Dam removal is increasingly considered as a river restoration tool for impoundments that harm the environment or have exceeded their lifespan. However, few studies report the ecological consequences of large dam removal. We performed a multiple Before-After/Control-Impact (mBACI) study to investigate the consequences of the decommissioning of a large dam (42 m high) on instream habitat and invertebrate communities in a temperate, forested catchment of northern Spain. Before decommissioning, a lack of fine sediments and high concentrations of manganese and iron occurred below the dam but decreased downstream. Invertebrate taxa richness and diversity were reduced, and pollution-sensitive taxa were missing just below the dam. The drawdown of the reservoir, the first step towards its decommissioning, mobilized stored sediments causing frequent turbidity peaks downstream, which nevertheless, caused no detrimental effects on macroinvertebrate communities. One year after the drawdown, the communities downstream from the dam, as well as those in the newly formed stream in the area formerly impounded by the reservoir, became very similar to those in control reaches, showing a successful restoration project.

Synthesis and applications: Dam decommissioning helps restore instream habitats and facilitates the recovery of invertebrate communities in a very short time frame if there are nearby sources of potential colonizers. Slow drawdown reduces the transport of the sediments accumulated in the reservoir and their potential downstream impacts, even more if before drawdown the reservoir is kept full for years to promote the retention of sediments in marginal areas that will later be readily colonized by trees.

Experimental design: Our study followed a multiple before-after / control-impact (mBACI) design (Underwood, 1994). We defined four control sites, one (C1) upstream from the dam and three (C2 to C4) in free-flowing tributaries, as well as four impact sites (I1 to I4) at increasing distances downstream from the dam. All eight reaches were sampled before (before the period), during (drawdown period), and after (after the period) the drawdown. Additionally, three sites located within the reservoir (R1 to R3) were also sampled in the after period (Fig. 1). These sites could not be sampled either before the period when the reservoir was full or during the drawdown, when safe wading in the newly carved stream channel was still impossible.

Habitat characteristics: We characterized benthic substrate composition following Wolman (1954), with 100 substrate particles collected per site and period while zig-zagging along 100 m-long reaches.

Benthic invertebrates: Invertebrates were sampled in November 2017 (before), October 2019 (drawdown), and November 2020 (after), one year after the end of the drawdown. On each occasion, we took 5 random samples with a Surber net (0.09 m², 500 µm-mesh) and preserved them in 70% ethanol. Invertebrates were sorted in the laboratory, counted, and identified under a binocular microscope (Tachet et al., 2010). Most invertebrates were identified to genus level (74.6% of the taxa), but some Amphipoda, Coleoptera, Ephemeroptera, Trichoptera, Mollusca, and Diptera were only identified to family level (13.6%). Acari and Oligochaeta were left at these taxonomic levels (Table S4). We estimated taxa richness (S), Shannon-Wiener diversity index (H´), and total density (TD, individuals m^-2) for each sample. We also estimated the IASPT (Iberian Average Score Per Taxon) index, which is widely used in Spanish biomonitoring programs to represent the average sensitivity to pollution of the taxa found (Guareschi et al., 2017). IASPT is calculated as the IBMWP value (Alba-Tercedor, 2002) divided by the number of scoring families present.