Journal of Sea Research 65 (2011) 19–27

Contents lists available at ScienceDirect

Journal of Sea Research

j ourna l homepage: www.e lsev ie r.com/ locate /seares

Zostera muelleri as a structuring agent of benthic communities in a large intertidalsandflat in New Zealand

P.F. Battley a,⁎, D.S. Melville b, R. Schuckard c, P.F. Ballance d,1

a Ecology Group, Massey University, Private Bag 11-222, Palmerston North 4442, New Zealandb Dovedale, RD2, Wakefield, Nelson 7096, New Zealandc 4351 Croisilles French Pass Road, RD3, French Pass 7193, New Zealandd Department of Geology, University of Auckland, Private Bag 92019, Auckland, New Zealand

⁎ Corresponding author. Tel.: +64 6 356 9099; fax: +E-mail address: p.battley@massey.ac.nz (P.F. Battley

1 Deceased, October 2009.

1385-1101/$ – see front matter © 2010 Elsevier B.V. Aldoi:10.1016/j.seares.2010.06.005

a b s t r a c t

a r t i c l e i n f o

Article history:Received 6 July 2009Received in revised form 25 June 2010Accepted 25 June 2010Available online 30 June 2010

Keywords:Zostera MuelleriSeagrassIntertidal BenthosNew Zealand

The influence of seagrass beds on intertidal infaunal communities has been widely studied, with vegetatedareas typically having higher diversity and abundances than adjacent bare sand patches. Such “seagrass–sand” comparisons, however, do not reflect the gradient of seagrass cover that may exist across largelandscapes. We studied the large-scale distribution of intertidal macrozoobenthos over approximately10,000 ha of sandflat on Farewell Spit, New Zealand. The benthic fauna, sediment composition and surfacecover of the seagrass Zostera muelleri were studied at 192 sites evenly spaced along 30 transects covering thelength of the 30 km spit. Most sites had Zostera present, generally at low densities (1–25% surface cover).Overall, invertebrate taxon diversity increased with Zostera cover, from a median of 4 taxa at sites with noZostera to 23 at sites with high Zostera cover. Multivariate analyses of 37 frequently occurring taxa (of the 91recognised) indicated that there was a site gradient of taxon abundances that reflected seagrass cover, with23 taxa increasing as Zostera cover increased. Only three taxa tended to be found more where Zostera wasscarce. Seventeen taxa were identified as being significant indicators of Zostera cover; in all cases abundancespeaked with high Zostera scores. Cluster analysis revealed a number of major groupings. One group wasassociated with low Zostera; two were strongly associated with high Zostera cover; a fourth was probablydistinguished by low tidal elevation and proximity to channels. On the Farewell Spit tidal flats, large-scalepatterns of abundance seem to be largely structured by the presence and density of Zostera.

64 6 350 5623.).

l rights reserved.

© 2010 Elsevier B.V. All rights reserved.

1. Introduction

Invertebrate assemblages on intertidal flats have been welldocumented to vary with the presence or absence of intertidalseagrasses such as Zostera spp (reviewed in for example Boströmet al., 2006). Most studies, however, treat seagrass beds as a singularentity: samples are either within or outside a seagrass patch (e.g.Honkoop et al., 2008; van Houte-Howes et al., 2004). Seagrass may,however, occur across intertidal landscapes in varying densitiesranging from sparse to complete cover. This variation may be partlycaptured by studies that measure seagrass biomass at sampling sites,but the number of sites (and hence number of patches included) isoften limited and patches may be categorised as simply “high” or“low” biomass Zostera. Studies focussing on edge effects in inverte-brate numbers or diversity will, for practical reasons, focus on welldefined seagrass beds (e.g. Bologna and Heck, 2002; Tanner, 2005)and where seagrass beds have become fragmented or comprise very

small patches (Hirst and Attrill, 2008), that approach may beappropriate. But not all Zostera beds are well defined and high-density — large areas can be covered by quite ‘light’, otherwisecontinuous, seagrass cover.

Analysing benthic invertebrate occurrence in relation to a gradientin seagrass densities or cover may give much stronger insights intothe role that seagrass beds play in structuring intertidal communitiesthan simply comparing high-density beds with bare sand. Such patch-bare sand comparisons often find that dissimilarities in the benthicassemblages are due to a few common taxa; these analyses are oftenrestricted to a small subset of the commoner taxa present (e.g. vanHoute-Howes et al., 2004). This does not reveal much about howcommunities as a whole are influenced by the presence and density ofseagrass.

A key issue is whether the invertebrates associated with Zosterabeds are part of a wider tidal flat community and respond favourablyto the presence of Zostera, or whether there are specialist Zostera taxafound largely or exclusively in the seagrass. Conversely, whether thereare species that truly favour non-vegetated habitats is seldomestablished, due in part to limited spatial sampling and limitedsample sizes. In this study we present data from a study of the

Fig. 2. Zostera fresh (squeeze-dried) mass in relation to surface cover on the FarewellSpit tidal flats. Numbers above the boxes give the sample size per category.

20 P.F. Battley et al. / Journal of Sea Research 65 (2011) 19–27

intertidal fauna of one of New Zealand's largest tidal flats, thec. 9500–10,000 ha sandflats of Farewell Spit, Northwest Nelson(Fig. 1). Farewell Spit extends approximately 30 km eastwards fromthe northernmost point of the South Island, and has tidal flatsextending up to 6–7 km into Golden Bay on the spit's southern side.These tidal flats contain extensive beds of the seagrass Zostera muelleri(Jacobs et al., 2006) that are readily evident in satellite images (e.g.http://earthobservatory.nasa.gov/IOTD/view.php?id=5754); manyareas, however, are devoid of vegetation and the total area of seagrasspresent has not been estimated. We used a transect survey across theflats to sample 192 regularly spaced sites across the entire flats. Thecoarse level of spatial resolution required to cover the entire tidal flats(transects every 1 km, samples every 500 m) cannot address fine-scale variability in seagrass and non-seagrass faunas. It does, however,enable us to determine whether large-scale patterns in macrozoo-benthos abundance and diversity exist and whether these reflectdifferences in Zostera coverage.

2. Materials and methods

The Farewell Spit Nature Reserve is situated at the northwestcorner of the South Island of New Zealand (40°31′S, 172°45′E to40°35′S 173°04′E). Extensive Z. muelleri beds occur on the sandflats onthe southern side of the spit, especially at lower tidal elevations andadjacent to major channels.

2.1. Sample collection and processing

Using a large team of volunteers, we sampled macrozoobenthos,sediments and Zostera at 192 sites in a grid survey over 10 days inMarch 2003 (Fig. 1) (see Battley et al., 2005). Transects were run in anorth–south direction every km along the spit, with samples takenevery 500 m. At each site, three 100 mm-diameter benthos coresweretaken to a depth of 250 mm, sieved in the field through a 1 mm sieve(earlier work found that large amounts of sand were retained in asmaller sieve: Battley, 1996), stored in plastic bags and sorted that

Fig. 1.Map and location of Farewell Spit, New Zealand. The thick line marks the approximatethe thin line shows the ‘dry’ land part of the spit. The compass arrow points north.

night. All invertebrates were retained and stored in 5% formalin inseawater. They were later identified to the lowest practicabletaxonomic level given the available equipment and expertise, whichvaried amongst groups. In general molluscs were identified to speciesor genus, polychaetes and small Crustacea to family. Some referencespecimens of groups that we only coarsely identified were identifiedto higher levels (by Rod Asher, Cawthron Institute, New Zealand; seeTable 1). For analyses, the three cores were combined. Any live Zostera(both above- and below-ground) taken with the benthos cores wassorted in the laboratory, squeeze-dried by hand and weighed to thenearest gram to give an index of Zostera biomass (there were onlytrivial amounts of algae or epiphytes on leaves).

To estimate seagrass surface cover at each sampling station,percent cover of Zostera in a 50 cm×50 cm quadrat was estimatedwith reference to a standard set of photographs, this being the

spring low-tide edge of the tidal flats (as judged fromwater cover during fieldwork) and

21P.F. Battley et al. / Journal of Sea Research 65 (2011) 19–27

recommended method of the SeagrassNet project (Short et al., 2001).A six-category scale was used, based on the Braun–Blanquet scale(Mueller-Dombois and Ellenberg, 1974). The 1–6 scale recorded in thefield corresponded roughly with Zostera cover of 0%, 1–5%, 6–25%,26–50%, 51–75%, and 75–100%. Observations were made at low tide,when the plants were lying flat.

A 25 mm diameter×100 mm deep core was also taken for sedimentgrain size analysis. A wet-washing sieving method suitable for fieldconditionswasdevised (seeBallance et al., 2006 for details anddiscussionof accuracy). For analyses here, sediments have been grouped into threecategories: fine sand (over 0.25 mm), medium sand (0.25–0.5 mm) andcoarse sand (over 0.5 mm). As initial field-sieving suggested there waslittle variation between samples, and the processing proved to berelatively time-consuming (0.5 h per sample), not all samples wereanalysed at the time. On some transects only every second sample wasanalysed, so that 145 samples were analysed fully. The coarse sandfraction was subsequently measured on 42 of the remaining sites.

2.2. Analysis

We used four main techniques to explore structure in the speciesand habitat data: Principal Components Analysis (PCA), IndicatorSpecies Analysis, Cluster Analysis and Detrended CorrespondenceAnalysis (DCA). All analyses were performed in PC-Ord, version 4.0(McCune and Mefford, 1999). Analyses used two matrices. The firstwas the site×taxon matrix, in which the densities were log10(n+1)transformed to reduce the scale of the differences in abundances. Thesecondwas a habitatmatrix,which contained the Zostera surface coverscore, Zostera mass, and the percent coarse sand in the substrate, foreach sample site. Four taxa were split into subclasses for analysis:the bivalve Austrovenus stutchburyi into five size-classes (1–10 mm,11–20 mm, 21–30 mm, 31–40 mm, and greater than 40 mm); thebivalve Paphies australis into two size-classes (1–13 mm, greater than13 mm); the limpetNotoacmea helmsi into two subspecies (helmsi andscapha); and polychaetes of the family Maldanidae into ‘lower’ and‘upper’ flat taxa (‘Maldanidae 1’ and ‘Maldanidae 2’ respectively: seeBattley et al., 2005). Because many taxa were recorded rarely, thedataset was trimmed of uncommon taxa. Initially, all taxa thatoccurred in only one site were removed, as was the one site that hadno animals present, and any sites that were lacking sediment data. Thedataset was further reduced by removing taxa that totalled fiveindividuals or fewer, and then those totalling 10 or fewer. Trialanalyses were done with all three datasets; results were very similar

Fig. 3. Fresh mass of Zostera present in the three core

for the three datasets, but the smallest one had fewer taxa that wereseemingly significant mainly because of their rarity. Analysesdiscussed here were done on the dataset containing all taxa in whichmore than 10 individuals were recorded (187 sites and 45 taxa,including size classes).

3. Results

3.1. Seagrass cover and sediments

Z. muelleri was present at 120 of the 192 sample sites on the tidalflats. Fifty-five sites had visual cover scores of 2 (equating to about 1–5% cover), 33 a score of 3 (6–25% cover), 14 a score of 4 (26–50%cover), 13 a score of 5 (51–75% cover), and five with a score of 6 (over76% cover). Surface seagrass cover was therefore mostly light, withfew sites having dense beds. Squeeze-dried Zostera mass from thecore samples was well correlated with the surface cover estimate(Fig. 2; ANOVA, F5,184=80.721, Pb0.0001, R2=0.685), with all butcategories 5 and 6 having significantly different Zostera masses(Bonferroni post-hoc test). The areas of moderate to high seagrassbiomass occurred across a 15 km stretch of the central tidal flats of thespit, particularly in the mid- to lower-level flats (Fig. 3).

For the 189 sites for which we had coarse sand (particle size0.5 mm and above) data, there was a tendency for few sites to have asubstantial biomass of Zostera and also high coarse sand proportions(Fig. 4). As the data were largely concentrated along both zero axes,we categorised sites as high or low coarse sediment (cut-off of 5%) andhigh or low Zostera mass (cut-off of 10 g fresh mass). A contingencytable analysis confirmed that frequencies differed significantly fromexpected, with the coarse sand/high Zostera combination occurring43% less frequently than expected (Chi-square test, χ2=18.12,Pb0.001).

More detailed sediment data (proportion of fine, medium, andcoarse grains) are available for 144 sites. There was no consistentpattern of variation between Zostera biomass and these sedimentcategories. Only a negative relationship between coarse sediment(arcsine-transformed proportion) and Zostera biomass (log trans-formed) approached significance (F143=2.775, P=0.098).

3.2. Macrozoobenthos

In total, 12,839 individuals of 91 taxa were recorded in samples(Table 1), but six taxa dominated the samples numerically (the

samples per site on the Farewell Spit tidal flats.

Fig. 4. Coarse sediment (N0.5 mm) in relation to Zostera fresh mass. Point sizes areproportional to the number of sites with that combination of values (range 1–14 sites).

Table 1Taxon list for those analysed in detail, with abbreviations used in figures.

Phylum Class Order/family Genus/species Abbreviation

Cnidaria Anthozoa Actinaria Anthopleuraaureoradiata

antho

Edwardsia tricolor edwardNemertea nemertAnnelida Polychaeta Capitellidae Capitella capitata capitel

Heteromastusfiliformis

Cirratulidae cirratGlyceridae Hemipodus sp. glycerMaldanidae Clymenella sp. maldan 1

Macroclymene sp. maldan 2Nephtyidae Aglaophamus sp. nephty

Nephtys sp.Nereididae nereidOrbiniidae Orbinia papillosa orbinOweniidae Owenia fusiformis oweniiScalibregmatidae unidentified

speciesscali

Spionidae Aonides sp. spionidLaonice sp.Polydora/Boccardia sp.Prionospio sp.Scolecolepidesbenhami

Syllidae syllidMollusca Bivalvia Lasaeidae Arthritica bifurca arthrit

Mesodesmatidae Paphies australis paphiesMytilidae Xenostrobus pulex xenosNuculidae Nucula hartvigiana nuculaTellinidae Macomona liliana macomoVeneridae Austrovenus

stutchburyiaustro

Gastropoda Batillariidae Zeacumantuslutulensis

zeacum

Zeacumantussubcarinatus

Buccinidae Cominellaglandiformis

cominel

Eatoniellidae Eatoniella cf.lambata

eaton

Lottiidae Notoacmea helmsiscapha

notoscap

Notoacmea helmsihelmsi

notohelm

Olividae Amalda sp. amaldaTrochidae Diloma

bicanaliculatadiloma

Diloma zeylandicamicrelenMicrelenchus

tenebrosusPolyplacophora Chitonidae Chiton glaucus chiton

Arthropoda Maxillopoda Cirripedia,Balanidae

Elminius modestus elimin

Malacostraca Amphipoda amphiCaprellidae caprelCumacea cumacIsopoda Isocladus spicatus flabelFlabellifera

Stomatopoda Squilla armata squillaSquillidae

Decapoda Halicarcinus cookii halicarOcypodidae Halicarcinus whitei macropPinotheridae Macrophthalmus

hirtipesEchinodermata Holothuroidea Apodida Trochidota dendyi holoth

Stelleroidea Ophiuroidea Patiriella regularis patirel

22 P.F. Battley et al. / Journal of Sea Research 65 (2011) 19–27

bivalves A. stutchburyi and P. australis, spionid polychaete worms,amphipods, the barnacle Elminius modestus and isopods). These 6accounted for almost 70% of the individuals recorded. Most sites had2–14 taxa, with the mode of 7 being close to the median of 8 taxa. Themost diverse site had 31 taxa.

Spatially, the most diverse sites were those of the large central flats,with the least diverse areas being those on the fairly narrow tidal flatsalong the inner part of the spit and those on outer areas on the extremeeastward end of the spit (Fig. 5). This partly reflects the extent of Zosteraon the central flats (Fig. 3), as the number of species at a site increasedwith Zostera biomass (linear regression of number of taxa against (log+1) Zostera fresh mass, F191=140.32, Pb0.001, R2=0.43; Fig. 6). Interms of surface cover scores, diversity increased from a median of 4.6(average 4) species at score 1 to 23 (average 20.6) at score 6.

3.3. Community analyses

3.3.1. Principal Components AnalysisA PCA was performed on the variance–covariance matrix. The first

three axes generated explained 40.9% of the variance in the data (22.9%by axis 1, 10.9% by axis 2, and 8.4% by axis 3). Eigenvalues for the firstfour axes were larger than the corresponding broken-stick eigenvalues,indicating that they were, in effect, significant axes. Although plots ofthe sampling sites in relation to Axes 1–3 indicated there were nodiscrete groups of sites, there was nevertheless a gradient of sites thatreflected the presence of Zostera. This ismost obvious in the plot of Axes1 and 3 (Fig. 7), where Zostera biomass increased from the right-handside (no or little seagrass) to the lower left corner (high seagrassbiomass) of the main plot (Zostera mass was strongly negativelycorrelatedwith axis 1: r=−0.650),while the proportion of coarse sandgrainswas positively correlated to it (r=0.191; sites with a high coarsesand component sit towards the right-hand end; Fig. 8). The third axisprovidedmore separation, based on theproportion of coarse sand grainsbut not on Zosteramass (percent coarse sand, r=−0.192; Zosteramass,r=−0.099). Neither variable was strongly correlated with Axis 2(Zosteramass, r=−0.108; percent coarse sand, r=−0.019).

The main species data can also be correlated to the PCA axes(Table 2), and suggest that the distribution of some taxa reflectsdifferences in Zostera and sediments. Of the 23 taxa with correlationsof 0.33 or above (an arbitrary cut-off point), all correlations with Axis1 were negative. This indicates that abundances of a substantialproportion of the fauna increase with the amount of Zostera present.Three taxa had positive, though lower, correlations with Axis 1:Amalda at 0.227, Maldanidae 1 at 0.256, and Nephtyidae at 0.244.

There was a mix of positive and negative correlations with Axes 2and 3.

3.3.2. Indicator Species AnalysisThis tests whether any taxa are particularly good indicators of

certain environmental conditions. Environmental variables were

Fig. 5. Map of species diversity along Farewell Spit.

23P.F. Battley et al. / Journal of Sea Research 65 (2011) 19–27

summarised as categories, and each taxon's occurrence at sites inthose categories summarised. First, for each taxon the relativeabundance in each category was calculated (i.e. what percentage ofthe total individuals occurred in each category). Secondly, the relativefrequency in each group was calculated (what percentage of the sitesin each category the taxon occurred in). Finally an indicator value wascalculated, which summarises the relative abundance and relativefrequency for each taxon across the categories. This is the ‘percentageof perfect indication’; a value of 100 would mean that all individualsoccurred in that category, and all sites in that category contained thattaxon. A Monte-Carlo simulation was also run (1000 times) tocalculate the probability that the indicator scores could occur bychance (Dufrene and Legendre, 1997).

Zostera surface cover score (1–6) was used as the environmentalgrouping. Seventeen taxa were statistically significant indicators(Table 3). All of these taxa increased with Zostera score (Anthopleuraand Austrovenus 21–30 mm and 31–40 mm peaking at score 5). Thismay partly be a consequence of the unequal number of sites in eachcategory –with only five sites with score 6, it is easier for a taxon to berecorded at a high proportion of these than it is in a group with large

Fig. 6. Number of taxa per site in relation to biomass of Zostera. Point sizes areproportional to the number of sites with that combination of values (range 1–11 sites).

numbers of sites – but most of these taxa clearly increase inoccurrence and abundance as the density of Zostera increases. Theonly taxon to be recorded only at the highest Zostera level was thecrabMacrophthalmus. As only 11 individuals were recorded, it is hardto know whether this apparent restriction to high-density seagrass isreal or whether they hide in seagrass over low tide. Other taxa thatseem to be especially strongly associated with high seagrass levels(and were never recorded without some Zostera present) were thelimpet N. helmsi scapha (a known Zostera associate), the tube-buildingpolychaetes Oweniidae, the nutshell Nucula hartvigiana, and the stoutpolychaetes Scalibregmatidae.

No taxa were significantly indicative of bare sand, though twoinfrequently recorded taxa were recorded primarily in areas withlittle seagrass. The olive snails (Amalda sp., 10 individuals at eightsites), and the skeleton shrimps Caprellidae (11 individuals at foursites), occurred only at sites with Zostera scores of 1–3. Theirinfrequent occurrence, however, makes them statistically poorindicators of those habitats.

Fig. 7. Correlations of the mass of Zostera (“zostmass”) to Axes 1 and 3 of a PrincipalComponent Analysis. Points in the main plot are proportional to Zostera mass; theirsymbols represent visual surface cover scores.

Fig. 8. Correlations of the percent of coarse sediment (“sedi05”) to Axes 1 and 3 of aPrincipal Component Analysis. Points in the main plot are proportional to the coarsesand proportion; their symbols represent visual surface cover scores.

Table 3Significant indicator taxa, based on an Indicator Species Analysis. Only taxa with Pb0.05are shown. Zostera group with the highest score is given in bold. For reference, thenumber of sites and total number of individuals recorded is given.

Taxon P Indicator scores per Zosteracategory

N N

1 2 3 4 5 6 Sites Indiv

Anthopleura 0.001 0 3 3 9 36 2 45 210Austrovenus 1–10 mm 0.036 0 6 15 16 18 25 87 532Austrovenus 21–30 mm 0.007 1 5 5 17 30 9 73 514Austrovenus 31–40 mm 0.038 1 2 2 12 24 0 32 125Capitellidae 0.003 0 3 13 4 13 39 65 357Cominella 0.015 0 4 5 4 20 27 54 106Diloma 0.018 0 0 1 3 13 25 20 32Glyceridae 0.008 6 2 2 2 6 32 50 96Halicarcinus 0.006 0 2 10 1 12 33 42 82Macrophthalmus 0.001 0 0 0 0 0 55 6 11Micrelenchus 0.002 0 0 3 4 24 33 34 229Notoacmea subsp. scapha 0.001 0 0 1 0 1 58 13 19Nucula 0.001 0 0 7 6 6 51 42 630Oweniidae 0.001 0 0 1 0 4 69 15 270Scalibregmatidae 0.001 0 0 4 4 13 50 30 155Spionidae 0.003 1 3 9 9 16 35 80 1737Syllidae 0.031 0 1 0 3 0 22 14 34

24 P.F. Battley et al. / Journal of Sea Research 65 (2011) 19–27

3.3.3. Cluster AnalysisCluster analysis defines groupings of species based on their

similarity of occurrence. This was performed on the main data matrix,which had been transposed so that taxa, rather than sites, were theitems being grouped. Alternative dendrograms were generated usingthe Sorensen (Bray–Curtis) distance measure, and a variety of linkagemethods (Nearest Neighbour, Farthest Neighbour, Group Average,and Centroid). Of these methods, the Farthest Neighbour linkage gavethe lowest chaining (sequential addition of small groups, so that fewgroups are evident; 24.8% c.f. 28.6–84.7%) and is shown here.

Table 2Correlations between individual taxa and the first three Principal Component axes of aPCA. Only correlations of 0.33 and above are shown.

Taxon Axis 1 Axis 2 Axis 3

Amphipoda 0.453Anthopleura −0.382 0.436 −0.420Arthritica −0.344Austrovenus 0–10 mm −0.779Austrovenus 11–20 mm −0.773Austrovenus 21–30 mm −0.584 0.396 −0.467Austrovenus 31–40 mm −0.522Capitellidae −0.686Cominella −0.588Cumacea −0.372 0.505Diloma −0.422Eatoniella −0.362Elminius −0.364 0.453 −0.395Flabellifera −0.369 0.669 0.337Glyceridae −0.335Halicarcinus −0.556Macomona −0.450Micrelenchus −0.553 −0.398Nereididae −0.449 0.384Notoacmea subsp. scapha −0.375Nucula −0.644 −0.351Oweniidae −0.357Paphies 1–13 mm 0.528Paphies N13 mm 0.610Scalibregmatidae −0.470Spionidae −0.659 −0.473Zeacumantus −0.691 0.331

The resulting dendrogram (Fig. 9) revealed a number of groupings.

(1) A quartet that was well separated from all others: Amalda,Nephtyidae, Glyceridae, and the ‘seaward’ Maldanidae 1.

(2) A tight grouping of Amphipoda, Flabellifera, Nereididae, andCumacea.

(3) A slightly less-tightly bunched group of small A. stutchburyi(1–20 mm), Zeacumantus, Cominella glandiformis, Halicarcinus,and Macomona liliana, with Capitellidae, Spionidae, and N.hartvigiana closely linked nearby, while Eatoniella,Micrelenchustenebrosus and Scalibregmatidae branched off the same stem.

(4) A group in the lower half of the plot of Anthopleura aureoradiataand A. stutchburyi 21–40 mm, P. australis, E. modestus, andXenostrobus pulex.

3.3.4. Detrended Correspondence Analysis (DCA)The dendrogram groupings were then used as identifiers in a DCA

of the taxon data (Figs. 10 and 11). The two plots show Axes 1 and 2,and 1 and 3, respectively. Most taxa occurred in the same general ‘axisspace’. Some exceptions were clear, though: Amalda, Nephtyidae, andMaldanidae 1 were well separated, with Caprellidae, Notoacmea h.helmsi, Squilla armata, Edwardsia tricolor, Holothuroidea, P. australis,and A. stutchburyi over 40 mm also peripheral to the main groupingvia either Axis 2 or 3. These seem to be the ‘sandy’ taxa.

4. Discussion

The large-scale survey of intertidal macrozoobenthos across theFarewell Spit tidal flats confirmed that there was extensive, oftenquite light, cover of Zostera on the tidal flats, and that the diversity andabundance of invertebrates was strongly related to the presence anddensity of Zostera. This in itself is not surprising — many studies havefound higher invertebrate numbers and diversity in seagrass bedsthan in bare sand (e.g. Boström et al., 2006, and references therein).However, here we show that the influence of Zostera on invertebrateswas a general increase in diversity and abundance of many taxa acrossa wide gradient of Zostera coverage and biomass. Forty-three percentof the variance in diversity across sites was explained simply by thebiomass of Zostera present. Up to half of the taxa analysed in detailincreased in number as Zostera increased (Tables 2 and 3), and theDCA plots showed that most taxa grouped in the same general area ofthe plots (Figs. 10 and 11). It is notable that these relationships areevident despite our not having accounted for many other factors that

Fig. 9. Cluster dendrogram of groupings of common taxa on the Farewell Spit tidal flats, generated with the Sorensen distance and Farthest Neighbour linkage method. For taxonabbreviations see Table 1.

25P.F. Battley et al. / Journal of Sea Research 65 (2011) 19–27

must influence the invertebrate communities, such as tidal elevation,emersion time, sediment temperatures and current speeds (e.g.Legendre et al., 1997; Turner et al., 1999).

Most of the taxa that the Indicator Species Analysis recorded assignificant occurred in four ormore Zostera classes. These taxa seem tobe Zostera ‘generalists’, occurring wherever Zostera is present but

Fig. 10.OrdinationplotofAxes1and2ofaDetrendedCorrespondenceAnalysisof the speciesdata, with groupings from the Cluster Analysis shown. Abbreviations take the first four to sixletters of the family or genus. Notoscap and notohelm refer to Notoacmea helmsi scapha andN. h. helmsi respectively.Numbers refer to size-classes (Paphies, 0=1–10mm,10=N10mm;Austrovenus, 0=1–10 mm, 10=11–20mm, 20=21–30 mm, 30=31–40mm, 40+=over 40mm).

being most abundant in dense beds. Four taxa could be considered tobe Zostera specialists (and are well known for this association: Mortonand Miller, 1973) and were never recorded without some seagrasspresent: the Zostera limpet N. helmsi scapha, the nutshell Nucula, andpolychaetes of the family Oweniidae and Scalibregmatidae (note thatthe taxon found during the survey was not Travisia olens, which hadearlier been found in sandy sediments in the outer spit: Battley, 1996;Battley et al., 2005).

Some taxa evidently did respond to factors other than Zostera, andthese showed some separation from other taxa in the DCA plots.

Fig. 11. Ordination plot of Axes 1 and 3 of a Detrended Correspondence Analysis of thespecies data, with groupings from the Cluster Analysis shown.

26 P.F. Battley et al. / Journal of Sea Research 65 (2011) 19–27

Group 1 taxa (of the Cluster Analysis: Amalda, Nephtyidae, Glyceridae,and the seaward Maldanidae 1) were distinctive largely because oftheir limited distribution on the flats and their occurrence primarilywhere there was little seagrass. Amalda, Nephtyidae, and Maldanidae1 were found only near the spring low-tide waterline. Although allwere found at sites with eelgrass present (maximum Zostera scores of3, 5, 5 and 6 respectively), all were commonest where eelgrass wasabsent.

Group 4 taxa (the anemone A. aureoradiata and A. stutchburyi 21–30 mm and 31–40 mm, P. australis, the barnacle E. modestus, and themussel Xenostrobus pulex) showed high correlations with Axis 2 of thePCA, and all but Paphies also had moderate to strong negativecorrelations with Axis 1 (Xenostrobus, not in Table 2, was –0.271).The similarity in response of these taxa is not surprising: Anthopleuralives predominantly on large Austrovenus shells, and Xenostrobus andElminius both require firm substrates for attachment. Xenostrobuswasgenerally found in localised hummocky areas of firm eelgrass. They allwere found low on the tidal flat or near to major tidal channels(see maps in Battley et al., 2005). For Austrovenus 21–40 mm, thisdistribution was a consequence of a seaward shift with age (Battleyet al., 2005). In contrast, Paphies spat were concentrated in a fewextremely high-density sites at the extreme lower edge of the tidalflats. Seven of the eight sites with densities of more than 1000 smallPaphies per square metre had no eelgrass present, and all but one sitehad high coarse sand content (average 37.6%, range 13.3–73.3%; theremaining site had 0.4% coarse sand). Many spat were present in threeadjacent sites along a sand island in a channel near the spring low-tidemark; next to this was a dense patch of larger Paphies, and thepresence of large numbers of eleven-armed seastars (Coscinasteriascalamaria) and eagle rays (Myliobatis tenuicaudatus) suggest thatsubstantial populations of large Paphies were also present nearby(and subtidally). Hence, although Axis 2 of the PCA was not wellexplained by Zostera mass or coarse sediment, features of Group 4taxa suggest that low tidal elevation and proximity to channels maybe reflected in it. Cole et al. (2000) likewise recorded high densitiesof Paphies spat subtidally near a major channel in Tauranga Harbour,New Zealand.

There are several limitations to this study. The taxonomicresolution employed in this study was low in this study. Polychaetes,for example, are grouped by family, despite the diversity in lifehistories and ecologies that are probably present between species. Noseasonal variation is accounted for, even though this must be presentand could affect the strength of the Zostera–invertebrate relationships(e.g. Rueda and Salas, 2008; Turner et al., 1999). Tidal elevation wasnot measured for the sampling sites. The tidal flats have several largechannels crossing them, and seagrass beds tend to occur along theedge of these channels and their branches. The intervening flats (up to1.5 km wide) are often much higher in elevation, subject to a strongdrying predominantly westerly wind, and may become desiccatedover a daytime low-tide period.

We have no data on whether samples were taken from withindiscrete patches, and if so, where they were situated relative to theedge. Dense patches certainly do occur, cover very large areas of tidalflat, and are highly stable over time in their location and surface cover(P.F. Battley, pers. obs, from 1993 to present). But much of theseagrass on the Farewell Spit tidal flats occurs in quite low densitiesand is not readily identifiable as patches (pers. obs). It is also possibleto find areas of degrading Zostera (established beds eroding back),apparently colonising Zostera (light surface cover but withoutextensive build-up of subsurface root or other organic matter), andareas that presumably have been had seagrass present in the past(bare sand but with large volumes of brown organic material presentin the sediment). Such variation in ‘history’ is present but unrecog-nised in the data.

Finally, the sediment analysis method used in this study was notsensitive enough to accurately divide the sediments up into standard

fine-scale components. Consequently, we restricted our analyses tothe most reliably measured component, the proportion of coarsesand grains (0.5 mm and above). Seagrass beds tend to accumulatefine-particle sediments because of decreased water velocities,production of organic matter, and different infauna to unvegetatedareas (Heiss et al., 2000; Little, 2000). A corollary of this is that thereshould be a lower proportion of coarse sand grains with higherseagrass levels, and sites with high Zostera biomass on Farewell Spittended to have little coarse sediment (Fig. 4).

Despite these limitations, clear relationships between the benthicfauna and the biotic (seagrass) and physical (sediment) environ-ment were detected. These would likely have been stronger had wemeasured sediments more sensitively and analysed seagrass struc-ture in detail. As it is, the proportion of coarse sand grains wasweakly related to the PCA axes, indicating that the gradient ininvertebrate composition to a degree reflected sediment structure, inaddition to the stronger effects of Zostera content. Given the absenceof obvious west–east gradients in macrozoobenthos distribution(see maps in Battley et al., 2005) and the strength of the relation-ships between macrozoobenthos and Zostera, the distribution ofZostera is arguably the dominant structuring agent in the intertidalbenthic communities across the 10,000 ha tidal flats of the spit. Thedistribution of high-density seagrass beds is decidedly non-random,being found mainly adjacent to large channels over about a 16-kmstretch of the spit, as well as at the base. The sediments of the tidalflats are thought to be composed largely of sand grains brought upthe West Coast of the South Island by longshore movement andblown south onto the flats, augmented by coarse sand grainspossibly brought into the system by floating trees (Ballance et al.,2006). Fine sand grains are recirculated by waves and currents, butcoarse grains are irregularly distributed on the flats (being mostly onthe western and far eastern flats) and are probably transported lessaround the flats. Those few taxa that were associated with low orabsent Zostera were found in sandy sediments, on areas of firmersubstrate, or near the extreme lower edge of the tidal flats. Indirectly(for “Zostera taxa”) or directly (for others), then, the large-scaledistribution of invertebrates must also be related to the localgeomorphology, particularly the tidal drainage patterns across theflats.

Our study addressed the seagrass–macrozoobenthos relationshipdifferently to most others, which focus on edge effects or patch sizesof seagrass beds (e.g. Bowden et al., 2001; Connolly, 1997) and tend tocover small spatial scales (Hirst and Attrill, 2008) or concentratesampling effort within sites (Turner et al., 1999; van Houte-Howes etal., 2004). Our finding that over a large spatial scale communitiesresponded across a wide gradient of Zostera covers suggests that thedichotomous Zostera-bare sand comparisons prevalent in the litera-ture are biased representations of seagrass beds — the choice of well-demarcated patches to study restricts our knowledge to a subset ofwhat is present on tidal flats.

Acknowledgements

This project was funded by the Ministry of Fisheries ProjectZBD2002-18 scheme to the Ornithological Society of New Zealand.Thanks to the New Zealand Department of Conservation (DOC) forpermission to work within the Farewell Spit Scientific Reserve, and toDOC and the Maritime Safety Authority for logistical assistance. Thesurvey relied on a large amount of volunteer labour, especially fromthe Nelson-Marlborough Institute of Technology's Trainee Rangerprogramme. Thanks to Rod Asher, Cawthron Institute, Nelson, NewZealand, for taxonomic assistance, Ian Henderson, Massey University,Palmerston North, New Zealand, for statistical advice, Matt Irwin fromMassey University for map-making and two anonymous referees forhelpful comments on the manuscript.

27P.F. Battley et al. / Journal of Sea Research 65 (2011) 19–27

References

Ballance, P.F., Schuckard, R., Melville, D.S., Battley, P.F., 2006. Dual sand sources onFarewell Spit intertidal sand flats, New Zealand: partitioning during redistribution.New Zeal. J. Geol. Geophys. 49, 91–99.

Battley, P.F., 1996. Ecology of Migrant Shorebirds in New Zealand, Focussing onFarewell Spit, North-West Nelson. MSc. thesis, Massey University, PalmerstonNorth, New Zealand.

Battley, P.F., Melville, D.S., Schuckard, R., Ballance, P.F., 2005. Quantitative Survey of theIntertidal Benthos of Farewell Spit, Golden Bay.Ministry of Fisheries,Wellington, p. 119.

Bologna, P.A.X., Heck Jr, K.L., 2002. Impact of habitat edges on density and secondaryproduction of seagrass-associated fauna. Estuaries 25, 1033–1044.

Boström, C., Jackson, E.L., Simenstad, C.A., 2006. Seagrass landscapes and their effects onassociated fauna: a review. Estuar. Coast. Shelf Sci. 68, 383–403.

Bowden, D.A., Rowden, A.A., Attrill, M.J., 2001. Effect of patch size and in-patch locationon the infaunal macroinvertebrate assemblages of Zostera marina seagrass beds.J. Exp. Mar. Biol. Ecol. 259, 133–154.

Cole, R.G., P.J., H., Healy, T.R., 2000. Assemblage structure, spatial patterns, recruitment,and post-settlement mortality of subtidal bivalve molluscs in a large harbour innorth-eastern New Zealand. N.Z. J. Mar. Freshwat. Res. 34, 317–329.

Connolly, R.W., 1997. Differences in composition of small, motile invertebrateassemblages from seagrass and unvegetated habitats in a southern Australianestuary. Hydrobiologia 346, 137–148.

Dufrene, M., Legendre, P., 1997. Species assemblages and indicator species: the need fora flexible asymmetrical approach. Ecol. Monogr. 67, 345–366.

Heiss, W.M., Smith, A.M., Probert, P.K., 2000. Influence of the small intertidal seagrassZostera novazelandica on linear water flow and sediment texture. N.Z. J. Mar.Freshwat. Res. 34, 689–694.

Hirst, J.A., Attrill, M.J., 2008. Small is beautiful: an inverted view of habitatfragmentation in seagrass beds. Estuar. Coast. Shelf Sci. 78, 811–818.

Honkoop, P.J.C., Berghuis, E.M., Holthuijsen, S., Lavaleye, M.S.S., Piersma, T., 2008.Molluscan assemblages of seagrass-covered and bare intertidal flats on the Banc

d'Arguin, Mauritania, in relation to characteristics of sediment and organic matter.J. Sea Res. 60, 255–263.

Jacobs, S.W.L., Les, D.H., Moody, M.L., 2006. New combinations in Australasian Zostera(Zosteraceae). Telopea 11, 127–128.

Legendre, P., Thrush, S.F., Cummings, V.J., Dayton, P.K., Grant, J., Hewitt, J.E., Hines, A.H.,McArdle, B.H., Pridmore, R.D., Schneider, D.C., Turner, S.J., Whitlatch, R.B.,Wilkinson, M.R., 1997. Spatial structure of bivalves in a sandflat: scale andgenerating processes. J. Exp. Mar. Biol. Ecol. 216, 99–128.

Little, C., 2000. The Biology of Soft Shores and Estuaries. Oxford University Press,Oxford.

McCune, B., Mefford, M.J., 1999. Multivariate Analysis of Ecological Data, Version 4.0.MjM Software. Gleneden Beach, Oregon, USA.

Morton, J.E., Miller, M.C., 1973. The New Zealand Sea Shore, 2nd edition. Collins,Auckland.

Mueller-Dombois, D.R., Ellenberg, H., 1974. Aims and Methods of Vegetation Ecology.Wiley, New York.

Rueda, J.L., Salas, C., 2008. Molluscs associated with a subtidal Zostera marina L. bed insouthern Spain: linking seasonal changes of fauna and environmental variables.Estuar. Coast. Shelf Sci. 79, 157–167.

Short, F.T., McKenzie, L.J., Coles, R.G., Vidler, K.P., 2001. SeagrassNet Manual forScientific Monitoring of Seagrass Habitat. Queensland Department of PrimaryIndustries, Cairns.

Tanner, J.E., 2005. Edge effects on fauna in fragmented seagrass meadows. Aust. Ecol.30, 210–218.

Turner, S.J., Hewitt, J.E., Wilkinson, M.R., Morrisey, D.J., Thrush, S.F., Cummings, V.J.,Funnell, G., 1999. Seagrass patches and landscapes: the influence of wind-wavedynamics and hierarchical arrangements of spatial structure on macrofaunalseagrass communities. Estuaries 22, 1016–1032.

van Houte-Howes, K.S.S., Turner, S.J., Pilditch, C.A., 2004. Spatial differences inmacroinvertebrate communities in intertidal seagrass habitats and unvegetatedsediment in three New Zealand estuaries. Estuaries Coasts 27, 945–957.