Aim: The physical characteristics of biogenic habitats and environmental conditions are important determinants of biodiversity, yet their relative importance can change across spatial scales. We aimed to understand how relationships between the physical characteristics of macroalgal habitats and their invertebrate communities varied across spatial scales and whether general ecological patterns occurred across two countries.

Location: 18 sites across the temperate east coasts of Australia (over 1,300 km) and New Zealand (over 1,000 km), with the latitudinal gradient in the two countries overlapping by 6.73 decimal degrees.

Time period: January to early April 2012.

Major taxa studied: Three intertidal macroalgal habitats in each country and the invertebrate communities within them.

Methods: We measured variation in patch- and individual-level characteristics of macroalgal habitats and their invertebrate communities. Patterns in macroalgal characteristics and communities were compared across latitude, and at smaller spatial scales, and correlated with 26 abiotic environmental variables using multiple multivariate analyses.

Results: Separately, macroalgal habitat characteristics and communities showed unusual but consistent non-linear latitudinal patterns, with greater similarity among sites at the edges of the sampled distribution (i.e. north and south) than at centrally-located sites. Macroalgal characteristics did not correlate with a particular set of environmental variables, however, communities were structured by sea surface temperature at the country scale, and by macroalgal habitat type and biomass within countries. Anthropogenic variables were also important and may have contributed to the unusual non-linear patterns observed between macroalgal characteristics and communities across latitude.

Main conclusions: Our results support other studies showing that large-scale patterns can emerge from systems where there is high local-scale variability. The results show that communities within macroalgal habitats respond to both the physical characteristics of the habitat and external environmental conditions (e.g. temperature). Suggesting that local-scale environmental factors, including anthropogenic stressors may modulate environmental gradients over larger scales.

Study sites

We sampled a single rock platform (25 – 75 m long) at each of 18 sites across the temperate east coasts of Australia and New Zealand (Fig. 1). In Australia, we surveyed 10 sites from Bonny Hills in Northern NSW to Eaglehawk Neck in Tasmania, ranging across more than 1,300 km (linear distance) (Fig. 1). In New Zealand, we surveyed eight sites from Leigh (northern New Zealand) to Shag Point, Otago (southern New Zealand) ranging across more than 1,000 km (Fig. 1). We selected Bonny Hills as the upper latitudinal limit of the study, as this coincides with the transition from temperate to sub-tropical climate, based on Köppen climate classes (Bureau of Meteorology, 2014). Within countries, sites were at least 10 km apart; however, sites were generally over 100 km linear distance from each other (Fig. 1). The latitudinal gradients sampled in both countries overlapped by 6.73 decimal degrees (Fig. 1). The east coasts of both countries were suitable for comparison as they have similar macroalgae and physiographic conditions – including large, flat rock platforms and moderate wave climates (National Institute of Water and Atmospheric Research, 2016; Shand & Carley, 2011).

Study organisms

We sampled three macroalgal species/complexes with distinctive physical characteristics in each country; two of which were shared between countries: Hormosira banksii (Turner) and red turfing algae (hereafter Coralline) (see Appendix Fig. S1 in Supporting Information). Hormosira banksii is a prostrate brown alga with beaded vesicles that are connected in chains 10 – 30 cm long. It is distributed in Tasmania and NSW in Australia and is widely distributed on both islands in New Zealand (Edgar, 2008). Coralline included several morphologically similar species from the family Corallinaceae (e.g. Corallina officinalis, Amphiroa spp., Jania spp.). Species from the family Corallinaceae are widely distributed in temperate Australia and New Zealand (Atlas of living Australia website, 2018 at http://www.ala.org.au. Accessed 01 July 2018) and different species occurred interchangeably throughout the study area. Grouping of Coralline species at the family level as a morphologically similar complex has also been done in other similar studies on habitat-community associations (Kelaher, 2002, 2003a). The third macroalgal habitat sampled was Sargassum spp. (hereafter Sargassum) in Australia and Cystophora spp. (hereafter Cystophora) in New Zealand, which are closely related brown algae that occur in the lower intertidal zone (Edgar, 2008). These two habitats are both brown frondose, branching seaweeds, with receptacles either on branches in Sargassum or on vegetative fronds in Cystophora (Edgar, 2008). Cystophora was sampled at the genus level as multiple species occurred throughout the study area (e.g. Cystophora retroflexa, Cystophora scalaris, and Cystophora torulosa) that would provide a broadly similar physical habitat structure compared to the other habitats, as fucoids with branching fronds. For Sargassum, numerous, morphologically similar, species co-occur in Australia and accurate identification is difficult, being based on the seasonal size and shape of receptacles (reproductive structures at the end of the algal branches) (Edgar, 2008); therefore, this habitat was categorised to genus. Sargassum are broadly distributed in Australia (though absent at some specific study sites, see below) and Cystophora are widely distributed in New Zealand (Atlas of living Australia website, 2018 at http://www.ala.org.au. Accessed 01 July 2018; Edgar, 2008). At each site, we sampled three macroalgal habitats except in: 1) Leigh and Picton in New Zealand where H. banksii was absent, 2) Cook’s Beach in New Zealand where Cystophora was absent, 3) Coles Bay in Australia where Coralline was absent, and 4) Eden and the two Tasmanian sites in Australia where Sargassum was absent (Fig. 1).

Spatial patterns in macroalgal habitat characteristics

All macroalgal taxa were surveyed from January 2012 to early April 2012. Australian sites were sampled in a random order between January and April and New Zealand sites were sampled over a three-week period in February. As ocean temperatures lag seasonally, the sampling period represented summer water temperatures. At each site we sampled six replicate patches of each macroalgal habitat during low tide across the length of the rock platform (Fig. 2). The habitat patches selected occurred as discrete mono-specific patches with less than 10% of other habitat-forming organisms present.

Two patch-level characteristics were measured (patch area and percentage cover), plus two individual-level characteristics (frond length and biomass) of the macroalgae. Patch area was estimated by multiplying the longest and widest dimensions of each patch. Frond length was determined from the average of 10 randomly selected fronds measured at the patch centre. Percentage cover of algae was approximated using a grid of regularly spaced points in a 25 x 25 cm quadrat. Macroalgal biomass was determined from two replicate core samples per patch. PVC cores (10 cm diameter) were collected near the centre of each patch, with algae scraped off at the rock surface with a paint scraper and placed into labelled plastic bags (Kelaher, Castilla, & Seed, 2004; Thrush et al., 2011). Biomass samples were rinsed over a 1 mm sieve to remove trapped sediment. After excess water was drained, the algae were weighed in the field on digital scales (nearest 1 g). The two samples were then pooled to determine patch biomass. To ensure wet weight was an appropriate measure of biomass, samples of each macroalgal taxa were taken back to the lab and oven dried at 60°C for 48 hours to determine dry weight (n=12 cores/habitat). For each macroalga, wet and dry weights all were significantly correlated (Pearson’s Correlation coefficient; r >0.90).

Spatial patterns in communities

Invertebrate communities in macroalgal patches were sampled using one of the PVC cores and collecting the invertebrates retained on the 1 mm sieve. To capture large or benthic invertebrates that may not be collected in the cores, a 25 x 25 cm quadrat with a 5 x 5 cm grid was used to survey larger, benthic invertebrates in each patch. The survey was conducted by searching the fronds and substrate in each of the quadrat grid cells for macroinvertebrates (>2cm). All invertebrates from each patch (core + quadrat) were combined in a labelled plastic bag to capture one composite replicate patch (Fig. 2). The sample was later fixed in 7% formalin for a minimum of one week before being washed and transferred to 80% ethanol for preservation.

In the laboratory, all animals were identified and counted under a dissecting microscope. Molluscs were identified to family level and below (down to species), polychaetes to family level, crustaceans to order or suborder, echinoderms to class, Anthozoa to order, and foraminifera to phylum. The level of taxonomic identification related to the taxonomic group’s dominance among samples and the condition of the samples required for fine scale identification (e.g. although amphipods were a dominant group due to the high volume of collections and the time required to process the samples, some diagnostic features degraded after collection). It was also deemed more useful to include a large range of taxonomic groups identified to a coarse taxonomic level rather than a smaller range of taxa identified to species level in order to maximise chances of detecting habitat-community associations (Anderson, Connell, et al., 2005). Although our sampling methods may not capture all invertebrate taxa (e.g. barnacles, tube dwelling polychaetes, and colonial species such as sponges and bryozoans; Kelaher & Castilla, 2005), these did not appear common when sampling, most likely due to an absence of the bare rock they need for colonisation (Edgar, 2008), and were excluded from the dataset.

Environmental variables

We sourced data on 40 environmental variables (13 of which were later excluded) related to coastal abiotic conditions (natural and anthropogenic) for each site from publicly-available databases (Halpern et al., 2008), satellite images (e.g. NASA/NOAA; Meeus, 1991), and field observations (Table 1). All variables were sourced or determined at the site scale.