Early life stages of Dreissena polymorpha ( zebra mussel ) : the importance of long-term datasets in invasion ecology

The early life stages of Dreissena polymorpha were studied in Lough Key, Co. Roscommon, Ireland during the reproductive season, 1998-2003. This involved weekly sampling of larval/veliger density, size distributions and settlement density. Low larval and adult densities of Dreissena in 1998 indicated that this was the first year of significant reproduction in Lough Key. Variation existed in seasonal larval densities, larval size distributions and juvenile settlement patterns among sampling weeks, years and monitoring sites from 1998 to 2003. In the early years of invasion (1998-2000) annual variations were observed in larval density and juvenile settlement. Increasing levels of larval density and settlement from 1998 to 2000 were typical for the early exponential growth phase of Dreissena invasions. The high level of successful recruitment in those years was evident from the exponential increase of the adult zebra mussel population. In subsequent years, these differences were likely related to environmental variations in summer water temperatures and food availability for veligers. In 2001, settlement estimates were extremely low relative to the larval densities present. Prolonged warm water temperatures in summer 2003, resulted in a long reproductive season and also high settlement rates. The variation between annual datasets observed in this study, implies that long term studies of the early life stages of Dreissena polymorpha is required to monitor the dynamics of this species within lake ecosystems.


Introduction
Zebra mussels are thought to have colonised Ireland about 1993/1994, arriving as fouling on the hulls of leisure craft from Britain (Minchin et al. 2005, Pollux et al. 2003).Colonisation of Lough Key is believed to have taken place in 1996 (Lucy 2005).Zebra mussel surveys have taken place in Lough Key, since 1998, when adult zebra mussels were initially found.
It is a fundamental ecological principle that both successful reproduction and the survival of sufficient early life stages to maturity are necessary for the success of any species' population.
Dreissena polymorpha shares characteristics of many successful invasive species: rapid growth, prolific reproduction at an early age (0-1+) and a relatively short life span.This usually provides the potential for rapid exponential growth of zebra mussel populations within a waterbody (Ehrlich 1989, McMahon 1992, Nichols 1996).

F . L u c y
Researching the dynamics of early Dreissena life stages involved sampling larval/veliger densities, measuring larval size distributions and estimating settlement of juveniles (plantigrade stage).This was carried out annually in Lough Key from 1998 to 2003.The aim was to provide information on larval development within the plankton and also survival to a settled juvenile stage throughout the sampling seasons.Annual datasets were interrogated to determine the varying dynamics of early life stages in Lough Key over a six year period (1998)(1999)(2000)(2001)(2002)(2003).
The commencement of research in 1998, when populations were still in their early invasive stage, provided an opportunity to monitor the exponential growth of zebra mussels in Lough Key at both larval and adult stages in the lake population.Continued studies until 2003 also provided data on the annual status of Dreissena polymorpha in Lough Key.

Site description
These studies have involved research on veliger and juvenile zebra mussels (density and size), adult zebra mussels (density and biomass) and also on the nutrient status of the lake (Lucy and Sullivan 2001, Sullivan et al. 2003, Lucy 2005, Lucy et al. 2005).Lough Key (9km 2 ) is located in the upper Shannon River catchment (Graczyk et al. 2004) at an altitude of 45m O.D., and is relatively square in shape with a maximum length of 4.2km and a maximum width of 4km (Figure 1).It has an estimated volume of 45.97 m 3 x 106 and a maximum depth of 23.5m (Toner 1979).The lake is relatively shallow with a mean depth of 4.5m (Lucy et al. 2005).There are 13 islands on the lake.Most near shore areas of the lake have a stoney substrate.

Sampling methods
Sampling of early zebra mussel life stages (larvae/veligers and settled juveniles) has been largely adapted from a monitoring programme designed by Marsden (1992).The scheduled sampling programme was designed to maximize sampling during the summer season.Weekly sampling was generally implemented from July to September.Weeks for summer months are assigned numbers, weeks 1-4 for each month, in the format July week 1.During 2001-2003 occasional sampling was also carried out earlier and later in the year.
A temperature datalogger (Vemco minilog V3.04) was deployed at Site D to record temperatures every four hours between 2001 and 2003.Temperature was also recorded with an alcohol thermometer at Site A and Site D at each monitoring (1998)(1999)(2000)(2001)(2002)(2003).
Three-metre vertical tows were carried out at four monitoring sites (A-D) in 1998 and 1999 and at five sites (A-E) from 2000-2003 (Figure 1).A 64 µm mesh plankton net with a 30cm diameter opening was used for veliger sampling, allowing collection of all veligers >70 µm.
For the monitoring programme each sample was examined under a high magnification (80-100x) (Olympus CX-41), using cross polarized light (Johnson 1995) using one-ml sub-samples in a Sedwick-Rafter Counting cell.Three one-ml subsamples were counted and the veliger density per ml of concentrated sample calculated from the mean as below: Number of veligers/ml x 25 a Volume of Sample b a = Volume in 25ml tube b = 3 x 3.14 x (0.3) 2 = length of tow x  π x radius of net mouth 2 = 848L The microscope was fitted with an optical micrometer and size distributions of veligers from Sites B, C and E were measured to the nearest 10 µm.Veliger density and size distribu-tions were then analysed (Ni Chonmhara 1999, Commons 2000, Duggan 2001, Marshall 2002, O'Mahony 2003, Lohan 2004).A mean density value was obtained for each sample week from the different lake sites.
Settled juveniles were sampled using 15x15 cm grey PVC plates with three plates deployed in series at each site as in Lucy and Sullivan (2001).The rope was tied at the top to a navigational buoy at monitoring sites A, B, C and E and held in position from a jetty at Site D. The settlement plates were suspended from a concrete anchor weight, which held the plates firmly in mid-water (3m depth approx).All plates were conditioned for one week to allow a bio-film to build up, thus encouraging settlement.On each sampling date beginning early/mid June either the bottom or the middle plate was changed on a two-week rotation.Once recovered, the plates were placed in a sectioned sample container and then examined using a microscope for the presence of newly settled juveniles.These were clearly visible on the surface of the plates.The mean of 30 x 1 cm 2 quadrats was obtained for each plate and an estimated settlement density/m 2 was extrapolated from this data.A new settlement plate was replaced each time a plate was recovered.

Statistical analysis
Mean veliger densities were calculated from data for Sites A-D (1998-2003).It was considered more representative to take mean values for all sites rather than site-specific ones, due to the known patchy distribution of larvae (Lucy 2005).
Size distribution results were compared statistically using descriptive statistics.As datasets were not normally distributed and the variances were heterogeneous (Levene statistic, p<0.05),Kruskal-Wallis and post-hoc Mann-Whitney test tests (p<0.05) were applied to compare sample distributions.
Statistical analysis determined that the size distributions of veligers were not significantly different (Mann-Whitney test, P>0.05) at sampling sites within sampling weeks.Weekly datasets from different sites were therefore pooled for analysis to give a larger sample size for each week (>250).Samples taken in consecutive weeks were also analysed to determine whether they varied with regard to size distribution (P<0.05).Where significant differences were found, post-hoc tests were carried out to determine which paired samples were significantly different (P<0.05).The percentage of veligers greater than 200µm/sample was calculated for each sample week (1998)(1999)(2000)(2001)(2002)(2003) to determine the proportion of veligers close to settlement.
Settlement analysis was based on cumulative means for Sites A to D. Correlation analysis (Pearson's correlation coeffecient, significance level P<0.05) was used to compare the following: • Veliger density (week n) vs settlement (following week, n+1) • Percentage of veligers > 200µm (week n) vs settlement (week n+1) The statistical package SPSS 11 was used to carry out the various statistical analyses (Howit and Cramer 2003).

Larval Densities in Lough Key
In terms of seasonal appearance, zebra mussel larvae were detected in the plankton as early as February 2000 (7.2 o C) and March, 1999 (11 o C) These were probably overwintering larvae as observed by Kirpichenko (1971).The first definite seasonal appearance of larvae occurred in May corresponding with site water temperatures greater than 12.5 o C. Water temperature however, was generally in excess of 15 o C during the reproductive season.Larval densities from 1998-2003 are shown with associated water temperatures (Figure 2).The annual start of significant spawning in the lake was early July, with larvae appearing in samples until the beginning to the middle of October.Peak annual spawning was typically from June week 4 to August week 4.
Larval densities (veligers/L) showed seasonal variation between years.In 1998 spawning was not detected at any site until July week 4.In 1999 and 2000 the spawning season was longer, particularly in 2000 when significant densities were recorded from June week 4.The spawning season was particularly long in 2003, with larvae detected from the second week in June and with significant numbers of larvae still present in the plankton in mid September 2003.Individual recorded peak densities increased yearly, from 1998 to 2003, with the exception of a decrease in 2001 (Figure 3).
There appeared to be no relationship between water temperatures and larval densities in 1998 or 1999.In 2000 the highest larval density occurred the week following peak water temperatures.The maximum larval density coincided with peak temperatures in both 2001 and 2003 and appeared to be also related to temperature rises in 2002 (Figure 2).At each site larval numbers decreased during autumn period, at the end of the spawning period and in October were detected only in low numbers.

Larval Size Distributions
Statistical analysis determined that there was often a significant difference between size distributions in consecutive weeks in 2002 and  2003 (Table 1).This was rarely observed in 2001.Zebra mussel larval size distribution in 2003 was detailed (Figure 4).Small D veligers (70-80µm) were not well represented in the majority of samples.In most sampling weeks, veligers >260µm were not generally present in the plankton.Larger individuals were occasionally recovered, including one at 840µm sampled in September 2002.The percentage of veligers >200µm (Sites A and B combined) is given in Table 2.In general, the proportion of larger veligers increased from July week 4 to August week 4, with an elevated percentage also occurring during the first two weeks of September in 1998 and 2002.

Settlement
Cumulative settlement during the spawning season (1998)(1999)(2000)(2001)(2002)(2003) was estimated (Figure 5).Annually, the highest settlement occurred during the entire month of August (weeks 1 to 4).The low settlement in 1998 was in the early invasive stages of Dreissena colonisation in Lough Key (Lucy and Sullivan 2001).The high settlement in 1999 and 2000 coincided with the exponential increase in the lake's dreissenid population (Lucy et al. 2005a, b).Overall settlement was very poor

Veliger Density
The length of the spawning season (June-September) was typical for zebra mussel infested lakes in FSU, Europe and North America (Sprung 1989).The presence of occasional large postveligers outside the normal spawning season may have resulted either from prolonged duration in the plankton, with delayed development due to low temperature (Lewandowski 1982) or more likely from resuspension from the substrate as reported by Martel (1993).The latter could easily occur following stormy weather incidents, where settled juveniles could get washed off stones in the littoral zone, becoming resuspended in the plankton.
Veliger densities in this study fall within the wide range reported in the literature (Kirpichenko 1964, Lewandowski 1982b, Haag and Garton 1992, Nalepa et al. 1995).Results from Lough Key indicated that peak densities were found at three separate sites in summers 2000-2004.Overall results from 1998 to 2000 seem to indicate that veliger densities and length of spawning season are influenced by three main factors; parent population size; maturity of parent population and temperature of lake water.It is interesting to note that even though the temperature range is small in terms of other international studies carried out in continental climates, it still appears to have an impact on spawning patterns.
Nineteen ninety eight was considered the first year of significant spawning in Lough Key (Lucy and Sullivan 2001).Overall densities were very low in that year, as could be expected from a small parent population comprised of mainly 1+ individuals.Figure 7 shows peak veliger density versus percentage adult zebra mussel cover on stone indicating the exponential increase in the  The early peak in 2000 (June week 4), probably reflected the high water temperature present at that time (19 o C).This temperature is relatively unusual at that early stage of the summer.The early peak also indicated that mature zebra mussels were present.By 2000 there was a well-established cohort of 2+ zebra mussels and these would certainly have been ready to reproduce in addition to early maturing 1+ individuals.Peak larval density that year (July week 4) succeeded the second week of peak temperature in that year.Both these events indicate how important temperature is as a trigger to mass spawning.Temperature effects may be considered direct, with growth of adult mussels occurring when temperatures reach 10-12 o C (Mackie 1991, Jantz and Neumann 1992, Karatayev et al. 2006), or indirect due to increased availability of food (Mantecca et al. 2003).Larval peaks in 2002 and 2003 were closely related to maximal water temperatures.In 2003 the continued high water temperatures in late summer resulted in increased densities in early September, with larvae well represented until the end of sampling in September week 3.
There may be several reasons why the spawning cycle is prolonged over the summer and early autumn.Variation in spawning patterns has been reported in a number of comparative studies (Gist et al. 1997).a) In spawning, each zebra mussel in principle is able to release gametes over a period of six to eight weeks (Walz 1973, Borcherding 1991).So the same individuals may be spawning sequentially, which could result in bimodal patterns in size distribution.b) Separate age cohorts of zebra mussels may develop at different rates and smaller (1+) individuals may spawn later in the season than older (2+) zebra mussels.One-year-old zebra mussels are known to spawn in Irish waters (Juhel et al. 2003) c) Similar age cohorts may develop at different rates and spawn over the entire summer season.Smaller 1+ individuals become mature and spawn later in the season.In one study Borcherding (1991) observed this phenomenon in zebra mussels and similar results have also been derived for marine species.

Veliger Size Distribution
Seasonal size distribution patterns actually provide more useful information than larval densities, because it is possible to trace larval size patterns through to settlement stage over the summer period.This provides a bridge in information between the veliger/larval and settlement stages.High proportions of small veligers (less than 110µm) were present early in each season with corresponding low numbers of larger veligers, corresponding with the start of the spawning season.In both 2002 and 2003 mean size decreased in August week four due to a pulse in spawning increasing the proportion of small veligers (less than 110µm).This was particularly apparent in 2003, where the mean size was only 96.6µm (Figure 4).The week of highest larval density was two weeks earlier in August week 2 of 2003, which recorded the greatest size range of veligers (80-290µm).This suggests that the highest densities sampled represent the maximal spread in larval development resulting from continual spawning rather than from peaks achieved from single spawning incidents.Seasonal reduction in peak densities represent both increased rate of settlement (as in 2003) and a reduction in overall spawning.
As statistical analysis showed, there was often variation in size distributions between different sequential weeks during the sampling season, particularly from August onwards.This would appear to indicate that veligers were growing and settling, with more appearing at the smaller cohorts of the range, following new spawning incidents.The increase in the proportion of larger veligers as the season progressed indicated the development of veligers from the D-shaped stage to larger pediveligers.
The majority of veligers measured in samples over the three-year period were between 100 and 150 µm (mainly umbonate).This may relate to early settlement, dispersal or is more likely due to high mortality rates prior to metamorphosis into the juvenile (Sprung 1989).Stoeckel et al. (2004) found highest larval mortality during the transition from D stage to umbonal stage, supporting the suggestion of a developmental bottleneck as found in previous field studies (Schneider et al. 2003).It may also relate to variation in size of same-aged larvae, which have been found to differ by as much as 120μm (Stoeckel et al. 2004).
The presence of significant numbers of larvae in late September 2003 was a new recording event for Lough Key and a definite new peak in veliger density at that time was probably related to the high water temperatures during the summer.This increase could easily be explained by factors (a) or (c) above.There is no doubt that the spawning season was prolonged in 2003.Theoretically, if temperatures were maintained above 15 o C, spawning would continue indefinitely as more zebra mussels matured and began to reproduce.Warmer Irish summers may become a future consequence of global warming, impacting on the reproductive cycle of Dreissena.
The strong correlation between seasonal veliger density and the percentage of veligers greater than 200µm in 2003 also showed that the higher densities reflected the presence of a large size range of veligers, some as great as 290µm.The correlation could also indicate increased survival of larvae or shorter development time due to increased water temperatures.Larvae appearing in samples at the beginning to the middle of October of sampling years were usually larger (>200um) indicating that the spawning season was over for the year.Data from settlement plates indicate that postveligers in Lough Key settled from 200µm upwards.Veligers larger that 200µm were, however, well represented in samples, indicating that settlement takes place generally between 200 and 260µm.

Settlement
Settlement rates reflect the survival of veligers through the postveliger stage to the settled juvenile.Following mortality at the trochophore to D-stage (Sprung, 1989), and from the D stage to the umbonal stage (Stoeckel et al. 2004), studies indicate that larval mortality can also occur late in the cycle during metamorphosis and settlement (Stanczykowska 1977, Lewandowski 1982a).Given that the size range of settled juveniles on plates varied from 200-400µm, it is probable that the low proportion of larger veligers in the plankton reflected a combination of both early settlement and mortality.Mortality (range 20-100%) could be due to a number of factors such as water turbulences (Rehmann et al. 2003), predation by fish (Molloy et al. 1997), filtering by copepods or by adult Dreissena (MacIsaac et al. 1991), bacterial infection, egg quality or starvation (Sprung 1989).
Settlement plate data did provide some indication of overall survival of veligers to the settlement stage during each sampling season (1998)(1999)(2000)(2001)(2002)(2003).Size distribution analysis of adults from stoney substrate in the years consecutive to settlement also provided information on settlement rates and survival to the 1+ stage (Lucy 2005).
It is not possible to determine exactly how long veligers remain in the plankton due to a number of reasons:

•
Spawning is ongoing throughout the summer, providing a number of cohorts.Development rate is also known to differ within cohorts (Sprung 1989).This results in size overlap, which complicates exact estimation of cohort settlement.

•
The data in this study suggest that postveligers settle at different sizes as recorded in the literature (Ackerman et al. 1994, Nichols 1996).

•
There were a low number of veligers greater than 200µm in most larval samples and also a high variation in the sample proportion of these 'close-to-settlement' veligers.

•
Veligers were already present in the plankton when seasonal sampling began each summer.It was not possible therefore to gauge time of first seasonal settlement.
The datasets and correlation analysis (settlement vs veliger density; settlement vs veligers greater than 200µm) did, however, definitely indicate that a development cycle exists.Most veligers were greater than or equal to 100µm in weeks of peak density and would be well into their first week of life, according to growth rate of 4µm/day from a trochophore larva (70µm) (Sprung 1989, Kern et al. 1994).By combining this information with the size distribution analysis in Lough Key, a two to three-week development time for 2001 and 2002, and a one to two-week development time for 2003 can be estimated.The amount of time required for larval development varies inversely with temperature (Nichols 1996).Higher water temperatures may therefore have resulted in a shorter development time for 2003 as noted in another study by Borcherding (1991).
The high settlement rate in 2003 may have been due to a decrease in development time related to the high water temperatures (degree-day related), which lead to an increased survival rate.Faster development would decrease the opportunities for predation, as larvae would be present in the water column for a shorter length of time.The higher temperatures in 2003 also resulted in an increase in productivity (food availability) which would also give higher settlement.The two to four-week development time concurs with Sprung (1989) but estimates vary within the literature with some as high as five weeks (Walz 1973).Some European studies show zebra mussel larvae present in the plankton for not more than ten days (Kachanova 1961, Hilbricht-Ilkowska and Stanczkowska 1969, Lvova 1977and 1980).
Recruitment on settlement plates in 2002 was poorer than in any other year.Zero + individuals were nevertheless well represented as the first age cohort in adult samples taken in early 2003 (Lucy 2005).This highlights the fact that only a very small proportion of zebra mussels (2 settled juveniles out of 10,000s of veligers produced) need to survive to juvenile stage to maintain their numbers in the lake.These inconsistencies in the abundance of developing life stages and in successful recruitment are consistent with other studies (Haag and Garton 1992, Doka 1994, Nalepa 1995).
In the early years of invasion (1998)(1999)(2000), during the exponential growth phase, extensive settlement was noted on aquatic plants, on buoy chains to a depth of 7.5m and also on the ropes used for attaching settlement plates.While settle-ment can be defined as the transfer from the pelagic phase to the benthic, recruitment shows settlement combined with post-settlement mortality.The increase in adult zebra mussel densities in Lough Key in 1999 and 2000 indicates that recruitment was high in 1998 and 1999.Even though veliger densities continued to increase in 2001 and 2003, the phenomenon of widespread settlement on various substrates noted above was not noted from 2001 onwards, suggesting lower overall recruitment.It would seem likely that the overall observed reduction in settlement was due to lack of food resource for larvae in the plankton as they depend on food particles (phytoplankton, Cyanobacteria, bacteria and detritus) of between 1-4µm and may starve when phytoplankton is dominated by larger or smaller algal species (Sprung 1989).

Conclusions
Early life stage monitoring is often used to detect the presence of zebra mussels in waterways and in such instances is used as a short term measure.This Lough Key study shows the variation in larval density, size distributions and settlement over a six year period from the early invasive stages through the exponential rise in the zebra mussel population.Future changes may occur due to lake ecosystem factors or due to external environmental pressures, e.g.global warming.With the increasing development of computerized ecosystem modeling, it is important that long term variation be taken into account rather than using single annual datasets.This variation can only be measured by long term monitoring as exemplified by this research.

Figure 1 .
Figure 1.Map of Lough Key

F
and 2001.In 1998 the late start of the spawning season may relate to the reproductive development of the 1+ parent generation, many of which may have taken until August to reach sexual maturity.