Silica phytoliths in angiosperms: phylogeny and early evolutionary history

Significant silicon accumulation and the formation of phytoliths (microscopic amorphous silica bodies) occur in various pteridophytes, gymnosperms and angiosperms, but vary greatly among taxonomic groups (Hodson et al., 2005; Piperno, 2006; Katz, 2014; Trembath-Reichert et al., 2015). Among the angiosperms, silicon and phytoliths are more abundant in members of the Poaceae (grass family), while in many other families phytoliths occur in smaller, sometimes even negligible amounts (Hodson et al., 2005; Piperno, 2006). Intra-familial variations of phytolith abundance are sometimes large (Katz, 2014), for example, reaching four or five orders of magnitude in the Asteraceae (Katz et al., 2013, 2014). The sources of this inter- and intra-familial variability are not fully understood (Katz, 2014). Silicon is beneficial to many plant species, providing them with mechanical support and protection from a variety of abiotic and biotic stresses, including water deficit, metal toxicity, pathogenic infections and herbivory (Richmond & Sussman, 2003; Cooke & Leishman, 2011; Guntzer et al., 2012). Recent studies suggest that some silicon-rich species partially substitute carbon with silicon, hence possibly allowing them to allocate carbon resources more efficiently (Schoelynck et al., 2010; Cooke & Leishman, 2012; Schaller et al., 2012). Thus, plant silicon accumulation and phytolith formation have been suggested implicitly (Cooke & Leishman, 2011; Katz et al., 2014) and explicitly (Schoelynck et al., 2014) to be a neglected functional trait. One of the most commonly discussed issues in the ecology of phytoliths is the dominance and proliferation of the silicon- and phytolith-rich Poaceae family in many open-habitat ecosystems, most notably grasslands (Beerling & Osborne, 2006; Bouchenak-Khelladi et al., 2009; Strömberg, 2011) which comprise c. 40% of the Earth's unfrozen land surface. Since the Oligocene (34–23 million years ago; henceforth, Ma) and into the Miocene (23–5 Ma), a drying climate (Beerling & Osborne, 2006; Strömberg, 2011) and the evolution of large mammalian grazers (Bouchenak-Khelladi et al., 2009; Strömberg, 2011) may have favoured greater phytolith formation in plants, owing to silicon and phytoliths providing plants with protection from and alleviation of the effects of these stresses. Vice versa, silicon- and phytolith-rich vegetation plays an important role in terrestrial and marine silicon cycles and the coupled carbon cycle (Street-Perrott & Barker, 2008; Li et al., 2011; Carey & Fulweiler, 2012; Song et al., 2012). Thus, grassland expansion and the evolution of large mammalian grazers probably accelerated silicon cycling (Falkowski et al., 2004; Vandevenne et al., 2013) and thus increased carbon sequestration (Street-Perrott & Barker, 2008; Li et al., 2011), reinforcing this climate change and feeding into an already complex positive feedback network (Beerling & Osborne, 2006). Thus, silicon and phytoliths play a multitude of significant roles in plant and global ecology. However, since most research in the field focuses on a small number of silicon- and phytolith-rich families (Katz, 2014) it is unknown whether silicon- and phytolith-rich taxa share an evolutionary history. Here, I review the occurrence or absence of phytoliths in families across the angiosperm phylogenetic tree and the emergence times of major phytolith-rich clades, and propose hypotheses as to what drove the early evolution of phytolith formation in angiosperms. To provide a good representation of as many angiosperm families as possible, I have combined mostly qualitative data from Piperno's (2006) extensive review of phytolith abundance in seed plants, to which I added information from Prychid et al.'s (2004) extensive review of the monocots and other regional-scale studies (Lanning & Eleuterius, 1983; Carnelli et al., 2001; Wallis, 2003; Tsartsidou et al., 2007). Families were classified into three categories: families in which phytoliths are rare or absent, families in which phytoliths are abundant, and families with mixed data. The latter category was used for families within which there is large variability or if there is no agreement between sources, following a similar category used by Piperno (2006). As a rule of thumb, in view of the qualitative nature of most available data and the categorization used here, phytoliths were considered to be abundant if they comprise > 1% of the plant dry weight or if there are > 1000 phytoliths per gram dry matter. Phylogeny, validity of families and assignment of families to orders or terminal clades followed the APG III (2009). Out of the 412 angiosperm families (Fig. 1; Supporting Information Table S1), in 42 families (10.2%) phytoliths are abundant, in 99 families (24.0%) phytoliths are rare, and for 29 families (7.0%) there is mixed data. No data were found for the remaining 242 families (58.7%), including 218 of 304 (71.7%) eudicot families, compared to only 4 out of 77 (5.2%) monocot families. Albeit restricting for analysis of eudicot clades, this lack of data highlights taxonomic representation biases in phytolith studies (Katz, 2014). Although phytolith-rich families occur across the entire angiosperm phylogenetic tree, and in all major clades, some large clades consist of more phytolith-rich or phytolith-poor families than others (e.g. Magnoliales, commelinids and Rosales for the former, Nymphaeales, Zygophyllales and Solanales for the latter) (Fig. 1; Table S1). Thus, phytolith formation is homoplasic among angiosperms, sharing deeper evolutionary origins (see also Trembath-Reichert et al., 2015). Three dating sources were used to study evolutionary history, fending for biases associated with both the fossil record (Crepet et al., 2004) and molecular dating (Rodriguez-Trelles et al., 2002). Magallon & Castillo (2009) present molecular dates calibrated against early fossil evidence for the crown groups of seed plants at 350 Ma and crown groups of Eudicots at 125 Ma (Fig. 2a), Bell et al. (2010) present molecular dates calibrated against a large fossil record (Fig. 2b), and Magallon & Sanderson (2001) and Friis et al. (2006) present data on the earliest fossil records of angiosperm orders (Fig. 2c). Magallon & Castillo's (2009) stem group ages (Fig. 2a) suggest that there was a pronounced increase in the emergence of orders that include phytolith-rich taxa during the Albian (113–100 Ma), which probably diversified mostly during the Late Cretaceous (Fig. S1). The Albian angiosperm diversification peak is largely attributed to phytolith-rich orders (e.g. Commelinales, Zingiberales and Poales). Bell et al.'s (2010) average ages (Fig. 2b) provide younger dates than Magallon & Castillo's (2009), suggesting that orders that include phytolith-rich taxa have emerged mostly during the Late Cretaceous (predominantly 85–75 Ma). Both dating sources show that while phytolith-poor orders emergence and diversification were largely completed by the middle of the Late Cretaceous, orders that include phytolith-rich taxa have continued to emerge and diversify into the Palaeocene and early Eocene (Figs 2a,b, S1). The fossil evidence presents a slightly different picture (Fig. 2c), in which orders that include phytolith-rich taxa tend to first appear in the fossil record during the Berriasian (145–139 Ma), Cenomanian (100–93.9 Ma) and Eocene (56–33.9 Ma). The Oligocene–Miocene expansion of grasslands and the phytolith-rich Poaceae family is often associated with declining CO2 concentrations, aridification and the evolution of large mammalian grazers possessing abrasion-adapted dentition (i.e. potentially adapted to phytolith-rich plants) (Beerling & Osborne, 2006; Bouchenak-Khelladi et al., 2009; Strömberg, 2011). Since some of these processes were already initiated during the later stages of the Eocene (Janis, 1993), they provide a possible explanation for the emergence and diversification of orders that include phytolith-rich taxa during the Eocene. However, the relevance of such processes to the Late Cretaceous is doubtful. First, although atmospheric CO2 concentrations have been declining since the Early Cretaceous (Berner, 2006), only during the Oligocene did they decline below plants' CO2 saturation point (Gerhart & Ward, 2010; Leakey & Lau, 2012). Second, at least 86% of epiphytic angiosperm species, mostly common in humid forests, belong to phytolith-rich clades such as the Epidendroideae (Orchidaceae) and Bromeliaceae (Zotz, 2013). Third, large mammalian grazers with abrasion-adapted dentition evolved only since the Oligocene–Miocene boundary (Bouchenak-Khelladi et al., 2009; Strömberg, 2011). Nonetheless, Late Cretaceous coevolution of plant silicon and herbivore dentition is plausible. Pteridophytes, and especially the Sphenopsida (horsetails), are known for their very high phytolith contents (Hodson et al., 2005). These were abundant during the Cretaceous (Channing et al., 2011), and probably comprised a significant portion of herbivorous dinosaurs' diets (Williams et al., 2009). During the Late Cretaceous, abrasion-adapted dentition has evolved in both Laurasian Hadrosaurideae (Williams et al., 2009; Erickson et al., 2012; Prieto-Marquez, 2014) and some early Gondwanan mammals (Verma et al., 2012; but see Schultz et al., 2014). Thus, coevolution of phytolith formation in pteridophytes alongside abrasion-adapted dentition in dinosaurs and early mammals could have triggered the radiation of phytolith-rich angiosperm clades that have emerged at approximately that time. It is plausible that earlier large herbivorous dinosaurs had similar effects on Sphenopsida and basal angiosperms since the Triassic (e.g. Chloranthaceae and magnoliids; Fig. 2a; Table S1); however, they did not possess a sophisticated abrasion-adapted dentition. Consumption of pteridophytes and other phytolith-rich plants by herbivorous dinosaurs and early mammals during the Late Cretaceous has possibly accelerated silicon cycling and enriched the soil with plant-available silicon (for a modern analogy see Vandevenne et al., 2013). High silicon-accumulating plant species take up more silicon if its availability is high (Mitani & Ma, 2005). Therefore, once species or taxa have evolved the ability to take up large quantities of silicon and utilize it for their own benefits they are expected to profit more from its greater availability. Likewise, Falkowski et al. (2004) suggested that the Tertiary coevolution of grasslands and mammalian grazers enriched the oceans with silicon and promoted the evolution and diversification of the silicon-rich diatoms. This hypothesis also fits Berendse & Scheffer's (2009) suggestion that angiosperms, which commonly have shorter life spans and faster decomposition rates than gymnosperms, have promoted their own radiation and diversification by accelerating nutrient cycling and increasing nutrient availability. In the case of the phytolith-rich angiosperms, abrasion-adapted herbivores may have simply been an additional catalyst for such processes. Phytoliths play multiple and significant roles in the plant species in which they occur, and have a long evolutionary history in angiosperms. This trait appears to be intimately related to angiosperm–herbivore coevolution and to the evolution of ecosystems and the silicon cycle for c. 100 million years. However, the large variability of phytolith abundance among angiosperms is little understood. The evidence I present here suggests that this trait did not become common among angiosperms until the Late Cretaceous, and its evolution during the Late Cretaceous may be related to the contemporaneous evolution of abrasion-adapted dentition in some herbivores. This association suggests direct plant–herbivore arms-race or large-scale effects of plant–herbivore interactions on silicon cycling and availability, which require empirical paleontological, physiological and ecological evidence. Understanding the broader aspects of how phytolith formation is involved in the evolution of plants, ecosystems and biomes has much to contribute to our understanding of plants and their past, present and future place in the world. The author thanks Simcha Lev-Yadun (University of Haifa, Israel), Andrew J. Tanentzap (University of Cambridge, UK) and two anonymous reviewers for their fruitful comments on the manuscript. This work was supported by the Israel Science Foundation (grant 103/11) and a Short-term Post-Doctoral Fellowship from the Kreitman School of Advanced Graduate Studies at Ben-Gurion University of the Negev. 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. Fig. S1 Crown group ages of angiosperm orders by phytolith occurrence. Table S1 Complete phytolith abundance data including molecular dates 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.