Skistodiaptomus pallidus ( Copepoda : Diaptomidae ) establishment in New Zealand natural lakes , and its effects on zooplankton community composition

The North American calanoid copepod Skistodiaptomus pallidus is an emerging invader globally, with non-indigenous populations recorded from constructed waters in New Zealand, Germany and Mexico since 2000. We examined the effects of S. pallidus establishment on the zooplankton community of a natural lake, Lake Kereta, where it was first recorded in late-2008, coincident with releases of domestically cultured grass carp (Ctenopharyngodon idella) . Although not present in any of our samples prior to August 2008, S. pallidus was found in all samples collected in the subsequent five years. ANOSIM indicated zooplankton community composition significantly differed between samples collected before and after S. pallidus invasion, whether the invader was included in the analysis or not. Zooplankton species affected most greatly were the copepods Calamoecia lucasi and Mesocyclops sp., which decreased in their relative importance, and the cladocerans Bosmina meridionalis and Daphnia galeata, which increased. Rotifer species were relatively unaffected. As the length of grass carp released were >6.5 cm, direct predatory effects by this species on the zooplankton community are unlikely. Associated reductions in macrophyte biomass could explain increases in the relative abundances of planktonic cladocerans (B. meridionalis and D. galeata). However, the effect of macrophyte reduction by grass carp on zooplankton communities is considered to be limited elsewhere, while the reduced macrophyte biomass cannot explain the decrease in relative abundance of the native planktonic calanoid copepod C. lucasi. Competition between C. lucasi and S. pallidus is the most compelling explanation for the reduction in importance of the native calanoid copepod species. Skistodiaptomus pallidus appears to have undergone a “boom-and-bust” cycle in Lake Kereta, increasing in relative abundance in the first three years following establishment, before declining in importance.


Introduction
A common pattern observed in aquatic invasions is the relative ease of establishment in constructed waters (e.g., retired quarry and mine pits, dams constructed for water supply or electricity generation, and ornamental ponds) compared with natural waters.New water bodies have increased the number, area and spatial distribution of lakes and ponds in many areas globally, and have seemingly aided in the establishment and spread of non-indigenous species (Havel et al. 2005;Johnson et al. 2008;Banks and Duggan 2009;Parkes and Duggan 2012).The African cladoceran Daphnia lumholtzi G.O. Sars, 1885, for example, spread through the United States after first establishing in a Texan reservoir, primarily with dams at the invasion front (Havel et al. 2005).Johnson et al. (2008), similarly, found five other widespread, high profile, invaders in Wisconsin and Michigan, USA (Eurasian watermilfoil, zebra mussel, rusty crayfish, spiny waterflea and rainbow smelt) to have higher rates of occurrence in dams than natural waters.New Zealand has similar examples, with four calanoid copepod species recorded exclusively in constructed water bodies (Banks and Duggan 2009).Parkes and Duggan (2012) found zooplankton community composition to vary between natural and constructed waters in New Zealand; natural waters housed species better adapted to pelagic conditions, with composition largely governed by trophic state, while constructed waters had more varied assemblages and composition related to opportunity for colonisation (i.e., closeness to other lakes and number of water bodies nearby were an important determinant).Reduced biotic resistance from poorly adapted species in newer water bodies may allow for easier establishment of non-indigenous species.Once established in constructed waters, however, these water bodies can provide sites of high propagule supply to other water bodies, allowing greater opportunities for invasion of natural waters through large release sizes and frequencies.As such, species can commonly spread to natural water bodies following invasion of constructed water bodies.For example, D. lumholtzi spread from dams to natural waters in North America, while the calanoid copepod Boeckella triarticulata (Thomson, 1883) first invaded constructed ponds in northern Italy before spreading to other Italian constructed and natural waters (Ferrari 1991;Ferrari and Rossetti 2006;Alfonso and Belmonte 2008).
Here we report on the invasion of the North American diaptomid copepod Skistodiaptomus pallidus (Herrick, 1879) into two natural North Island, New Zealand, lakes, a species previously only recorded there from constructed waters.We also provide the first record for this species in the South Island, from a constructed lake.All of the new distribution records are in lakes where domestically produced grass carp (Ctenopharyngodon idella (Valenciennes, 1844)) have been introduced for aquatic macrophyte control.Skistodiaptomus pallidus is an invader of emerging importance globally, having expanded its geographical distribution in North America (Byron and Saunders 1981), and recently establishing populations in constructed waters in Germany and Mexico (Brandorff 2011;Suárez-Morales and Arroyo-Bustos 2012).In New Zealand, S. pallidus was first recorded in 2000 in ponds at the Auckland Regional Botanic Gardens, but may have been present for some time prior to this (Duggan et al. 2006).Banks and Duggan (2009) expanded these records, finding the species in other New Zealand localities (again, all constructed).As there are few studies documenting the effects of non-indigenous freshwater zooplankton globally, including S. pallidus, we examined the consequences of establishment of this species over a six year period following invasion on the zooplankton community in one natural lake, Lake Kereta.

Methods
Skistodiaptomus pallidus was recorded from a series of samples from Lake Kereta, North Island (36°35'34.66"S,174°16'50.26"E)from late 2008, from a one-off sample from Lake Omapere, North Island (35°20'40.58"S,173°47'21.76"E)from mid-June 2012, and a one-off sample from Lake Hood, South Island (43°58'00.3"S,171°46'16.4"E)on 18 January 2014.Lake Kereta is a eutrophic lake, with a maximum depth of 1.5 m.Zooplankton samples were collected from Lake Kereta briefly in 1997, and then quarterly from September 2001, using vertical hauls from a central lake station with a 40 µm plankton net.Samples were preserved in ethanol (>50% final concentration).While samples have been lost in the intervening period, 13 samples from Lake Kereta were available prior to the first record of S. pallidus on 13 August 2008, with 20 samples available to 20 May 2013.Where possible, samples were enumerated in aliquots until at least 300 zooplankton individuals were counted.Because net hauls provide only semi-quantitative results, we calculated the relative abundances of zooplankton in each sample for all analyses.Nonmetric multi-dimensional scaling (MDS) and analysis of similarities (ANOSIM) were used to infer whether changes have occurred in the Lake Kereta zooplankton community composition before and after the invasion of S. pallidus (Primer v.6.1.13;Primer-E Ltd 2009).MDS is a multivariate ordination technique that builds a 2-D 'map' of samples based on their similarity to each other as defined by a distance metric; a stress value measures the goodness of fit of the map, showing how well correlated the distances on the map are to those of the underlying similarity matrix.We performed MDS on the Bray-Curtis similarity matrix.Zooplankton data were log(x+1) transformed to down weigh the influence of dominant species in the analyses.Only zooplankton species found in two or more samples were included in our analyses, to reduce the influence of rare species potentially sampled by chance.ANOSIM was applied to the similarity matrix to test whether differences in zooplankton community composition between samples collected before and after the first detection of S. pallidus were statistically significant.For all tests, 999 permutations were executed.SIMPER analysis was used to explore the contribution of individual species to the average dissimilarity of samples collected before and after Skistodiaptomus pallidus (Herrick, 1879) Macrocyclops albidus (Jurine, 1820) Ostracods invasion of S. pallidus.Analyses were undertaken using S. pallidus included (a combined total of copepodites and adults), to determine whether the presence of this species significantly influenced community composition.The analyses were repeated with S. pallidus excluded, to infer whether the invasion of this species has altered the community composition of other zooplankton present.Copepod nauplii were excluded from all analyses as these could not be confidently ascribed to species.

Results
The last samples collected without S. pallidus in Lake Kereta were on 2 April 2008, with its first detection on 13 August 2008.Skistodiaptomus pallidus was present in all samples collected following this date.Fifty-five taxa were observed in the lake during the study period, comprising 41 rotifers, 8 cladocerans and 5 copepod species (Table 1).Prior to the invasion of S. pallidus, the native calanoid copepod Calamoecia lucasi Brady, 1906 was typically present, and occasionally dominated the community (e.g., March 2003, January 2005; Figure 1).At various times, rotifers, cyclopoid copepods or cladocerans dominated the community.Following invasion, S. pallidus generally increased in relative abundance through time, commonly dominating the community (>25% of numbers), and particularly so in August 2010 (94.1%) and June 2011 (91.8% of individuals counted).Following this time, the relative abundance of S. pallidus decreased (always <25% of total numbers over the last 13 months).Excluding S. pallidus, cladocerans and rotifers (as a group) became relatively more important post-invasion, and other copepods (including C. lucasi) decreased in their relative importance, compared to before the invasion.The MDS undertaken with S. pallidus included revealed a clear separation between samples before and after S. pallidus invasion (Figure 2).ANOSIM indicated community composition significantly differed before and after the invasion (Global R=0.501; P=0.001).SIMPER analysis indicated that species contributing most greatly (>5%) to the average dissimilarity calculated between sample groups before and after invasion were crustaceans: S. pallidus (responsible for 13.5% of the dissimilarity between sample groups before and after invasion), Mesocyclops sp.(7.3%; decreased post-invasion), Bosmina meridionalis G.O. Sars, 1904 (7.2%; increase), Daphnia galeata G.O. Sars, 1864 (6.9%; increase) and C. lucasi (6.8%; decrease).This indicates the dominance of S. pallidus in many samples following its invasion was the most important species differentiating the sample groups before and after the invasion.
The MDS excluding S. pallidus showed a separation in samples before and after the S. pallidus invasion, with 'before' samples distributed primarily on the left of the ordination, and 'after' samples on the right.However, the separation of 'before' and 'after' samples was not as clearly defined as in the ordination where S. pallidus was included (Figure 2).Nevertheless, ANOSIM inferred community composition to significantly differ before and after the invasion (Global R=0.246; P=0.002).SIMPER analysis indicated that species contributing most greatly (>5%) to the dissimilarity calculated between samples before and after invasion were D. galeata (9.5%; increase), B. meridionalis (7.8%; increase), C. lucasi (7.3%; decrease), Mesocyclops sp.(6.8%; decrease), and Chydorus sp.(5.2%; decrease).Due to the clear statistical differences between community composition in samples collected before and after S. pallidus invasion, missing samples in the dataset were unlikely to have affected the statistical results.

Discussion
Skistodiaptomus pallidus was first recorded in Lake Kereta on 13 August 2008.Although not present in 13 samples prior to this date, it was present in all 20 samples collected in the following five years.First detection of S. pallidus in Lake Kereta coincided with releases of grass carp (Ctenopharyngodon idella) into the lake to aid in the control of hornwort (Ceratophyllum demersum L., 1753); hornwort had invaded the lake prior to 1999.Between March 2008 and April 2009, four grass carp releases were made into Lake Kereta, comprising almost 15000 fish (de Winton 2012).
Translocations in association with fish appears to be a common transportation vector for zooplankton invasions globally with, for example, Daphnia lumholtzi potentially having been released into North America with introductions of Nile Perch (Havel and Hebert 1993), while the first introduction of Boeckella triarticulata to northern Italy likely also occurred with introductions of fish (Ferrari 1991).Lake Omapere, where we also recorded S. pallidus, had over 40000 grass carp released as a method of controlling Brazilian waterweed Egeria densa Planch., 1849 in 2000 (Champion 2004;Ray et al. 2006), while Lake Hood had over 2000 grass carp released in 2005.Grass carp utilized in New Zealand are domestically produced, from stock originally derived from Hong Kong (McDowall 1990), and it may be that ponds for the stocking of carp are contaminated with S. pallidus.However, S. pallidus has also been recorded among live fish food (Daphnia carinata King, 1852) in aquarium stores (Duggan et al. 2006), and is present in several constructed Auckland lakes, leaving the possibility that it may have been spread by vectors other than grass carp.Skistodiaptomus pallidus has not yet been recorded from some lakes where grass carp have been released, for which we have zooplankton samples (Western Springs and Lake Wainamu; de Winton and Edwards 2012).However, the consistent timing of the first records of S. pallidus in Lake Kereta with grass carp releases appears compelling.
The first records of S. pallidus in the North Island of New Zealand were from constructed waters (Duggan et al. 2006;Banks et al. 2009).
Here we present the first record from the South Island, in a lake constructed for water sports activities such as fishing, rowing and water skiing.Taylor and Duggan (2012) demonstrated that S. pallidus could readily establish populations in experimental tanks where native New Zealand calanoid copepods (primarily C. lucasi) were absent, while invasions were repelled where calanoid copepods were present.However, these authors noted that any community could be invaded if presented with adequate propagule supplies, ultimately overwhelming biotic resistance (see also von Holle and Simberloff 2005).Abundant propagule supplies of S. pallidus could have been provided to the natural lakes in this study in association with large releases of grass carp, if ponds used in the culture of this fish species contained S. pallidus.While recent invasions of S. pallidus in Germany (Brandorff 2011) and tropical Mexico (Suárez-Morales and Arroyo-Bustos 2012) are currently restricted to constructed water bodies, it is likely a matter of time before they spread to natural waters also (as has also been the case for D. lumholtzi in North America, and B. triarticulata in Italy).In New Zealand, the Japanese calanoid copepod Sinodiaptomus valkanovi Kiefer, 1938, and Australian species' Boeckella minuta G.O. Sars, 1896and B. symmetrica G.O. Sars, 1908, are seemingly still confined to constructed waters (Banks and Duggan 2009;Makino et al. 2010), but may also establish in natural waters given time.
A significant change in zooplankton community composition coincided with the establishment of S. pallidus in Lake Kereta.When considering S. pallidus as part of the community, the assemblages before and after the invasion differed significantly, as S. pallidus commonly dominated.However, even when S. pallidus was removed from the analyses, the zooplankton assemblages were still found to be significantly different, primarily due to changes in the relative abundances of other crustacean species.Most notably, the native calanoid copepod C. lucasi decreased in relative abundance.Calanoid copepods of similar size, or with significant dietary overlap, cannot co-exist, such that in many lakes coexistence is rare unless they are of significantly different size (e.g., Hutchinson 1967;Maly and Maly 1997).Similar to the current study, the non-indigenous population of S. pallidus in the Tahoe Keys marina, southern Lake Tahoe, United States, was found to displace the native calanoid copepods Leptodiaptomus tyrelli (Poppe, 1888) and Epischura nevadensis Lilljeborg, 1889 (Byron and Saunders 1981).In the current study, the planktonic cladocerans Bosmina meridionalis and Daphnia galeata increased in relative abundance, and the cyclopoid copepod Mesocyclops sp. and the cladoceran Chydorus sp.decreased, while no rotifer species contributed greater than 5% to the average dissimilarity between zooplankton communities before and after S. pallidus invasion.
Complicating the identification of potential effects of S. pallidus on the zooplankton community was the concomitant release of grass carp and associated reduction in macrophyte biomass.However, studies of grass carp introductions have typically found limited effects on zooplankton communities.Grass carp introduced to Lake Kereta were all greater than 6 cm in length (de Winton 2012).Although grass carp fry utilize zooplankton as a primary food source, these were not released to the lake, and adults rarely eat anything except plant material (Richard et al. 1985).Grass carp transition from a diet of zooplankton or benthos to aquatic macrophytes at approximately 5.5 cm (He et al. 2013).For example, fish >6.3 cm in a Florida pond were found to be strict herbivores (Colle et al. 1978).Indeed, Prowse (1971) noted that adult grass carp may starve in ponds lacking vegetation even if they have high abundances of zooplankton.As such, direct effects of grass carp are unlikely.Macrophyte biomass decreased markedly following grass carp introduction.Observations in 2008 showed hornwort to have >75% coverage in the lake, while >99.9% of biomass was removed when the lake was resurveyed in February 2012 (de Winton 2012).Conditions may have thus become more favourable for some species by providing more open water for planktonic species, while the reduced extent of weedbeds may have reduced the abundances of typically littoral species.This reduction in macrophytes could explain increases in the planktonic cladoceran species, B. meridionalis and D. galeata, and the associated decrease in the littoral Chydorus sp.However, the reduction in macrophytes cannot provide an explanation for the decreased importance of the planktonic C. lucasi.Most studies on the indirect effects of grass carp on zooplankton, through macrophyte removal, have found limited effects also.Fry and Osbourne (1980) found grass carp additions to have little direct or indirect effect upon the zooplankton in three experimental Florida ponds, while Pípalová et al. (2009) found no changes in zooplankton composition or abundance associated with increasing grass carp densities from 29 kg/ha to 92 kg/ha in a Czech Republic pond, despite a reduction in macrophyte density from 109 g/m 2 to 33 g/m 2 .Kirkagac and Demir (2004) found increases in zooplankton abundance in ponds stocked with increasing densities of grass carp (>19.5 cm) in Turkey, largely as a result of a greater availability of nutrients for phytoplankton, providing food for zooplankton.Where effects on composition have been noted, it is usually related to subsequent increases in the abundances of planktivorous fishes following macrophyte reduction (e.g., Maceina et al. 1992;Richard et al. 2005).However, New Zealand zooplankton communities are likely to be under reduced pressure from zooplanktivorous fish relative to elsewhere (Chapman and Green 1987).Overall, competitive interactions with S. pallidus provide the most convincing explanation for the reduction in C. lucasi, rather than any direct or indirect effects of grass carp.
Skistodiaptomus pallidus appears to have undergone a "boom-and-bust" cycle in Lake Kereta.The species gradually increased in its numerical importance following establishment, peaked in relative abundance in late-2010 and mid-2011 when it comprised >90% of individuals counted, and thereafter declined dramatically in its importance, comprising <25% of the relative abundance over the last 13 months of study.Such boom-and-bust cycles are typically expected when food of the invader is previously under-utilised by the existing species, leading to a boom period for the invader, followed by a decline towards an equilibrium with available resources (bust) (Williamson 1996).Alternatively, but less likely, pathogens or predators may also have played a role in reducing the importance of this species.Further monitoring is required to determine whether S. pallidus remains integrated in the lake community at low densities, is extirpated, or if the species may continue to dominate in a cyclic or irregular manner (sensu Strayer and Malcom 2006).Such monitoring is also important to determine the long-term effects on the native calanoid copepod, C. lucasi, whose population had not attained their pre-invasion importance following the decline in S. pallidus.

Figure 1 .
Figure 1.Relative abundances of the major zooplankton groups in Lake Kereta between 1997 and 2013, with S. pallidus included (top) and excluded (bottom).

Figure 2 .
Figure 2. Multidimensional scaling plots showing change in species composition before and after invasion of Skistodiaptomus pallidus into Lake Kereta.Analyses were undertaken with S. pallidus included (above) and excluded (below) from the dataset.

Table 1 .
Zooplankton species observed in Lake Kereta during the study period.