New Idea The soil mosaic hypothesis : a synthesis of multi-trophic diversification via soil heterogeneity

Myriad unexplored mechanisms potentially drive ecological speciation and could help explain global variation in diversity. Here, we develop a novel hypothesis focused on variation in biotic, chemical, and physical properties of soil as a factor contributing to diversification in communities of plants and animals. The Soil Mosaic Hypothesis (SMH) suggests that differences in soil attributes can affect intraspecific variation in phytochemistry, leading to cascading ecological and evolutionary effects on higher trophic levels. To illustrate the potential importance of the SMH, we examine three underlying ideas: (1) plant species and species assemblages shift over time, exposing them to novel soil environments, which can lead to genetic differentiation; (2) differences in soil properties can alter phytochemistry via plasticity and local adaptation; (3) phytochemistry can drive herbivore diversification via divergent natural selection (i.e. ecological speciation). The SMH provides insight into the process of diversification in a variety of landscapes and at a variety of scales and may inform analyses of diversification at local, regional, and global scales.


Introduction
The idea that ecological interactions can influence evolution has been a major component of evolutionary theory since Darwin developed the concept of natural selection (Darwin 1859), and it is clear that ecologicallybased divergent natural selection is an important model of speciation (reviewed by Rundle andNosil 2005, Nosil 2012).Ecological speciation occurs when populations are exposed to contrasting environments and ecologicallybased divergent natural selection promotes (either directly or indirectly) the evolution of reproductive isolation.Divergent selection can result in differences in morphology, behavior, or other phenotypes, potentially leading to genetic isolation between populations comprised of ecologically specialized individuals (Schluter and McPhail 1992, Rundle and Nosil 2005, Matsubayashi et al. 2010, Soria-Carrasco et al. 2014).Research on ecological speciation has benefited from detailed studies of well-understood populations or taxa associated with different resources or habitats (e.g., Schluter and McPhail 1992, Rundle et al. 2000, Nosil et al. 2008, Matsubayashi et al. 2010, Nosil 2012, Soria-Carrasco et al. 2014), but there are still many unexplored axes of ecological variation that potentially contribute to diversification (Coley et al. 1985, Thompson 2005, Whitham et al. 2008, Dyer et al. 2014).Here, we consider a previously understudied potential source of divergent natural selection: variation in soils and associated effects on phytochemistry and plant and animal community structure.
The Soil Mosaic Hypothesis (SMH) posits that differences in soil properties (i.e.biotic, chemical and physical factors) can affect individual variation in plant primary and secondary metabolites, yielding a highly variable phytochemical landscape (sensu Hunter 2016) and leading to cascading ecological and evolutionary effects on autotroph and consumer trophic levels.Soil heterogeneity could lead by direct and indirect mechanisms to reproductive isolation in plant and herbivore populations.A direct mechanism could be, for example, adaptation by herbivores to plant populations with divergent phenology causing a shift in herbivore phenology (emergence time or peak abundance) that immediately isolates consumer populations.A less direct mechanism could be local adaptation to phytochemistry, and associated selection against hybrids that are maladapted to either of the chemical profiles experienced by the parents.The SMH is not entirely novel: in addition to theories of ecological speciation and host-associated differentiation (Stireman et al. 2005), the SMH can be considered a corollary to the following well-established theories: i) coevolution (Ehrlich and Raven 1964, Berenbaum and Feeny 1981, Agrawal et al. 2012), ii) the geographic mosaic (Thompson 1999, Thompson 2005), iii) plant defense theory (Bryant et al. 1983, Coley et al. 1985, Stamp 2003, Massad et al. 2011, Smilanich et al. 2016), iv) effects of environmental heterogeneity on communities and ecosystems (Ricklefs 1977, Whitham et al. 2006, 2008), v) the phytochemical landscape (Hunter 2016), and vi) plant-soil feedbacks (van Breeman and Finzi 1998, van der Putten et al. 2013, Schweitzer et al. 2014).The utility of the SMH is that it utilizes key components of existing theory that are usually limited in scope to average levels of chemical defense and bitrophic interactions to provide a focused, testable framework that includes a new perspective on phytochemical diversity, multi-trophic interactions, and abiotic selective drivers of diversification.Natural systems are adaptive landscapes of complex community dynamics.The SMH integrates both below-and above-ground processes when assessing patterns of ecological divergence and speciation.We examine three underlying postulates (Figure 1): (1) plant species and communities are exposed to diverse soil environments across multiple spatial and temporal scales, which can lead to divergence in plant populations; (2) differences in soil properties can alter plant primary and secondary metabolites; and (3) variation in phytochemistry can drive herbivore diversification via ecological speciation at fine geographic scales.
While these postulates could be used to help understand the link between soils, phytochemistry, and diversification in any ecosystem, such mechanistic relationships may be particularly evident in regions that encompass extreme habitat variation (e.g., serpentine soils, white-sands versus clay soils, dry versus wet tropical forests, or mountains) and which have a unique combination of soil diversity, movement dynamics, and biotic interactions.Below we will discuss each of the three aspects of the SMH to illuminate the process by which consumers adapt to phytochemical variation driven by changes in soil chemical and physical properties.

Plants disperse to novel soil environments.
Because the chemical and physical properties of soil are influenced by associated climate, parent material, topography, time, and biotic communities (Laliberté et al. 2013, van der Putten et al. 2013), distinct soil types can occur in close proximity, leading to a soil mosaic (Sollins et al. 1994).Soil mosaics characterize many landscapes at different spatial scales, and soil formation models predict that tropical soils in particular should have extreme heterogeneity in soil properties (e.g., Jenny 1980).
When plant ranges shift due to climatic changes or other factors, populations are likely to encounter novel soils because of the heterogeneity of most soil landscapes.Ecological processes that promote dispersal will also lead to an increased likelihood that a plant will The lowland tapir, Tapirus terrestris, acts as an important long distance disperser of the tropical palm Maximiliana maripa.Tapirs disperse palm seeds as far away as 2 km from parent trees, increasing survival rates of seeds to 98% for those that are dispersed compared to 17% for seeds close to parent trees (Fragoso 1997;Fragoso et al. 2003).Long distance dispersal events increase the probability of offspring shifting to a different soil type than the parent and could result in offspring with different phytochemical or nutritional properties from their parents, leading to increased morphological and genetic divergence (e.g., Barbosa et al. 2013, Misiewicz andFine 2014).

Soils affect phytochemistry.
Changes in biotic and abiotic factors, including soil microbes and nutrients, are known to cause significant changes in plant chemistry, and the magnitude of these changes are likely to affect many biotic interactions that are mediated by chemistry (Hunter and Price 1992, van Breeman and Finzi 1998, Dyer et al. 2004, Massad and Dyer 2010).There is a vast literature on the plasticity of plant secondary metabolites, and while there is little consensus on the directions and magnitudes of these responses to changes in soil nutrients, there is ample evidence of large shifts in phytochemical profiles in response to soil nutrient variation (reviewed by Massad et al. 2012).
Phytochemical diversity maintains plant function and fitness across diverse environments, affecting herbivore communities (Richards et al. 2015, Glassmire et al. 2016) and habitat specialization by plants (Fine et al. 2013).Phytochemical plasticity could be more beneficial when there are resource pulses or outbreaks of herbivory in resource-limited environments, and empirical studies have shown that plants allocate more resources to defense versus growth under such conditions to prevent the loss of leaf tissue, which is costly to replace (reviewed in Endara & Coley 2011).Thus, the adaptive nature of phytochemical plasticity can depend on resource availability and the presence of biotic interactions (Coley et al. 1985, Dyer et al. 2004, Hunter 2016).Similarly, the growth-defense trade-off hypothesis suggests plants have different phytochemical defense strategies across soil gradients (Coley et al. 1985), potentially promoting soil specialization and adaptation over evolutionary time (Fine et al. 2013).
Thus, soil mosaics can provide an adaptive landscape promoting edaphic specialization and plant diversification.For example, patterns of spatial genetic structure in the tropical tree Protium subserratum (Burseracae) in the Ducke Reserve Brazil are significantly influenced by soil type, which is highly heterogeneous, with soil clay composition ranging from 2% to 80% in a 250 m 2 area (Barbosa et al. 2013).Edaphic specialization has been posited as a mechanism of diversification for multiple plant lineages; two prominent examples are diversification of Protieae species (Burseraceae) shifting from clay to sand soils (Fine et al. 2005, Fine et al. 2014) and endemism of streptanthoid species (Brassicaceae) transitioning from bare to serpentine soils (Cacho and Strauss 2014).Furthermore, studies have shown that some populations of Mimulus guttatus monkeyflowers (Family: Phrymaceae) have adapted to the copper-rich soils near copper mines, resulting in ecological speciation (Macnair and Christie 1983).

Phytochemistry affects herbivore diversity.
While diversification in herbivorous insects is often thought to involve shifts in host plant use (e.g., Powell et al. 2013, Soria-Carrasco et al. 2014), recent evidence from a diverse tropical system raises the possibility that diversification can also occur without host shifts and also without major geographic barriers, supporting the hypothesis that intraspecific variation in phytochemistry may play a role in insect diversification at relatively small spatial scales (Wilson et al. 2012, Glassmire et al. 2016).Furthermore, intraspecific variation in edaphicassociated phytochemistry sheds light on the documented phenomenon that distinct insect communities are associated with soil ecotypes, as in P. subserratum (Fine et al. 2013).It is well known that phytochemical variation can influence insect herbivores by affecting oviposition preference (Carlsson et al. 2011), larval performance, mortality (Richards et al. 2010), and the ability of an herbivore to defend itself against predators and parasitoids (Smilanich et al. 2009).The SMH suggests that as plants experience new soil environments and respond with altered phytochemical properties, associated herbivore communities will experience strong selection pressure based on these new phytochemical environments, which can lead to diversification and speciation.Richards et al. (2015) found that phytochemical variation affects entire host-associated communities, including the diet breadth and diversity of herbivores.

Future Studies
The three main tenets of the SMH described here were inspired by pondering the theoretical framework of the phytochemical landscape (Hunter 2016) as well as considering our own work showing evidence of rapid diversification within one genus of herbivores (Eois, Geometridae: Larentiinae) that includes multiple sister species consuming the same host plant species in close geographic proximity (Wilson et al. 2012, Glassmire et al. 2016).The SMH incorporates a combination of ecological and evolutionary processes associated with plants colonizing novel soils, followed by diversification of taxa at higher trophic levels.Below we provide several examples of future studies that would test specific hypotheses generated by the SMH to elucidate how soil interacts in a multi-trophic framework.
First, transplant studies should be conducted to examine how differences in soil nutrient availability influence phytochemical profiles, and how this impacts performance of the associated arthropod communities (Fine et al. 2013).These studies should be accompanied by feeding assays to examine herbivore preference and performance on phytochemically distinct plants.Second, controlled experiments should investigate the mechanisms by which soil resource availability affects phytochemical plasticity in the presence of natural enemies.This would involve a fully-crossed experimental design including manipulated abiotic (addition of soil resources) and biotic factors (exclusion of herbivores and natural enemies).Associated with these manipulations of soil resources, the richness and abundance of soil biotic properties could be manipulated to examine the influence of soil biotic diversity on phytochemical diversity.For example, one could experimentally alter the diversity of arbuscular mycorrhizal fungi, bacteria, and soil arthropods, and quantify differences in phytochemistry.One important response variable for both types of soil manipulation experiments is the concentration of individual secondary metabolites, allowing for responses to these key questions: 1) How important are soil resources, arthropod communities, and the interaction between these factors for structuring plant secondary metabolomes?2) What are the norms of reaction for individual secondary metabolites and phytochemical diversity in response to soil nutrients and arthropods? 3) Are any metabolites fixed with respect to variation in soil and arthropods?In experiments of this kind, arthropod communities can both be manipulated as treatments and measured as response variables, including behavior (e.g., oviposition preferences) of focal herbivores.
Finally, as an extension of the SMH, future studies could investigate the possibility of plant defensive profiles becoming fixed by genetic assimilation (Waddington 1953, Crispo 2007).If the colonization of a novel edaphic environment results in the developmentally-plastic production of a distinct phytochemical profile that is favored by natural selection, theory suggests that the novel phenotype could eventually become fixed because plasticity to produce that phenotype would be selected against at that location.An interesting outcome of this process would be the retention of phytochemical diversity at the species or metapopulation scale (Figure 1), even in the face of range shifts associated with climatic fluctuations, since the previously plastic phenotype is fixed.The conversion from plasticity to fixed phytochemical diversity also opens the possibility for another mechanistic component of classic coevolutionary dynamics between plants and herbivores.While the potential importance of genetic assimilation for micro and macroevolutionary processes is well recognized (West-Eberhard 2003, Ehrenreich and Pfennig 2016), we know very little about the potential for assimilation to affect phytochemical phenotypes or associated arthropod communities.Perhaps the best systems for utilizing this approach would be wellresolved foundation species, such as Populus, for which there are documented networks of interacting soil microbes and herbivore communities, as well as documented effects from genes to ecosystems (Whitham et al. 2006, 2008, Lau et al. 2016).

Conclusion
Variation in phytochemical profiles can arise in plants following the colonization of novel soil types (Fine et al. 2006, Fine et al. 2013, Cacho and Strauss 2014), allowing for phytochemically-associated adaptation and divergence to occur in herbivores at a fine geographic scale (Glassmire et al. 2016).Future studies investigating previously unrecognized mechanisms of diversification, such as the processes comprising the SMH, will shed light on the origin and maintenance of biodiversity.Testing this hypothesis should be a part of the general goal to understand the extent to which ecological processes influence diversification in a multi-trophic framework.

Figure 1 .
Figure 1.A graphical model of the Soil Mosaic Hypothesis: 1) Soils are heterogeneous and plant populations move;2) Soils affect phytochemistry, creating subpopulations with different chemistry; 3) Phytochemistry affects herbivore diversification.The first column shows that plants (open circles) move across a landscape over time.The second column shows how soils with distinct characteristics (shaded regions) influence plant phytochemistry.Plastic changes in phytochemistry are represented by different symbols within each circlethese changes can also be followed by genetic assimilation.The third column shows how herbivores track the movement of plant subpopulations over time.The pattern on the wings of the herbivores represents divergent characters that are linked to adaptations to unique phytochemical profiles of associated host plants.