Research Functional biogeography of angiosperms: life at the extremes Amy E. Zanne1, William D. Pearse2, William K. Cornwell3, Daniel J. McGlinn4, Ian J. Wright5 and Josef C. Uyeda6 1Department of Biological Sciences, George Washington University, Washington, DC 20052, USA; 2Ecology Center and Department of Biology, Utah State University, Logan, UT 84322, USA; 3Evolution & Ecology Research Centre, School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney, NSW 2052, Australia; 4Biology Department, College of Charleston, Charleston, SC 29424, USA; 5Department of Biological Sciences, Macquarie University, Sydney, NSW 2109, Australia; 6Department of Biological Sciences, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, USA Summary Author for correspondence: Nonlinear relationships between species and their environments are believed common in Amy E. Zanne ecology and evolution, including during angiosperms’ rise to dominance. Early angiosperms Tel: +1 202 994 8751 are thought of as woody evergreens restricted to warm, wet habitats. They have since Email: [email protected] expanded into numerous cold and dry places. This expansion may have included transitions Received: 7 December 2017 across important environmental thresholds. Accepted: 9 February 2018 To understand linear and nonlinear relationships between angiosperm structure and bio- geographic distributions, we integrated large datasets of growth habits, conduit sizes, leaf New Phytologist (2018) 218: 1697–1709 phenologies, evolutionary histories, and environmental limits. We consider current-day pat- doi: 10.1111/nph.15114 terns and develop a new evolutionary model to investigate processes that created them. The macroecological pattern was clear: herbs had lower minimum temperature and precipi- Key words: angiosperms, conduit size, tation limits. In woody species, conduit sizes were smaller in evergreens and related to environmental limits and thresholds, growth species’ minimum temperatures. Across evolutionary timescales, our new modeling approach form, leaf phenology, macroevolution, found conduit sizes in deciduous species decreased linearly with minimum temperature limits. minimum temperature, nonlinearity. By contrast, evergreen species had a sigmoidal relationship with minimum temperature limits and an inflection overlapping freezing. These results suggest freezing represented an important threshold for evergreen but not deciduous woody angiosperms. Global success of angiosperms appears tied to a small set of alternative solutions when faced with a novel environmental threshold. change points occur is a vital question to be addressed in Introduction macroevolution and macroecology (Brown, 1984; Ogle & A plant’s structural characteristics determine the size of the spatial Reynolds, 2004; Tomkins & Hazel, 2007; Andersen et al., and temporal windows over which it experiences its local envi- 2009). However, current methods in comparative biology are ronment. Small ephemeral herbs may live their short lives poorly suited to detecting these transitions. Rather, most meth- between challenging events – fires, physical disturbances, ods either require researchers to discretize traits a priori (Lewis, droughts, or annual freezes – but such short lifespans preclude 2001; Felsenstein, 2012) or examine simple linear relationships tall stature. Tall woody trees are successful competitors for light between continuous predictors and traits (Butler & King, 2004; across much of the world, maintaining an aboveground presence Hansen et al., 2008). In this paper, we introduce a novel across years and changing environments (Schippers et al., 2001; approach to detecting nonlinear relationships between environ- Westoby et al., 2002). Running between these extremes is the ment and traits (i.e. abrupt changes in species distributions and breadth of morphologies that constitute our modern flora (Moles trait values) that allows researchers to test for change points and et al., 2009; Cornwell et al., 2014). The ability of plants with dif- directly estimate their values in an adaptive evolutionary frame- ferent body plans to tolerate changing environments determines work. where and when we see them. In considerations of species’ trait responses to environmental While much of the variation in critical traits underpinning pressures, an additional dimension in any comparative analysis is plant ecological strategies may be continuous (e.g. adult height), how to represent the environmental space a species inhabits other key properties contain obvious discontinuities. For exam- (Brown, 1984). In most cases, a given species exists across a range ple, transitions between woody and herbaceous stems, annual of spatial and temporal environments; the existence of range lim- and perennial life histories, and deciduous and evergreen habits its themselves suggests nonlinear responses of species to their are abrupt (even while intermediate forms exist) and suggest criti- environments (Whittaker, 1965; Bridle & Vines, 2007). There cal change points (thresholds) in how organisms adapt to a given has been debate as to the degree to which a species’ presence in a set of conditions. Understanding when, where, and why these given location is driven by average vs extreme environmental Ó 2018 The Authors New Phytologist (2018) 218: 1697–1709 1697 New Phytologist Ó 2018 New Phytologist Trust www.newphytologist.com New 1698 Research Phytologist conditions at that place (Gutschick & BassiriRad, 2003; Lloret Existence of such precise a priori biophysical limits to conduit et al., 2012; Coyle et al., 2013; Coyle & Hurlbert, 2016). How- size (at least with colder temperatures) with clear consequences for ever, focusing on distribution of a species, and not the species at a other structural traits allows us to set up tests at macroecological specific location, offers hope of synthesizing across biogeography, and macroevolutionary scales to examine the importance of ecolog- macroecology, and macroevolution. Biogeographers and other ical limits and change-point relationships in shaping trait–environ- modelers of species’ ecological niches focus on limits, not average ment relationships. First, we performed a set of macroecological conditions, to understand how species’ tolerances are shaped by analyses to detect predictors of herbaceousness and, for woody their structural features and further filtered by biotic interactions species, conduit size for species at limits of their biogeographic (Soberon & Nakamura, 2009; Araujo & Peterson, 2012). ranges and with different leaf phenologies. Building from these Macroevolutionary biologists, by comparison, often examine results, we explored macroevolution of conduit size in woody plants species’ average tolerances or environmental conditions (Felsen- in concert with leaf phenology and minimum temperature across stein et al., 2008). Focusing on modeling limits rather than species’ range limits. Both leaf phenology and minimum tempera- averages (or central tendencies) offers hope of resolving this ongo- ture help define the ‘adaptive regime’ for a given lineage, which is ing tension by focusing modeling efforts on the same properties a set of (admittedly incomplete) predictor traits mapped on the across ecological and evolutionary studies. phylogeny and used to reconstruct and assign lineages into discrete By exploring which species had minimum temperature range categorizations. These regimes attempt to capture the essence of limits crossing the freezing threshold, Zanne et al. (2014) identi- shifts in Simpsonian adaptive zones on phylogenies (Simpson, fied three functional traits that likely facilitated radiation of 1944; Hansen, 1997). However, when determining what predic- angiosperms (flowering plants) from their beginnings in warm, tors should be used to define regimes, it can be unclear what wet tropics into freezing environments. We focus on the same percentile of a species’ environmental range best captures the limit traits here: size of water-conducting conduits (average cross- to which species are responding. In combination with leaf phenol- sectional area of vessels and tracheids in wood), leaf phenology ogy, we therefore examined a range of minimum temperature (evergreen vs deciduous), and growth habit (woody vs herba- percentiles (from lower limits, when species are infrequent to ceous). The earliest angiosperms were probably woody trees with central tendencies,toupperlimits, as species are frequent at that evergreen leaves and a vascular water-transport network with minimum temperature) to best predict evolution of conduit size. larger conduits than their gymnosperm relatives (Sinnott & Bai- Then, using a novel macroevolutionary analysis, we tested the ley, 1915; Wing & Boucher, 1998; Feild et al., 2004). functional form of the relationship between minimum temperature Large conduits allow for fast flow rates, as flow increases to the and conduit size to determine whether freezing imposes a nonlin- second power of conduit cross-sectional area (and fourth power ear shift in tempo and mode of plant trait evolution. By framing of conduit diameter), but larger conduits have a greater risk of our approach in terms of limits of species’ niches, we provide an embolism formation (air bubbles blocking the vascular stream) in integrative description of ecological and evolutionary responses of freezing (as air comes out of solution), and possibly drought (as plant species to avoid or persist through environmental extremes. air