Sampling of macrobenthic fauna and environmental variables was conducted during austral summer 2012 in Kaipara, Manukau, and Tauranga harbours, North Island, New Zealand (Table 1). Physical descriptions of each of these areas can be sourced from a large number of publications by Simon F. Thrush and co-workers (e.g. Thrush et al., 2003). In each harbour we took 400 cores (13 cm diameter, 20 cm deep) on a predetermined grid (four 1000 m transects, spaced at 100 m) on foot during low tide (n=3⋅400; 1200 in total), thereby covering the area from the high- to low-water mark (Fig. 1 for an illustration). Sampling points along transects were spaced at distances of 30 cm, 1 m, 5 m, 10 m, 30 m, 50 m, 100 m, 500 m, and 1000 m (see Fig. 1 in Kraan et al., 2015), located by using a measuring tape and handheld GPS. Given the close proximity of sampling locations we provide sampling coordinates in NZTM (New Zealand Transverse Mercator; Geodetic CRS: NZGD2000; unit: metres) at the data publisher PANGAEA (Kraan et al., 2019). Cores were sieved in the field (500 µm mesh) and the residue preserved with 70 % isopropyl alcohol.

Prior to destructive sampling, we took a photograph of 50 cm × 50 cm at each sampling point (n=960) to assess the coverage of seagrass (Zostera muelleri), bare sand, and shell hash. In addition, at each point (n=960), we pooled three surface sediment cores (2 cm diameter, 2 cm deep) for sediment grain-size analyses (median grain-size and sediment fractions), chlorophyll a measurements, and determination of the organic content of the sediment (Table 1). These samples were stored in the dark on ice immediately after collecting. Note that at the smallest spatial scale, i.e. 30 cm, we took three adjoining benthic cores, but we limited ourselves to taking one photograph and one sediment sample to represent the environmental features for these three locations. This was done to economically manage our time in the field and our financial budget for processing samples, leading to 320 photographs and 320 sediment samples per harbour. See Kraan et al. (2015, 2019) or Greenfield et al. (2016) for details.

In the laboratory, rose-bengal-stained (2 %) taxa were identified to the lowest practical taxonomic resolution and their abundance assessed. In total we identified 146 species, mostly bivalves, polychaetes, and crustaceans, encompassing 73 813 individuals (Table 1; Supplement, Kraan et al., 2019). For bivalves, the longest shell axis was also measured, allowing adults and juveniles to be distinguished, because habitat preferences can differ between adults and juveniles (Kraan et al., 2010, 2013). Size classes were categorized as <1, 1–5, 5–10, 10–15, 15–20, 20–25, 25–30, 30–35, 35–40, and >40 mm. Each sample was sorted and its taxa identified by Casper Kraan, after which Barry L. Greenfield verified species identifications on each sample (Kraan et al., 2019).

Sediment samples were freeze-dried upon arrival in the laboratory. Prior to freeze-drying, seagrass and bivalves were removed. For measuring, 0.1 g of sediment was weighed, topped up with 90 % acetone buffer, and centrifuged for 10 min at 3300 rpm. Chlorophyll a and pheophytin concentrations (n=960) were determined with a fluorometer using standard methods (see Kraan et al., 2015). The first sample was measured on 10 May 2012, and the last sample was measured on 28 June 2012, avoiding degradation of samples over time (see Kraan et al., 2019).

To determine sediment median grain-size and sediment fractions (silt <63 µm, very fine 63–125 µm, fine 125–250 µm, medium 250–500 µm, and coarse >500 µm), sediment grain sizes were measured (n=960), following standard methods for using a Malvern Mastersizer 2000 with a particle range of 0.02–2000 µm (see Kraan et al., 2015). This involved digesting about a teaspoon of sediment by adding 10 % hydrogen peroxide to remove organic content from the sediment, leaving it to digest for 7 d, and stirring every couple of days.

Organic content (n=960) was determined after burning a teaspoon of freeze-dried sediment for 5.5 h in a furnace at 560 ∘C, i.e. the loss-on-ignition approach.

The coverage of seagrass, shell hash, and bare sand within each photograph (n=960) was estimated based on 75 random points within a photograph using the software CPCe (Kohler and Gill, 2006) (see Kraan et al., 2019).