Creating new littoral zones in a shallow lake to forward-restore an aquatic food web

Freshwater fish communities typically thrive in heterogenous ecosystems that offer various abiotic conditions. However, human impact increasingly leads to loss of this natural heterogeneity and its associated rich fish communities. To reverse this trend, we need guidelines on how to effectively restore or recreate habitats for multiple fish species. Lake Markermeer in the Netherlands is a human-created 70,000-ha lake with a uniform 4 m-water depth, steep shorelines, high wind-induced turbidity, and a declining fish community. In 2016, a forward-looking restoration project newly created a 1000-ha five-island archipelago in this degrading lake, which offered new sheltered shallow waters and deep sand excavations to the fish community. In 2020, we assessed how omnivorous and piscivorous fish species used these new habitats by tracking 78 adult fish of five key species across local and lake scales. We monitored spring arrival of adult fish and assessed local macro-invertebrate and young-of-the-year fish densities. Adult omnivorous Cyprinidae and piscivorous Percidae arrived at the archipelago in early spring, corresponding with expected spawning movements. During the productive summer season, 12 species of young-of-the-year fish appeared along the sheltered shorelines, with particularly high densities of common roach (Rutilus rutilus) and European perch (Perca fluviatilis). This suggests the sheltered, shallow, vegetated waters formed new suitable spawning and recruitment habitat for the fish community. Despite the highest food densities for adult fish in the shallowest habitats (< 2-m), adult fish preferred minimally 2-m deep water. After spawning, most Cyprinidae left the archipelago and moved long distances through the lake system, while most Percidae remained resident. This may be related to (1) high densities of young-of-the-year fish as food for piscivores, (2) medium food densities for omnivores compared to elsewhere in the lake-system, or (3) the attractiveness of 30-m deep sand excavations that were newly created and frequently used by one-third of all tracked fish. New littoral zones and a deep sand excavation constructed in a uniform shallow lake that lacked these habitat types attracted omnivorous and piscivorous fish species within four years. Both feeding guilds used the littoral zones for reproduction and nursery, and notably piscivorous fish became residents year-round.

Methods
Eighteen different locations at the shorelines of the Marker Wadden archipelago in the Netherlands were sampled monthly between May and August 2020. Three different taxonomic groups were collected by various sampling techniques: zooplankton, macroinvertebrates and young-of-the-year fish.

Meso-zooplankton densities were assessed by filtering 30 L of water through an 80-μm mesh size plankton net and condensing the sample into a 50 ml tube. Rotifera densities were assessed by filtering 1 L of water through a 30-μm mesh size plankton net into a 50 ml tube.

Densities of pelagic macroinvertebrates in the water column were assessed by placing an open metal cylinder (ø 50 cm, 50 cm high) within 5 m from the waterline in the water at a depth between 20 and 30 cm and catching all pelagic macroinvertebrates in this enclosed area with a small dip net (13 × 10 cm, 1 mm mesh size). Benthic macroinvertebrates were sampled by taking one sediment core (ø 6 cm, depth 10 cm) in the metal cylinder and sieving the sample over a 0.72 mm mesh.

Young-of-the-year (YOY) fish catch-per-unit-effort (CPUE) was assessed by means of multiple fishing gears that increased in mesh sizes over the season to accommodate for the increasing body lengths and swimming speeds of the fish over the season. In May, two types of dip nets (21 × 21 cm, mesh 0.5 mm, and 55 × 70 cm, 3 mm mesh) were used, to which from June until August additional seine netting (length 15 m, height 2.5 m, mesh-size 10 mm from knot to knot) was added. Sampling always occurred during daytime and at wind speeds <4 Beaufort.

Data were analysed for spatial and temporal variation in generalized linear mixed-effects models as described in the two linked publications.