Towards a consistent benchmark for plant mycorrhizal association databases

Abstract

Mycorrhizal symbiosis, comprising functionally distinctive plant–fungus associations, mediates key plant population and community processes, and ultimately the functioning of terrestrial ecosystems (Tedersoo et al., 2020). It is estimated that c. 90% of the world’s vascular flora forms mycorrhizal symbioses with soil fungi (Smith & Read, 2008; Brundrett & Tedersoo, 2018). Although this general estimate is probably adequate, there is a severe shortage of empirical information about mycorrhizal associations at the plant species level, with only c. 5% of the world’s flora explored (Moora, 2014; Bueno et al., 2019b). Several database developments have emerged since the seminal work of Harley & Harley (1987; HH); extending both the number and the geography of plant species covered (Wang & Qiu, 2006, WQ; Akhmetzhanova et al., 2012, MID), and defining and describing some key mycorrhizal traits of plant species – ‘mycorrhizal type’ and ‘mycorrhizal status’ (Hempel et al., 2013, MF; Moora, 2014). Nonetheless, this expansion poses new challenges connected with the compilation of global data based on heterogeneous sources with different practical and conceptual frameworks (Bueno et al., 2019b; Kattge et al., 2020). Careful work developing consistent definitions and standardizing field and laboratory protocols is essential for harmonizing database content and avoiding critical inconsistencies (Pérez-Harguindeguy et al., 2013; Schneider et al., 2019). Recently, Soudzilovskaia et al. (2020) presented the largest compilation to date of empirical information about mycorrhizal associations in plants based on scientific literature (FungalRoot). Although this is a valuable and unique effort, we note three critical aspects that seriously hamper consistent data harmonization and should be addressed before the FungalRoot database can be considered as a standard reference in the field. Namely: (1) conceptual inconsistency in the designation of plant mycorrhizal associations; (2) incoherent application of plant mycorrhizal trait concepts; and (3) limited transparency in the incorporation of expert opinion. As we explain below, these issues appear particularly problematic in the cases of arbuscular mycorrhizal (AM) and nonmycorrhizal (NM) plants, and perhaps of lesser concern in the cases of other mycorrhizal types. However, given the high share of AM and NM plant species in the previous largest mycorrhizal database (73.1% and 18.0%, respectively, in WQ) and in the FungalRoot database (76.6% for AM, 15.5% for NM), these are critical issues to be resolved. FungalRoot (Soudzilovskaia et al., 2020) is a large compilation of plant mycorrhizal information with rich ancillary data, which potentially constitutes a valuable step forward. The database defines rules for plant mycorrhizal trait assessment (plant mycorrhizal types and statuses, defined in the next section), claims to apply them, and proposes that they be followed by future studies. FungalRoot defines AM and NM plant species based on the presence or absence of arbuscules (hereafter the arbuscule criterion; Table 1). However, a careful check of the criteria and definitions stated in the previous databases that feed into FungalRoot (HH, WQ, MID, MF), suggests that the new additional references included in FungalRoot do not match this criterion. All of the reference studies from previous databases and 79% of the new references included in FungalRoot (Fig. 1a,b) consider the mere presence or absence of AM fungi, irrespective of whether other mycorrhizal structures are apparent in plant roots, to define AM and NM plant species (hereafter the colonization criterion; Table 1). This criterion is also used in those studies using molecular approaches to detect AM fungi in plant roots, which will clearly increase in the future. Among the new references included in FungalRoot, only c. 5% can be diagnosed AM according to the arbuscule criterion, whereas c. 13% of references did not specify their mycorrhizal diagnosis criteria. Consequently, although the authors of FungalRoot define rules for determining AM and NM plant species, the criteria are not consistent with the large majority of sources in the database. This inconsistent application of criteria leads to misinterpretation of the original data and ultimately hampers the usability and credibility of the FungalRoot database. It should be stressed that we do not challenge the quality of the original data in the FungalRoot database; rather, we challenge the suitability of the arbuscule criterion for defining AM and NM plants when it is inconsistent with the majority of the database content. Therefore, we suggest a reconsideration of the criteria for defining AM and NM plants, and acknowledgement of the wide use of the colonization criterion. The database should maintain consistency with the source data and avoid unsupported assumptions about the functionality of mycorrhizal symbiosis (Jones & Smith, 2004; Bueno et al., 2019a; Albornoz et al., 2021) by applying the colonization criterion throughout, potentially with the arbuscule criterion applied separately when the data allow. There has long been confusion about the terminology used to describe different characteristics of plant mycorrhizal associations (Moora, 2014). Unfortunately, like many earlier authors, FungalRoot fails to clearly distinguish the plant mycorrhizal traits, ‘mycorrhizal type’ and ‘mycorrhizal status’. Mycorrhizal type refers to the identity of the symbiotic fungal partner (NM can be considered a special case; a plant with no fungal partners), whereas mycorrhizal status (obligately or facultatively mycorrhizal) is a distinct trait indicating the frequency of occurrence of mycorrhizal symbiosis in a plant species (Smith & Read, 2008; Moora, 2014). The criteria used to characterize the traits have been defined somewhat differently, depending on the author’s perspective. For example, Brundrett (2017), although not explicitly separating mycorrhizal type and status, defines facultative mycorrhizal species as ‘AM-facultative: Roots have AM in all cases but colonization levels are low and inconsistent (generally 5–40% of root length)’; whereas Smith & Read (2008) do distinguish mycorrhizal type and status and define facultative mycorrhizal species (on p. 28) as ‘Species which are sometimes, but not always, colonized are often referred to as ‘facultatively mycorrhizal’, to distinguish them from those ‘obligately mycorrhizal’ species that are consistently colonized’. We argue that these two traits describe different aspects of plant species’ mycorrhizal associations and should be clearly defined and considered separately. Indeed, although Soudzilovskaia et al. (2020) agree with the concept of facultative mycorrhizal plants, following Smith & Read’s definition, they do not define plant mycorrhizal status and often use it as a synonym of plant mycorrhizal type. It would only benefit the FungalRoot database and its potential users if the terminology of plant mycorrhizal traits were applied consistently. In fact, many empirical observations of AM plant species indicate that plant individuals present at a particular location or time do not associate with mycorrhizal fungi, while other individuals do. This observation does not challenge the mycorrhizal type designation, as the plant species is able to associate with AM fungi; rather, it indicates that the species is not obligately but instead facultatively mycorrhizal. Not applying this status concept can thus lead to misleading designations of mycorrhizal traits. For example, in FungalRoot many NM observations of AM plant species are challenged, for instance, but not exclusively, in Alopecurus aequalis and Arabis alpina. These species should be considered as facultatively AM species, given that both present individuals with AM and NM types. Observations of NM individuals among otherwise mycorrhizal species should not be systematically challenged unless evidence of misclassification or other clear reasoning is offered (see the next section on expert opinion). Expertise is a valuable resource, especially in areas where knowledge gaps are wide and sound empirical alternatives are unavailable. Uncertainty can be credibly minimized by collecting, organizing and structuring expert opinion and the corresponding bodies of evidence (Kaplan, 1992). However, expert opinion without evidence may hamper clear communication to an audience (e.g. by not clarifying the lines of evidence and logic used). This, in turn, can lead to insufficient scrutiny and potential bias, hindering further development (Burgman et al., 2011). The FungalRoot database relies on expert opinion at two levels, but in neither case is it accompanied by clear evidence and reasoning. First, expert opinion is used to assign plant mycorrhizal type and status to many genera (> 70% of genera assigned to a mycorrhizal category in FungalRoot; Supporting Information Table S2) where empirical data on mycorrhizal associations at the species, genus or even family level are lacking. In particular, 71% of genus assignments proposed (10 347 of 14 541 genera; see detailed list in Table S2) have no empirical support at species, genus or even family level. Thus, when empirical information was unavailable at plant species or genus levels, the authors relied on a previous taxonomic extrapolation of mycorrhizal traits applied to entire families (Brundrett, 2017): once a family-level mycorrhizal trait category is estimated by expert extrapolation, then all genera within the family are assigned the same category. This approach has been shown to produce estimates that diverge strongly from empirical measurements of AM, NM and facultatively mycorrhizal (FM) plant species (Bueno et al., 2019b). Yet, FungalRoot applies taxonomic extrapolation of the AM category at an even higher taxonomic level, using order-wide values where no data are available at the family level. The only study assessing the precision of taxonomic extrapolation at the family level suggested abandoning the practice (Bueno et al., 2019b), and even worse discrepancy can be expected at higher taxonomic levels (Bueno et al., 2019a). This problem is clearly critical in the case of AM and NM plants but may also introduce errors in the assignment of other mycorrhizal types. Therefore, if it is to be used at all, we advocate relying only on genus-level extrapolation of plant mycorrhizal type when there is empirical species-level evidence, but avoiding extrapolations at the order and family levels altogether. Thus, we strongly suggest avoiding use of the extrapolated genus-level assignments in FungalRoot (see Table S1), and instead to actively enhance efforts to investigate the roots of species, genera and families where little or no empirical evidence is available. It is also important to recognize that for the remaining plant genera in the FungalRoot species list (< 30%), where assignment was based on empirical evidence at the species level, the authors used a highly questionable approach to filter supporting studies. The authors considered a study to have low reliability, and thus excluded it, if > 20% of its records diverged from or ‘conflicted’ with other studies. Considering the low average abundance of records per plant species (2.4 records per species in FungalRoot) and provided that studies follow transparent, consistent and appropriate methodology, differences among studies in plant mycorrhizal trait assignments should not be considered to be conflicting unless they can be clearly demonstrated empirically. The practice of dismissing certain results because they disagree with findings collected elsewhere significantly increases the risk that the true natural dynamics of mycorrhizal associations are not captured. Finally, FungalRoot offers the opinions of three experts on specific records. Where these opinions disagree with a published record, this essentially represents a disagreement between experts: the original authors and the assessing expert. To solve such differences, the areas of disagreement along with the corresponding reasoning of the experts and the original author are needed. Unfortunately, the expert commentaries in FungalRoot lack transparent reasoning, with only brief statements of expert opinion provided. As such, this leaves database users uninformed when deciding whether to include or omit records: either blindly following unreasoned opinion or ignoring potentially valuable guidance. We therefore suggest adding the specific reasoning and logic behind each assessment, so that the opinion will be scientifically transparent and understandable to database users. Developing a robust understanding of the role of mycorrhizal symbiosis in ecosystems requires consolidated, high-quality global databases for both symbiotic partner groups. To unify and harmonize different sources in databases, it is essential to discuss, develop and follow strict conceptual baselines to avoid critical inconsistencies. There is an urgent need to accommodate alternative perspectives about plant mycorrhizal traits into a consistent and transparent reference database. Indeed, FungalRoot represents a helpful advance, but substantial limitations need to be carefully considered. In large part, these are connected with acknowledging diversity in mycorrhizal functioning, which is critical to deciphering the ecological role of mycorrhizal associations in changing and dynamic ecosystems. We have identified key aspects that can improve such plant mycorrhizal trait databases, but there remains an urgent need to collect empirical information about many poorly studied plant species. These communal efforts can greatly advance our knowledge of mycorrhizal ecology. This research was funded by the European Regional Development Fund (Centre of Excellence EcolChange) and University of Tartu (PLTOM20903). Table S1 List of references contained only in FungalRoot supporting AM and NM plant species, along with the assignation of their diagnosis criteria and rationale used for each reference. Table S2 List of genera with mycorrhizal type assignation presented in FungalRoot, without supporting evidence at the species, genus or even family levels. Please note: Wiley Blackwell are not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

Topics & Concepts