Dispersal of native species from the regional pool can recover invaded communities to a pre-invaded state by supplementing declining populations or providing resistant species. However, dispersal may also exacerbate the negative effects of an invader. Introduced species can open or create new niche space, which could facilitate the establishment of competitors or predators that previously could not succeed in the uninvaded local community. To investigate the interaction between dispersal and invasion by a non-native consumer, we conducted a field mesocosm experiment that introduced zebra mussels into native zooplankton communities. Regional zooplankton were collected and added to both invaded and uninvaded communities. In uninvaded communities, zooplankton dispersal reduced cladoceran diversity by ∼40%, rotifer abundance by ∼65% and copepod nauplii abundance by ∼80%. In invaded communities, dispersal increased cladoceran diversity by ∼60%, but also further exacerbated the negative effects of zebra mussels on rotifer abundance. This experiment illustrates the potential for dispersal to both positively and negatively affect local communities, and how these effects may change with disturbance and the taxa or community metric of study.
We conducted a field mesocosm experiment at the Queen's University Biology Station (QUBS, 44.5788°N, –76.3804°W) in Ontario, Canada, from 9 May 2012 to 23 Aug 2012. We employed a 2 × 2 factorial design to observe zooplankton community response to (i) zebra mussel invasion, and (ii) dispersal of regional zooplankton. On 9–10 May, 378 L cattle tanks (134 × 78 × 63 cm) were filled with 350 L of Lake Opinicon (see Supplementary data, Table SI) water filtered through a 50 µm mesh to remove zooplankton while allowing most phytoplankton taxa to pass through. Tanks were then randomly assigned (to ensure that treatments were spatially interspersed) to zebra mussel (Uninvaded/Invaded) and dispersal (No dispersal/Dispersal) treatments.
Each mesocosm was stocked with an ambient density of zooplankton collected from Warner Lake (uninvaded, Supplementary data, Table SI) on 24 May, with 70% of the zooplankton collected from the pelagic zone and 30% from the littoral region.
Zebra mussels were collected from Lake Opinicon on 20–21 May. Mussels were transferred to fibreglass aquarium tanks and sorted into large (>20 mm), medium (10–20 mm) and small (<10 mm) size classes. Treatment densities were standardized such that each mussel treatment comprised 35% small, 60% medium and 5% large mussels. Invaded treatments received a mussel density of 1.0 mussels L−1, chosen as a volumetric approximation of a natural, high-density mussel invasion (Mellina et al., 1995). In the aquarium, mussels were supplied with a constant flow of Lake Opinicon water while they attached to cylindrical, plastic mesh supports 8 cm high and 20 cm in diameter weighed down with a 13 cm2 ceramic tile. Supports were used to raise the mussels above the bottom of the tanks and provide an easy attachment surface. Zebra mussels were left to acclimatize for 10 days then transferred to the mesocosms on 31 May; supports without zebra mussels were also added to the Uninvaded treatments. A diaphragm pump (51 L min−1 at 10.3 kPa, JEHM Co., Lambertville, NJ, USA) was used to provide mild aeration (via a 5.1 cm airstone) to each mesocosm for the duration of the experiment. Mussel mortality was checked every 3 days during the experiment by raising supports in both Invaded and Uninvaded treatments to the water surface. Dead mussels in the Invaded treatment were collected and replaced with live mussels of the same size class from a stock that was maintained in the aquarium. Mortality per week remained low for the first 9 weeks of the experiment (<10%), and increased to ∼10–15% in the final weeks, with no differences in mortality between treatments (Supplementary data, Fig. S1).
Zooplankton dispersers were first collected and added on 7 June, then again every 2 weeks until conclusion of the experiment. Volume of dispersers collected was based on an amount equal to 2% of the volume of all mesocosms. Zooplankton dispersers were collected with a 50 µm, 15 cm diameter closing net from the deepest point (within the epilimnion but below the first meter) in six uninvaded lakes (Buck, Elbow, Lindsay, Long, Round and Upper Rock; Supplementary data, Table SI) located <30 km from Warner Lake. Zooplankton were pooled and split using a Folsom plankton splitter, which produced 32 equal subsamples. Twenty were randomly chosen, of which 10 were added to the Dispersal treatments and the remaining 10 were first heat-killed in a microwave then added to the No dispersal treatments.
Two additions of 0.147 µM KH2PO4, and two additions of 2.941 μM NaNO3 combined with 4.674 µM NH4Cl (Fisher Scientific, Pittsburgh, PA, USA), were added to all tanks on 16 June and 1 August to account for nutrient losses to periphyton on tank walls.