Loss of reproductive output caused by an invasive species

Transcription

1 rsos.roylsocietypublishing.org Downloded from on April 7, 216 Loss of reproductive output cused by n invsive species Reserch Cite this rticle: Trembly MEM, Morris TJ, Ackermn JD. 216 Loss of reproductive output cused by n invsive species. R. Soc. open sci. 3: MudeE.M.Trembly 1,ToddJ.Morris 2 nd Josef D. Ackermn 1 1 Deprtment of Integrtive Biology, University of Guelph, Guelph, Ontrio, Cnd N1G 2W1 2 Gret Lkes Lbortory for Fisheries nd Aqutic Sciences, Fisheries nd Ocens Cnd, 867 Lkeshore Rod, Burlington, Ontrio, Cnd L7S 1A1 Received: 13 September 215 Accepted: 3 Mrch 216 Subject Ctegory: Biology (whole orgnism) Subject Ares: ecology Keywords: unionid mussels, species t risk, Neogobius melnostomus,gretlkes Author for correspondence: Josef D. Ackermn e-mil: ckermn@uoguelph.c Electronic supplementry mteril is vilble t or vi We investigted whether Neogobius melnostomus, n invder of biodiversity hot-spots in the Lurentin Gret Lkes region, fcilittes or inhibits unionid mussel recruitment by serving s host or sink for their prsitic lrve (glochidi). Infesttion nd metmorphosis rtes of four mussel species with t-risk (conservtion) sttus (Epioblsm torulos rngin, Epioblsm triquetr, Lmpsilis fsciol nd Villos iris) ndone common species (Actinonis ligmentin) on N. melnostomus were compred with rtes on known primry nd mrginl hosts in the lbortory. All species successfully infested N. melnostomus, but only E. triquetr, V. iris nd A. ligmentin successfully metmorphosed into juveniles, lbeit t very low rtes well below those seen on even the mrginl hosts. Neogobius melnostomus collected from res of unionid occurrence in the Grnd nd Sydenhm rivers (Ontrio, Cnd) exhibited glochidil infection rtes of 39.4% nd 5.1%, respectively, with up to 3 glochidi representing s mny s six unionid species per fish. A mthemticl model suggests tht N. melnostomus serve more s sink for glochidi thn s host for unionids, thereby limiting recruitment success. This represents novel method by which n invsive species ffects ntive species. 1. Introduction The Lurentin Gret Lkes hve experienced number of threts nd impcts including those from invsive dreissenid mussels in the lte 198s [1,2], which extirpted ntive unionid mussels by interfering with their locomotion nd burrowing nd preventing them from closing their vlves [3]. Although unionids remin in some tributry nd embyment refugi of the Gret Lkes [4 6], 15 of 41 unionid species in Ontrio hve been ssessed s t risk for extinction or extirption by the Committee 216 The Authors. Published by the Royl Society under the terms of the Cretive Commons Attribution License which permits unrestricted use, provided the originl uthor nd source re credited.

2 on the Sttus of Endngered Wildlife in Cnd [6]. Severl lrge rivers in southwestern Ontrio re considered species conservtion hot-spots [5] becuse of the high diversity of unionids present. The recent expnsion of the round goby (Neogobius melnostomus) into the lower reches of these rivers is cuse for concern [6,7] becuse of the potentil for predtion on juvenile unionids nd competition with nd/or predtion on ntive host fishes required by unionids for their lrvl development [6,8,9]. However, such ssessments hve not considered whether unionid mussels cn use N. melnostomus s hosts to complete their life cycle nd mitigte these hypothesized impcts. Unionids hve n obligte prsitic lrvl stge (glochidium), which requires vertebrte host (usully fish) to fcilitte metmorphosis into free-living juvenile mussels [1,11]. Among unionids, there re host generlists (ble to use mny host fishes) nd host specilists (only use few host fish species) nd mny species hve evolved elborte species-specific strtegies of host ttrction [9]. For fish to be considered host, both glochidil infection nd metmorphosis must occur; primry hosts hve high infesttion rtes (intensity of infection or the proportion of glochidi tht ttch) nd high metmorphosis rtes (proportion of ttched glochidi tht metmorphose into juveniles), wheres mrginl hosts provide lower rtes, prticulrly metmorphosis rtes, for given mussel species [11]. Neogobius melnostomus my be mrginl host for severl unionid species in the lbortory [12] (J.D.A. 21, personl communiction) nd glochidil prsitism ws reported for single individul in the field [13]. Given the cpcity for high post-invsion bundnce of N. melnostomus [14], N. melnostomus my either fcilitte the recruitment of unionid mussels s host fish or limit successful unionid recruitment by removing lrge proportions of glochidi through initil ttchment followed by low or unsuccessful metmorphosis (i.e. serving s sink for glochidi) s observed for non-host species [15]. This my contribute to the biotic homogeniztion of the region, which cn decrese the vilbility of ntive host fishes to mussels [16]. The role of N. melnostomus s host for unionids or s sink for their glochidi remins to be determined. Thepurposeof this studyis to determinewhether N. melnostomus serve s hosts for unionid mussels or s sinks for their glochidi. As hosts, they should fcilitte glochidil infection nd metmorphosis of lrge numbers of juvenile mussels, nd thus propgte unionids. As sinks, glochidi should ttch but not metmorphose (or do so t low rtes), thereby reducing unionid mussel recruitment. We investigted the host qulity of N. melnostomus in the lbortory using controlled infection experiments, nd in the field where nturl glochidi infections of N. melnostomus were evluted to determine the bundnce nd species of glochidi Mteril nd methods 2.1. Assessment of infection nd metmorphosis on Neogobius melnostomus in the lbortory Grvid femle Epioblsm torulos rngin (Le 1838) (n endngered species), nd Actinonis ligmentin (Lmrck 1819) ( common species) were collected from the Sydenhm River t Florence, Ontrio ( , ) nd Epioblsm triquetr (Rfinesque 182) (n endngered species) were collected t Dwn Mills, Ontrio (42.669, ). Lmpsilis fsciol Rfinesque 182 ( species of specil concern) nd Villos iris (Le 1829) (n endngered species) were collected from the Thmes River t Thmesford, Ontrio (43.561, ). Three femles of ech species nd similr size were trnsported in erted coolers to the Hgen Aqulb (University of Guelph) where they were cclimted to C nd moved to circulr flow-through tnk contining well wter t pproximtely 11 C. Mussels were fed n lgl diet (2 1 8 cells l 1 of Nnno 36 nd shellfish diet, Reed Mriculture, Cmpbell, CA, USA) three times per week. Young-of-the-yer fish, whenever possible, were collected from wter bodies in Guelph, Burlington, Dunds, nd Nigr Flls, Ontrio tht did not contin the mussel species of interest (i.e. to reduce immunologicl responses due to previous glochidil encystment [17]). Species included N. melnostomus (Plls, 1814), Cottus birdii Girrd, 185, known mrginl host for ll mussel species nd ecologiclly similr to N. melnostomus [9,18], nd primry fish hosts of the respective mussel species (Micropterus slmoides (Lcepède, 182) for A. ligmentin; Etheostom exile (Girrd, 1859) for E. t. rngin; Percin cprodes (Rfinesque, 1818) for E. triquetr; Micropterus dolomieu Lcepède, 182 for L. fsciol nd Ambloplites rupestris (Rfinesque, 1817) for V. iris). Fish were trnsported in erted buckets nd cclimted to C prior to use in experiments (21) or were treted with Melfix (Mrs Fishcre, Chlfont, USA;.125 ml l 1 te tree oil) for 3 dys nd qurntined for t lest one week prior to experiments (211). Fish were mintined on bloodworms, cryfish, brine shrimp nd/or smelt (Metro Ontrio, Inc.) s pproprite severl times weekly.

3 Lbortory infections, which followed the techniques in [11], provided n opportunity to evlute the cpcity for successful host-glochidi infection mong fish species in controlled mnner regrdless of the mechnism used by prticulr mussel species (e.g. brodcsting, host ttrction, direct contct) [9]; it ws not designed to estimte infection rtes by drifting glochidi rther it is closer to tht of host ttrction nd direct contct strtegies given the concentrtion of glochidi used. Briefly, glochidi from hlf the gill of ech femle mussel were ssessed for vibility [19,2] before being divided into three portions nd used to infest four conspecific fish of similr totl length (TL) t concentrtion of c 5 glochidi l 1 in n erted 1. l tnk for 1 h in the drk. Ech group of four fish from specific mussel fish infesttion (i.e. replicte) ws plced in one of three Aqutic Hbitt (AHAB) units (Pentir Aqutic Hbitts, Apopk, USA) operted t 18 2 C using 2 µm-filtered well wter. All juvenile mussels nd excysted glochidi were removed nd counted fter being flushed into 1 µm mesh cps locted in ech AHAB tnk (twice weekly) or fter of the wter volume of ech tnk ws siphoned through 1 µm sieve (weekly). Once the recovery cps were devoid of juveniles for seven consecutive dys, fish were inspected for glochidil infection under tricine methnesulfonte (MS-222; pprox. 23 mg l 1 ), nd if none were found, they were euthnized (MS-222; pprox. 1 mg l 1 ) nd dissected to confirm the bsence of encysted glochidi Assessmentofnturlinfection ofneogobius melnostomus in the field Neogobius melnostomus were collected, geo-referenced, fixed in formlin nd preserved in ethnol from the lower Grnd River (between Middleport nd Hldimnd, Ontrio; , to , ; June July 21) nd the Est Sydenhm River ( , to , ; August 21). The former is slow flowing, deep, nd hs substrte composed of cly nd mud [2], nd the ltter is low grdient with high diversity of hbitts (including grvel nd snd), exhibits distinct riffles nd pools nd high turbidity likely resulting from griculturl ctivities (85% of lnd use) [21]. A totl of 127 N. melnostomus from the Grnd River (N = 127) nd 79 from the Sydenhm River were mesured for TL, the body nd fins were exmined for glochidil encystment under 32 dissecting microscope (Nikon SMZ-2 T, Nikon Jpn), nd glochidil encystment on excised gills ws determined using cross polriztion. All glochidi were photogrphed (3.34MP CoolPix 995 Nikon, Jpn) nd the length, height nd hinge length of ech glochidium were mesured vi imge nlysis (ImgeJ 1.38, US Ntionl Institutes of Helth) Sttisticl nlyses Assessment of infection nd metmorphosis on Neogobius melnostomus in the lbortory The infesttion rte (R I ) ws clculted s the proportion of glochidi tht ttched to the fish (G A ) from the number used to infest the fish (G T )(R I = G A /G T )[11], wheres the metmorphosis rte (R M ) ws clculted s the proportion of glochidi tht metmorphosed into juvenile mussels (G M )fromg A (R M = G M /G A )[11]. Glochidi encysted on fishes tht died prior to the end of the experiment were included in G A for determining R I, but were excluded from G A for the determintion of R M. One-wy nlysis of vrince (ANOVA) ws used to compre R I, R M nd the number of juveniles produced per fish, respectively, with fish species s the fixed effect. A two-wy min effects ANOVA, with fish species nd femle mussel s the fixed effects, ws used if the vibility of glochidi differed mong femle unionids [18]. Assumption of normlity (Shpiro Wilks test) nd homogeneity of vrince (Levene s test) were exmined nd dt were rcsine squre root or log + 1 trnsformed s necessry [22]. A non-prmetric Kruskll Wllis test ws used if the dt were not norml but stisfied homoscedsticity [22]. Significnt pirwise tretments were exmined using post-hoc Tukey tests [22]. STATISTICA v. 6. (Sttsoft, Tuls, OK, USA) ws used for ll nlyses. An independent smples t-test ws used to compre the proportion of fish infected with glochidi (i.e. prevlence of infection) between rivers. A logistic regression ws conducted using TL s the predictor nd glochidi infection (i.e. = not infected, 1 = glochidi infected) s the dependent vrible. All nlyses were undertken using SPSS v. 19 (IBM, Armonk, NY, USA) Identifiction of glochidi from the field-collected fish Glochidi were ssigned to species vi discriminnt function nlysis (DFA) [23] using river specific models bsed on glochidi length, width nd hinge length dt collected for ll 36 species of unionids found in these rivers (n = 83 ± 16 (men ± s.e.) glochidi/species) [24]. Clssifiction success ws 78 ± 4%

4 nd 75 ± 4% for the Grnd nd Sydenhm rivers, respectively, including five species within two gener tht were prticulrly difficult to resolve [24] Modelled contribution ofneogobius melnostomus to unionid recruitment A model ws used to ssess the role of N. melnostomus s host fish for unionids versus sink for their glochidi using the bundnce (U) nd fecundity (f; the number of glochidi per femle [25,26] or estimted (i.e. V. iris)) of unionids, the encounter rte between glochidi nd hosts (R e ;from[27]), nd the R I nd R M mesured in this study. The infection (I N )onn. melnostomus is given by I N = U f R e R I, (2.1) wheres the number of juvenile unionid mussels produced (J) is given by J = I N R M N = (U f R e R I ) R M N, (2.2) where N is the reltive bundnce of N. melnostomus to known hosts. By extension, the number of glochidi lost from potentil recruitment (D g ) is given by D g = I N (1 R M ) N = (U f R e R I ) (1 R M ) N. (2.3) The difference between equtions (2.2) nd (2.3) is R M whereby higher R M leds to more juveniles produced nd fewer glochidi lost. The rtio of glochidi loss to juvenile mussel production (D g : J) for N. melnostomus ws compred to model results for the primry s well s the mrginl host (C. birdii). Although the model is bsed on prmeters derived from controlled lbortory experiments tht do not incorporte specific host fish interctions [9] nor the behviour of N. melnostomus, it does provide n the opportunity to evlute D g : J. 3. Results 3.1. Infection nd metmorphosis on Neogobius melnostomus in the lbortory Infesttion (R I ) nd metmorphosis rtes (R M ) were generlly highest on the primry host, followed by the mrginl host nd then N. melnostomus (figure 1,b). This pttern ws lso seen in juvenile mussel production (figure 1c). Considerble vrition in ech of these vlues ws observed including the number of glochidi tht successfully ttched to individul fish, which ws reveled by dissections of fishes tht died prior to completion of the experiments (mortlity rte for primry hosts: 33 ± 16%; C. birdii:2± 2% nd N. melnostomus:2± 1%) Actinonis ligmentin Infesttion rtes differed significntly mong fish species (Kruskll Wllis; χ2 2 = 6.5, p =.39) nd were higher on the primry host (M. slmoides) versus the mrginl host (C. birdii; p =.34). Metmorphosis rtes lso differed (ANOVA F 2,4 = 8., p =.4) with the highest rtes on the primry host versus mrginl hosts (post-hoc Tukey test, p =.44) nd N. melnostomus (p =.64). Juvenile production did not differ significntly (χ2 2 = 4.23, p =.12), but this ws likely the result of reduced power due to fish mortlity s differences were detected under ANOVA (F 2,4 = 177, p =.1). Ten juvenile A. ligmentin were produced on N. melnostomus Epioblsm torulos rngin Infesttion rtes did not differ significntly (F 2,6 = 3.6, p =.12), but mrginl differences in R M were observed (F 2,6 = 4.3, p =.78) with the highest rtes on the mrginl host (C. birdii), followed by the primry host (E. exile) ndn. melnostomus. Juvenile production differed significntly (F 2,6 = 5.2, p =.49), with the highest production on the mrginl host, followed by the primry host nd N. melnostomus. Significnt differences were found between the mrginl host (C. birdii) nd N. melnostomus (p =.42). No juvenile E. t. rngin were produced on N. melnostomus.

5 () infesttion rte, R l (%) (b) metmorphosis rte, R M (%) b b b b b b primry host mrginl host round goby 5 (c) no. juveniles per fish b b b b b b b b b c A. ligmentin E. t. rngin E. triquetr L. fsciol V. iris Figure 1. Experimentl vlues (men± s.e.) for ll mussel species nd ll host types (primry host, mrginl host nd round goby, Neogobius melnostomus). Primry hosts of: Actinonis ligmentin = Micropterus slmoides; Epioblsm torulos rngin = Etheostom exile; Epioblsm triquetr = Percin cprodes; Lmpsilis. fsciol = Micropterus dolomieu; Villos iris = Ambloplites rupestris). Cottus birdii served s the mrginl host for ll species. indictes non-significnt differences; differentlettersbovetwohosttypeswithinmusselspeciesindictesignificntdifferences. () Infesttion rtes, (b) metmorphosis rtes nd (c) number of juveniles produced per fish Epioblsm triquetr Infesttion rtes did not differ significntly (F 2,6 = 3.11, p =.12), nor did R M (F 2,6 = 1.37, p =.32); however, the highest rtes were on the primry host (P. cprodes), followed by the mrginl host (C.birdii) nd then N. melnostomus. Juvenile production ws mrginlly significntly different (χ2 2 = 6.1, p =.5) with mrginl differences between the primry host nd N. melnostomus (p =.5). Four juvenile E. triquetr were produced on N. melnostomus Lmpsilis fsciol Infesttion rtes did not differ significntly (F 2,6 = 2.84, p =.14), wheres R M did (F 2,5 = 13.2, p =.1) with the highest rtes on the primry host (M. dolomieu), followed by the mrginl host (C. birdii), nd then N. melnostomus. Significnt differences were found between the primry host nd N. melnostomus (p =.9) nd mrginlly with the mrginl host (p =.56). Juvenile production differed significntly (F 2,5 = 32.2, p =.1), with the highest production on the primry host, followed by the mrginl host

6 nd N. melnostomus. Differences were detected between the primry host nd N. melnostomus (p =.1) nd the mrginl host (p =.5). No juvenile L. fsciol were produced on N. melnostomus Villos iris Significnt differences in the vibility of glochidi mong femle mussels were observed, so two-wy ANOVAs were conducted. Infesttion rtes did not differ significntly mong fish species or femle mussels (fish fctor: F 2,4 = 3.24, p =.15; mussel fctor: F 2,4 =.31, p =.75). One-wy ANOVAs were conducted to ssess R M nd the number of juvenile mussels produced per fish s fish mortlity reduced the number of replictes per femle. Metmorphosis rtes did not differ significntly (F 2,4 = 1.49, p =.33), but the number of juveniles produced per fish did (F 2,4 = 34., p =.3) with the highest production on the primry host (A. rupestris), followed by the mrginl host (C. birdii) nd N. melnostomus (p <.4 for ll pirwise comprisons). One juvenile V. iris ws produced on N. melnostomus. There ws high mortlity of N. melnostomus with ttched glochidi ( 19 per fish) Nturl infection of Neogobius melnostomus in the field There ws significntly higher prevlence of glochidi infection on N. melnostomus in the Grnd versus the Sydenhm river (5/127 (39.3%) versus 4/79 (5.1%), of fish, respectively; χ1 2 = 34.88, p <.1, N = 26 using non-prmetric medin test following rcsine squre root trnsformtion; figure 2), nd glochidi were only found ttched to the gills. Individul fish lso exhibited different intensity of glochidi infection (i.e. prsite burden), with up to 3 glochidi/fish in the Grnd River versus up to 6 glochidi/fish in the Sydenhm River (figure 2,b). Although N. melnostomus from the Grnd River were significntly smller thn those from the Sydenhm River (Mnn Whitney U = 6112, p =.3, N = 24, two individuls were missing tils; figure 2c), the logistic regression of glochidi infection (i.e. = not infected, 1 = glochidi infected) on TL ws not significnt (Ngelkerke s R 2 =.19). Sixteen of the glochidi found encysted on N. melnostomus from Grnd River site GR-d ( , ) nd 171 of the glochidi found encysted from site GR- ( , 8.469) were removed, mesured nd clssified ccording to the DFA model s: A. ligmentin (n = 13), Alsmidont viridis (n = 1), Amblem plict (n = 68), Elliptio diltt (n = 17), Epioblsm triquetr (n = 15), Fusconi flv (n = 9), Lsmigon complnt (n = 1), Obovri reflex (n = 36), Obovri olivri (n = 5), Pleurobem sintoxi (n = 7), Qudrul pustulos (n = 1), Qudrul qudrul (n = 2) nd Toxolsm prvum (n = 12). Four glochidi removed from N. melnostomus t the three sites on the Sydenhm River (( , ), ( , ), ( , )) were similrly clssified s: A. plict (n = 1),E.t.rngin (n = 1), Ligumi rect (n = 1) nd O. reflex (n = 1). The glochidi from s mny s six unionid species were encysted on given fish from the Grnd River Modelled contribution ofneogobius melnostomus to unionid recruitment The rtio of the slopes of glochidi lost from potentil recruitment (D g ) to the slope of juvenile mussels produced (J) versus the reltive bundnce of N. melnostomus to known suitble fish hosts (N; proportion between nd 1.) ws ssessed using species-specific prmeter vlues for E. triquetr, V. iris nd A. ligmentin (electronic supplementry mteril, tble S1). The loss of glochidi by N. melnostomus incresed 6.41 fster thn juvenile production for E. triquetr (figure 3), 394 fster for V. iris nd 3.35 fster for A. ligmentin (figure 3b). These D g : J rtios were lwys lowest for the primry hosts (1.1, 1.36 nd.33, respectively, for E. triquetr, V. iris nd A. ligmentin), were generlly higher for the mrginl host (C. birdii; 2.68, 2.18 nd 8.34, respectively, for E. triquetr, V. iris nd A. ligmentin) nd highest for N. melnostomus (figure 3b). The trend between the mrginl host nd N. melnostomus ws reversed on A. ligmentin, likely due to the fct tht very few C. birdii survived until the end of the metmorphosis period, which limited the ssessment of R M. In the cse of E. t. rngin nd L. fsciol, none of the 164 nd 1827 encysted glochidi, respectively, metmorphosed on N. melnostomus thus representing complete loss of reproductive output. 4. Discussion The results ofthis studyrevel novelwy in whichn. melnostomus ffect unionid mussels. Specificlly, reltively high lbortory infesttion rtes confirmed by the occurrence of glochidi infection on N. melnostomus in nture, combined with reltively low metmorphosis rtes in the lbortory support

7 () 8 (b) (c) totl length (cm) frequency 1 2 no. glochidi per fish N = 126 Grnd N =79 river 3 Sydenhm N =78 frequency (d) 12.5 totl length (cm) N = no. glochidi per fish body burden bsent present frequency totl length (cm) Figure 2. Frequency distributions of glochidi infection on Neogobius melnostomus in the field. Glochidi per individul N. melnostomus in the () Sydenhm nd (b) Grnd rivers, (c) totl lengths of N. melnostomus with infections from both rivers (boxes indicte the spred of totl lengths; horizontl lines within the boxes indicte medins; whiskers indicte the smllest nd lrgest vlues; nd outliers re represented by open circles; two fish missing tils were not included) nd (d) totl lengths nd the proportion of fish infected with glochidi (present or bsent) of N. melnostomus collected from both rivers. the hypothesis tht N. melnostomus re sink for unionid glochidi. Model results indicte tht N. melnostomus contribute more to the loss of reproductive potentil in unionid mussels thn to their recruitment. To the best of our knowledge, this represents hitherto unknown mnner by which n invsive species ffects ntive species. This system in which unionid glochidi re intercepted by n invsive host, nd subsequently metmorphose t low rtes or not t ll, is nlogous to tht of ntive plnts whose pollen is trnsported to incomptible invsive species nd essentilly wsted [26]. It differs from prsite spillbck systems [28] becuse incresed prsitism of invsive fish hosts represents sink for glochidi nd loss to ntive mussel communities. Generlly, glochidil infesttion rtes on N. melnostomus were similr to those on known primry nd mrginl hosts; however, metmorphosis rtes nd juvenile mussel production were generlly higher on primry hosts. Moreover, metmorphosis on N. melnostomus ws only observed for three of the five unionid species exmined (endngered V. iris nd E. triquetr, nd common species, A. ligmentin). This indictes tht lthough initil ttchment of glochidi to N. melnostomus occurs, glochidi do not metmorphose t rtes comprble to their primry or mrginl hosts. Initil ttchment is not surprising, given tht glochidi ttch to non-host fish [15] s well s to innimte objects [29]. Longer term ttchment nd encystment, necessry for successful juvenile mussel production, requires some set of species-specific chemicl cues [29] tht did not occur for N. melnostomus. The infesttion rtes on primry nd mrginl hosts re consistent with previous experiments tht hve exmined these mussel-host fish using individuls from the northern limit of their respective species rnges [3] (K. Loftus 214, Ontrio Ministry of Nturl Resources nd Forestry, personl

8 () no. individuls (b) (D g /J) glochidil loss juvenile production Epioblsm triquetr juveniles produced glochidi lost N. melnostomus (proportion of hosts) primry host C. birdii N. melnostomus E. triquetr V. iris A. ligmentin unionid species 8 Figure 3. Modelled juvenile production nd glochidil loss due to unionid infection on Neogobius melnostomus using equtions (2.2) nd (2.3) nd the prmeter vlues listed in the electronic supplementry mteril, tble S1. The model could not be pplied to two species whose glochidi successfully infested but filed to metmorphose on N. melnostomus. () Number of juvenile Epioblsm triquetr produced nd glochidi lost s N. melnostomus dominted the proportion of host fish in the system. (b) The rtio of glochidi loss to juvenile mussel produced (i.e. D g /J; note log scle) on N. melnostomus versus primry fish hosts (Percin cprodes for E. triquetr; Ambloplites rupestris for V. iris; nd Micropterus slmoides for A. ligmentin) nd Cottus birdii ( known mrginl host for these unionids). communiction), s is the vrible number of encysted glochidi on n individul fish for given mussel fish combintion [31]. Regrdless, the lbortory experiments permitted quntittive ssessment of which unionid species re ble to use N. melnostomus s host. This is the first study, to our knowledge, which hs specificlly ssessed the nturl glochidil infection of N. melnostomus in North Americ. Glochidi infection rtes of pproximtely 4% in the Grnd River indicte tht N. melnostomus encounter nd become infected with glochidi t rtes comprble to known hosts: infection rtes of 34% hve been reported on M. dolomieu by L. fsciol or nother Lmpsilis species (Morris nd Grndos 21, Fisheries nd Ocens Cnd, personl communiction); infection rtes of 4% were found on Morone mericn, whichishostofleptode ochrce, ndlmpsilis crios [32]; nd infection rtes of 46 71% were found on P. cprodes nd A. rupestris, which re known hosts of some rre unionid mussels (D. Woolnough 211, Centrl Michign University, personl communiction). The reduced infection rtes observed in the Sydenhm River my be reflective of the recent invsion of this system by N. melnostomus s subsequent smples obtined in 213 nd 214 hve shown glochidi infection rtes of 36% (29 of 79 fish) nd 81% (5 of 61 fish) (T.J.M. 215, personl communiction). Although the sptil nd temporl nture of the unionid N. melnostomus encounters is not known, the nlysis of N. melnostomus from the Grnd River provides some vluble informtion. The mjority of encysted glochidi included A. ligmentin, A. plict, E. diltt, E. triquetr nd O. reflex. Actinonis ligmentin, the common species exmined in the lbortory experiments, is host generlist tht brodcsts its glochidi into the wter column [33], so glochidi encounter with N. melnostomus is not surprising. Amblem plict is lso host generlist [33] ndis common. Elliptio diltt, whichis common in Ontrio (S Rnk = 5[34]) nd host generlist [33], is not bundnt in the lower Grnd River where

9 N. melnostomus were collected [35]. Conversely, E. triquetr nd O. reflex re host specilists [33] nd rre in Ontrio [6]. Interestingly, O. reflex relese glochidi in worm-like conglutintes (i.e. pckges of glochid formed by the femle mussel) [33] nd A. plict relese fish-like conglutintes. A benthic fish such s N. melnostomus would likely consider conglutintes s prey items nd hence become infested. There my be n ssocition between glochidil dispersl/host ttrction mechnism nd infection on N. melnostomus, which suggests tht the effect of N. melnostomus on unionid species my not be uniform cross the txon. This wrrnts dditionl study. The reltive contribution of N. melnostomus to juvenile mussel production (J) versus glochidi loss (D g ) is determined primrily by the metmorphosis rte (R M ) of glochidi on N. melnostomus (eqution (2.3)/(2.2)) given by D g J = 1 R M R M. (4.1) Both J nd D g increse s the proportion of N. melnostomus increse, but higher R M leds to more J. This model does not ccount for differences in host ttrction strtegies mong unionids [9], nor predtion or competition for conglutintes with nturl host fishes by N. melnostomus, which could result in N. melnostomus-medited recruitment being the sole source of J for some species whose hosts experience decline or re extirpted. Neogobius melnostomus my reduce the pool of effective hosts for unionids where they co-occur, interfere with the norml reproductive cycle of unionids when they chieve high popultion densities [16], nd consequently hve greter impct on these biodiversity hot-spots thn ws predicted [5]. Such indirect effects hve been recently demonstrted in Andont ntin ( host generlist), which hd lower metmorphosis on non-ntive fish hosts nd lower success overll (i.e % fewer fish species) s result of biotic homogeniztion [16]. Although there is error ssocited with our model, especilly the error ssocited with estimte of R M for reltively poor host species, nd simplifiction of the mussel host interctions in the lbortory, the impliction of glochidi loss compred with primry hosts remins fundmentl finding, which will be excerbted s N. melnostomus domintes the ecosystem in biodiversity hot-spots. It is relevnt to note tht significnt differences in infection rtes hve been found between recently isolted mussel popultions [36], nd such differences my lso occur between geogrphiclly distinct lineges of host fish s well s within lineges of mussels nd their fish hosts [37]. Consequently, it is possible tht the effects of invsive N. melnostomus on host prsite reltionships my vry sptilly nd temporlly due to popultion-specific ttributes rising from locl dpttions nd fine-scle coevolutionry dynmics. The introduction of new species into n ecosystem hs the potentil to cuse mny unnticipted effects s result of the high vribility in the rtes of spred, types of impcts nd species interctions [38]. The results of this study indicte tht N. melnostomus re likely cting s sink for glochidi, whereby they prevent glochidi from reching their intended hosts. This hs negtive implictions for unionid species tht exhibit high rtes of infection nd poor/no metmorphosis on N. melnostomus, prticulrly those species whose popultions re limited geogrphiclly to res with lrge popultions of this invsive fish (e.g. riverine refugi of the Lurentin Gret Lkes). A thorough understnding of the effects of species invsion on n ecosystem cn help to predict the effects of its invsion elsewhere, t lest in terms of type nd direction [39]. These predictions, in turn, dictte how much effort should be directed towrds hlting the invsion nd spred of new species, nd more generlly, contribute to our understnding of the ecology of invsions. 9 Ethics. Fish nd mussels were collected under Licence to Collect Fish for Scientific Purposes OMNR permit nd cred for ccording to University of Guelph Animl Utiliztion Protocol 8RO91. All mussels were collected under Species t Risk Act, 24 permit SECT 73 SARA C&A: 1-12 nd 11-2 nd Endngered Species Act, 27 permit SR-B-4-1 nd AY-B Dt ccessibility. The supporting dt re vilble on Dryd: Authors contributions. M.E.M.T., T.J.M. nd J.D.A. designed the reserch; M.E.M.T. performed reserch; M.E.M.T. nd J.D.A. wrote the pper. All uthors gve finl pprovl for publiction. Competing interests. The uthors hve no competing interests. Funding. Support ws provided by Fisheries nd Ocens Cnd (DFO), Ontrio Ministry of Nturl Resources (OMNR), Cndin Wildlife Federtion (CWF) nd Nturl Sciences nd Engineering Reserch Council of Cnd (NSERC) to J.D.A. Acknowledgements. The uthors would like to thnk Dr Gerry Mckie, Dr Mike Nishizki, Kelly McNichols-O Rourke, Rosrio Cstñon, Amnd Conwy, Anthony Mernte, Bob Frnk, Mtt Cornish, John Schwindt (Upper Thmes Region Conservtion Authority), Dr Mrk Poesche (University of Albert) nd Json Brnucz nd his field crew t DFO.

10 References 1 1. Munwr M, Munwr IF, Mndrk NE, Fitzptrick M, Dermott R, Lech J. 25 An overview of the impct of non-indigenous species on the food web integrity of North Americn Gret Lkes: Lke Erie exmple. Aqut. Ecosyst. Hel. Mng. 8, (doi:1.18/ ) 2. Nlep TF, Schloesser DW (eds). 214 Qugg nd zebr mussels: biology, impcts, nd control, 2nd edn. Boc Rton, FL: CRC Press. 3. Mckie G Biology of the exotic zebr mussel, Dreissen polymorph, in reltion to ntive bivlves nd its potentil impct in Lke St. Clir. Hydrobiologi 219, (doi:1.17/bf 24759) 4. Nlep TF, Hrtson DJ, Gostenik GW, Fnslow DL, Lng GA Chnges in the freshwter mussel communityoflkest.clir:fromunionideto Dreissen polymorph in eight yers. J. Gret Lkes Res. 22, (doi:1.116/s38-133(96) ) 5. Poos M, Dextrse A, Schwlb AN, Ackermn JD. 21 Secondry invsion of the round goby into high diversity Gret Lkes tributries nd species t risk hotspots: potentil new concerns for endngered freshwter species. Biol. Invsions 12, (doi:1.17/s x) 6. COSEWIC. 214 Wildlife Species Serch. Dtbse: Committee on the Sttus of Endngered Wildlife in Cnd. 7. Jude DJ, Jnssen J, Crwford G Ecology, distribution nd impct of the newly introduced round nd tubenose gobies on the biot of the St. Clir nd Detroit rivers. In TheLkeHuronecosystem: ecology, fisheries nd mngement (eds M Munwr, T Edsll, J Lech). Ecovision World Monogrph Series. The Hgue, The Netherlnds: SPB Acdemic Publishing. 8. French JRP, Jude DJ. 21 Diets nd diet overlp of nonindigenous gobies nd smll benthic ntive fishes co-inhbiting the St. Clir River, Michign. J. Gret Lkes Res. 27, (doi:1.116/s38-133(1)7645-4) 9. Brnhrt MC, Hg WR, Roston WN. 28 Adpttion to host infection nd lrvl prsitism in Unionoid. J. North Am. Benthol. Soc. 27, (doi:1.1899/7-93.1) 1. Hg WR. 212 NorthAmericnfreshwtermussels: nturl history, ecology, nd conservtion. Cmbridge, UK: Cmbridge University Press. 11. McNichols KA, Mckie GL, Ackermn JD. 211 Host fish qulity my explin the sttus of endngered Epioblsm torulos rngin nd Lmpsilis fsciol (Bivlvi: Unionide) in Cnd. J. North Am. Benthol. Soc. 3, 6 7. (doi:1.1899/1-63.1) 12. Wtters GT, Menker T, Thoms S, Kuehnl K. 25 Host identifictions or confirmtions. Ellipsri 7, Muzzll PM, Peebles CR, Thoms M Prsites of the round goby, Neogobius melnostomus,nd tubenose goby, Proterorhinus mrmortus (Perciformes: Gobiide), from the St. Clir River nd Lke St. Clir, Michign. J. Helminthol. Soc. Wshingt. 62, Kornis MS, Mercdo-Silv N, Vnder Znden MJ. 212 Twenty yers of invsion: review of round goby Neogobius melnostomus biology, spred nd ecologicl implictions. J. Fish Biol. 8, (doi:1.1111/j x) 15. Jnsen W, Buer G, Zhner-Meike E. 21 Glochidil mortlity in freshwter mussels. In Ecology nd evolution of the freshwter mussels Unionoid (eds GBuer,KWchtler),pp Berlin,Germny: Springer. 16. Doud K, Lopes-Lim M, Hinzmnn M, Mchdo J, Vrnds S, Teixeir A, Sous R. 213 Biotic homogeniztion s thret to ntive ffilite species: fish introductions dilute freshwter mussel s host resources.divers. Distrib. 19, (doi:1.1111/ddi.1244) 17. Dodd BJ, Brnhrt MC, Rogers-Lowery CL, Fobin TB, Ronld VJ. 25 Cross-resistnce of lrgemouth bss to glochidi of unionid mussels. J. Prsitol. 91, (doi:1.1645/ge-511r.1) 18. Huebner JD, Pynnonen KS Vibility of glochidi of two species of Anodont exposed to low ph nd selected metls. Cn. J. Zool. 7, (doi:1.1139/z92-315) 19. ASTM. 212 Stndrd guide for conducting lbortory toxicity tests with freshwter mussels (E2455-6). Annu. B. ASTM Stnd, E Kidd BT Unionide of the Grnd River dringe. MSc thesis, Crleton University, Ottw, Ontrio, Cnd. 21. Stton SK, Dextrse A, Metclfe-Smith JL, Di Mio J, Nelson M, Prish J, Kilgour B, Holm E. 23 Sttus nd trends of Ontrio s Sydenhm River ecosystem in reltion to qutic species t risk. Environ. Monit. Assess. 88, (doi:1.123/A: ) 22. Zr JH Sttisticl nlysis. Englewood Cliffs, NJ: Prentice Hll. 23. Quinn GP, Keough MJ. 22 Experimentl design nddtnlysisforbiologists. Cmbridge, UK: Cmbridge University Press. 24. Trembly MEM, Morris TJ, Ackermn JD. 215 A multivrite pproch to the identifiction of unionid glochidi with emphsis on Species t Risk in Southern Ontrio. Cn. Mnuscr. Rep. Fish Aqut. Sci. 357,vii McNichols KA. 27 Implementing recovery strtegies for mussel species t risk in Ontrio. MSc thesis, University of Guelph, Ontrio, Cnd. 26. Bjerknes A, Totlnd O, Heglnd SJ, Nielsen A. 27 Do lien plnt invsions relly ffect pollintion success in ntive plnt species? Conserv. Biol. 138, (doi:1.116/j.biocon ) 27. Schwlb AN, Grvie M, Ackermn JD. 21 Dispersion of freshwter mussel lrve in lowlnd river. Limnol. Ocenogr. 55, (doi:1.4319/lo ) 28. Kelly DW, Pterson RA, Townsend CR, Poulin R, Tompkins DM. 29 Prsite spillbck: neglected concept in invsion ecology? Ecology 9, (doi:1.189/ ) 29. Wood EM Some mechnisms involved in host recognition nd ttchment of the glochidium lrv of Anodont cygne (Mollusc: Bivlvi). J. Zool. 173, (doi:1.1111/j tb1744.x) 3. Brnhrt MC, Bird MS. 2 Fish hosts nd culture of mussel species of specil concern: nnul report for Columbus, MO: US Fish nd Wildlife Service nd Missouri Deprtment of Conservtion. 31. Riusech FA, Brnhrt MC. 2 Host suitbility nd utiliztion in Venustconch ellipsiformis nd Venustconch plesii (Bivlvi:Unionide) from the Ozrk Plteus. In Proc. Conservtion, Cptive Cre nd Propgtion of Freshwter Mussels Symp. 1998, pp Columbus, OH: Ohio Biologicl Survey. 32. Kneelnd SC, Rhymer JM. 28 Determintion of fish host use by wild popultions of rre freshwter mussels using moleculr identifiction key to identify glochidi. J. North Am. Benthol. Soc. 27, (doi:1.1899/7-36.1) 33. Wtters GT, Hoggrth MA, Stnsbery DH. 29 The freshwter mussels of Ohio. Columbus, OH: Ohio Stte University Press. 34. Cordeiro J. 29 Comprehensive Report Species Elliptio diltt. Dtbse: NtureServe. 35. Metclfe-Smith JL, Mckie GL, Di Mio J, Stton SK. 2 Chnges over time in the diversity nd distribution of freshwter mussels (Unionide) in the Grnd River. J. Gret Lkes Res. 26, (doi:1.116/s38-133()777-6) 36. Doud K, Sell J, Kubíková-Peláková L, Horky P, Kczmrczyk A, Mioduchowsk M. 214 Host comptibility s criticl fctor in mngement unit recognition: popultion-level differences in mussel fish reltionships. J. Appl. Ecol. 51, (doi:1.1111/ ) 37. Reichrd M et l. 215 Popultion-specific responses to n invsive species. Proc.R.Soc.B282, (doi:1.198/rspb ) 38. Melbourne BA, Hstings A. 29 Highly vrible spred rtes in replicted biologicl invsions: fundmentl limits to predictbility. Science 325, (doi:1.1126/science ) 39. Kulhnek SA, Riccirdi A, Leung B. 211 Is invsion history useful tool for predicting the impcts of the world s worst qutic invsive species? Ecol. Appl. 21, (doi:1.189/ )