh bag seine (18.3-m long and 6-mm mesh) determine which means differed. An index of fullness, was used to supplement the August collection (37 which measures food intake relative to fish size rudds were captured during daylight hours on August (Hyslop, 1980; Jude et al., 1987), was calculated for 25), and electrofishing (Smith-Root boom shocking each rudd as: (weight of gut contents/total fish boat, pulsed DC, 500–1,000 V, 6–12 amps) was used weight) * 100. Differences in mean values of the on November 3 (51 rudds were captured during index of fullness among sexes (males, females, and daylight hours). Water temperatures were recorded on rudds of unknown sex) and among months were each capture date. Most rudds (n = 194) were stored determined via a two-factor ANOVA; Tukey’s multi- on ice until they could be transferred to the lab, ple comparisons test was used to determine which measured for total length (mm) and weight (g), means differed. examined for body color (categorized as olive or We used the Chi square goodness of fit test to orange/gold), and frozen. The 37 rudds captured while determine if the observed female:male sex ratio seining in August were measured for length in the differed from a 1:1 expectation. Similarly, we com- field, and gut contents were flushed out via gastric pared the proportion of two observed body color lavage and identified as algae, macrophytes, fish, or patterns (olive or orange/gold) between sexes with a invertebrates. All of the 194 frozen rudds were later two-tailed Fisher’s exact test to determine if body thawed and processed, which included scale collection color was sex-linked, as male fishes are often more and dissection for sex determination and removal of brightly colored than females (Helfman et al., 1997). gut contents, ovaries, and otoliths. Gut contents were Fecundity could not be predicted using the poly- removed from the gastrointestinal tract, weighed to the nomial equation reported by Tarkan (2006) because it nearest 0.01 g (wet weight), identified as algae, returned negative fecundity values for the heavier macrophytes, or fish (no invertebrates were observed), rudds in our data set. Therefore, we used linear and the percentage of each food category was visually regression to determine the relationship between the estimated. Ovaries were removed and weighed to the age-specific mean values reported by Tarkan (2006) nearest 0.01 g (wet weight), and otoliths were for fecundity vs. rudd weight (r2 = 0.990), and removed and stored with scales in envelopes. estimated the fecundity of female rudds from the The proportions of female and male rudds with food upper Niagara River from their body weight using the in their guts were tested for differences within each formula: month with Fisher’s exact test for two proportions F ¼ 92:099 W þ 117:688 ð2Þ (Zar, 1999; a = 0.05 for all tests herein). Next, all samples were pooled regardless of sex, and a Chi- where F is fecundity (number of eggs) and W is the square test was used to determine if the proportion of total body weight of rudd (g). Next, we calculated the rudds with food in their guts varied by month. gonadosomatic index (GSI) for each female rudd as: Consequently, pair-wise monthly comparisons of the GSI ¼ G=W 100 ð3Þ proportions of rudds with food in their guts were made with the normal approximation of the Chi-square test where G is the weight of the gonads (g) and W is the to determine which monthly proportions differed (Zar, total body weight (g). We then calculated monthly 1999). For each month, the frequency of occurrence of mean GSI values and compared them using ANOVA; each food type was tabulated and the mean proportion Tukey’s multiple comparisons test was used to by weight (MWi) was calculated: determine which means differed. ! We used the following power function to model XP 1 Wij rudd weight as a function of length: MWi ¼ P ð1Þ P Q b j¼1 i¼1 Wij W ¼ aL ð4Þ

123 172 Hydrobiologia (2012) 693:169–181

Fig. 1 Map of Buffalo Harbor (Lake Erie) and the upper Niagara River indicating electrofishing (EF), trap-net (T), and seining (S) locations sampled for rudds for dietary analyses (D) and YOY rudds (Y)

where W is total body weight, L is total length, and otolith was mounted in clear, two-part epoxy, and the a and b are species-specific constants (Anderson & section around the nucleus was removed via two Neumann, 1996). The constants a and b were esti- transverse cuts along the dorsoventral plane with a mated by linear regression of logarithmically trans- Buehler Isomet low-speed saw. The sections were formed weight–length data. Data were initially ground on each side by affixing them to the forefinger analyzed separately for males and females, but the with double-sided tape and moving the section over a weight–length relations did not differ (ANCOVA; series of lapping papers (30, 15, 12, 9, and 3 lm) that sex * log10(L) not significant in general linear model; were wetted with water. Each section was mounted on TM P = 0.4789). Therefore, a single weight–length rela- a glass microscope slide in Crystalbond (heated on a tion was quantified from data collected on all rudds, hotplate) and further ground until an age could be regardless of sex. estimated by viewing through a compound or dissect- Single otoliths from each of 17 male and 24 female ing microscope. Ages were calculated assuming that rudds were prepared for age determination. Each rudd hatched on July 1 and that annuli (opaque bands) 123 Hydrobiologia (2012) 693:169–181 173 were formed on February 1 each year (midway Results through the winter or slow-growth period). For example, if two annuli were observed for a rudd A total of 231 rudds were examined for gut contents captured on 21 May, then the calculated age was (194 by dissection and 37 by gastric lavage), of which 1.89 years. We used the length-at-age data to estimate 174 (75%) contained food. The percentage of female a von Bertalanffy growth model for male and female versus male rudds with food in their guts only differed rudds: in November (Table 1). The percentage of all rudds examined with food in their gut varied by month; Kðtt0Þ Lt ¼ L1ð1 e Þð5Þ rudds collected in May, June, and August had the where Lt is length at age t, L? is the asymptotic highest percentage of guts that contained food average maximum length, K describes the rate at (92–95%), followed by rudds collected in November which Lt approaches L?, and t0 is hypothetical age (63%), and then July (26%; Table 1). when Lt is zero (Quinn & Deriso, 1999). We used Macrophytes were the largest component of the the Gauss–Newton least-squares nonlinear regres- rudd diet by both frequency of occurrence (present in sion method (SAS 9.2, SAS Institute Inc.) to 88% of all guts that contained food) and mean estimate each parameter and their 95% confidence proportion by weight (67%; Table 2). Furthermore, limits (CLs). Growth of males and females was macrophytes were recovered from rudd guts in each modeled separately because Zerunian et al. (1986) month examined, whereas algae were not found in suggested that the rudd exhibits sexually dimorphic August and fishes were not found in July or August growth. (Table 2). Macrophyte material was sufficiently mas- We used catches of YOY rudds in fine-mesh ticated to make identification to the species level (9.14-m long and 1.6-mm mesh) and large-mesh impractical, but the vast majority ([99%) appeared to (18.3-m long and 6.4-mm mesh) bag seine hauls to be Stuckenia pectinata (Linnaeus) (possibly mixed document whether or not spawning was successful with Potamogeton pusillus Linnaeus), with very small at two nearshore sites in Buffalo Harbor and 17 sites amounts of Vallisneria americana Michaux rarely in the upper Niagara River during 2007–2010. The present. The algae consumed by rudd were almost linear distance of each bag seine haul was 30.5 m, exclusively filamentous, with the macroalgae Chara and the number of hauls conducted at each of the 19 vulgaris Linnaeus or Nitella spp. present in only one sites ranged from 2 to 10. Catches of YOY rudds in rudd gut. Similarly, the fish consumed by rudd the fine-mesh seine between July 27 and August 7 at appeared to be of a single species, the emerald shiner standardized sites (see Kapuscinski et al., 2010 for Notropis atherinoides Rafinesque, although digestion details) were used to quantify the relative abundance prohibited definitive identification of all fish. of YOY rudd each year (i.e., the relative success of The frequency of occurrence of algae, macro- natural reproduction) and to make comparisons phytes, and fish recovered from rudd guts varied by between sites with different flow conditions. The month, as did the mean proportion by weight of algae linear distance of each seine haul was 30.5 m, six (ANOVA, df = 118, P \ 0.001), macrophytes sites were sampled in 2007, 10 sites were sampled (ANOVA, df = 118, P \ 0.001), and fish (ANOVA, during 2008–2010, and the number of hauls per site df = 118, P = 0.001; Table 2). Mean values for the ranged from 3 to 10. A two-sample, two-tailed t test index of fullness also differed among months assuming unequal variances was used to determine (ANOVA, df = 116, P \ 0.001), with the highest if mean catch rates differed between sites where mean observed in November (4.5%), intermediate measurable stream flow was present or absent values in May (2.5%) and June (3.1%), and the lowest (measured with a Swoffer Model 3000 flow meter). value in July (0.7%; Table 1). However, mean values To characterize flow conditions at lotic sites, water of the index of fullness did not differ among males, velocity was quantified at five sites in the upper females, and rudds of unknown sex (ANOVA, df = 2, Niagara River in 2009 as the average of three point P = 0.488), and there was no detectable interaction measurements of water velocity, each the average of between month and sex (ANOVA, df = 5, three readings. P = 0.827). The maximum weight of food in an

123 174 Hydrobiologia (2012) 693:169–181

Table 1 Sample sizes of female (NF), male (NM), and all rudds guts for which contents were weighed (NW), and the weight of (NAll) examined for gut contents, the percent of female (F), gut contents expressed as a percentage of rudd body weight male (M), and all rudds with food in their gut, the number of (index of fullness for all rudds that contained food)

Month NF NM NA Percent with food NW Index of fullness F M All

May 23 26 50 96a 88a 92z 38 2.5 ± 1.4z June 25 24 50 88a 96a 92z 45 3.1 ± 1.2z July 9 30 39 33a 23a 26 9 0.7 ± 0.5 August – – 41 – – 95z 0 – November 33 7 51 48a 100b 63 14 4.5 ± 1.9 Percentages for females and males followed by the same letter (a or b) are not statistically different for a given month (Fisher’s exact test), and monthly values for the percent of all rudds with food (normal approximation of the Chi-square test) and the index of fullness (ANOVA) followed by the same letter (z) are not statistically different (a = 0.05)

Table 2 Dietary analysis of 231 rudds captured in the upper Niagara River in 2009

Method of Month Number with Food Frequency of Number Total MWi examination food category occurrence weighed weight

Dissection May 47 Algae 0.74 40 302 0.44 ± 0.08 Macrophytes 0.85 193 0.36 ± 0.08 Fish 0.43 95 0.20 ± 0.06 Dissection June 46 Algae 0.11 46 17 0.03 ± 0.02 Macrophytes 0.96 1,240 0.94 ± 0.04 Fish 0.07 60 0.04 ± 0.03 Dissection July 12 Algae 0.50 10 17 0.23 ± 0.13 Macrophytes 0.83 38 0.77 ± 0.13 Fish 0 0 0 Lavage August 36 Algae 0 0 – – Macrophytes 1 – – Fish 0 – – Leech 0.03 – – Dissection November 33 Algae 0.27 23 353 0.30 ± 0.10 Macrophytes 0.70 280 0.65 ± 0.10 Fish 0.27 12 0.05 ± 0.04 Total 174 Algae 0.32 119 688 0.23 Macrophytes 0.88 1,745 0.67 Fish 0.18 171 0.09 Leech 0.01 – – –

The number weighed represents the number of guts examined for which the contents were weighed, and MWi is the mean proportion by weight (±SE). Note: algae were mostly filamentous (one rudd consumed a non-filamentous macroalgae) individual gut as a percentage of rudd body weight was from 1:1 (Chi square goodness of fit test, P = 0.764). 8.5% (observed in November). The body color of 90 female rudds and 87 male rudds Sex was determined for 178 rudds that were did not differ between sexes (79% of females were captured in trap-nets and dissected, with a resulting olive and 21% were orange/gold, whereas 87% of sex ratio of 91 females to 87 males that did not differ males were olive and 13% were orange/gold; Fisher’s

123 Hydrobiologia (2012) 693:169–181 175

Table 3 Monthly values (mean – SE) of the GSI for female 0:3027ðtþ1:3258Þ Lt ¼ 391:7 ð1 e Þ rudds collected from the upper Niagara River in 2009 Month N GSI and the model for 24 female rudds as: 0:4592ðt0:1090Þ May 22 8.39 ± 0.69a Lt ¼ 417:2 ð1 e Þ: June 25 16.06 ± 0.51b A broad range of ages were used to estimate growth July 7 6.55 ± 0.95a,c of male (range 1.89–11.08 years, mean 5.34 years) November 33 5.37 ± 0.47c and female (range 1.89–15.25 years, mean 7.37 years) Mean values followed by the same letter were not statistically rudds (Fig. 3). The 95% CIs on estimates of L?, K, different (ANOVA, Tukey’s test, a = 0.05) and t0 for males and females overlapped (Table 4). YOY rudds were captured in seine hauls at 14 of 17 exact test, P = 0.162), indicating that coloration was sites surveyed in the upper Niagara River during not sex-linked. 2007–2010, indicating that adult rudds successfully Fecundity estimates ranged from 12,827 to 156,775 reproduced at or near these locations. No YOY rudds eggs per fish for 86 female rudds that ranged from 138 were captured at two sites sampled in Buffalo Harbor, to 1,701 g in weight; a sub-sample (n = 22) ranged even though adult rudds comprised 23% of all fish from 1.89 to 15.25-years old. Mean monthly GSI captured in trap-nets during spring in 2007 and 24% in values differed (ANOVA, df = 86, P \ 0.001), with 2008; rudd comprised 51% of the trap-net catch from the highest mean GSI observed in June (Table 3). The the upper Niagara River in 2007 and 72% in 2008. The mean GSI value for May did not differ from July, but relative abundance of YOY rudds captured in a was higher than the November mean; mean GSI values standardized seining survey at lotic sites of the upper from July and November did not differ. Niagara River was extremely variable during The weight–length relation of rudds in the upper 2007–2010 (0–110.8 caught per seine haul), whereas Niagara River was estimated as: catches were either nonexistent or very low at sites that W ¼ 7:144 107 L3:539: lacked measurable stream flow (range 0–1.5 caught per seine haul; Table 5). The total number of rudds -7 The 95% CIs for a ranged from 4.7308 9 10 to caught per seine haul during 2007–2010 was 10.1 from -6 1.0788 9 10 and for b ranged from 3.4681 to lotic sites but only 0.2 from sites that lacked measur- 3.6091. This relation was estimated from 190 obser- able stream flow; mean catch rates only differed vations of rudds that ranged from 138 to 445 mm in between the two flow regimes in 2009 (Table 5). length and 28 to 1,701 g in weight (Fig. 2). Water velocities at five lotic sites of the upper Niagara We estimated the von Bertalanffy growth model for River ranged from 0.001 to 0.108 m/s (mean = 17 male rudds as: 0.056).

1,750

1,500 Discussion

1,250

-07 3.5386 Rudds collected from the upper Niagara River were 1,000 W = 7.1439*10 *L r2 = 0.981 mostly herbivorous, with macrophytes dominating the 750 diet and comprising a larger percentage of the diet in

Weight (g) warmer months. The positive relationship between 500 water temperature and assimilation efficiency (nutri- 250 tional value) of macrophytes has been suggested as the 0 mechanism controlling diet switches to plants in 100 200 300 400 omnivorous fishes (Behrens & Lafferty, 2007). Prejs Total length (mm) (1984) concluded that intense herbivory by rudd, defined as macrophytes comprising [50% of the diet Fig. 2 Weight–length relation for rudds in the upper Niagara River, estimated from 190 rudds captured in 2009 that ranged by weight, should occur at water temperatures above from 138 to 445 mm in length and 28 to 1,701 g in weight 16°C. Our results are consistent with this, as 123 176 Hydrobiologia (2012) 693:169–181

450 36% during May when water temperature was 14°Cat 400 a main river site (where 47 of 50 rudds were collected) 350 and 15–18°C at a still-water site (where the remaining 300 three rudds were collected). Furthermore, macro- 250 phytes comprised 65% of the mean proportion by 200 weight of the rudd diet during November when the -0.3027(t+1.3258) water temperature was 14 C. Similar reductions in 150 Lt = 391.7(1-e ) °

Total length (mm) feeding intensity have been reported for the herbivo- 100 2 r = 0.8418 rous grass carp Ctenopharyngodon idella (Valenci- 50 ennes), which significantly reduces consumption at 0 0481216water temperatures below 14°C (Colle et al., 1978). Age (years) Seasonal availability of macrophytes likely affected the diet composition of the rudds sampled in our study, 450 but this was not tested. 400 Herbivory by rudd has been reported for popula- 350 tions both within (Prejs & Jackowska, 1978; Nieder- 300 holzer & Hofer, 1980; Prejs, 1984) and outside 250 (Garcia-Berthou & Moreno-Amich, 2000; Hicks, 200 2003; Nurminen et al., 2003) its native range, but the -4592(t-0.1090) 150 Lt = 417.2(1-e ) breadth of the rudd diet varies greatly among popu-

Total length (mm) 100 r2 = 0.9297 lations and is likely influenced by macrophyte avail- 50 ability and water temperatures. In Lake Hiidenvesi, 0 Finland, where the biomass and diversity of sub- 0481216merged macrophytes was low, rudds consumed sig- Age (years) nificant amounts of detritus and invertebrates from Fig. 3 Length-at-age of male (top) and female (bottom) rudds May 15 to August 15; invertebrates composed about from the upper Niagara River in 2009. Solid lines indicate von 50% of the diet during mid-summer (Nurminen et al., Bertalanffy growth curves, whereas open circles represent ages 2003). The diet of rudds in Lake Banyoles, Spain, was estimated from otoliths. Mean length-at-age data from the dominated by detritus, macrophytes, and algae; the St. Lawrence River (triangles; J. M. Farrell, unpublished data) and Lake Sapanca, Turkey (squares; Tarkan, 2006) are provided percent biomass of plant material (31.6%) in the diet for comparison was probably lower than that reported for some populations due to low availability (Garcia-Berthou macrophytes comprised C78% of the mean proportion & Moreno-Amich, 2000). Conversely, rudds in New by weight during June, July, and August when water Zealand are considered obligate herbivores after they temperatures were [16°C (range 17–27°C), but only reach 200 mm fork length (Hicks, 2003). The diet of

Table 4 Growth parameters estimated from additive error von Bertalanffy growth models for 17 male and 24 female rudds collected from the upper Niagara River in 2009 Sex Age range Parameter Estimate SE LCL UCL

M 1.89–11.08 L? 391.7 22.5 343.3 440.0 K 0.3027 0.1017 0.0846 0.5208

t0 -1.3258 0.9742 -3.4153 0.7637

F 1.89–15.25 L? 417.2 6.2 404.4 430.1 K 0.4592 0.0179 0.2109 0.2817

t0 0.1090 0.3088 -0.5331 0.7511 The range of ages (years) in the samples, approximate standard errors SE, and approximate 95% lower LCL and upper UCLs for each parameter are also given

123 Hydrobiologia (2012) 693:169–181 177

Table 5 Summarized data from standardized seining surveys at nearshore sites where stream flow was absent (immeasurable with a flow meter) or present Year Flow absenta Flow presentb P-value # Sites # Rudds # Hauls Total C/H Mean C/H # Sites # Rudds # Hauls Total C/H Mean C/H

2007 3 0 16 0.00 0.00 3 17 17 1.00 0.83 0.1994 2008 3 6 12 0.50 0.50 7 678 36 18.83 23.62 0.1940 2009 3 0 14 0.00 0.00 7 41 42 0.98 0.82 0.0427 2010 3 5 12 0.42 0.42 7 502 28 17.93 17.93 0.1856 Total 11 54 0.20 0.23 1,238 123 10.07 12.46 0.1565 The total number of YOY rudds caught per seine haul (Total C/H), mean number caught per seine haul among sites (Mean C/H), and P-values of t tests (two-tailed assuming unequal variances) comparing Mean C/H between sites where flow was absent or present are provided a Includes two sites in Buffalo Harbor and one in the upper Niagara River b Includes seven sites in the upper Niagara River rudds in the upper Niagara River seems most similar to the spawning period, which was delayed in 2009 due that of the New Zealand populations. Even the diet of to below-average temperatures (National Oceanic and young rudds (age 1.89–2.94 years; n = 6) was dom- Atmospheric Administration, 2009). These data sug- inated by macrophytes, contrary to findings for gest an approximate 66% reduction in feeding activity similarly aged rudds in Finland (Nurminen et al., by rudds during the spawning period relative to the 2003). However, filamentous algae and fish were remainder of the summer. The index of fullness was important in diets of rudds during spring and fall in the also lowest in July (0.7%), providing further evidence upper Niagara River, probably because new growth of that rudd feeding is greatly reduced during the macrophytes was limited and water temperatures were spawning period. The index of fullness was probably unfavorable for digestion of macrophytes. The north highest during November (4.5%) because the gonads temperate climate of the upper Niagara River likely were lighter in November (mean GSI = 5.37) than forces rudds to broaden their diet to include fish during May (mean GSI = 8.39) and June (mean GSI = cold-water periods when material is more 16.06), so the same weight of food in the gut efficiently digested (Behrens & Lafferty, 2007), comprised a higher percentage of the total body whereas rudds in the more moderate New Zealand weight. climate can behave as obligate herbivores. The Although we used passive and active sampling absence of fish in the diet of rudds from the upper gears to collect rudd, and lavage during August but Niagara River in July–August, a period of high density dissection during other months to determine stomach of small fishes (especially in littoral habitats with high contents, it is unlikely that these different methods macrophyte densities), suggests that rudds preferen- caused the observed differences in diet composition or tially feed on macrophytes when temperatures are feeding intensity. For example, the proportion of rudd conducive for digestion. with empty stomachs was higher in July than May or Our results indicate that the feeding activity of June, even though trap-nets and dissection were used rudds decreases at cold temperatures and is signifi- in all 3 months. In addition, the proportion of empty cantly reduced during the spawning period. The stomachs was higher in November than in May or percentage of rudd guts that contained food was June, even though an active gear (electrofishing) was higher in May (92%), June (92%), and August (95%) used in November and a passive gear (trap-netting) than in November (63%), which is expected because was used in May and June, which may have allowed feeding intensity by rudds is greater at higher temper- rudd to digest food and present a higher proportion of atures (Hofer & Niederholzer, 1980; Prejs, 1984). empty stomachs. It is unlikely that the absence of algae However, the percentage of rudd guts that contained and fish in rudd stomachs in August resulted from food was lowest during July (26%), concurrent with stomach contents being retrieved via gastric lavage

123 178 Hydrobiologia (2012) 693:169–181 rather than dissection, because evaluations of gastric relative to other populations for which published data lavage have shown it to be highly efficient at retrieving exist. The functional exponent b = 3.5386 and its 95% stomach contents (Light et al., 1983). lower CL = 3.4681 were greater for rudds from the Selective feeding by the herbivorous grass carp has upper Niagara River than for rudds from Lake Sapanca, been shown to alter macrophyte assemblages Turkey (b = 3.189, 95% upper CL = 3.2990, esti- (McKnight & Hepp, 1995). Similarly, selective feeding mated from mean length and weight-at-age data by rudds may pose a significant threat to native reported by Tarkan, 2006), Lake Hiidenvesi, Finland macrophyte assemblages and habitat restoration efforts (b = 3.305; Nurminen et al., 2003), and Lakes Karap- that seek to reestablish macrophytes. Rudds in Poland iro and Ngaroto, New Zealand (b = 3.301; Lake et al., exhibited both selection for Elodea canadensis Mich- 2002). Age-4 male rudds from the upper Niagara River aux and avoidance of macrophytes that were unpalat- averaged 314 mm in length, whereas age-4 male rudds able or contained toxic substances (Prejs, 1984). Some from Lake Bracciano, Italy, averaged about 140 mm abundant macrophyte species in Lake Hiidenvesi, (standard length; Zerunian et al., 1986), and age-4 Finland, were absent from rudd guts (Nurminen et al., rudds of unknown sex averaged 203 mm from the St. 2003), indicating avoidance by rudds. In New Zealand, Lawrence River (J. M. Farrell, unpublished data), rudds fed selectively on native macrophyte and mac- 190 mm from Lake Sapanca, Turkey (Tarkan, 2006), roalgae species, consumed up to 22% of their body about 140 mm from Lake Hiidenvesi, Finland weight per day, and consumption rates were higher for (Nurminen et al., 2003), and 79–140 mm for seven larger rudds (Lake et al., 2002; Hicks, 2003). Rudd Irish waters (fork length; Kennedy & Fitzmaurice, populations in New Zealand and The Netherlands were 1974). Female rudds from the upper Niagara River also implicated in facilitating changes in macrophyte averaged 347 mm in length at age 4, whereas female assemblages (some toward domination by a non-native rudds from Lake Bracciano, Italy, averaged about species), and shifts in lakes from a clear-water, 155 mm (standard length; Zerunian et al., 1986). macrophyte-dominated state to a turbid, phytoplank- Eklov & Hamrin (1989) found that northern pike ton-dominated state (van Donk & Gulati, 1995; van Esox lucius Linnaeus (length range 217–261 mm) Donk & Otte, 1996; Wells, 1999; Hicks, 2003). The preferred rudds (length range 65–79 mm) to perch population of large, well-conditioned rudds (see below) Perca fluviatilis Linnaeus (length range 66–75 mm) as in the upper Niagara River might also alter macrophyte prey, because the surface feeding behavior of rudds assemblages by selectively consuming large amounts and their lack of spiny fin rays made them especially of preferred species, thereby threatening habitat resto- vulnerable to predation. However, the rapid growth, ration projects that are underway or planned for this large ultimate size, and laterally compressed body area (USAFERC, 2007). For example, rudds probably form of rudds probably allows them to attain sizes fed selectively on S. pectinata, which comprised about invulnerable to predation by all but the largest 99% of the macrophytes consumed, but only dominated predators, enabling rudds to thrive in waters contain- 58% of the 154.1-m2 quadrats that were examined in a ing significant piscivore populations. In Buffalo habitat survey (conducted adjacent to rudd capture sites Harbor and the upper Niagara River, large piscivorous during July 27–29; J. M. Farrell, unpublished data); fishes (bass Micropterus spp., bowfin Amia calva rudds apparently avoided V. americana, which com- Linnaeus, longnose gar Lepisosteus osseus Linnaeus, prised \1% of the macrophytes consumed, yet domi- muskellunge, northern pike, rainbow trout Oncorhyn- nated 33% of quadrats. Contrary to our results, rudds in chus mykiss Walbaum, tiger muskellunge E. masqu- Vrana Lake, Croatia, consumed few macrophytes, even inongy 9 E. lucius, and walleye Sander vitreus though ‘‘large beds’’ of S. pectinata were available (Mitchill)) comprised 16% of the fish assemblage (Tomec et al., 2003). Research is needed to determine captured in trap-nets in spring 2007 and 12% in 2008, which macrophyte species are preferred by rudds in yet rudd was the dominant species (38% of the total waters outside their native range, and to estimate trap-net catch in 2007 and 56% in 2008). Rudds also population-level effects of herbivory if efforts that seek have a long life span relative to many native cyprinids, to restore nearshore macrophytes are to be successful. which allows adults to contribute young to multiple Rudds from the upper Niagara River were in year-classes and increases the probability of popula- superior condition and exhibited more rapid growth tion persistence in the face of predation. 123 Hydrobiologia (2012) 693:169–181 179

An increase in the abundance of rudds in the upper predation, (3) a long life span that allows population Niagara River seems likely because density-dependent persistence despite consecutive years of poor repro- consequences (reductions) to condition and growth duction, and (4) the ability to reproduce in lentic and were apparently absent. The robust condition and lotic habitats. In addition, the ability of rudds to obtain rapid growth of rudds suggests that food resources are nutrients from algae, macrophytes, and detritus, which not limiting and competition is minimal for the age- is considered an adaptation to avoid competition and size-classes examined. Furthermore, rudds should within their native range (Johansson, 1987), also be buffered from both top-down and bottom-up appears to facilitate their establishment in waters controls because their large body size makes them outside their native range by exploiting abundant food less vulnerable to predation than native cyprinids, and resources that are not used by native fishes. The rudd’s their ability to consume macrophytes and detritus popularity as a baitfish, and in some instances as a allows them to avoid food limitations (as described for sport fish (Hicks, 2003), has further promoted its gizzard shad by Stein et al., 1995). An increase in rudd spread from its native European range to waters as far abundance and range expansion from Buffalo Harbor away as the United States and New Zealand. These and the upper Niagara River into Lake Erie and other extensive human introductions of rudds occurred even areas of the Great Lakes basin could negatively affect though such activity was often illegal. The seemingly littoral habitat and fish assemblages, especially in inevitable spread of rudds and their potential to alter nearshore, vegetated areas. aquatic ecosystems requires that research be con- Our results suggest that the paradigm of optimal ducted to understand the effects of rudds on native rudd habitat requires reconsideration. The rudd is often aquatic resources and to identify methods that reduce described as a littoral species that prefers lentic habitats or eliminate such effects. (Johansson, 1987; Lake et al., 2002), with few references made to rudd occupying slow-flowing lotic Acknowledgments The authors thank Andrew Panczykowski habitats (Cadwallader, 1978; Zerunian et al., 1986; for his assistance in the field and laboratory. Jonathan Sztukowski and Derek Crane also provided significant Hicks, 2003). Furthermore, most studies of wild rudd assistance in the field. Alexander Karatayev and Mark were focused on lake populations (Prejs, 1984; Garcia- Clapsadl (Buffalo State-State University of New York, Great Berthou & Moreno-Amich, 2000; Vila-Gispert & Lakes Center) provided storage, laboratory, and boat launch Moreno-Amich, 2000; Popovic´ et al., 2001; Nurminen facilities. The authors are grateful to Karin Limburg for providing assistance and expertise in otolith preparation, et al., 2003; Tarkan, 2006). The rudd population in the Gillian AvRuskin for assistance with figure creation, and upper Niagara River not only exhibited superior Stephen Stehman for statistical advice. Three anonymous condition and more rapid growth compared to lentic reviewers provided comments that greatly improved this populations, but reproduction was successful at lotic manuscript. This study was funded by Federal Aid in Sport Fish Restoration Grants administered by the New York State sites and very poor at sites with very low flow Department of Environmental Conservation, a grant from the (immeasurable with a flow meter). In addition, the Greenway Ecological Fund Standing Committee, and is a rudd was first detected in the St. Lawrence River in contribution of the Thousand Islands Biological Station. 1989 (Klindt, 1990), but has remained rare (0.06% of total spring trap-net catches in 1993 and 2006–2009; J. M. Farrell, unpublished data) in embayments with References very low flow. The rudd’s ability to thrive in lotic habitats requires us to expand our concept of the waters Allendorf, F. W. & G. Luikart, 2007. Conservation and the vulnerable to invasion and to gain an understanding of Genetics of Populations. Blackwell Publishing, Malden, the mechanisms controlling rudd population dynamics. MA: 642 pp. Anderson, R. O. & R. M. Neumann, 1996. Length, weight, and associated structural indices. In Murphy, B. R. & D. W. Willis (eds), Fisheries Techniques, 2nd ed. Ameri- Conclusion can Fisheries Society, Bethesda, MD: 447–482. Behrens, M. D. & K. D. Lafferty, 2007. Temperature and diet Several adaptations make the rudd a successful effects on omnivorous fish performance: implications for the latitudinal diversity gradient in herbivorous fishes. invader, including (1) early maturation, (2) rapid Canadian Journal of Fisheries and Aquatic Sciences 64: growth and large body size, which reduce the risk of 867–873. 123 180 Hydrobiologia (2012) 693:169–181

Burkhead, N. M. & J. D. Williams, 1991. An intergeneric hybrid muskellunge (2008) in the Buffalo Harbor, Lake Erie, and of a native minnow, the golden shiner, and an exotic the upper Niagara River. In: Einhouse, D. (ed), NYSDEC minnow, the rudd. Transactions of the American Fisheries Lake Erie 2008 annual report. New York State Department Society 120: 781–795. of Environmental Conservation, Albany, section R. Cadwallader, P. L., 1978. Acclimatisation of rudd Scardinius Available from: http://www.dec.ny.gov/docs/fish_marine_ erythrophthalmus (Pisces: Cyprinidae), in North Island of pdf/2008leurpt.pdf. New Zealand (NOTE). New Zealand Journal of Marine and Kapuscinski, K. L., M. A. Wilkinson & J. M. Farrell, 2010. Freshwater Research 12: 81–82. Sampling for muskellunge, rudd, and the nearshore fish Colle, D. E., J. V. Shireman & R. W. Rottmann, 1978. Food community of the Buffalo Harbor (Lake Erie) and the selection by grass carp fingerlings in a vegetated pond. upper Niagara River, 2009. In: Einhouse, D. (ed), Transactions of the American Fisheries Society 107: NYSDEC Lake Erie 2009 annual report. New York State 149–152. Department of Environmental Conservation, Albany, sec- Crossman, E. J., E. Holm, R. Cholmondeley & K. Tuininga, tion Q. Available from: http://www.dec.ny.gov/docs/fish_ 1992. First record for Canada of the rudd, Scardinius ery- marine_pdf/2009lerpt.pdf. throphthalmus, and notes on the introduced round goby, Kapuscinski, K. L., J. M. Farrell & M. A. Wilkinson, 2012. First Neogobius melanostomus. Canadian Field-Naturalist 106: report of abundant rudd populations in North America. 206–209. North American Journal of Fisheries Management 32: Eklov, P. & S. F. Hamrin, 1989. Predatory efficiency and prey 82–86. selection: interactions between pike Esox lucius, perch Kennedy, M. & P. Fitzmaurice, 1974. Biology of the rudd Perca fluviatilis and rudd Scardinius erythrophthalmus. Scardinius erythrophthalmus (L) in Irish waters. Proceed- OIKOS 56: 149–156. ings of the Royal Irish Academy 74: 245–303. Environmental Protection Agency, 2009. Niagara River area of Klindt, R, 1990. Distribution of rudd in the St. Lawrence concern. Available from: http://www.epa.gov/glnpo/aoc/ River. Report to the Great Lakes Fishery Commission, niagara.html [accessed October 2010]. St. Lawrence River Subcommittee, pp. 71–72. Garcia-Berthou, E. & R. Moreno-Amich, 2000. Rudd (Scardinius Kwon, T., S. W. Fisher, G. W. Kim, H. Hwang & J. Kim, 2006. erythrophthalmus) introduced to the Iberian Peninsula: feed- Trophic transfer and biotransformation of polychlorinated ing ecology in Lake Banyoles. Hydrobiologia 436: 159–164. biphenyls in zebra mussel, round goby, and smallmouth Hansson, L., L. Johansson & L. Persson, 1987. Effects of fish bass in Lake Erie, USA. Environmental Toxicology and grazing on nutrient release and succession of primary Chemistry 25: 1068–1078. producers. Limnology and Oceanography 32: 723–729. Lake, M. D., B. J. Hicks, R. D. S. Wells & T. M. Dugdale, 2002. Helfman, G. S., B. B. Collette & D. E. Facey, 1997. The Consumption of submerged aquatic macrophytes by rudd diversity of fishes. Blackwell Science, Inc., Malden, MA: (Scardinius erythrophthalmus L.) in New Zealand. Hyd- 528 pp. robiologia 470: 13–22. Hicks, B. J., 2003. Biology and potential impacts of rudd Light, R. W., P. H. Alder & D. E. Arnold, 1983. Evaluation of (Scardinius erythrophthalmus L.) in New Zealand. In: gastric lavage for stomach analyses. North American Managing Invasive Freshwater Fish in New Zealand: Journal of Fisheries Management 3(1): 81–85. proceedings of a workshop hosted by Department of McKnight, S. K. & G. R. Hepp, 1995. Potential effect of grass Conservation, Wellington, New Zealand, pp. 49–58. carp herbivory on waterfowl foods. Journal of Wildlife Hofer, R. & R. Niederholzer, 1980. The feeding of roach (Ru- Management 59: 720–727. tilus rutilus L.) and rudd (Scardinius erythrophthalmus L.). Mills, E. L., J. H. Leach, J. T. Carlton & C. L. Secor, 1993. II. Feeding experiments in the laboratory. Ekologia Polska Exotic species in the Great Lakes: a history of biotic crises 28: 61–70. and anthropogenic introductions. Journal of Great Lakes Hyslop, E. J., 1980. Stomach contents analysis—a review of Research 19: 1–54. methods and their application. Journal of Fish Biology 17: National Oceanic and Atmospheric Administration, 2009. 411–429. NOAA: summer temperatures below average for U.S. Johansson, L., 1987. Experimental evidence for interactive [online]. Available from http://www.noaanews.noaa.gov/ habitat segregation between roach (Rutilus rutilus) and stories2009/ 20090910_summerstats.html [accessed 15 rudd (Scardinius erythrophthalmus) in a shallow eutrophic September 2010]. lake. Oecologia 73: 21–27. Nico, L., P. Fuller & G. Jacobs, 2010. Scardinius erythroph- Jude, D. J., F. J. Tesar, S. F. Deboe & T. J. Miller, 1987. Diet and thalmus. USGS Nonindigenous Aquatic Species Database, selection of major prey species by Lake Michigan Salmo- Gainesville, FL. Available from http://nas3.er.usgs.gov/ nines, 1973–1982. Transactions of the American Fisheries queries/FactSheet.aspx?speciesID=648 [accessed March Society 116: 677–691. 2010]. Kapuscinski, K. L. & M. A. Wilkinson, 2008. Adult muskel- Niederholzer, R. & R. Hofer, 1980. The feeding of roach lunge sampling activities in the Buffalo Harbor and upper (Rutilus rutilus L.) and rudd (Scardinius erythrophthalmus Niagara River May and June, 2007. In: Einhouse, D. (ed), L.). I. Studies on natural populations. Ekologia Polska 28: NYSDEC Lake Erie Unit 2007 annual report. New York 45–59. State Department of Environmental Conservation, Albany, Nurminen, L., J. Horppila, J. Lappalanien & T. Malinen, 2003. section U. Implications of rudd (Scardinius erythrophthalmus) her- Kapuscinski, K. L., M. A. Wilkinson & J. M. Farrell, 2009. bivory on submerged macrophytes in a shallow eutrophic Sampling efforts for young-of-year (2006–2008) and adult lake. Hydrobiologia 506–509: 511–518. 123 Hydrobiologia (2012) 693:169–181 181

Popovic´, N. T., M. Hacmanjek & E. Teskeredzˇic´, 2001. Health Tomec, M., Z. Teskerdzˇic´ & E. Teskerdzˇic´, 2003. Food and status of rudd (Scardinius erythrophthalmus hesperidicus nutritive value of gut contents of rudd (Scardinius erythr- H.) in Lake Vrana on the Island of Cres Croatia. Journal of ophthalmus L.) from Vrana Lake, Cres Island, Croatia. Applied Ichthyology 17: 43–45. Czech Journal of Animal Science 48: 28–34. Prejs, A., 1984. Herbivory by temperate freshwater fishes and its USAFERC (U.S.A. Federal Energy Regulatory Commission), consequences. Environmental Biology of Fishes 10: 2007. Order on offer of settlement and issuing new license. 281–296. Project No. 2216-066. Prejs, A. & H. Jackowska, 1978. Lake macrophytes as the food van Donk, E. & R. D. Gulati, 1995. Transition of a lake to turbid of roach (Rutilus rutilus L.) and rudd (Scardinius erythr- state six years after biomanipulation: mechanisms and ophthalmus L.). I. Species composition and dominance pathways. Water Science and Technology 32: 197–206. relations in the lake and food. Ekologia Polska 26: van Donk, E. & A. Otte, 1996. Effects of grazing by fish and 429–438. waterfowl on the biomass and species composition of Quinn, T. J., II & R. B. Deriso, 1999. Quantitative fish dynamics. submerged macrophytes. Hydrobiologia 340: 285–290. Oxford University Press, New York, NY: 542 pp. Vanni, M. J., 2002. Nutrient cycling by in freshwater Schlueter, L, 2010. Aquatic nuisance species—the animals. ecosystems. Annual Review of Ecology and Systematics North Dakota Game and Fish Department. Available from 33: 341–370. http://gf.nd.gov/fishing/ans-animals.html#fishes [accessed Vila-Gispert, A. & R. Moreno-Amich, 2000. Fecundity and March 2010]. spawning mode of three introduced fish species in Lake Sˇprem, N., D. Matulic´, T. Treer & I. Anicˇic´, 2010. A new Banyoles (Catalunya, Spain) in comparison with other maximum length and weight for Scardinius erythroph- localities. Aquatic Sciences 61: 154–166. thalmus. Journal of Applied Ichthyology 26: 618–619. Wells, R., 1999. Are rudd a threat to water plants? Water and Stein, R. A., D. R. DeVries & J. M. Dettmers, 1995. Food-web Atmosphere 7(4): 11–13. regulation by a planktivore: exploring the generality of the Zar, J. H., 1999. Biostatistical Analysis, 4th ed. Prentice Hall, trophic cascade hypothesis. Canadian Journal of Fisheries Englewood Cliffs, NJ. and Aquatic Sciences 52: 2518–2526. Zerunian, S., L. Valentini & G. Gibertini, 1986. Growth and Tarkan, A. S., 2006. Reproductive ecology of two cyprinid reproduction of rudd and red-eye roach (Pisces, Cyprini- fishes in an oligotrophic lake near the southern limits of dae) in Lake Bracciano. Italian Journal of Zoology 53: their distribution range. Ecology of Freshwater Fish 15: 91–95. 131–138.

123