The Footprint of Continental-Scale Ocean Currents on the Biogeography of Seaweeds

Explaining spatial patterns of biological organisation remains a central challenge for biogeographic studies. In marine systems, large-scale ocean currents can modify broad-scale biological patterns by simultaneously connecting environmental (e.g. temperature, salinity and nutrients) and biological (e.g. amounts and types of dispersed propagules) properties of adjacent and distant regions. For example, steep environmental gradients and highly variable, disrupted flow should lead to heterogeneity in regional communities and high species turnover. In this study, we investigated the possible imprint of the Leeuwin (LC) and East Australia (EAC) Currents on seaweed communities across ~7,000 km of coastline in temperate Australia. These currents flow poleward along the west and east coasts of Australia, respectively, but have markedly different characteristics. We tested the hypothesis that, regional seaweed communities show serial change in the direction of current flow and that, because the LC is characterised by a weaker temperature gradient and more un-interrupted along-shore flow compared to the EAC, then coasts influenced by the LC have less variable seaweed communities and lower species turnover across regions than the EAC. This hypothesis was supported. We suggest that this pattern is likely caused by a combination of seaweed temperature tolerances and current-driven dispersal. In conclusion, our findings support the idea that the characteristics of continental-scale currents can influence regional community organisation, and that the coupling of ocean currents and marine biological structure is a general feature that transcends taxa and spatial scales.


Introduction
Understanding how regional-scale processes contribute to geographic structuring of biodiversity is challenging because many possible environmental drivers often correlate across broad spatial scales. A starting point is the recognition that increasing geographical distances between observations reveals increasing dissimilarities among biological communities, a phenomenon reflected in distance decay curves [1]. In addition to documenting biogeographic patterns, distance decay curves also provide insights into the nature of the processes which underpin these patterns, through variation in the rates of distance decay of community similarity -species turnover -between geographical regions [1,2].
Marine biogeographers have traditionally held that broadscale patterns of species distributions result primarily from species-specific responses (e.g., larval biology, thermal physiology) to environmental clines such as those associated with continental-scale current systems [3,4]. Temperature, in particular, has been invoked to explain biogeographic patterns for centuries [5,6]. These early accounts are supported by contemporary quantitative studies which show how temperature can be a strong predictor of spatial changes in species composition and biodiversity [7][8][9][10]. Whilst temperature clearly is a powerful driver of biogeographic patterns in marine systems, the interpretation of patterns in relation to alternate drivers that might by superimposed onto the temperature signal have received comparatively less attention. Recently, however, marine biogeographers have drawn attention to the role of physical connectivity mediated by current flow and water movement, even across large distances [11][12][13][14].
Marine systems differ from terrestrial systems in typically being more 'open', with fewer physical barriers to biological exchange between regions [15]. Community structures are therefore often strongly influenced by the passive dispersal of microscopic propagules (spores, gametes, larvae) and macroscopic drifting or swimming organisms, from adjacent areas, with delivery mediated by water movement [e.g., 16,17]. Major ocean currents can provide a link between neighbouring (kilometres) and distant (100's kilometres) habitats. Model calculations have shown that, in some places, flow patterns alone can explain species range-limits [13]. Evidence on how oceanic flow structures such as up-welling, down-welling, eddies and along-shore water transport influence the recruitment and population structure of marine invertebrates across spatial scales up to 100's of kilometres have gained increasing support [16][17][18]. Recent analyses of genetic connectivity in sea urchins [19] and seaweeds [12,20] provide further evidence of significant correlations between water movement, oceanographic connectivity, species distributions, and gene-flow. However, such links between oceanographic connectivity and biological structures should also be evident across larger spatial scales (i.e. continents) and higher levels of biological organisation (i.e. communities within entire seaweed floras).
The coastal waters of temperate Australia are dominated by two major boundary currents: the Leeuwin Current (LC, [21]) and the East Australia Current (EAC, [22]). Both currents are relatively old in geological terms, warm and both flow polewards across approximately the same latitudes on either side of the Australian continent ( Figure 1). Whilst temperature ranges are relatively similar between the LC and the EAC, the LC covers >3 times the geographical distance of the EAC (Figure 1), implying that the spatial temperature gradient is much steeper along the EAC. Such a steep environmental gradient should be reflected in narrow distribution ranges of individual species and thereby relatively high turnover of species between communities [1]. In addition, the two currents also differ in flow structure. The EAC is a stronger current which is periodically disrupted by eddy formation where pockets of water are spinning offshore (see Figure 1). In contrast, the LC is a weaker but relatively undisrupted and unidirectional current [12,21,23,24]. Connectivity modeling has shown that these differences in current characteristics can have important consequences for local retention, off-shore advection and cross-shore transport of marine organisms [24].
Here we use an extensive database of seaweed herbarium records for biogeographic analyses aimed at identifying and understanding continental-scale patterns in seaweed community organisation. Given the predominant direction of dispersal and connectivity we hypothesised that seaweed communities would change serially across regions, in the direction of current flow. Moreover, given that the LC has a weaker temperature gradient and more uniform uninterrupted along-shore flow, compared to the EAC's steep temperature gradient and disrupted flow, we hypothesised that seaweed communities within the LC would show less spatial structure and lower species turnover compared to seaweed communities within the EAC.

Methods
This study encompassed the entire temperate coastline of Australia, which is swept by the LC and the EAC (Figure 1). Our spatial units of analysis were 17 'bioregions', 10 associated with the LC and 7 associated with the EAC ( Figure  1, Table S1). These mesoscale bioregions were derived from the National Marine Bioregionalisation of Australia (IMCRA) [25], designed as marine management units based on geomorphology, oceanography and biological communities. We used these IMCRA regions as sampling units because they represent units of community characterisation independent of spatial extent of coast. Seaweed communities for each region were constructed by compiling species presences from records of seaweeds (marine macroalgae) lodged in the Australian Virtual Herbarium [26] as of 14 September 2009; experienced seaweed taxonomists (G. T. Kraft and C. F. D. Gurgel) inspected all downloaded electronic records [27] and carried out a detailed taxonomic revision, updating and standardizing nomenclature according to Algaebase, an online resource for seaweed taxonomy [28]. While there are many possible challenges associated with the analysis and interpretation of herbarium data, they represent a viable and comprehensive source of information for continental-scale analyses; the very large size of our units of analysis reduce the limitations associated with false absences, uneven recording effort and the dynamic nature of species distributions [29]. For example, recent changes to the distribution of seaweeds have been recorded along all coastlines in Australia [30][31][32][33], but at distances substantially smaller than our bioregion-scale of analysis (Table S1).
We used the Bray-Curtis similarity index based on presenceabsences to compare community structure among bioregions within current systems. A series of multivariate analyses were performed in PRIMER 6.1.10 & PERMANOVA+ for PRIMER [34 for technical details,tests refer to routines in this program, see 35]. Principal Coordinates Ordination (PCO) was used to visualise patterns of community similarity among regions, as this unconstrained metric ordination, show optimal interrelationships between data in a non-preconceived way. Multivariate analysis of variance by permutation (PERMANOVA) and Permutational analysis of multivariate dispersion (PERMDISP) tested if seaweed communities were different, and had different levels of variation, between the LC and EAC. We also tested the strength of serial rank-correlation in community structure between successive bioregions for each of the two current systems (RELATE, using the default model matrix for serial change). Finally, we plotted community similarity as a function of spatial separation distance (measured as linear coastal distances from the centre of each region) and calculated species turnover as the slope of loglinear regression lines. Analysis of covariance (ANCOVA) tested if turnover differed between the two current systems (small slope = adjacent communities are similar = low species turnover).

Results
There were a total of 80,188 records of seaweeds in AVH for the 17 bioregions, with an average (± SE) of 4,717 ± 1,152 records per region. There was relatively similar sampling effort (= number of herbarium specimens) between current systems, with 9.1 ± 1.7 (n = 10 regions) and 4.8 ± 1.1 (n = 7 regions) specimens km -1 coastline within the LC and the EAC systems, respectively (t-test for unequal variances, t 10,7 = 2.04, P = 0.061, see also Table S1). This effort resulted in a total of 1,499 recorded species with an average of 490 ± 54 species per region and similar species densities of 0.96 ± 0.19 and 0.88 ± 0.12 species per km of coastline within the LC and the EAC, respectively (t-test for unequal variances, t 10,7 = 0.35, P = 0.733).
Ordination of seaweed communities placed the 17 bioregions into two groups, clearly separating regions influenced by the LC from those influenced by the EAC (Figure 2). Overall, the LC and EAC groups were significantly different (F 1,15 = 5.2, P = 0.001), with highly distinct assemblages between the northern, subtropical regions in Western Australia and New South Wales, converging in the cool-temperate regions of Tasmania (see  also Table S2). Moreover, the ordination show the contrasting patterns of serial correlation between regions within the two current systems, with 'short-narrow' relations between LC regions and 'long-wide' relations between EAC regions ( Figure  2). In both current systems, the serial change from region to region was significant (ρ LC = -0.611, ρ EAC = -0.762, P < 0.004), but the multivariate dispersion between regions was significantly smaller within the LC (29.0 ± 2.0, 'narrow' arrow) than the EAC (41.0 ± 2.7, 'wide' arrow) (F 1,13 = 13.2, P = 0.003).
Log-linear regression analyses revealed highly significant negative relationships between spatial separation distance and similarity in seaweed community structure within both current systems (Figure 3, r 2 LC = 0.36, r 2 EAC = 0.36, P < 0.001). Species turnover (slope of the regression lines) was significantly different between the two current system (F 1,69 = 14.87, P =

Discussion
Understanding the processes that underpin broad-scale patterns in species distributions is becoming increasingly important because of the pressing need to integrate information across broader spatial scales to resolve issues of human impacts (e.g., climate change, fishing, trophic cascades, eutrophication) [36]. The present study shows a strong imprint of continental-scale ocean currents on the biological community structure of seaweeds across temperate Australia, with high community variability and rapid regional species turnover coinciding with large-scale heterogeneity of current patterns and a strong temperature gradient. Our study thus adds to the increasing support for the idea that broad-scale patterns of species distribution and community structure can be mediated by a combination of well-known temperature gradients and less-studied flow patterns [11,13].
It is well-established that high species turnover (rates of decay in community similarity) is associated with steep temperature (and other environmental) gradients in terrestrial and freshwater systems [1,[37][38][39], but fewer marine studies Figure 1 for spatial arrangement of regions. The green and blue arrows indicate the current systems super-imposed onto the ordination to illustrate the magnitude of floristic change across the current system (length of arrow) and the variation in community structure among regions within the system (width of arrow). There was significant sequential change in community structure along both current systems. have quantified change in community structure across large spatial scales [but see 40 for an example]. In marine systems, flow structure as a driver of biogeographic patterns has previously been associated with regional differences in abundance, recruitment [16][17][18]41] and genetic structure [19,42] of invertebrates and a single seaweed species [12]; our results extend these findings to encompass continental-scale patterns in entire seaweed communities. Collectively, these studies demonstrate how the physical structure of ocean circulation act as footprints that impart structure transcending taxa and spatial scales.

Influence of temperature
Marine organisms tolerate a limited range of temperatures for their survival, growth and reproduction. Temperature variation is therefore a fundamental physiological constraint to global-scale distribution of seaweeds [8,43]. This wellestablished rule is possibly amplified by the present day distribution of coastlines that generally orientate north to south (e.g., E/W Atlantic Ocean, E/W Pacific Oceans, E/W) so that most seaweed communities live within steep spatial temperature gradients. Hence, most seaweed biogeography has been interpreted as temperature driven [8,[44][45][46][47][48]. Still, the causal factors underlying marine biogeographic patterns may not always be so straightforward [49,50], and this difficulty has been particularly evident in southern Australia where temperature gradients are weak [e.g., 51]. As oceanographic currents often drive changes in temperature regimes, marine biogeographers have often interpreted associated biogeographic transitions in terms of temperature [50], to the exclusion of alternative explanations such as those that involve flow connectivity alone [13]. For example, Schils & Wilson [7]

Figure 3. Species turnover between bioregions swept by the Leeuwin current (LC = green triangles) and the East Australian current (EAC = blue triangles).
Dashed lines represent linear regression of community similarity against coastal distance between regions. Dotted lines are 95% confidence limits around the regression lines. Turnover (slope × 10 4 ) within the LC system is significantly lower than within the EAC system. At the same time, regional seaweed communities within the LC were more similar (similarity values were higher), and associated with less region-to-region variability (less spread around the regression) than within the EAC. doi: 10.1371/journal.pone.0080168.g003 attributed very sharp floristic changes in the Arabian Sea to temperature, but did not consider the potential contribution from effects of oceanographic circulation. To date, most biogeographic seaweed studies have focused on how large scale latitude-temperature patterns influence species identities and species richness [43,47,48,52], rather than how temperature gradients modify species turnover. We documented patterns in seaweed communities likely to reflect continental-scale temperature gradients with strong serial community transitions within both of the major current systems influencing temperate Australia. Moreover, species turnover was higher within the EAC (steep temperature gradient) than within the LC (flat temperature gradient). Nevertheless, temperature might not account for all region-to-region seaweed species turnover or the level of variability among regions. Indeed, a study of species turnover in marine invertebrates reported no direct relationships with temperature, although they did not present any alternative explanation for the biogeographic patterns detected [51]. We propose that oceanographic transport of microscopic propagules and drifting reproductive fragments could also be a key contributing structuring process in Australia's seaweed biogeography [12,24,53], and that this model might have broad biological relevance beyond Australia.

Influence of flow
Oceanographically mediated dispersal has been considered to shape spatial patterns of community similarity of coastal fish and invertebrates [11,18]. Ocean circulation may directly link with patterns of propagule dispersal, particularly around biogeographic disjunctions [13]. Given that both the LC and EAC flow southward and that seaweeds are passive dispersers that rely strongly on transport of drifting adult thalli for longrange dispersal (microscopic propagules typically settle metres from the parent plant) [43,[54][55][56][57][58], it seems plausible that different serial correlation in community structure and species turnover are, at least partly, explained by flow. Moreover, offshore currents have previously been shown to be a barrier to dispersal [24,41,59] and could contribute to differences in floristic heterogeneity among regions within the LC and EAC, with drifting propagules more likely to be transported away from the coastline within the EAC [24].
The co-variation between flow patterns and temperature gradients prevent an unambiguous separation of their influences on broad-scale patterns in seaweed community structure. However, comparative studies, testing the physiological and ecological performance of species and populations across latitudinal temperature regimes [60,61,62] might shed light on the extent to which the patterns are primarily temperature or flow driven. Such comparisons will be particularly important in terms of understanding the extent to which impending ocean warming might cause shifts in species distribution, relative to other climate stressors in this system [36,63,64].

Influence of other factors
Temperature gradients and off-shore flow might not be the only mechanisms that contribute to the extant biogeographic patterns. Historical factors, such as the relative stability of the current systems and lack of large-scale disruptions (e.g., glaciation) over millions of years, which contrasts most other temperate regions in the world, are likely to have had a major influence on the evolution and biogeography of the temperate seaweed flora of Australia [46,65]. However, while these factors might explain the mega-richness of the flora and lack of major disjunctions, they are less likely to be a cause of regional patterns in species turnover and community structure. Additional environmental heterogeneity unrelated to ocean currents (e.g., estuaries), human activities [66,67], or biological interactions such as overgrazing by sea urchins [60] could also add to community heterogeneity. However, most of these features operate at spatial scales smaller than our units of analyses and over relatively short time-scales. Differences in niche breadths or dispersal capabilities of seaweeds within the LC and EAC would also influence conclusions about species turnover [1]. However, there are no reasons to suspect that such systematic differences should exist between current systems. Finally, the EAC is characterized by faster current speed and transports a greater volume of water compared to LC [68]. However, these flow characteristics should decrease species turnover along the east coast (opposite to our findings) and are therefore unlikely explanatory models independent of differences in current variability and flow structure. Thus, while we suggest that temperature gradients and broad-scale flow structure are key mechanisms that drive species turnover and community seriation and heterogeneity we cannot exclude the possible influence of additional factors.

Conclusion
We found strong regional seriation of seaweed communities along the LC and the EAC in temperate Australia. Communities within the LC had lower species turnover and less variation than communities situated within the EAC. These patterns are likely to be predominantly determined by a combination of seaweed temperature tolerances and current-driven dispersal.
In conclusion, our findings support the hypothesis that continental-scale currents impart a strong footprint on regional community organisation, and that the coupling of ocean currents and biological structure is a general process that transcends taxa and spatial scales.