Introduction
The symbiotic association between roots and mycorrhizal fungi has fascinated botanists, mycologists and ecologists for centuries (Cairney 2000). In this symbiosis, plants host fungi inside or around cortical tissue in the root tips, providing habitat and C assimilates to the fungi, whereas fungi improve plant defense and nutrient and water acquisition (Fitter 2005, Selosse et al. 2015). Approximately 90% of all extant plant species engage in some type of mycorrhizal association, making this symbiosis arguably the most widespread and ecologically important mutualism in nature (Brundrett 2009, Maheraliet al. 2016). The origin of mycorrhizal associations predates plant colonization of terrestrial ecosystems. The ancestral association started when the Glomeromycota formed the first arbuscular mycorrhizal (AM) association with the primitive subterranean tissues of ancestral plants approximately 500 million years ago (Redecker et al. 2000, Field & Pressel 2018 but see Orchard et al. 2017).
Root morphological traits have evolved continuously since the initial colonization of terrestrial habitats. Currently, plants can show 20-fold differences in root diameter, root length per unit dry mass (specific root length; SRL) and root dry mass to root volume (root tissue density; RTD) (Guo et al. 2008, Comas & Eissenstat 2009, Comas et al. 2014). These traits are frequently used as proxies for strategies in resource acquisition and conservation in root systems, and the way they relate varies by ancestry and habit form, suggesting different evolutionary pressures shaping the interaction between traits in different groups of plant independently (Ma et al. 2018, Konget al. 2019). In addition, many species evolved particular structures such as rhizomorphs, beaded roots, root clusters, root hairs, dauciform roots and root nodules (Dohoux et al. 2000, Markmann & Paniske 2009, Datta et al. 2011), usually associated with the adaptation of plants to limiting environmental conditions (Lamberset al. 2006, Oliviera et al. 2015).
Historically, botanists considered switches to novel fungal symbiosis intrinsically associated with morphological changes in roots (Brundrett 2002). For instance, Pinales and Gnetales within the Gymnosperms, were the first to depart from the ancestral AM association, developing ectomycorrhizal (ECM) associations with Dikarya fungi during the Early Cretaceous (Wang et al. 2000), which corresponded with thinner roots and higher SRL than AM plants (Brundrett & Tedersoo 2018). In Angiosperms, there was greater diversification in mycorrhizal associations, with possibly 22 independent transitions to form associations with Dikarya fungi (ECM widespread in Eudicots and rarely Monocots, Brundrett 2002; Ericoid restricted to Ericales, Setaroet al. 2006). Many of the species in these groups too are characterized for having roots with high SRL, thin diameters and denser tissues (see Valverde-Barrantes et al. 2017 for a review on root traits by phylogenetic groups). Other angiosperm clades evolved to exclude the mycorrhizal association altogether (non-mycorrhizal, NM), notably within the Caryophyllales, but also in the Proteales, Poales, Brassicales, Dipsacales and Lamiales (Lambers et al. 2009). The evolution of the NM habit also appears to be associated with thinner roots, lower density and higher SRL (Laliberté 2016, Freschet et al. 2017).
Two alternative hypotheses have been formulated to explain this evolutionary decrease in root diameter. The “mycorrhizal” hypothesis suggests that climatic changes during the Cretaceous impeded tissue decay, increasing the proportion of nutrients trapped in organic compounds and forcing plants to acquire novel fungal associations (Comaset al. 2012). In particular, the transition from the ancestral AM to novel ECM fungal mutualism might have been caused by the ability of ECM fungi to produce enzymes that directly degrade organic matter (Brzostek & Finzi 2011) and increase net N mineralization (Phillips & Fahey 2006), which is not present in AM fungi. Since ECM fungi do not require cortical tissue and their growth is limited to root tips, the adoption of an ECM affiliation may have accelerated the reduction in root cortical tissue, as well as increased SRL and root tip abundance (Read & Perez‐Moreno 2003, Chen et al. 2016). Further, in some groups the reduction in cortical area extended to the point that some species abandon mycorrhizal affiliations altogether. Thus, the evolution of high SRL, thin root systems appeared as the results to the acquisition of novel fungal affiliations that required less cortical tissue, with the additional benefit of more efficient soil exploration (Eissenstat et al. 2000).
The “integrated” hypothesis proposes that changes in root morphology could be a consequence of changes in physiological leaf demands (Konget al. 2017) and the emergence of new growth habits (Freschetet al. 2017). During the radiation of angiosperms in the Cretaceous, leaf venation and foliar area increased exponentially due to the drop in CO2 levels (Feild et al. 2011). Some plant clades evolved leaves with higher vein density and stomatal abundance, allowing a 174 % increase in photosynthetic rates in angiosperms relative to gymnosperms (Brodribb et al. 2010). Although increased carbon assimilation rates made Angiosperms more competitive, it also incurred a higher demand for water and nutrients. Kong et al. (2017) hypothesized that increases in water expenses selected for a reduction in the relative proportion of cortical to vascular tissue, resulting in thinner roots, higher SRL and higher proportion of vascular tissue, increasing RTD (Li et al. 2015). Thinner roots with a reduced cortical area likely have less impedance to water movement (Eissenstat & Achor 1999), thus providing greater hydraulic conductivity (Huang & Eissenstat 2000; Solari et al.2006, Hernández et al. 2009) that could compensate for the increasing demands of higher photosynthetic rates. Moreover, many groups of Angiosperms also lost the ancestral woody habit as a strategy to avoid new climatic extremes during the Late Cretaceous (Zanne et al. 2014), which is also associated with thinner roots, greater SRL and lower RTD (Ostonen et al. 2007, Prieto et al. 2015)
Both hypotheses suggest that climatic changes during the time of Angiosperm diversification influenced the evolution of root systems and mycorrhizal associations in seed plants. Nonetheless, the order of these evolutionary changes is not clear. It remains unknown whether reduced diameter and increased SRL appeared independently of the mycorrhizal state of ancient plants but in coordination with changes at the entire plant level (Valverde-Barrantes et al. 2017), or whether mycorrhizal affiliation played an active role in root morphology changes (Baylis 1975, Comas et al. 2012, Chen et al. 2013, Maherali 2014). In addition, higher RTD is usually associated with ECM roots whereas low RTD is linked to NM species, but little evolutionary context have been provided to these observations (Freschet et al.2017, Valverde-Barrates et al. 2018). Although it has been suggested that the evolutionary associations between mycorrhizal state, leaf function and root morphology could have occurred simultaneously (Comas et al. 2012, 2014, Kong et al. 2017), this is unlikely. Fossil calibrations and ancestral reconstruction studies suggest that vein densities in angiosperms increased from 3.3 to ~ 8.6 mm mm-2 during the Late Cretaceous (90-110 MYA, Hickey & Doyle 1977, Brodribb et al.2010). In contrast, fossil records from ECM roots date back to the Lower Eocene from both Gymnosperm and Angiosperm plants (60-50 MYA, LePageet al. 1997, Beimforde et al. 2011). Similarly, typical NM families such as Brassicaceae, Caryophyllaceae, Cyperaceae, Dipsacaceae, Proteaceae, and Ranunculaceae evolved in the last 50 MYA (Hill & Brodribb 2006, Magallon et al. 2015, Bell et al. 2010, Iles et al. 2015, Hohmann et al. 2015). Unfortunately, equivalent studies comparing changes in root morphology based on fossil records have proven elusive due to the lack of reliable root fossil data, paucity of cortical tissue preservation and the impossibility to identify root fossils to species level in most cases (Heteringtonet al. 2016).
In this study, we test the “mycorrhizal” and “integrated” hypotheses based on the extant patterns of leaf, root and mycorrhizal information of seed plants (Brundett 2002). If A) morphological root trait evolution was associated with changes in mycorrhizal symbiosis, we predicted A.1) the evolutionary changes in diameter and SRL must correspond with the transition from the ancestral AM to other mycorrhizal groups and A.2) those changes in root traits occurred later than the changes in leaf vein density. Alternatively, B) morphological root trait evolution was driven by changes in aboveground strategies, thus before major switches in mycorrhizal affiliation. In this case, we predicted that a shift to B.1) the ECM state or non-mycorrhizal state appeared later, after the formation of thin, high SRL and denser root systems and B.2) shifts in aboveground traits were concomitant with root changes. To test these predictions, we reconstructed the evolutionary history of root traits, habit, leaf vein density, and mycorrhizal state using a ∼600 taxon database and tested for associations among these variables using phylogenetically informed statistical analyses. Since changes in mycorrhizal associations vary largely among seed plant clades (Brundrett 2002, 2009), the analysis was done across all seed plants and within their major clades (gymnosperms, magnoliids, monocots, rosids, and asterids).