A comprehensive functional trait database of seagrasses in tropical Queensland

Introduction

Functional traits are specific characteristics of individual organisms that directly affect how well they perform or survive in their environment (Violle et al. 2007; Nock et al. 2016). These traits are crucial ecological indicators that can predict ecological dynamics with a well-established cause and effect relationships between the traits and the targeted ecological process (Streit and Bellwood 2023). For instance, in Eucalyptus trees, bark traits relate to flammability and shifting of bark traits of a forest can indicate fire-frequency changes (Grootemaat et al. 2017). Establishing clear cause-and-effect relationships can help us understand ecosystem services and monitor processes such as increasing fire frequency as a result of climate change. Monitoring traits rather than species alone advances our understanding of ecosystem function and health as different species might share similar traits, known as trait redundancy (Laliberté and Legendre 2010; Nash et al. 2016). Trait-based methods are widely applied in disciplines such as ecology and biological conservation, including the assessment of the vulnerability of ecosystem services to a changing environment (LeRoy Poff et al. 2006; Butt and Gallagher 2018; Wood and Dupras 2021).

The application of trait-based approaches to seagrass ecology has been very limited despite its widespread application in other communities (Moreira-Saporiti et al. 2023). Seagrass meadows are globally distributed and play a crucial role in marine ecosystems (Larkum et al. 2006; Dunic et al. 2021). They serve as nursery grounds for fish and prawn species (Jinks et al. 2019; Jones et al. 2021), and provide feeding habitats for herbivorous marine mammals and sea turtles (Kelkar et al. 2013; Tol et al. 2016). Additionally, seagrass meadows offer valuable ecosystem services such as coastal protection through wave attenuation and significant carbon storage in sediment (Fonseca and Cahalan 1992; Lavery et al. 2013; Johnson et al. 2017; Twomey et al. 2020).

In the tropical Queensland region of Australia (Fig. 1), a diverse range of 12–15 seagrass species have been documented, dependent on inclusion of rare species and definition of species complex (Waycott et al. 2004; Carter et al. 2021a). The total area of seagrass meadows in the Great Barrier Reef (GBR) region of tropical Queensland (Fig. 1) estimated by habitat modelling is from 35,000 (Coles et al. 2015) to 88,331 km² (Carter et al. 2021b). These meadows are mixed-species meadows and exhibit dynamic changes over time and often with evidence of herbivory (Fig. 2). Several large-scale seagrass-loss events have been recorded in Queensland over the past 30 years (Poiner et al. 1993; Preen et al. 1995; Rasheed et al. 2014; McKenna et al. 2015; James Cook University 2022). These losses have sometimes been caused by a single hazard event, but more commonly are a result of multiple impacts, such as consecutive weather events that inhibit growth and/or cause direct physical damage to meadows (McKenna et al. 2015). Notably, not all seagrass meadows respond to individual or multiple events in the same way; for instance, meadows in the southern GBR displayed a much faster recovery rate than did meadows in the central GBR during a series of hazard events in 2011 (McKenzie et al. 2016). Trait-based methods offer a promising solution to understand the high variability in seagrass meadows and the consequence of meadow dynamics on service delivery. Rather than characterising a seagrass meadow solely on the basis of their species composition, these methods focus on the combination of traits exhibited by the seagrass present in the meadow. This approach allows for a direct link between the characteristics of the seagrass meadow and its performance in response to environmental conditions.

Fig. 1.

Seagrass presence records in tropical Queensland from 1985 to 2018. Seagrass records are sourced from the TropWATER, James Cook University seagrass monitoring programs (Carter et al. 2021a, 2024).

Fig. 2.

Mixed-species seagrass meadows at Cape Pallarenda, Townsville (19°11′01.3″S, 146°45′53.7″E), showcasing trait diversity and the ecosystem services they provide. (a) General view of these intertidal seagrass meadows. (b) Clear and dense dugong feeding trails indicated with the orange bars. (c) A mixed meadow composed mainly of Halophila ovalis (oval-shaped leaves) and Halodule uninervis (thin needle-like leaves). (d) Halodule uninervis (thin, green needle-like leaves, indicated with a yellow circle) co-exists with Zostera muelleri subsp. Capricorni (brown strap-like leaves, indicated with red star) in this mixed meadow.

There are two significant knowledge gaps that need to be addressed to effectively apply trait-based approaches in tropical Queensland. First, establishing solid causal relationships between traits and their functional outcomes is crucial (Funk et al. 2017; Streit and Bellwood 2023). Second, there is a lack of comprehensive trait data for seagrass species, which has limited previous research, such as a large-scale coastal protection estimation (De Battisti 2021). Existing curated plant-trait databases, such as TRY Plant Trait Database (Kattge et al. 2020) and Austraits (Falster et al. 2021), have limited information related to aquatic plants, especially submerged seagrass species. Although the current trend of using trait-based approaches is increasing (Moreira-Saporiti et al. 2023), global or local seagrass trait data are hard to gather. Recently, two publicly accessible databases, namely Seagrass TraitDB (De los Santos et al. 2022) and a peer-reviewed paper focused on seagrass meadow structural traits (Strydom et al. 2023), have emerged. Seagrass TraitDB aims to integrate global empirical data for seagrass traits and is in a data-collection stage. Strydom et al. (2023) used a literature-based approach to gather information from global seagrass research to provide 11 traits at the genus level. Collecting traits at a genus level is useful for understanding general patterns or trends within a group of related species, but does not support finer-grained analysis of characteristics specific to individual organisms, including mechanisms that are crucial to resilience, such as reproductive success.

The objective of our study was to establish a robust knowledge base for the application of trait-based methods to tropical seagrass species in Queensland through the development of a functional-traits database. To accomplish this, our approach involved several key steps. First, we identified a comprehensive set of seagrass traits that have the potential to influence seagrass recovery, resistance, and ecosystem functioning. Subsequently, we used a combination of literature, database and botanical information searches and structured expert elicitation to collect data on the values of the identified traits. Expert elicitation allowed us to overcome gaps in published information by collecting knowledge systematically and transparently from experts (Martin et al. 2012; Hemming et al. 2018a). Through appropriate question design, the results obtained from expert elicitation can provide quantitative estimations with associated uncertainty (Adams-Hosking et al. 2016). It is common practice to supplement information when empirical published data are unavailable, as has been seen in fields such as conservation (Gallagher et al. 2021) and ecological modelling (James et al. 2010; Kuhnert et al. 2010). In our study, this approach allowed us to complement the information gathered from the literature search and further enhance our understanding of seagrass traits in a comprehensive and robust manner. Finally, we collaborated with existing platforms to make this set of seagrass traits readily available for further scientific research and management. By establishing this solid foundation of seagrass traits, we have facilitated future advancements in understanding and managing tropical seagrass ecosystems using trait-based methods.

Methods

The study extent covered the tropical Queensland coastline (Fig. 1), starting in the south at Gladstone, Queensland, Australia (23°50′32″S, 151°17′19″E), and extending to the western end at Burketown, Queensland, Australia (16°32′11″S, 138°00′12″E). We focused on seagrass species in shallow meadows because of limited information for deep-water meadows (deeper than 10–15 m). We included the following 13 well-defined seagrass species occurring in tropical Queensland: Cymodocea rotundata (Ascherson & Schweinfurth, 1870), Oceana serrulata (Byng & Christenh, 2018), Enhalus acoroides (Royle, 1839), Halophila capricorni (Larkum, 1995), Halophila decipiens (Ostenfeld, 1902), Halophila ovalis (Hooker f., 1858), Halophila spinulosa (Ascherson, 1875), Halophila tricostata (M.Greenway, 1979), Halodule uninervis (Ascherson 1882), Syringodium isoetifolium (Dandy, 1939), Thalassodendron ciliatum (Hartog 1970), Thalassia hemprichii (Ascherson, 1871), and Zostera muelleri subsp. Capricorni (S.W.L.Jacobs, 2006). Regarding species complexes, Halophila ovata, Halophila minor, and Halophila colesi/australis are grouped into Halophila ovalis because of their morphological similarities, often making them indistinguishable in the field surveys. Similarly, Halodule pinifolia is included in Halodule uninervis for the same reason (Carter et al. 2021a).

To maximise the usability of our tropical Queensland seagrass database, we follow the guidelines of creating a generic plant-trait database (Kattge et al. 2011). We developed a comprehensive list of targeted traits on the basis of their relevance to seagrass recovery, resistance, and ecosystem services, drawing insights from previous studies (Larkum et al. 2018; O’Brien et al. 2018; Johnson et al. 2021). This selection was guided by our research focus on understanding the dynamics of seagrass meadows in this region, their conservation purpose, and the value of the seagrass meadows through their ecosystem services. Seventeen traits were chosen and their definitions along with their relevance to recovery, resistance, or ecosystem services are detailed in Table 1. Additionally, these traits are cross-referenced with full trait names and TraitID in the TRY Plant Trait Database whenever applicable. In this format, our database facilitates accessibility for other researchers and mitigates the issue of unstandardised plant traits. A few traits were considered important in the seagrass ecology discipline but not included in the TRY Plant Trait Database; so, they were included in this project without providing the TraitID. Recognising that traits can exhibit variations within a single species owing to factors such as habitat condition, seasonality, or genetics (Albert et al. 2011; Bolnick et al. 2011), supplementary information associated with the data collection (e.g. date of collection, measurement method, and coordinates of the location site) is required to allow further examination of the traits. We reported the biotic and abiotic content of the collected trait when applicable.

We reviewed three types of existing seagrass-trait resources (published literature, trait database, and botanical information). First, a systematic literature search was performed using Scopus (https://www.scopus.com) and Web of Science (https://www.webofscience.com) to retrieve relevant articles published between 1980 and 2021. The search used a combination of keywords, namely ‘each seagrass species name’, ‘resistance’, ‘recovery’, ‘resilience’, ‘functioning’, and ‘each trait name’ (as listed in Table 1). The articles from the two databases were merged and duplicates removed. An initial screening was performed by assessing the title and abstract to remove studies associated with species not included in our study. Further screening was then conducted to categorise studies on the basis of their geographic region and measurement method. Priority was given to studies or measurements conducted in tropical Queensland, emphasising their inclusion in the database. Additionally, preference was given to data that recoded trait variation, such as seasonal fluctuations or variations in water depth, because these are valuable for capturing the breadth of seagrass traits. Second, the following trait databases were searched for relevant information: TRY Plant Trait Database (Kattge et al. 2020), Austraits (Falster et al. 2021), and Seagrass TraitDB (De los Santos et al. 2022). Trait data were extracted following the protocol of each database. Finally, we sourced botanical information from Flora of Australia (2024). There could be a potential bias associated with missing seagrass studies that are not reported in English. However, the area of interest for this study was tropical Queensland and all the published studies in Australia are reported in English.

An expert elicitation process was employed to gather information on functional traits and meadow properties that were not documented in existing literature. We used the IDEA protocol method developed by Hemming et al. (2018a), which comprises the following four stages: ‘investigate,’ ‘discussion,’ ‘estimate,’ and ‘aggregate’. The expert elicitation was conducted under the approval from the JCU Human Research Ethics Committee (H8879). The experts were identified as any person that had sufficient field-work experience of seagrasses in tropical Queensland. All experts were provided with an information sheet, and participation was completely voluntary (Supplementary File S1). The participants responded to the survey anonymously, but a unique ID was recorded to trace multiple responses from one person. Two general background questions were included in the survey, including, what sector the experts work, and their experience on seagrass.

To address the challenges of low response rate for particular species and the lengthy survey duration, we made slight modifications to the IDEA protocol in our study. The ‘investigate’ stage was conducted online by using an online survey platform SurveyMonkey (Momentive Inc., https://www.surveymonkey.com/). To make sure the experts share the same understanding of the purpose of the survey, we presented a detailed introduction with survey location maps, brief seagrass species information with figures, and a trait definition for each question (Fig. S1). In total, 28 trait/meadow properties from 13 species were included in the survey. Every question followed the four-step question format (Speirs-Bridge et al. 2010), which asked the participant to answer the highest plausible, lowest plausible, best guess, and confidence level. This question setup is intended to mitigate overconfidence in estimation, and confidence levels were applied to standardise material in the discussion section (Hemming et al. 2018b). The survey period spanned from 17 October 2022 to 10 March 2023. The experts were asked to base their answer on meadows they encountered along the tropical Queensland coast, from Gladstone to Burketown.

All expert responses were standardised using the formula outlined in the IDEA protocol (Hemming et al. 2018a). The standardised and anonymous results were presented in figures and shared with the participants in the ‘discussion’ phase to collect their feedback regarding the group average (Supplementary Table S1, File S3). Because of a high consensus among experts at the ‘discussion’ phase, the ‘estimate’ stage, which involved a second-round survey, was omitted. The ‘aggregate’ stage adhered to the instructions of the IDEA protocol and generated the final results.

Results

Table 2 presents the trait values obtained from a comprehensive review of the literature, trait databases, and botanical information. In total, 629 articles were initially retrieved from literature databases, and subsequent filters were applied to exclude studies conducted outside tropical regions, as well as those focused on seagrass-related organisms such as adhesive algae and bacterial communities. This initial filtering resulted in a final set of 59 articles. From these articles, 16 were selected for inclusion on the basis of the availability of trait measurements. In addition to peer-reviewed articles, trait data from trait databases along with botanical information were also included. Overall, 151 trait values were collected, with 39 sourced from botanical information and 117 from literature. Five trait values were derived from dual data sources to provide a value range. The Austraits database contained the leaf width and length values that are sourced from the Flora of Australia, as cited in Table 2. The ideal database comprises species trait values accompanied by measures of variation such as standard error and quantile. Nevertheless, owing to variable data quality, less preferred data such as genus-level data were included with additional annotation (Table 2).

In total, 16 experts were invited to participate in the survey, and we gathered five responses, all from researchers in academic institutions, with experience ranging from 5 to >20 years. Of the 28 questions asked, 23 questions received multiple responses, varying from two to five individual responses. The group-aggregated averages, derived from these responses, are presented in Table 3. On reviewing the group-aggregated average values (File S2) at the ‘discussion’ phase, the experts generally agreed that most of the elicited values were reasonable.

Discussion

Our project aims to establish a robust foundation for a trait-based approach to seagrass resilience and services research in tropical Queensland. In pursuit of this goal, we focused on 17 functional traits associated with recovery, resistance, and ecosystem services for 13 seagrass species in tropical Queensland. Of 221 traits, data were collected for 78% (174 traits values). Among the compiled traits, 23 trait values (13%) were obtained from structured expert elicitation, demonstrating the importance of expert knowledge as a source for ecological or conservation studies (Martin et al. 2012; Mallen-Cooper et al. 2020). By publishing, our results can assist future seagrass trait-based research in tropical Queensland and, potentially, regional comparison projects, given that the suite of species is common to much of the tropical Indo Pacific region. The missing 22% of traits also indicate important knowledge gaps and areas for further research and data collection on seagrass traits.

The collated trait values obtained from our study reflect the diversity of morphological and life-history strategies of seagrass species in tropical Queensland. For example, morphologically, the canopy height of E. acorodies is 28 times greater than that of H. decipiens. The life-history trait of low-light tolerance ranged from 13 days for H. decipiens to 133 days for T. hemprichii. This diversity of traits observed across species is an indication of the wide array of ecosystem function and services provided by seagrass meadows in tropical Queensland (Mason et al. 2005; Moreira-Saporiti et al. 2023). For instance, structural diversity, such as aboveground structure, plays important roles for supporting diverse fish assemblages, (Jinks et al. 2019; Jones et al. 2021). Additionally, biomass influences the foraging behaviour of dugongs (Tol et al. 2016) and determines the blue carbon storage capabilities of meadows (Lavery et al. 2013; Johnson et al. 2017). Furthermore, some species exhibit similar values in certain traits; for instance, H. uninervis, S. isoetifolium, and H. spinulosa all have biomass within the range of 60–70 mg/m². Trait redundancy indicates the resilience of this ecosystem because, even though species composition might alter because of interference, the overall functional diversity might stay the same if multiple species play a similar functional role in the community (Laliberte et al. 2010). In the case of tropical seagrass, the similar biomass of H. uninervis, S. isoetifolium, and H. spinulosa could stabilise the function provided by the meadow, even if one species decreases in number, because other species might take over the space and bring the meadow to a new state with similar functioning.

Despite our integration of the literature, trait databases, botanical information and expert elicitation, some species or traits still have limited information. Of the 13 species in this study, H. capricorni, H. decipiens, and H. tricostata had the least information available in the literature. Furthermore, in the expert elicitation, H. capricorni did not receive more than two expert responses for any of the questions. This species-specific lack of data is likely a result of the low abundance of these species in tropical Queensland (Waycott et al. 2004). Traits that can be measured by a single individual morphological property, such as leaf dimension or rhizome features, are more commonly documented in existing data and are typically easier for experts to respond to. In contrast, traits that involve biotic or abiotic factors, such as herbivory pressure, often have less available information in the literature or the responses from experts are more inconsistent (Table S1). For instance, the nutrient content of leaves presents challenges in data collection, with only one published study reporting this value, and the raw data being inaccessible, leaving only average values available for reference (Duarte 1990). Experts reported a lack of confidence in providing information on nutrient content. The lack of information reflects the further need for research into seagrass traits associated with nutrient content, low-light tolerance, and seed dormancy. Additionally, traits that influence interactions between herbivores and seagrasses are under-represented in the literature. These traits often required long-term measurements, such as, shoot formation rate and leaf elongation rate, or specialised tools, (e.g. leaf toughness; De los Santos et al. 2016), which limits available data. However, some of difficult-to-measure traits (hard traits) can be inferred from easier-to-measure traits (soft traits) (Hodgson et al. 1999). This approach, employed in previous research, showed that certain hard traits exhibit strong correlation with the other soft traits and this link offers a pathway to reduce the effort required for collecting hard traits, particularly physiological traits such as nutrient content and air-exposure tolerance (Belluau and Shipley 2018). We encourage future research to include a broader range of trait measurements and explore the relationship between soft and hard traits.

Our database has the potential to inform seagrass trait-based studies in other tropical regions that share the same species. For example, our database provides essential structural trait information for assessing the value of seagrass meadows in supporting fisheries production (Jones et al. 2021) or estimating coastal protection capacity (Christianen et al. 2013). When integrated with other databases, our local database provides useful information for broader-scale regional comparisons (Vandewalle et al. 2010). For example, assessing the impact of land-use changes on functional diversity across eco-regions or exploring functional diversity in habitats with size and connectivity variation worldwide (Jacquet et al. 2017). Future research endeavours should aim to explore causal relationships between functional traits and ecosystem dynamics. For instance, by quantifying recovery and resistance abilities of seagrass meadows on the basis of functional-trait values and meadow composition, we can elucidate meadow dynamics in a trait-based context. However, it is also important to recognise the limitation of a trait-based approach, particularly when interaction between organisms and the environment are driven by location-specific factors, such as epiphyte abundance (Balata et al. 2007). In cases where epiphytes contribute significantly to meadow recovery, traits alone may not fully explain resilience, and other methods may be required.

Various studies have demonstrated the importance of incorporating intraspecific trait variation in plant-trait studies (Bolnick et al. 2011; Westerband et al. 2021). Plants have the potential to vary in their traits according to environmental conditions or genetic composition, which can change the performance of the plant or affect the habitat it forms (Hughes 2014). In some cases, our database reported traits from literature that publish only a single value for each trait, such as rhizome elongation rate and shoot weight. This limitation has the potential to restrict the application of our database. To improve trait-data quality and its broader application in research, we encourage researchers to publish raw data or at least provide more comprehensive details when publishing trait information, such as minimum and maximum values, sample date, and sample size. One highly comprehensive approach for raw data is using the free and open-source platform, Open Science Framework (OSF), which encompasses data management plans, version control, and data sharing (Sullivan et al. 2019). It is also beneficial to condense raw data in a more compact format and provide it to other types of repositories that are available under an open Creative Commons licence, such as eAtlas in Queensland (https://eatlas.org.au/home) and PANGAEA (Felden et al. 2023).

Conclusions

Our project has provided a fundamental resource for implementing a trait-based approach for seagrass studies within tropical Queensland and other tropical areas that share similar species assemblages. Our primary focus was to understand seagrass resilience and the provision of ecosystem services, with the careful selection of 17 traits on the basis of their relevance to these processes. The innovative mixed-method approach we employed involved creating a plant-trait database by amalgamating literature, databases, botanical information and expert elicitation. This method allowed us to collect 78% of our targeted traits, affirming the cost-effectiveness of gathering trait information from experts when it is unavailable for specific species or traits in a published form. The uneven distribution of collected traits among species highlights knowledge gaps, underscoring the significance of further empirical data collection targeted to traits. Moreover, although our database is designed for tropical Queensland, it also serves as a valuable source for broader geographic-scale studies.

Supplementary material

Supplementary material is available online.