The addition of round goby to the Lake Ontario prey fish community can, because it appears to lack thiaminase, ameliorate the effects of a diet high in thiaminase-rich alewives associated with a thiamine deficiency in Lake Trout. However, the effects of round goby predation may negate such effects. We conducted studies to assess the proportion, by weight, of round gobies in the diet of contemporary Lake Ontario Lake Trout and predicted its effect on egg thiamine concentration, and early mortality syndrome (EMS) and growth retardation, two indicators of thiamine deficiency. We compared these parameters to that for the historic prey community, as well as possible future diet scenarios with progressively greater proportions of round goby (e.g. 50, 75 and 100%). To assess the combined and separate effects of the thiamine deficiency we used a modeling approach to determine the effect of EMS (70%) typical of Lake Ontario Lake Trout, growth retardation, density of Lake Trout egg (e.g. 100, 500, 1000, and 5000 eggs m−2) and round goby (e.g. 1, 9, 18, 35, 75, 150 gobies m−2), three separate durations of round goby residence on Lake Trout spawning habitat (e.g. 30, 60 and 213 days), and at either a shallow water or deepwater temperature regime. Eight years after first invading Lake Ontario, the proportion of round gobies in Lake Trout diets remains low, although the highest observed (25%) would result in significant amelioration of the thiamine deficiency from past conditions. Modeling indicated that the negative effects associated with round goby predation on eggs and fry could easily exceed any positive effects resulting from amelioration of the thiamine deficiency by a diet containing round goby. The effects of goby predation were found to be dependent on density of both Lake Trout eggs and round gobies, and while predation effects were minimally affected by temperature regime, they were highly dependent on the period of round goby residence on Lake Trout spawning habitat.
The Great Lakes face an ongoing and apparently increasing threat from the effects of aquatic invasive species (AIS) (Ricciardi et al., 2001), yet the emphasis of invasion biology (Davis and Thompson, 2001) has been more on the invasiveness of a particular species (Kolar and Lodge, 2001) than on impacts; impacts are often inferred from invasiveness by policy-makers and stakeholders (Falk-Petersen et al., 2006). For example, the Canadian government considers invasive species as “harmful alien organisms whose introduction or spread threatens the environment” (Environment Canada, 2004). Such an approach is unlikely to produce valid predictions of impacts since the invasiveness of a particular species does not necessarily predict its impact, and it is only through the detailed knowledge of the ecology of target species that the impacts of AISs can be adequately assessed (Ricciardi and Cohen, 2007). Many species having a high potential to invade, have had few recognizable impacts once they invaded (Williamson and Fitter, 1996; Parker et al., 1999); while others whose distribution is limited like the Asiatic clam Potamocorbula amurensis found only in San Francisco Bay, have exerted strong impacts (Kimmerer et al., 1994). In addition, the impacts of a particular invader may take a prolonged period to occur and/or be recognized. For example, for alewife Alosa pseudoharengus, although an invader to the Great Lakes in the late 1800s, it was not until the late 1900s that its effect as a significant predator on Lake Trout fry (Krueger et al., 1995) and as the cause of a thiamine deficiency amongst top predators such as salmon, trout, walleye, and American eel (Fitzsimons and Brown, 1998; S. Brown, Environment Canada, pers. comm., unpublished data), were recognized.
Interactions between the impacts of individual invaders have received relatively little attention; but the potential for facilitation as well as amelioration exists. For example, the rapid growth, spread, and ultimate impacts of round gobies Apollonia melanostoma, (herein called gobies), throughout the Great Lakes, have been linked in part to their ability to use dreissenids, represented by Dreissena polymorpha and D. bugensis (Johnson et al., 2005). These invaded and spread throughout the Great Lakes a decade earlier (Griffiths et al., 1991), and were initially viewed as an energy sink and a loss of production for pelagic systems (Johannsson et al., 2000). Gobies however, resulted in the remobilization of production formerly thought lost to dreissenids, such that the dreissenids have facilitated gobies. Amelioration where the impacts of one invader are potentially cancelled out or reduced by another invader is also possible. Invasive prey species such as alewives or rainbow smelt Osmerus mordax, herein called smelt, containing high thiaminase activity (Tillitt et al., 2005), have been implicated in the development of thiamine deficiencies and related effects in their predators (Fitzsimons and Brown, 1998; Fitzsimons et al., 1999; Honeyfield et al., 2005), such that an invading prey species lacking such activity and capable of displacing species like alewives and smelt in salmonine diets, could potentially reduce or ameliorate the impacts of these thiaminase containing species.
Alewives, which contain high levels of thiaminase activity (Tillitt et al., 2005) while having normal concentration of thiamine (Fitzsimons et al., 1998), are amongst the most significant sources of thiamine deficiencies in the Great Lakes Basin, and are associated with significant reductions in the egg thiamine levels of Lake Trout Salvelinus namaycush (Fitzsimons and Brown, 1998; Honeyfield et al., 2005; Fisher et al., 1996). The resulting thiamine deficiency impairs reproduction through early mortality syndrome (EMS) (Fitzsimons, 1995 a,b; Fitzsimons et al., 1999; Brown et al., 1998) and reduced growth, predator avoidance and foraging at the larval stage (unpublished data). In Lake Ontario, where alewives have dominated the diet of Lake Trout for over three decades (Lantry, 2001; Olsen et al., 1988; Rand and Stewart, 1998), the thiamine deficiency and associated effects on reproduction is thought to be a major impediment to natural reproduction (Brown et al., 2005; Fitzsimons et al., 2003; Fitzsimons et al., 2007a).
Gobies, also an invasive species that was first found in the Great Lakes in 1990 (Charlebois et al., 1997), unlike alewives, have virtually no thiaminase according to extensive surveys of prey fish in Lake Michigan (Tillitt et al., 2005). Consequently a diet dominated by gobies instead of alewives could provide relief for Lake Ontario salmonines like Lake Trout, from the thiamine deficiency caused by alewives (Fitzsimons et al., 2007). The expected improvement in reproduction (Honeyfield et al., 2005) would represent a type of amelioration for the negative impacts of alewives. As early as 2003 there was evidence that Lake Ontario Lake Trout were consuming gobies, being the second most important prey species in terms of abundance (20%) and mass (36%) to alewives (Dietrich et al., 2006). That an increasing proportion of gobies in the diet may improve thiamine levels was suggested by data from Lake Huron where gobies have also invaded, and Lake Trout stocks feeding mostly on gobies had higher levels of thiamine in their eggs than those feeding mostly on alewives (D. Honeyfield, USGS, pers. comm.; J. Johnson, MDNR, pers. comm.).
Although gobies may be capable of improving recruitment by Lake Trout through increased thiamine levels and corresponding reductions in EMS and other thiamine deficiency impacts, gobies are also a significant predator on Lake Trout eggs and fry (Chotkowski and Marsden, 1999; Fitzsimons et al., 2006). Based on modeling, the predation effects of an interstitial egg predator like gobies are expected to have significant effects on Lake Trout (Jones et al., 1995). A single goby in the laboratory, can consume up to 7 Lake Trout eggs per day depending on holding conditions (Fitzsimons et al., 2006). The same cobble habitat that is preferred by goby (Jude et al., 1995; Ray and Corkum, 2001) is also the preferred spawning habitat for Lake Trout (Marsden et al., 1995) such that a high degree of habitat overlap between gobies and Lake Trout eggs and fry is expected with potentially significant effects on egg and fry survival. Hence the negative effects associated with predation could reduce, cancel out, or in fact exceed any benefit associated with goby diet mediated relief from the thiamine deficiency.
To evaluate the effect of goby in the diet on the thiamine deficiency, we used a range of proportions of gobies in the diet based on past (1998–1999) and present (2005) diet surveys, as well as possible future diet scenarios, and predicted egg thiamine concentrations for these diets from published studies that related the amount of thiaminase in the diet of a Lake Trout to the thiamine concentration in its eggs, that was in turn related to amounts of EMS and growth retardation. To assess the possible outcome of increasing goby abundance in the prey community of Lake Ontario on Lake Trout reproduction, we modeled the combined effect of the thiamine deficiency, and its possible diminution by a goby diet, and the effect of goby predation on Lake Trout eggs and fry, under a number of scenarios, on Lake Trout egg and fry mortality. For goby predation, we used six densities of gobies to reflect what has been observed at Lake Trout spawning reefs in lakes Michigan (Jonas et al., 2005; J. Jonas and R. Claramunt, MDNR, pers. comm.) and Ontario (this study). The effect of the period of residence of gobies on nearshore Lake Trout spawning reefs, was assessed for 30, 60 and 213 day periods to reflect the variation observed for nearshore habitats in the Great Lakes (J. Jonas and R. Claramunt, MDNR, Charlevoix, pers. comm.; Walsh et al., 2007). As goby predation on Lake Trout eggs is temperature dependent (Fitzsimons et al., 2006) we modeled effects for the temperature regime at representative shallow water (e.g. 5 m) (Fitzsimons, 1995b) and deep water (e.g. 40–50 m) (Janssen et al., 2006) reefs, having dynamic and relatively constant temperature regimes (Janssen et al., 2006) respectively. Information from published egg (Fitzsimons et al., 2006) and fry (Chotkowski and Marsden, 1999) predation studies were used to further parameterize the predation model. To incorporate the combined effect of thiamine deficiency with predation, we included EMS or growth impairment effects with predation effects.
To test our model predictions, we conducted intensive studies at a small spawning reef in western Lake Ontario (Port Weller) where egg and fry density had been well characterized (Fitzsimons 1995b; Fitzsimons et al., 2003) before gobies colonized this location, likely in 2001, and compared expected egg and fry densities with observed post goby invasion.
Diet composition of Lake Trout in Lake Ontario
We used the proportion of prey fish species in stomach contents of Lake Trout collected by gill nets from four individual sites in Lake Ontario (e.g. Niagara on the Lake, Port Credit, Cobourg and Oswego) during May and September of 2005 to determine Lake Trout diets. Eight 45 m panels of multifilament gill net ranging in size from 91 to 151 mm (1.8 m high) were set overnight, perpendicular to depth contours in 25–50 m of water depending on bottom temperature. During isothermal conditions in the spring, nets were set from 25 m outward; when the water column was stratified, nets were set outward from the 9°C isotherm. The shoreline distance between any two locations exceeded that of the estimated home range for Lake Trout (see Fitzsimons, 1995b) so fish from each location were considered to represent distinct non-overlapping feeding areas. Thiamine is deposited into the Lake Trout egg throughout the period of oogenesis that starts in April and is not complete until after September (unpublished data), so that thiamine contributions to the developing eggs during the early (e.g. May) and latter (e.g. September) part of oogenesis, likely contribute to the final egg thiamine concentration at the time of spawning. After collection, Lake Trout were measured for length (mm) and weight (gm) and the number and length of all prey items in each individual Lake Trout's stomach determined. The wet weight (gm) of each prey item was estimated from a power relationship (e.g. weight = a(length)b) developed between the weight and length of each prey species based on pooled samples collected from western and eastern Lake Ontario during 2005. The percent composition in the diet for each prey species was based on the proportion of its total wet weight to the total wet weight of diet items in an individual Lake Trout's stomach. Data was analyzed by size category (e.g. 601–700, 701–800 and 801–900 mm) so as to include only mature females; female Lake Trout mature at 600 mm in Lake Ontario (O'Gorman et al., 1998). An average for each size category was calculated as the mean of the two seasonal means and from this, a grand mean was calculated across either two or three size categories depending on availability. We used two-way ANOVA to determine whether there was an effect of location or season for any of the four prey species. When a significant effect of either factor was detected, we used a Bonferoni pairwise comparison to locate these differences. For this we used only the proportions for the two smallest size categories (e.g. 600–700, 701–800 mm) as fish in these size categories were consistently present at all locations and for both seasons. To stabilize variance and achieve normality all proportions were arc-sine transformed prior to analysis and significance set at p ≤ 0.05.
We selected six goby densities to model (e.g. 1, 9, 18, 35, 75 and 150·m−2) based on spatial and temporal variation in goby density at a single spawning reef in Lake Ontario at Port Weller, and four reefs located in northeastern Lake Michigan (Jonas et al., 2005) (Table 1: available at www.aehms.org/Journal/12_3_Fitzsimons_Appendix.pdf). Gobies in all cases were collected using standard Lake Trout egg nets and associated protocols (Fitzsimons et al., 2002; Jonas et al., 2005) with the following exception: at Port Weller, during the fall of 2002 and 2003, 1-mm nets were placed over egg nets by divers and the nets taken to the surface before the nets contents were sorted and the number and sizes of gobies present determined. This was necessitated by the observation that during the spring of 2002 when divers emptied nets of substrate on the bottom, they noticed gobies escaping from nets although the number of escapees was included in net counts. Since not all gobies escaping nets at this time were likely seen by divers we consider this count to be conservative. This behaviour was not noted in the Lake Michigan collections and nets were all emptied on bottom by divers.
Based on trawling, gobies in Lake Ontario show offshore migration during the fall (Walsh et al., 2007), although based on scuba diver observations, gobies are present on spawning reefs in eastern and western Lake Ontario up until at least December and as early as April at temperatures of approximately 3–4°C (J. Lantry, New York Department of Environmental Conservation, Cape Vincent, NY., pers. comm., unpublished data). In Lake Michigan gobies are at Lake Trout spawning reefs in northeastern Lake Michigan at least until November when temperatures are approximately 6–10°C (S. Lennart, Little Traverse Bay Band of Odawa Indians, Bay Harbour, MI, pers. comm.). In southwestern Lake Michigan, gobies are still on spawning reefs until at least December when temperature is approximately 4°C (J. Dettmers, GLFC, Ann Arbor, MI, pers. comm.). To reflect seasonal variation in the observed period of residence of gobies on Lake Trout spawning habitat in lakes Ontario and Michigan; we modeled the effects for 30 (October 15 to November 14), 60 (October 15 to December 14), and 213 (October 15 to May 15) day periods of goby residence on the spawning reef, assuming that for the first two periods that density declined to zero immediately after the time period had elapsed because gobies had migrated offshore.
Goby egg and fry consumption
To calculate individual goby egg consumption over the period of residence on a reef, we used the formula:Fitzsimons et al. (2006) relating egg consumption by a single goby (total length (75 ± 1 mm; mean ± SEM) to temperature. We assumed all eggs were deposited on October 15. We assumed that consumption of sac-fry by gobies was the same as for eggs, as Chotkowski and Marsden (1999) found no difference in the predation rate on these embryonic stages by gobies although it was significantly less than for emergent fry. Consequently at the onset of emergence, which we set at April 15 and which lasted for 30 days, we used a predation rate of one emergent fry consumed · goby−1 day–1. Lacking information on the temperature dependency of emergent fry predation by goby, we assumed it was temperature independent. We assumed that the rate of egg consumption by gobies was unaffected by either egg or goby density. Fitzsimons et al., (2007) found that although the number of Lake Trout eggs eaten per individual per day by one goby was higher than that of five gobies, consumption by five gobies was no different than that of ten gobies, suggesting that per individual consumption may reach an asymptote at a relatively low density of gobies such that per individual egg consumption may be similar across most of the range of goby densities included in the model. Further these authors found that egg consumption by gobies peaked at approximately 3000 eggs m−2. This egg density was intermediate to the two highest egg densities (e.g. 1000 and 5000 eggs m−2) used in modeling (see below) such that by selection of the egg densities used in the model, the effect of egg density on consumption was reduced although not eliminated.
To reflect a typical temperature regime during the egg incubation period of a relatively shallow and a relatively deep reef respectively, we used the 2001–2002 seasonal temperature cycle from October 15 to May 15 for Port Weller from previous studies (Fitzsimons et al., 2003), and the 1982–1983 temperature cycle for Sheboygan Reef (Gottlieb et al., 1989), the only period for which temperature data was available for this site (Figure 1: available at www.aehms.org/Journal/12_3_Fitzsimons_Appendix.pdf). We used October 15 as the starting date as it corresponded with the onset of spawning for the two temperature regimes. Fitzsimons et al. (1995b) gave the peak spawning date for Port Weller in 1992 as October 15. Although the peak spawning date for Sheboygan reef is not known, Janssen et al., (2006) found eyed eggs nearby in mid November of 2003. For the temperature regime at Sheboygan reef during October-November, when temperatures are approximately 5–6°C, the eyed egg stage would have been reached in approximately 30 days according to Balon (1980), and hence a spawning date of October 15 for Sheboygan seems reasonable.
The reef at Port Weller, located in western Lake Ontario adjacent to the Welland Canal, is man-made and has an average depth of 5 m (Fitzsimons, 1995b). Sheboygan Reef, a large natural reef located in west-central Lake Michigan, ranges in depth from 40 to 50 m, (Janssen et al., 2006). We were unable to locate temperature data for a deep reef in Lake Ontario although given the bathymetry and maximum depth of this lake (i.e. 244 m) it is possible that deep reefs could exist in this lake, as have been documented for other Great Lakes (Thibodeau and Kelso, 1990). We modeled the effects of gobies at four egg densities (e.g. 100, 500, 1000, and 5000 eggs m−2) to reflect egg densities that have been observed at Lake Trout spawning reefs in Lake Ontario (Fitzsimons, 1995b; Perkins and Krueger, 1995; Fitzsimons et al., 2003). To assess the combined effect of goby predation and the thiamine deficiency, we evaluated the effect of the four densities of gobies and either EMS, which we set to 70% mortality, typical of Lake Ontario in 1994 and probably representative of years since because of ongoing low egg thiamine levels (Brown et al., 1998; Fitzsimons et al., 2007), or growth retardation (see below), which we set at 100% mortality. The effect of either EMS or growth retardation, were added on the first day of emergence.
Effect of diet composition on the thiaminase content of the diet and corresponding Lake Trout egg thiamine concentration
To assess the effect of Lake Trout diet composition on the egg thiamine concentration of female Lake Trout, we used a relationship developed by Honeyfield et al. (2005) between dietary thiaminase in the maternal diet and resulting thiamine in their eggs:
The amount of thiaminase in the diet (e.g. pmol g−1 min−1) was based on the proportion of alewife, smelt, slimy sculpin Cottus cognatus (herein referred to as sculpin) and goby in the diet and corresponding thiaminase activity for each species. For alewives, we used the reported thiaminase content of Lake Ontario alewives (4236 pmol g−1 min−1; Fitzsimons et al., 2005) that was adjusted to take account of the seasonal variation in thiaminase activity noted for Lake Michigan alewives (Tillitt et al., 2005). This brought predicted egg thiamine concentrations in line with those recently reported for Lake Ontario Lake Trout (Fitzsimons et al., 2007a). For smelt, we used the average thiaminase content reported for Lake Michigan smelt (2636 pmol g−1 min−1; Tillitt et al., 2005) but with no adjustment for seasonal variation, since none was observed in this study. Finally for sculpins and gobies, we used the thiaminase activity reported by Tillitt et al. (200%) for deepwater sculpin Myoxocephalus thompsoni (172 pmol g−1 min−1) and goby (18 pmol g−1 min−1), respectively. We used Lake Trout diet information from Lantry (2001) to infer the proportion of alewives in the diet in 1998–1999 prior to gobies becoming widely distributed in Lake Ontario and still absent from the diet of Lake Trout. Similarly for 2005, eight years post goby invasion of Lake Ontario, we used the proportions of prey fish species based on diet surveys of Lake Ontario (see above) and corresponding thiaminase content estimated as above. In addition to the 2005 data, we also included three other scenarios in our modeling to reflect the potential future importance of gobies in the diet of Lake Trout. Deepwater sculpins, a potential historic ecological analog to the goby because of its size, benthic habitat and diet, once comprised close to 100% of the diet of Lake Trout (Dymond, 1928). Accordingly for these three future scenarios, we used the following proportions of alewife:goby in the diet: 50:50, 25:75 and 0:100%, where we assigned their thiaminase content as above.
Effect of egg thiamine on EMS mortality
We used the following dose-response relationship of Fitzsimons et al. (2007a) to predict EMS mortality for a given egg thiamine concentration:
The total mortality due to EMS was added on the first day of emergence (e.g. April 15).
Effect of egg thiamine on larval growth and associated mortality
To assess the effects of the thiamine deficiency on growth, we used the dose-response of Fitzsimons et al., (in press, 2009) between egg thiamine and specific growth rate (SGR),Morito et al., 1986). Since feeding is the only way of increasing thiamine levels and reversing the thiamine deficiency, the likely outcome of anorexia in this case is death.
Assessment of model predictions of effects of goby on egg abundance and fry emergence in the wild
To determine egg and fry abundance at the spawning reef at Port Weller (Fitzsimons, 1995b) up until and including 1999, prior to when gobies first invaded this site, we used the data given in Fitzsimons et al. (2003). Similarly, after 1999 we used the methods found in Fitzsimons et al. (2003) to assess egg and fry abundance in 2002, 2003, and 2004. Egg abundance was assessed using 14 to 17 egg nets (Fitzsimons, 1995b) buried in the substrate by scuba divers while fry emergence the following spring was assessed using 7 to 12 fry traps (Fitzsimons et al., 2003) that were located over the same area as where the egg nets had been deployed the fall of the previous year. Goby were first observed in trawl sampling near the reef in 2000 (unpublished data). In 2002 egg nets were recovered in April and December and in 2003 only in December. We had intended to recover egg nets during the fall of 2001 but because of inclement weather, had to wait until April 2002. As these nets were deployed in the fall of 2001 we considered that their numbers were representative of that period when retrieved in April 2002. Determination of goby size and abundance in egg nets followed the methods of Fitzsimons et al. (2002) as noted above. For all dates only gobies greater than or equal to 42 mm in length were included in the estimate of abundance since smaller gobies were found to be unable to consume Lake Trout eggs (Jonas et al., 2005).
As predation by gobies may have affected the number of eggs found in egg nets retrieved in 2002 and 2003, we estimated what this effect may have been. For this we developed a relationship between egg abundance at Port Weller and a lake wide average CPUE of mature female spawners (> 4000 gm) (Schneider et al., 1996; Lantry and Lantry, 2007) from the five years sampled (e.g. 1992, 1994, 1997, 1998, 1999 ) prior to gobies being present at this site, and used this to estimate egg abundance for 2002 and 2003, when gobies were present. We then compared these values with measured egg abundance for these years. For CPUE, we used the CPUE for the year prior to egg deposition, reasoning that not all spawners considered mature in September of a year would necessarily contribute to egg deposition one month later of that same year but assumed that would be the case for 13 months later. Perkins and Krueger (1995) reported that the egg viability of first year spawners was considerably less than that of females that had spawned previously, suggesting the eggs of first year spawners may be underrepresented in the eggs collected in egg nets, because of higher mortality and subsequent disintegration. We compared the number of eggs observed at this site in either 2002 or 2003 with the number of eggs predicted for that year from spawner CPUE.
ANOVA was used to assess temporal changes in the size and abundance of gobies and Lake Trout eggs and fry. Where a significant temporal effect occurred, a Bonferroni pairwise comparison was used to locate these differences. Data was log transformed where appropriate to stabilize variance and achieve normality with tests declared significant at P ≤ 0.05.
Lake Trout diet and relationship with thiaminase activity, egg thiamine, and EMS
There was significant spatial (F3,8 = 19.7, p < 0.001) and seasonal (F1,8 = 14.7, p = 0.005) variation in the proportion of alewives in the diet of Lake Trout during 2005 (Table 2: available at www.aehms.org/Journal/12_3_Fitzsimons_Appendix.pdf). Alewives comprised a significantly higher proportion of the diet in summer, and the proportion of alewives in the diet of Lake Trout collected at Oswego was significantly (p = 0.001) higher than at Niagara on the Lake and Port Credit; Cobourg was similar to Niagara on the Lake. Neither the proportion of smelt nor the proportion of goby in the diet showed seasonal or spatial variation. However, gobies were present in the diet of the two western Lake Ontario sites but absent from the eastern sites.
The proportion of sculpin in the diet was higher in the spring (F1,8 = 25.0, p = 0.001) and at Port Credit (F3,8 = 7.3, p = 0.011). The proportion of sculpins in the diet was significantly (p < 0.001) higher at Port Credit than at Niagara on the Lake that was similar to the proportion for Cobourg and Oswego.
Compared to 1998–1999, just after gobies were first observed in Lake Ontario, the diet of Lake Ontario Lake Trout in 2005 was similarly dominated by alewives although this was reduced considerably at Port Credit, the site having the highest proportion of gobies in Lake Trout diets based on the three size categories (Table 3: available at www.aehms.org/Journal/12_3_Fitzsimons_Appendix.pdf). Sculpins comprised a variable but low proportion of the diet throughout Lake Ontario in 2005 when the proportion in Lake Trout diets was, in general, reduced from 1998–1999.
The predicted thiaminase activity in the diet varied by almost two-fold in 2005 and bracketed that predicted for 1998–1999 (Table 3: available at www.aehms.org/Journal/12_3_Fitzsimons_Appendix.pdf). The lowest thiaminase activity in the diet during 2005 was predicted for Lake Trout from Port Credit that had the lowest proportion of alewife in their diet. Further reductions in the thiaminase content of the diet would be predicted if gobies comprised a higher proportion of the diet as shown for future scenarios 1–3 in Table 3: available at www.aehms.org/Journal/12_3_Fitzsimons_Appendix.pdf. During 2005, estimated egg thiamine varied by over three-fold with the highest levels predicted for Lake Trout from Port Credit, the site having the lowest proportion of alewife in Lake Trout diets. For the other three sites, predicted Lake Trout egg thiamine was lower than that predicted for 1998–1999 reflecting the higher proportion of smelt and alewives. With a greater proportion of gobies in the diet, as shown for future scenarios 1–3, a higher egg thiamine level would be expected. With variation in predicted egg thiamine levels, there was variation in predicted EMS, with almost no EMS predicted when the proportion of gobies in the diet was > 25%, as for Port Credit and the three future diet scenarios. In contrast, the amount of EMS predicted for the three sites in 2005 that had less than 25% gobies, was on average higher than for 1998–1999 and may reflect as above, the greater proportion of smelt and alewives in the diet. Variation in predicted amounts of EMS resulting from past, present or future diets, resulted in corresponding changes in embryo density on the reef relative to initial egg density (Figure 2).
Combined effect of goby predation and the EMS and negative growth effects of the thiamine deficiency
On the whole, the combined effects of an average EMS occurrence of 70% and goby predation (Figure 3) compared to the effects of goby predation alone for the shallow-water temperature regime (Figures 4, 5: available at www.aehms.org/Journal/12_3_Fitzsimons_Appendix.pdf), indicate that the EMS effects made the difference between some and no embryonic survival to the end of emergence but only under certain combinations in the density of eggs and gobies. At a goby density of nine gobies m−2, there was virtually no survival to emergence at an egg density of 1000 eggs m−2 when EMS was present compared to some survival (i.e. 30%) when EMS was absent. Similarly, at a goby density of 35 gobies m−2, there was virtually no survival to emergence at an egg density of 5000 eggs m−2 when EMS was present compared to appreciable survival (i.e. 72%), when EMS was absent. Of course, if all eggs were lost to predation by the time of emergence, as was the case for low egg density independent of goby density, or with high goby density and higher egg density, the presence of EMS had no effect on the number of embryos that survived to the end of emergence.
In contrast to the modest effect of EMS, the effect of thiamine deficiency related growth retardation that resulted in 100% mortality (Figure 6: available at www.aehms.org/Journal/12_3_Fitzsimons_Appendix.pdf), was dramatic: there was no survival past the day of emergence, regardless of egg or goby density.
Comparison of the effects of goby predation at shallow and deepwater temperature regimes
Although the pattern of temperature varied between the shallow water reef (SWR) and deep water reef (DWR) (Figure 1: available at www.aehms.org/Journal/12_3_Fitzsimons_Appendix.pdf), temperature had little effect on the number of embryos surviving to the end of the emergence period (Figures 4, 5: available at www.aehms.org/Journal/12_3_Fitzsimons_Appendix.pdf). Predicted differences became more evident with increasing goby density particularly at lower egg densities. At 18 gobies m−2, 100% mortality (FM) at an egg density of 500 eggs m−2 was reached on December 18 at SWR compared to 68 days later on February 24 at DWR. At a goby density of 35 gobies m−2, FM was reached on November 5 at SWR compared to 43 days later on December 17, at DWR for this same egg density. For this same density of gobies, at 1000 eggs m−2 FM was reached on December 21 at SWR compared to 69 days later on February 28 at DWR. Above a density of 35 gobies m−2, the effect of goby predation was so great that any advantage of the deepwater temperature regime was lost and FM was reached before the end of the 213 day incubation period.
Effect of the period of goby residence on the reef
The effect of a restricted period of residence of gobies on the spawning reef compared to residence for the entire period of embryonic development, showed interactions with both egg and goby abundance as well as the period of goby residence (Figure 7, Figure 8). At an egg density of 100 eggs m−2, there was no embryo survival to the end of emergence, regardless of the period of goby residence on the reef or the density of gobies. Increasing egg density to 500 eggs m−2 at goby densities of either 9 or 18 gobies m−2, resulted in some embryo survival to the end of the emergence period for a one month period of residence compared to none for over winter residence. After two months at this same egg density, no embryos survived to the end of the incubation period at a goby density of 18 gobies m−2. At a goby density of 35 gobies m−2, for 500 eggs m−2, there was no embryo survival regardless of the period of goby residence. Increasing egg density to 1000 eggs m−2 resulted in some survival for a one month period of goby residence although there was none for longer periods of residence. Increasing egg density further to 5000 eggs m−2, resulted in some embryo survival to the end of emergence up to a goby density of 35 gobies m−2 regardless of the period of goby residence on the reef, although at either 75 or 150 gobies m−2, this was only possible when the period of goby residence was less than or equal to two months.
Assessment of the effect of the goby invasion at Port Weller on predicted and observed Lake Trout eggs and fry survival
At Port Weller, goby abundance increased rapidly from being undetectable at this site in 1999, to 26 ± 7 (mean ± SEM) gobies m−2 in April of 2002 increasing significantly to 109 ± 15 gobies m−2 by December of that same year with a further although non-significant increase by December of 2003 to 142 ± 7 gobies m−2 (Figure 9).
The size of gobies over the period from spring 2002 to fall of 2003, showed significant temporal variability both on the basis of length (F2,288 = 25.8, p < 0.001) and weight (F2,288 = 15.3, p < 001). Gobies were significantly longer and heavier in the fall of 2002 (83.6 ± 2.5 mm; 10.6 ± 1.1 g) compared to the fall of 2003 (65.7 ± 1.1 mm; 4.7 ± 0.3 g) although in neither fall was the length or weight of gobies significantly different from those collected during the spring of 2002 (73.7 ± 2.9 mm; 5.8 ± 0.6 g).
Lake Trout emergence was the first to be affected by increasing goby abundance at Port Weller, being virtually non existent in the spring of 2002, 2003, and 2004 after gobies invaded the site, and the lowest abundance of emergent fry observed during the entire nine-year long time series (Figure 10). These observations were consistent with the model prediction of no survival to the onset of emergence for the observed densities of Lake Trout eggs and gobies. At a goby density of 75 gobies m−2, less than the average goby density observed at Port Weller during the fall of 2002, no embryo survival was predicted by the model for an egg density of 5000 eggs m−2 which was well above the observed egg density of 3141 ± 762 eggs m−2 (Figure 9) observed during the fall of 2002 if gobies are present for the entire incubation period. Similarly in 2003, when egg density declined to 1434 ± 390 eggs m−2 and there was a further albeit non-significant increase in goby density, the finding of no emergence of Lake Trout fry at Port Weller was consistent with model predictions.
Density of Lake Trout eggs (Figure 9) at Port Weller in December of 2002, when goby density was appreciable (109 ± 15 gobies m−2), was over 30% higher than expected based on the relationship developed between spawner CPUE of female and egg density (Figure 11: available at www.aehms.org/Journal/12_3_Fitzsimons_Appendix.pdf). Consequently, gobies did not appear to exert a significant effect on egg density at least, in 2002. However, by 2003, egg density was well below that predicted from spawner CPUE by almost 50%, and does show an impact from gobies (Figure 10).
It is evident, based on our analysis that the replacement of alewives by gobies in the diet of Lake Ontario Lake Trout, even as a relatively small proportion (e.g. 25%), will result in significantly lower thiaminase activity. Such a proportion is predicted to increase thiamine levels to 3.6 nmol·g−1, above the threshold for both EMS (2.63 nmol gm−1; Fitzsimons et al., 2007a), and for growth retardation (3 nmol g−1, unpublished data). Because of the relatively simple (i.e. few species) prey fish community of Lake Ontario, the abundant goby that have low levels of thiaminase activity, may provide the greatest opportunity for relief from the effect of the thiamine deficiency for Lake Trout. Prior to the expansion of gobies in Lake Ontario and their occurrence in the diet of Lake Trout, Lake Trout diets consisted of only one thiaminase-free species, the sculpin, but this comprised a relatively small proportion of the diet (i.e.14%) (Lantry, 2001). Slimy sculpin abundance has been declining (Owens et al., 2003) due to the combined effects of juvenile Lake Trout predation (Owens and Bergstedt, 1994), reduced lake productivity (Millard et al., 1996; Johannsson et al., 1998), and disappearance of Diporeia, its major diet item (Dermott, 2001; Owens and Weber, 1995). Other historic Lake Trout prey items that are low in thiaminase (lake herring Coregonus artedi and deepwater sculpin) (Tillitt et al., 2005) are also at low levels of abundance in the lake (Owens et al., 2003; Fitzsimons and O'Gorman, 2006).
Our predictions for Lake Ontario appear to be reflected in Lake Huron where the alewife stock recently crashed (Riley et al., 2007). In this lake, some Lake Trout stocks are now feeding heavily on gobies and these groups have higher egg thiamine concentrations which are consistent with lower thiaminase activity in their prey, than those feeding on alewives (J. Johnson, MDNR, pers. comm.; D. Honeyfield, USGS, pers. comm.). Nevertheless, the thiaminase activity we used in our modeling was only based on goby from one site in Lake Michigan (Tillitt et al., 2005) and since intra-specific population variability of thiaminase activity up to 20-fold has been documented for alewives, (Fitzsimons et al., 2005, unpublished data), there is a need to determine the thiaminase activity of Lake Ontario gobies to confirm our predictions.
Despite the fact that goby incorporation into Lake Trout diets may provide some relief from thiamine deficiency, the predicted effects of predation far exceed any positive effect on egg viability. Low egg densities (< 100 m−2) have been reported for several Lake Trout spawning reefs in lakes Ontario, Erie, and Michigan (Fitzsimons, 1995b; Fitzsimons and Williston, 2000; Jonas et al., 2005). As goby density increases, their predation effects are expected to result in the total loss of recruitment up to and exceeding the highest egg densities observed in the Great Lakes (e.g. 1000–5000 eggs m−2), such as for spawning reefs in lakes Huron, Ontario, and Superior (Peck, 1986; Jonas et al., 2005; Perkins and Krueger, 1995; Fitzsimons et al., 2003). For Port Weller, such high densities (e.g. 5000 eggs m−2) were apparently only achievable under conditions of high spawner CPUE but high spawner abundance has only rarely been obtained in the Great Lakes (Selgeby et al., 1995). Consequently Lake Trout egg deposition targets of 500 eggs m−2 and the implied spawner CPUE associated with this target that have been set by fisheries management agencies for lakes Michigan (Bronte et al., 2007) and Erie (Markham et al., 2007), seem too low to compensate for the expected effects of goby predation on reproduction.
Although there is considerable uncertainty about the factors regulating goby density on Lake Trout spawning reefs in the Great Lakes, our observation for Port Weller and the predictions of our model suggest that extreme goby density (i.e. 150 gobies m−2) will be catastrophic for Lake Trout recruitment. It is noteworthy at Port Weller that fry emergence was in decline owing to factors other than egg deposition, prior to colonization of gobies suggesting that other factors may have contributed to the failure of emergence post goby colonization. In Lake Michigan, goby densities remain considerably lower (< 20·m2) suggesting predation effects on Lake Trout there may be less. It is noteworthy, however, that the highest goby density reported for a location in the Great Lakes (133·m2; Chotkowski and Marsden, 1999) was for southern Lake Michigan. Claramunt et al. (2005) found that abundance of interstitial egg predators including gobies, on a shallow Lake Trout spawning reef in northern Lake Michigan, was directly related to water depth. For gobies this may have been related to the observed increase in abundance with depth of dreissenids, the preferred food of gobies (Jude et al., 1995).
A major uncertainty in the effects of goby predation is their period of temporal overlap with Lake Trout eggs and fry. As we have shown here, reducing the period of residence from the entire Lake Trout embryo incubation period to a period of one or two months significantly increased embryonic survival. In their native range, gobies reportedly move to greater depths up to 60 m. to over winter (Miller 1986). Our direct observations at Port Weller suggest goby are resident in December and April, when temperatures are as low as 4°C, but this does not confirm over winter residence. We can only infer this from a near absence of emergence during the spring of 2003 and 2004, that for the predicted egg density (e.g. 2000–3000 eggs m−2) and measured goby density (100–150 gobies m−2) would only have been possible according to our model, if gobies were present for much of the winter. Bottom trawling conducted in the immediate area of Port Weller indicated that large numbers of gobies move from inshore spawning habitat in autumn to over winter in the profundal zone and remain there through the early spring (M. Walsh, USGS Lake Ontario Biological Station, unpublished data). Further, Lee and Johnson (2005) and Fitzsimons et al., 2006 present data to suggest very low feeding rates at temperatures < 4°C. Clearly, work is needed to confirm if, and what proportion of the goby population remain on shallow water reefs throughout the winter. Additional winter work could verify if these gobies are feeding on Lake Trout eggs.
For deep water reefs, the overall abundance of gobies may be lower because of temperature effects on goby reproduction and this may have a greater overall impact on losses of eggs and fry than the direct effects of the temperature regime on predation rates. Differences in temperature between a shallow-water and a deepwater reef resulted in relatively little variation in the egg loss trajectories for a given level of goby abundance. However, the lower temperature threshold for reproduction by gobies (9°C) (MacInnis and Corkum, 2006) is approximately 1°C above the maximum temperature observed at the deepwater reef in 1982–1983. Consequently its unlikely gobies could reproduce at a deepwater reef such that their abundance there would be under the control of reproduction at and colonization from warmer and presumably much shallower and more distant nearshore reefs, where they reproduce (MacInnis and Corkum, 2006), and this would tend to reduce abundance at deepwater reefs considerably.
Our work represents the first time in the Great Lakes that Lake Trout spawner abundance has been related to egg abundance on a reef and reiterates the need by fisheries management agencies to increase spawner abundance to overcome losses to predation (Jonas et al., 2005) and encourage restoration. There is a greater likelihood of embryo survival at high egg abundance than at low egg abundance. It is unlikely, however, that the relationship between spawner and egg abundance generated for Port Weller would apply to other reefs. Port Weller has amongst the lowest wind fetch of spawning reefs documented in the Great Lakes and loss of eggs on spawning reefs has been directly related to wind fetch presumably through the effects of physical disturbance (Fitzsimons, 1995b; Fitzsimons et al., 2007b). Many of these eggs are likely lost immediately after spawning (Claramunt et al., 2005) before they have settled into the interstices. As a result, more exposed Lake Trout spawning reefs are expected to require much higher spawner abundance to achieve egg densities similar to Port Weller.
In conclusion, an increasing proportion of gobies in the diet of Lake Trout could provide significant relief from thiamine deficiency, even when gobies comprise only a small proportion (e.g. 25%) of the diet. The greatest effect is on relief from the negative growth effects of the thiamine deficiency, as opposed to the direct mortality imposed by EMS. Any positive effect of gobies on thiamine levels and ultimately Lake Trout recruitment, however, pales in comparison to the negative effects predicted from predation that are significant at low levels of goby abundance when egg abundance is low, and even at extraordinary levels of egg abundance when goby abundance is high. The predation effect of gobies will be highly dependant on their period of over winter residence on a Lake Trout spawning reef. When gobies only remain on spawning reefs for a one to two month period, predation effects are dramatically reduced and in such instances, gobies may provide a net reproductive benefit to Lake Trout.