Introduction

Many coastal people in the tropics and subtropics depend on functioning reef communities that provide critical ecosystem services such as nutrient cycling, food, and income from tourism and fisheries (Woodhead et al. 2019). Globally, coral reefs provide livelihoods for over one billion people, with this number increasing as the global population grows (Cinner 2014). Global warming and habitat degradation can, however, induce multi-species shifts across communities (Feary et al. 2014), reorganising reef community compositions (Kumagai et al. 2018; Precht and Aronson 2004; Vergés et al. 2019), and potentially their functions (Perry and Álvarez-Filip 2018; Vergés et al. 2014). Community reorganisation and species shifts can result in species extinctions and further community degradation (Agostini et al. 2021; Beger et al. 2014). With increasing sea surface temperatures due to climate change, species are expanding their ranges into higher latitudes from tropical to temperate ecoregions (Miller et al. 2023). This process, called tropicalisation, is altering marine ecosystems worldwide and is expected to occur at an accelerated rate with ongoing climate change (Vergés et al. 2014). However, tropicalisation may also create new community functions, such as increased stability (Fung et al. 2011). Currently, we are at a crucial point of change, where rapid ocean warming is leading to widespread localised extinctions and the reorganisation of marine communities (Bindoff et al. 2019). However, our understanding of how tropicalisation reorganises communities and their functioning is limited, and typically restricted to well-known taxa such as fishes (Floyd et al. 2020; Kumagai et al. 2018; Vergés et al. 2016); hard corals (Baird et al. 2012; Sommer et al. 2017) and molluscs (Floyd et al. 2020). Thus, any changes in lesser-known taxa, including for octocorals and zoantharians, remain not well examined.

In functional ecology, species with similar traits are expected to provide equivalent roles in ecosystems (Anderson et al. 2022; Miller et al. 2023), and therefore, considering such qualitative data is also beneficial when characterising community roles and identifying species assemblages. Functional diversity is an important metric when measuring the trait diversity in an ecosystem and for seeking to understand functional responses in communities experiencing environmental disturbances (Cadotte et al. 2011; Floyd et al. 2020; Sommer et al. 2017; Tilman 2001). Such functional information on reef communities can provide insights into the fate of ocean ecosystems when supported by enough data (Halpern and Floeter 2008). This is particularly powerful to know about in high-diversity ecosystems, such as coral reefs, where functional redundancy is high (Clay et al. 2023; Mouillot et al. 2014).

Evidence of coral and fish species' tropicalisation has been documented (Tuckett et al. 2017; Vergés et al. 2016; Yamasaki et al. 2014) and trait-based approaches can support our understanding of the persistence of tropical species in subtropical and temperate environments (Madin et al. 2016). For example, for scleractinian corals, traits linked to environmental tolerance and energy acquisition, such as morphology, determine the establishment potential of zooxanthellate corals on subtropical reefs in Australia (Madin et al. 2016; Mizerek et al. 2016; Sommer et al. 2014), suggesting these traits may also be important for coral reef octocorals and zoantharians along biogeographic transition zones. Marginal high-latitude coral communities tend to inhabit shallower habitats, due to comparatively lower levels of light, specifically demonstrated by the scleractinian zooxanthellate genus Acropora (Madin et al. 2012; Muir et al. 2018). Conditions in the subtropics, specifically low light levels and highly variable sea surface temperatures, limit poleward expansion, and therefore, organisms that are pre-adapted to such conditions may be more successful at shifting (Kahng et al. 2019). However, the presence of certain functional groups at higher latitudes under lower light levels may allow specialised species to evolve, due to functional fitness and adaptation of octocoral and zoantharian species (Hoegh-Guldberg et al. 2017).

In the marine environment, trait-based approaches have primarily focused on fish and hard corals, and given the vast diversity of taxa occupying reefs, trait-based studies for other taxa are urgently needed (Butt et al. 2022). Particularly, species-specific information is also lacking for many marine taxa (e.g., molluscs, sponges, octocorals and zoantharians), and broader metrics, such as functional diversity, can measure their functional roles in their place. Among benthic invertebrates, many non-scleractinian anthozoans such as octocorals and zoantharians can become dominant as coral communities degrade (Fabricius and McCorry 2006; Reimer et al. 2015, 2021; Rodríguez et al. 2020; Ruzicka et al. 2013; Yang et al. 2013). Specifically, many octocorals and zoantharians are predicted to expand into higher latitudes and marginal habitats, making it increasingly important to understand how they will respond to environmental change (Jenkins and Stevens 2022; Lasker et al. 2020; Reimer et al. 2021).

The marine transition zone of southern Japan is currently experiencing rapid ecosystem change (Kumagai et al. 2018; Yamano et al. 2011). Spanning the Ryukyu Islands and the southern Pacific coast of mainland Japan, this region is dominated by the northwardly flowing warm Kuroshio Current. The current brings warm waters to higher warm-temperate latitudes, allowing the influx and survival of larvae of tropical species, such as scleractinian corals, as high as approximately 35°N latitude (Matsumoto et al. 2016; Veron 1992). The environmental gradients across the tropical to warm-temperate Kuroshio region can facilitate coral and fish range expansion, and some high-latitude areas have shifted dramatically from temperate- to tropical-dominated ecosystems over the last 20 yrs (Vergés et al. 2014; Yamano et al. 2011). However, little is known about spatial patterns in community composition of other important invertebrate taxa, such as octocorals and zoantharians and whether they have changed over time. A recent paper found that the trait composition of molluscs varies along the tropical-to-temperate transition zone in Japan, suggesting that environmental gradients likely influence species composition and their ecosystem functions (Floyd et al. 2020). Such phenomena may also be currently occurring for other taxa, such as octocorals and zoantharians. As species range shifts can have large ecosystem and economic consequences, understanding the dynamics that influence species distributions of various taxonomic groups is critical including currently understudied habitat-forming coral reef taxa, such as octocorals and zoantharians.

In this study, we quantify spatial patterns of octocoral and zoantharian communities along the tropical-to-temperate transition zone in Japan. Assessing changes along a tropical to temperate latitudinal gradient (Kreyling et al. 2014), we specifically aimed to: (a) detect functional shifts in soft coral communities across a latitudinal gradient in the Kuroshio region of southern Japan via field surveys and data analyses; (b) create a trait database for 25 soft coral and zoantharian taxa that were found in our surveys; (c) classify taxa into functional groups to determine abundance changes along the latitudinal gradient; and, finally (d) identify which traits contributed most to variation of octocoral and zoantharian composition among sites. As most data on reefs in Japan are focused on scleractinian corals and fish (Kumagai et al. 2018), data for other, often neglected, taxa, such as octocorals and zoantharians, are critically needed (Reimer et al. 2019) to understand ecosystem dynamics in this biogeographic transition zone. This study highlights the challenges of working with these understudied and difficult-to-identify groups whilst providing the basis for continued future research in the tropicalisation of soft cnidarian taxa.

Methods

Replication statement

Scale of inference

Scale at which the factor of interest is applied

Number of replicates at the appropriate scale

Species

Groups of species

25

Sites

Sites

31

Octocoral and zoantharian data collection along Japan’s Pacific Coast

In the field, octocoral and zoantharian communities were recorded at 31 sites across the Pacific coast of Japan, ranging from tropical to temperate sites located between 24 and 35° N latitude. Most sites were surveyed in July 2015, except for sites located in Okinawa and Kochi Prefectures which were surveyed in July 2016. All sites were known locations of scleractinian coral communities, ranging from tropical reefs (Iriomote, Okinawa) to the northernmost warm-temperate limits of zooxanthellate coral distribution (Tateyama, Chiba; Veron 1992). Abundance data at each site were collected using photographs of 3–5 replicate belt-transects of 25 m length × 1 m belt widths. Site depths were standardised at 8 m, except for the most northerly site where coral communities only occur shallower, at 3 m.

After field surveys, all acquired photographs of octocorals and zoantharians were identified based on available literature (Fabricius and Alderslade 2001; Reimer 2010). For many octocorals, identifications were only possible to genus level, and in such cases, taxa were separated into morphological operational taxonomic units (OTUs) based on gross colony morphology. For each colony, the size (length × width to the nearest 5 cm) was noted. Average colony size was then calculated for each identified octocoral and zoantharian taxa across all sites (Fig. 1).

Fig. 1
figure 1

The position of the 31 survey sites along the southern Pacific coast of Japan where octocoral and zoantharian taxa were recorded

Full size image

Trait database construction

We constructed a trait database for octocorals and zoantharians to support our functional analyses. Traits describing morphological, physiological, anatomical and reproductive attributes, and feeding preferences were included in the database for the 25 taxa of octocorals and zoantharians identified in the surveys. Both numerical and categorical traits were chosen. Octocorals and zoantharians that were not identified below genus level were averaged and represented by their genus as ‘spp.’. Values for the database were found from the primary literature, databases, professional expertise and survey data; e.g., average and minimum size (Fabricius and Alderslade 2001) was particularly referred to for octocorals as it contains broad genus traits. Where possible, primary literature for octocorals and zoantharians in the Indo-Pacific was preferentially used (Alderslade 2000; Benayahu 1991; Benayahu et al. 2017; Benayahu et al. 1992; Choi and Song 2007; Dipper 2016; El-Bawab 2020; Fabricius 1995; Fabricius and Alderslade 2001; Fabricius et al. 1995; Garthwaite 2018; Geetha et al. 2018; Green and Shirley 1999; Hirose et al. 2011; Hutchings et al. 2008; Hwang and Song 2007; Irei et al. 2011; Janes 2013; Lewis 1982; Linné and Salvius 1758; Loya et al. 2019; McFadden et al. 2006; Ono et al. 2005; Palomares and Pauly 2022; Picton and Morrow 2016; Polak et al. 2011; Reimer et al. 2013; Reimer et al. 2006; Reimer 1971; Ross and Hodgson 1981; Ryland and Lancaster 2003; Samimi-Namin and van Ofwegen 2016; Samimi-Namin et al. 2016; Shiroma and Reimer 2010; Soka et al. 2011; Tuti and Ofwegen 2018; Utinomi 1954; van Oppen et al. 2005).

We compiled 42 traits relevant to the biology and ecology of soft corals and zoantharians, consisting of: colony form, average colony size; maximum colony size; tentacle shape; ability to fluoresce; polyp height; polyp diameter; number of polyp types; number of sclerite types; tissue type; axis type; depth (min, max and range); invasive; rafting; epibiont; relative competitive ranking; range size; artificial substrate; growth rate; habitat type; substrate type; genus of Symbiodiniaceae; Symbiodiniaceae acquisition method; flexibility in Symbiodiniaceae acquisition; temperature (min, max and range); bleaching threshold; sexual mode; asexual mode, reproduction type; sexuality; spawning timing; pelagic larvae duration; oocyte size; source preference; plankton size. Traits were checked and edited by taxonomic experts of each group (McFadden for octocorals, Reimer for zoantharians). References are found in the supplementary material (S1 Table 5).

Categorising species into functional groups

To evaluate whether our species belong to different functional groups based on their traits, we categorised species within our octocoral and zoantharian assemblage via trait cluster analysis. Seven traits were selected based upon ecological importance, data availability (based upon completeness in the database), and data type. A seven-trait data frame was constructed, including separate categoric levels (Tables 1, 2). All statistical analyses were conducted using R Studio Version 1.3.1073 (Team 2020). The function ‘gowdis’ from the mgcv package, version 1.8–33, in R (Wood 2004) was used to create the dissimilarity matrix using the Gower similarity coefficient. The function ‘hclust’ from the stats package, version 4.0.2, hierarchically clustered the data, which was used to create a dendrogram using the dendextend package, version 1.14.0 (Galili 2015). The dendrogram was cut to create five functional groups using the ‘cut tree’ function.

Table 1 Trait descriptions used for clustering code to create the functional groups for the dendrogram
Full size table
Table 2 Traits used to create a hierarchical cluster analysis for the octocoral and zoantharians functional groups
Full size table

Extracting environmental variables for functional group analyses

The environmental variables were chosen to reflect potential effects of thermal and other environmental stressors, and the distribution of these taxa along the biogeographic transition at these sites. Variables extracted were maximum and minimum sea surface temperature (maxSST; minSST, both in °C) and mean chlorophyll a as a proxy for productivity and water turbidity (S1 Table 3). Turbidity indicates organic matter and zooplankton in water, providing corals with an alternative pool of carbon to compensate for lack of photosynthesis during bleaching events (Houlbreque and Ferrier-Pages 2009). SST extreme events can cause bleaching and mortality in octocorals and zoantharians, and chlorophyll a concentrations directly relates to their opportunities for heterotrophic feeding (Seemann et al. 2012; Sully et al. 2019). Data were extracted from Bio-ORACLE using the R package sdmpredictors, version 0.2.9, obtained at a spatial resolution of 5 arcminutes (approximately 9.2 km at the equator) (Assis et al. 2018; Tyberghein et al. 2012). Variables were extracted from the year of the analysis (2019). Covariance between the environmental variables and latitude were checked before using them in regression analyses with the function ‘cov’ so highly correlated variables (> 0.6) could be discarded. No variables were discarded as none were calculated to have significant covariance. No model selection procedure was performed to eliminate non-significant predictors, and significant correlations were identified by given p values (p < 0.0001).

Table 3 GAM regression models describing the relationship between functional group abundance with several predictors: environmental variables: maximum sea surface temperature (Max SST) and minimum sea surface temperature (Min SST) and chlorophyll a concentrations
Full size table

Functional metric analyses against environmental and latitudinal gradients

We calculated functional group abundance and standard deviation across sites. Functional group abundance is defined as the absolute number of octocoral and zoantharian colonies in one functional group at one site. Generalised additive models (GAM) were created with the ‘gam’ function from the R package mgcv, version 1.8–33 (Wood 2017) to identify environmental drivers of abundances for each functional group across the environmental gradients (maxSST, minSST) and latitude. The smooth terms were created from the environmental variables for each site and significant relationships were identified.

Regression models with functional indices for significant correlations

Functional indices quantify the diversity of species traits in an ecosystem; here, we calculated them to show how species traits change across the latitudinal gradient. The following functional indices were calculated using the ‘dbFD’ function from the FD package, version 1.0–12, in R (Laliberté et al. 2014): Functional Richness (FRic) of each community, Functional Evenness (FEve), Functional diversity (FDiv), Rao’s quadratic entropy (RaoQ), and Functional Group Richness (FGR). FEve reflects how evenly distributed the individuals are in the trait space (Villeger et al. 2008), whilst FDiv characterises the diversity of traits in an ecosystem that may be used to identify the functional response to disturbances within a community (Mouillot et al. 2014; Floyd et al. 2020). Rao’s Q provides a functional diversity metric that combines both functional richness and divergence (Ricotta and Moretti 2011). Linear and polynomial regressions were used to analyse how the functional indices, species richness and functional group abundances correlated with latitude, maxSST, and minSST. Both linear and polynomial regressions were used to identify whether a significant linear correlation also fit a polynomial relationship. Models were created using the ‘lm’ function from the stats package, version 4.0.2, in R (Team 2020). In addition, functional group abundances were also regressed against FRic and RaoQ. The p value and adjusted R2 value were extracted from each model to assess their significance.

Trait composition analysis

We used the ‘dbFD’ function from the FD package, version 1.0–12, in R to calculate the community-weighted means (CWMs) for each trait and created a Bray–Curtis dissimilarity matrix from the CWMs to quantify variation in functional composition among sites. A trait-based multivariate analysis was generated using the ‘metaMDS’ function from the vegan package, version 2.5–6, in R, creating a non-metric multidimensional scaling (NMDS) ordination to identify the traits which cause the main composition change across sites (Floyd et al. 2020). We looked for outliers in environmental space and latitude which contributed most to the trait composition. Environmental space is defined by minimum and maximum sea surface temperature and chlorophyll a from a principal components analysis using the ‘prcomp’ function in the stats package, version 4.0.2, in R. The results from this analysis were bound into a data frame together with scores from the sites based upon a bound NMDS of the sites with coral abundance and the latitude of the sites. This was performed with the ‘cbind’ function from the base package, version 4.0.2. A ggplot was then created using the NMDS as the axes from the principal component analysis. This was to identify any traits which were associated with the cumulative effect of the 3 variables at their different levels.

Results

Database construction

We found five distinct groups with similar trait combinations (Fig. 2), which we call ‘functional groups’ as their trait similarity implies similar ecological responses (Gitay and Noble 1997).

Fig. 2
figure 2

Dendrogram showing the five functional groups of octocoral and zoantharian species based upon trait similarity among species

Full size image

Upon examination, species were divided into groups by their broad functional role in the ecosystem. For example, group 1 was azooxanthellate octocorals and zoantharians, group 2 was zooxanthellate Malacalcyonacea; group 3 was comprised of various “generalist” zooxanthellate zoantharians and octocorals; group 4 was sheets and whips corals, and group 5 was characterised as “specialist” zooxanthellate species. Unexpectedly, three zoantharian species in group 3 (Palythoa mutuki, Palythoa tuberculosa, Zoanthus sansibaricus) were placed in a different group from the two other zoantharian species (Zoanthus gigantus, Zoanthus kuroshio) in group 5.

Functional groups

The abundances of group 2 and group 3 showed statistically significant relationships with at least one of the environmental variables (Table 3). Relationships with maximum and minimum sea surface temperature showed nonlinear significant correlations with the abundance of group 3. In addition, functional group richness decreased with increasing latitude with a nonlinear relationship (Fig. 3, Table 4). Species richness decreased significantly with increasing latitude. Only functional group richness (FGR) and species richness (SR) showed significant linear and polynomial relationships with latitude (Table 4). Due to the low number of species/entities, with only one or two species/entities at sites, FRic, FEve, and FDiv could not be calculated for the two sites Izu and Kushimoto. RaoQ was zero for the two sites (S1 Table 1).

Fig. 3
figure 3

Regression models describing changes in species and functional group richness across increasing latitude. Generalised linear regressions for a functional group richness (FGR) across latitude, with a confidence level of 0.1 alpha (grey shaded region); b species richness across latitude

Full size image
Table 4 Linear and polynomial regressions were used to describe the relationships between latitude, functional group richness (FGR), functional richness (FRic), functional evenness (FEve), functional divergence (FDiv), and species richness (SR)
Full size table

Trait composition analyses

CWMs for each trait at the 27 sites were used to create NMDS plots. The NMDS axes showed that average colony size and depth range contributed most to differences in trait variation among the sites (Fig. 4). Regional outliers of trait composition were seen to have varying latitudes (Fig. 4B) suggesting latitude is not a definitive influencing factor upon trait composition. A lower environmental space value was seen to be associated with sites where colony size and form influenced trait composition the most (Fig. 4A). There was no correlation between environmental space and latitude.

Fig. 4
figure 4

Non-metric multidimensional scaling (NMDS) describes which traits contribute most to the variation of trait composition among sites associated with; a Latitude. Variables are categorised by a colour gradient. b Environmental Space, defined by the first axis of a principal coordinate axis with minimum sea surface temperature, maximum sea surface temperature, and chlorophyll A

Full size image

Discussion

The ongoing tropicalisation of subtropical and temperate reefs demands a better understanding of habitat-building taxa, particularly those that are typically neglected in research. The current trait-based analyses of soft coral communities on a tropical to temperate environmental gradient revealed some variation in community and functional composition with increasing latitude, as both species and functional richness decreased poleward. We identified five functional groups of octocorals and zoantharians based on their traits of colony form, average colony size, depth range, habitat type, type/genus of Symbiodiniaceae, reproduction type, and preferred food source. The abundance patterns of these various groups were driven by environmental parameters: minimum and maximum sea surface temperature and chlorophyll a concentration. Azooxanthellate octocorals and zoantharians (group 1), sheets and whips (group 4) and “specialist” zooxanthellate species (group 5) were non-significantly correlated with these parameters (Table 3, S1 Table 4). However, the abundance patterns of zooxanthellate Malacalcyonacea (group 2) and “generalist” octocorals and zoantharians (group 3) were seen to be significantly influenced by gradients in minimum sea surface temperature and chlorophyll a, in both positive and nonlinear fashions—group 2 fluctuated across increasing minimum sea surface temperature whilst group 3 increased and reached a plateau (S1 Table 4). Both groups 2 and 3 showed parabolic curves across mean chlorophyll a, suggesting presence of chlorophyll a at temperate sites was more significant upon functional groups compared to higher and lower latitudes (Table 3, S1 Table 4). Group 2 showed an inverted curve suggesting a lower chlorophyll a mean had a negative impact on the abundance of functional group 2 coral at temperate sites; group 3 showed a positive curve suggesting a greater concentration of chlorophyll a at temperatate sites were correlated with a greater abundance of functional group 3 coral (S1 Table 4). In addition, functional group richness and species richness were the only functional indices which showed statistically significant correlations with latitude.

The predictive capacity of our models was low due to the lack of specificity in our data, both in terms of taxonomy and the assignment of trait values. Many octocorals and zoantharians cannot reliably be identified to species or even genus level in the field (McFadden et al. 2022), and as such it was necessary to categorise multiple species into one taxonomic entity at coarser taxonomic resolution. Massive soft corals contain many cryptic species (McFadden et al. 2019). Species turnover may also have contributed to a lack of specificity within the functional groups; for example, if species were different along the gradient yet looked outwardly very similar, no functional change would be detected. Trait conservatism states the traits of closely related species should be very similar between species, proposing there may not be much difference in trait composition, explaining no detectable functional change (Sommer et al. 2018). Despite the low specificity of the dataset, our data present declines in both species’ richness and functional diversity at higher latitudes, as has been observed for hard corals (Denis et al. 2013).

The construction of the octocoral and zoantharian database highlighted the general lack of basic zoological literature and species-based studies on these groups. It should be noted that much of the data on traits in the constructed trait database were at best expert estimates due to unavailability of data. Additionally, some traits, for example tentacle shape, were generally uniform for many groups (e.g., octocorals). Regarding morphology, polyp length and diameter values in the literature tend to be measured from preserved specimens. However, these values are not indicative of in situ polyp size, as polyps tend to contract when preserved; for example, Sarcophyton polyps can be up to 10 cm in length when expanded but near non-existent when contracted in preservation (McFadden et al. 2022). As well, many sclerite types grade into one another and therefore is often unclear when distinguishing between types (McFadden et al. 2006). Sclerites play roles in the structure and defence of octocorals, although the specific function of the different types of sclerites remains to be elucidated (Kushida et al. 2020; Tentori and van Ofwegen 2011). In this study, average size measurements of colonies were generally based upon field measurements and then supplemented from the literature. Therefore, due to the difficulty in delineating the edges of some colonies, the trait ‘colony size’ lacks accuracy and may not be fully representative of the species. Additionally, the data for depth range were based upon available literature and field measurements, and we cannot discount the potential for some inaccuracy in this trait. As well, for some traits, more zoological data are clearly needed, particularly bleaching thresholds, sexual reproduction, larval dispersal information, and spawning time. More basic research into these areas could identify important datasets related to reproduction and new colony formation, such as spawning time and environmental tolerances, as in zooxanthellate scleractinian corals (Mizerek et al. 2016).

Despite clear data gaps, the information gathered in our database can be used as a starting point to continue species-based studies to fill in current data gaps. Basic zoological and ecological studies form the foundation of many downstream analyses, including trait-based ecology, and it is clear more efforts need to be made to understand the basic biology of many zoantharian and octocoral species. In this study, the data on the traits we obtained formed the basis for our clustering analyses (Schumm et al. 2019), which separated octocorals and zoantharians into five functional groups. The functional groups categorised the species/genera based upon their generalist/specialist nature, or the presence or absence of endosymbiotic Symbiodiniaceae. One unexpected result was the split between groups 3 and 5, suggesting a functionally different role between the members of these groups that had not previously been pointed out. Group 3 members were generalist species with wide distributions, containing two Stereonephthya species (S. japonica; S. osimaensis) along with Sclerophytum spp., two Palythoa species (P. tuberculosa; P. mutuki), and Zoanthus sansibaricus. Whilst little information is available for the octocorals, the zoantharians P. tuberculosa and Z. sansibaricus both have traits related to their wide distribution across many environments, such as endosymbiont flexibility (Kamezaki et al. 2012; Wee et al. 2021), nutritional strategy (Kamezaki et al. 2012; Santos et al. 2021), and morphological plasticity (Ong et al. 2013). This allows us to classify these two species as generalists. Indeed, it has been predicted that these species may increase in number in areas with decreasing water quality (Reimer et al. 2021). In many coral reef surveys, zoantharians and octocorals are often lumped together (English et al. 1997), implying that all members of each group have similar functional roles. Our results indicate that these groups should be separately identified. Both zoantharians and octocorals are speciose groups with hundreds to thousands of species, each with distinct morphologies and other traits, and it is not surprising that various species have different roles in coral reef ecosystems. In our results, we observed functional splits among zooxanthellate zoantharians (Palythoa mutuki, Palythoa tuberculosa, Zoanthus sansibaricus differed from Zoanthus gigantus, Zoanthus kuroshio), and among other O + Z (some Sclerophytum spp. differed from Lobophytum, Sarcophyton, Klyxum and Cladiella spp.). For these O + Z genera, there may yet be even further splits, as species identifications for many of these taxa remain difficult (McFadden et al. 2019).

Whilst functional groups have broadly similar traits, trait values within the groups can differ slightly. For example, the depth ranges within group 3, specialist zooxanthellate O + Z, are varied; however, the habitat types and generalist natures of the species in this group distinguish them from species in group 5. Stereonephthya spp. tend to prefer reef cracks, walls, and overhangs, whilst Sclerophytum spp., Palythoa tuberculosa and Zoanthus sansibaricus tend to occur in a wider range of reef habitats (Albinsky et al. 2018; Jeng et al. 2011; Santos et al. 2021). Interestingly, although Sclerophytum and Lobophytum have similar depth ranges and both harbour Cladocopium, they are separated into distinct groups based on different habitat type preferences. These non-intuitive results may be due to the lack and accuracy of data on octocoral and zoantharian species and outdated published sources, again highlighting the need for more research on these taxa. Additionally, Sclerophytum, Tubipora sp., and Xenia + Heteroxenia spp. were not grouped within the zooxanthellate Malacalcyonacea (group 2), despite being within Malacalcyonacea. This result may also be due to lack of species-specific data for Sclerophytum spp., whilst separating the specialists Tubipora sp. and Xenia + Heteroxenia sp. from other non-specialists within group 2. Moreover, group 3 was most abundant at higher latitudes (Table 3), suggesting their split from the group 5 specialist zooxanthellate species adds support that the two groups are functionally different (Fig. 2). Group 1 comprised azooxanthellate gorgonians, and although the species members look dissimilar, our analyses grouped the species according to their azooxanthellate nature. Group 2 was comprised of malacalcyonacean zooxanthellate species that preferred reef flats, slopes, and crest habitats, whilst group 4 was zooxanthellate Stolonifera, functional outliers based upon their distinct dissimilarities. Although our research is only a preliminary step, it is clear more investigation into the functional roles of these two groups is warranted.

Previous studies on marine taxa have shown gradients in functional richness to be both associated (Schumm et al. 2019) and not associated with latitude (Floyd et al. 2020). Our study builds upon both hypotheses, with significant correlations only for latitude in functional groups 2 and 3 for the temperature-based minimum sea surface temperature variable and water-turbidity measure chlorophyll a. Functional group abundance and species richness showed significant negative relationships with higher latitude, suggesting the functions of soft coral communities shift with latitude as species traits define their functions (Laureto et al. 2015). Declines in functional group abundance and species richness were seen at higher latitudes (Fig. 3), suggesting that higher abiotic stress in these colder subtropical-to-temperate regions leads to declines in species diversity and functional diversity of communities. Reflecting on this, in our study, the Izu Peninsula and Kushimoto, the highest latitude sites, had only one to two species recorded within the diverse groups 1 and 3 (S1 Table 2). Despite some regional outliers of variance observed at varying latitudes (Fig. 4B), azooxanthellate species were grouped within groups 1 and 3, which were abundant in higher latitude sites where light levels are lower. In contrast, zooxanthellate species were mostly observed in lower latitude sites that have higher light levels.

Some key challenges were encountered in this study, and our results represent a first attempt to understand trait-based distributions of octocorals and zoantharians. Firstly, environmental variables that likely shape the distribution and abundance of our target taxa were potentially incomplete. For example, nitrate levels and pH were excluded due to a lack of data availability. Additionally, due to availability, the environmental variables were not taken from the years of our study (2015–2016), but from the year of the analysis (2019). Unfortunately, Bio-Oracle cannot provide data per year (Tyberghein et al. 2012). Finally, we identified species based upon their morphological OTUs, and not DNA barcoding. Although some species identifications are based on previous molecular work in southern Japan (Reimer 2010), other groups, such as many octocoral families from this region, have not yet been completely DNA barcoded. Therefore, many of our species could only be identified to genus level or coarse taxonomic categories, such as ‘Xenia + Heteroxenia'. Incorporating DNA barcoding (McFadden et al. 2019) into future studies would increase the precision and potential accuracy of our results here, possibly resulting in more functional groups and allowing further in-depth analyses. More species-specific trait data are also needed to allow this.

The roles that octocorals and zoantharians play in coral reef communities are generally not well-characterised or understood (Benayahu et al. 2019; Reimer et al. 2021). Understanding their variation in community composition related to abiotic and biotic factors, in addition to their functional roles in reef ecosystems, will enable better characterisations of their responses to climate change. Importantly, this will help us to prepare efforts to benefit coastal communities which depend on coral ecosystems for their livelihoods. Our study considers patterns in the composition of octocorals and zoantharians along the tropical-to-temperate transition zone in southern Japan. The database created from this study has identified data gaps that should be filled in the future. The collection of such trait data will allow for more accurate analyses to characterise shifts in functional composition globally and enable more targeted management and conservation actions under climate change.