Establishing the South Australian Macrobenthic Traits (SAMT) database: A trait classification for functional assessments

Abstract Trait‐based approaches are increasingly used as a proxy for understanding the relationship between biodiversity and ecosystem functioning. Macrobenthic fauna are considered one of the major providers of ecosystem functions in marine soft sediments; however, several gaps persist in the knowledge of their trait classification, limiting the potential use of functional assessments. While trait databases are available for the well‐studied North Atlantic benthic fauna, no such trait classification system exists for Australia. Here, we present the South Australian Macrobenthic Traits (SAMT) database, the first comprehensive assessment of macrobenthic fauna traits in temperate Australian waters. The SAMT database includes 13 traits and 54 trait‐modalities (e.g., life history, morphology, physiology, and behavior), and is based on records of macrobenthic fauna from South Australia. We provide trait information for more than 250 macrobenthic taxa, including outcomes from a fuzzy coding procedure, as well as an R package for using and analyzing the SAMT database. The establishment of the SAMT constitutes the foundation for a comprehensive macrobenthic trait database for the wider southern Australian region that could facilitate future research on functional perspectives, such as assessments of functional diversity and changes to ecosystem functioning.

The use of traits has gained momentum in marine ecology with an growth in published research in recent years, which has improved the understanding of the functioning of marine ecosystems (Cano-Barbcil et al., 2019;Castro et al., 2019;Costello et al., 2015;Lam-Gordillo et al., 2020). The increased interest in traits has been particularly evident in the assessment of macrobenthic communities (Beauchard et al., 2017;Degen et al., 2018;Dissanayake et al., 2018;Lam-Gordillo et al., 2020). Macrobenthic invertebrates have long been recognized as important providers of ecological processes and ecosystem functions in soft sediments due to their capability to enhance recycling of nutrients, modifying sediment properties (e.g., bioturbation, exchange processes). They are also useful bioindicators of pollution and other environmental stressors (Dissanayake et al., 2019;Liu et al., 2019;Reiss et al., 2009;Dittmann et al. 2015;Shojaei et al., 2015).
Throughout the literature, several traits have been proposed to assess the relationship between macrobenthic fauna and ecosystem functioning; however, there are no standardized definitions for traits. In addition, the deficiency on species trait information, data accessibility, and different levels of taxonomic resolution make the selection and use of traits even harder (Lam-Gordillo et al., 2020).
In order to address these issues, some frameworks for assessing biological traits in marine fauna have been suggested, as well as standardized guidelines for the analysis and interpretation of this information (Beauchard et al., 2017;Degen et al., 2018;Lam-Gordillo et al., 2020).
The southern Australian coast is the longest east-west temperate coastline in the southern hemisphere with a diversity of sedimentary habitats (Short, 2020). However, information about traits of macrobenthic fauna from this region is scarce or nonexistent (Lam-Gordillo et al., 2020). The limited information about traits, combined with gaps in the taxonomic knowledge of southern Australian benthic species, has limited the use of functional assessments for management and conservation purposes, as well as understanding benthic ecosystem functioning in this part of the world.
Here, we present the South Australian Macrobenthic Traits database (SAMT), to advance trait-based approaches for southern temperate coastlines. The trait information provided is based on previous studies for comparability and presented in an easily accessible database for downloading and sharing among researchers (Beauchard et al., 2017;Costello et al., 2015;Degen et al., 2018;Lam-Gordillo et al., 2020). In addition, we present a flow chart detailing the step-by-step process of assessing ecosystem functioning and highlighting the utility of the SAMT database for accomplishing this task. This is the first comprehensive assessment of traits of the South Australian macrobenthic fauna, with the aim to facilitate further research across southern Australian temperate marine waters on functional perspectives, elucidating patterns on functional diversity and detect changes in ecosystem functioning.

| Data acquired
A dataset was compiled from previous projects led by the senior author on macrobenthic fauna in soft sediments of South Australia (Table S1), from 37 different localities within this region (Figure 1).
The dataset encompasses records from inter-and shallow subtidal soft sediments in coastal embayments, lagoons, and inverse estuaries, representative of coastal sedimentary habitats along the arid and warm temperate coastline of southern Australia.

| Selection of traits
Selection of traits was based on the most commonly used traits for assessing macrobenthic fauna (Lam-Gordillo et al., 2020), ensuring that the selected biological traits could be compared across studies (Degen et al., 2018), geographical areas (Bremner et al., 2006), and are applicable to most benthic taxa (Costello et al., 2015). The selected traits capture the four subject areas "Biology," "Habitat," "Lifehistory," and "Larval" introduced by Costello et al. (2015) to structure trait categories. In total, based on Lam-Gordillo et al. (2020), 13 traits and 54 trait-modalities were assessed (Table 1).

| Trait allocation
Trait data were gathered from various published online sources, depending on the availability of information for each taxon. When trait information on a particular taxon was missing, its trait values were inferred from the nearest phylogenetic neighbor. For example, if no trait information was available at the species level, trait information was used from another species within the same genus; if information was unavailable at genus level, we considered information at family level. Additional considerations such as taxa distribution, resemblance, and expert judgment were also applied (see Tables S2   and S3).

| Fuzzy coding of traits
Each of the taxa analyzed was scored depending on the affinity that a taxon displayed with a trait-modality using a fuzzy coding procedure (Bremner, 2008;Bremner et al., 2006;Chevenet et al., 1994).
A scoring range from 0 to 1 was used, with 0 being no affinity and 1 being high affinity to a trait. For example, coding the trait "Feeding mode" for Aglaophamus australiensis (Polychaeta), considered that A. australiensis is mostly a predatory species, however, it also exhibits some degree of subsurface deposit feeding, giving a fuzzy coding of 0.75 as predator, and 0.25 as subsurface deposit feeder, completing the full allocation of 1 for the feeding mode trait.

| Case study: assessment of the SAMT database
To elucidate the utility of the SAMT database on the assessment of ecosystem functioning, a functional assessment encompassing four main regions across South Australia was performed. The regions selected were Coffin Bay (locality 1, 3, 4, and 6), Spencer Gulf (locality 9-10), Gulf St. Vincent (locality 14-17), and the Coorong (locality 28, 31-33) (Figure 1). For this case study, we only selected information on macrobenthic fauna from intertidal mudflats. Trait selection was made in the context of ecosystem functioning; thus, we analyzed only traits that influence the functioning of ecosystems (i.e., effect traits) that included, bioturbator, body size, feeding mode, morphology, living habit, and sediment position (Lam-Gordillo et al., 2020).
Macrobenthic fauna were analyzed using both traditional biodiversity metric and functional approaches. The traditional biodiversity approaches included the analysis of taxonomic richness (S) and Simpson diversity index (1−λ) on macroinvertebrate abundances. For the functional approach, trait richness, Simpson index, and functional diversity (as Rao's quadratic entropy: RaoQ) were calculated on macroinvertebrate trait data. Diversity analyses and graphics were performed using R (R Core Team, 2017) and the packages "vegan" (Oksanen et al., 2019), "FD" (Laliberté et al., 2014), and "ggplot2" (Wickham, 2016). A univariate one-factor PERMutational ANalysis Of VAriance (PERMANOVA) using Euclidean distance for the single variable (either effect traits, taxa-or trait-based diversity index), permutation of residuals under a reduced model and 9,999 permutations was used to test for significant differences across  regions (Anderson et al., 2008). All PERMANOVA tests were carried out using PRIMER v7 with PERMANOVA + add on.

| Taxa included
In total, we generated trait information for 277 taxa (see Table S4 for a full list of taxa). The number of taxa varied (i.e., range from 4 to 142 per site, mean of 28) across the 37 localities of South Australia, with the greatest numbers from subtidal sediments in Gulf St Vincent ( Figure S1). Different levels of taxonomic identification were assessed, 152 at the species level, followed by 28 at genus level, 86 at family level, and the remaining 11 taxa at higher levels (order, class, or phyla; Figure S2a). The phylum with most records was Mollusca (112 records, 40% of all taxa), followed by Arthropoda (94 records, 34% of all taxa) and Annelida (45 records, 16% of all taxa), with the remaining 10% belonging to other taxa (Echinodermata 15 taxa, one to three taxa each for Chordata, Sipunculida, Nemertea, Cnidaria, Porifera, and Brachiopoda; Figure S2b). Although Mollusca was the phylum with the highest number of records overall, Annelida was the phylum with the most records across localities (i.e., 43% of all sites) (Figure 2).

| Data sources
The information on traits was retrieved from diverse peer reviewed and expert sources, and a database was generated for easy inter-   (Table 2). However, the source of trait information varied between types of traits (Figure 4a). Across taxonomic levels, most of the trait information was available at the family (42%) and species (38%) levels, with proportionally less at the order/class and genus levels (11% and 9% respectively; Figure 4b). It also emerged that the traits larval type, life span, reproductive frequency, and technique are less studied for the macrobenthic fauna from Australia ( Figure 4).

| The South Australian Macrobenthic Traits (SAMT) database
Functional trait information (i.e., traits and fuzzy coding classification) for the 277 macrobenthic taxa analyzed from the South Australian region is the basis for the SAMT database, which is available as an accessible resource at "https://doi.org/10.6084/m9.figsh are.12763154" (see Figure 5 for a screenshot of the SAMT database).
Along with the database resource, version 1.0.0 of the SAMT R package is provided for assistance in using and analyzing the SAMT database. The SAMT v1.0.0 R package is currently available on the repository https://github.com/Orlan doLam/ SAMT (see Appendix 1 for SAMT package user guide). The SAMT database is intended to progress with regular updates of new data by researchers conducting work across southern Australia for easy downloading and sharing.
To illustrate the utility of the SAMT database, we developed a flow chart showing the step-by-step process for assessing the contribution of macrobenthic fauna to ecosystem functioning ( Figure 6).
The first steps are to compile macrobenthic data from diverse sources (e.g., surveys, field sampling, collections, and online databases) and allocate the respective trait information to each taxon.
The SAMT database reduces the time needed for gathering and finding the taxa-trait information and provides the information in one place. Macrobenthic abundance data can be added to the database at any time, and the R package provided within SAMT database can be used for compiling a trait x sample matrix (LQ). Depending on the aim of the study, and with all the matrices compiled, different analyses can be performed using different software (e.g., R, PRIMER), from measuring trait patterns (LQ), relationships between species-traits and the environment, or modeling the interactions between species-traits and the environment (RQL), to calculating functional diversity as a proxy for assessing ecosystem functioning ( Figure 6).

| Case study using SAMT database: Preliminary functional perspectives for South Australia waters
The analysis of data from the SAMT database included, on average, Trait expression (i.e., the number of taxa that exhibit a determined trait) differed significantly across the regions (p < .01, Diversity, measured using the Simpson Index (Figure 8c), revealed significant differences for taxa and traits across regions (p < .01, Table 4). Coffin Bay was the most significantly different region compared to the other regions based on both taxa and traits (  Figure 8d).

In pairwise comparisons, functional diversity was different in
Spencer Gulf compared to the other three regions, and in Gulf St Vincent compared to the Coorong (p < .05, Table 5). The case study demonstrated the usefulness of the SAMT database for elucidating functional similarities for taxonomically different benthic assemblages across regions.
Compiling trait information of marine macrobenthic fauna is often considered time-consuming and difficult, due to knowledge gaps on the biology and ecology of many species, the lack of identification keys, as well as the scarcity of relevant data (Beauchard et al., 2017;Degen et al., 2018;Verissimo et al., 2012).

F I G U R E 4 Stacked bar graphs showing (a) the cumulative percentage of trait information sources, and (b) the cumulative percentage of trait information by taxonomic level
The SAMT database we present here aims to close the information gap by enabling a comprehensive assessment of traits for the South Australian macrobenthic fauna. SAMT, and the accompanying R package, will facilitate and enhance further research addressing ecosystem functioning and functional perspectives.
The SAMT database provides trait information for 277 macrobenthic taxa and a trait classification for South Australian temperate marine waters. This first iteration of the SAMT database can be F I G U R E 6 Flow chart showing step-by-step processes for assessing ecosystem functioning. Solid colored boxes (green, pink, blue, and black) represent the separate task for analyzing trait data, and black arrows indicate the logical order for the steps. Red box highlights the essential step for having a macrobenthic fauna trait database for southern Australia. Yellow box shows the complementary information needed. Blue dotted box and arrows show the information provided in this study, and the brown dotted box and arrow show the range of potential use of the information provided The SAMT database is available for easy downloading, sharing, and using. However, as in any trait classification, several limitations need to be considered: (a) The structure of the database represents TA B L E 3 Results from univariate pairwise test of bioturbator, body size, feeding mode, morphology, living habit, and sediment position across regions.

ACK N OWLED G M ENTS
We would like to thank all the research assistants and volunteers who have contributed to field sampling over the years. We also thank the Department for Environment and Water (DEW) for al- acknowledge the Editor, associate Editor, and two anonymous reviewers for helpful comments and suggestions that improved the manuscript.

CO N FLI C T O F I NTE R E S T
No potential conflict of interest was reported by the authors.

TA B L E 5
Results from univariate pairwise test of richness (S), Simpson index (1-Lambda'), and functional diversity (FD