Is experimental evolution relevant for botanical research?

Most researchers in botany acknowledge the importance of evolution in shaping plant traits, communities, or interactions with other organisms, but nevertheless implicitly assume that ongoing evolutionary change is not fast enough to impact the outcome or repeatability of their research. Although this may be true for some studies, especially those focusing on macroevolution, for others it is less clear, because more and more research shows that evolution can be rapid in some cases, especially when selection is strong and the generation time of the evolving organisms is short (Agrawal et al., 2012; Franks et al., 2016; Ramos and Schiestl, 2019). Therefore, single-generation studies, that merely capture a single frame out of the “evolutionary footage”, may fall short of documenting the evolutionary dimension, which is one of relentless change. For example, in community ecology experiments lasting for several plant generations, drift and selection may have changed the (epi)genotypic composition of the experimental populations, hence having an effect of the outcomes and repeatability of the experiment (van Moorsel et al., 2019). When studying floral traits including rates of selfing, one should be aware that they may evolve rapidly when pollinator communities change (Gervasi and Schiestl, 2017). Lastly, for decisions to be made in terms of how threatened plant populations can be conserved, their potential to adapt to changing environmental parameters may be a key consideration (Jump and Penuelas, 2005). I argue that we should more often consider the opportunity for addressing real-time evolution in our research, by including an evolutionary framework in our experimental design, and build in the evolutionary dimension in ecological- and plant conservation models.

There are different ways to study real-time evolution directly or by incorporating it into experiments with another primary focus. Perhaps the most commonly used way is to use comparative approaches between populations or long-term studies without manipulating environmental parameters. Resurrection experiments, done by growing stored “historical” seeds together with the recent ones (Franks et al., 2016) or comparing genomes of herbarium samples with those of recent populations (Kreiner et al., 2022), provide additional approaches for assessing intermediate- to short-term evolutionary changes. Experimental evolution, defined as “the study of evolutionary changes occurring in experimental populations as a consequence of conditions (environmental, demographic, genetic, social, and so forth) imposed by the experimenter” (Kawecki et al., 2012) strives to identify “natural” evolution by using realistic ecological factors (as opposed to studies with artificial selection), in natural, or semi-natural greenhouse-based conditions. The advantage of this approach is that it usually allows one to identify the causative factors for evolutionary change.

Whereas experimental evolution has proven highly successful, it has mostly focused on microbes and animals, especially insects, whereas plants have rarely been the focus of this research avenue (Kawecki et al., 2012). This matters for botany because results obtained with microbes and animals cannot easily be extrapolated to plants, because plants, with their autotroph feeding, sessile lifestyle, and unique pollination- and mating systems, evolve under different kinds of selection and face different constraints and trade-offs than those of mobile, heterotroph organisms, or those with more simple phenotypes like microbes. Long-term experiments with plants have mostly focused on aspects of plant community change with environmental parameters, and evolutionary questions therein being more a research side-line (van Moorsel et al., 2019). Only a handful of studies, however, have performed experiments that specifically targeted “real-time” evolution in plants (see Appendix S1).

Most likely, the scarcity of experimental evolution studies in plants is due to their comparatively long generation time, making evolution experiments last comparatively long, even when experiments run for only few generations.

Despite this challenge, plants provide powerful models for evolutionary experiments, because they are easier to work with than their animal counterparts in several aspects. For example, the lack of mobility in plants allows for an easy assessment of fitness components (e.g., seed set or survival) and enables the setup of experimental plots in the field for trans-generational research (Agrawal et al., 2012). The presence of seeds in spermatophytes or spores in ferns and mosses makes it possible to store a population of individuals, often for many decades, to use them as a backup of intermediate generations during an experiment, or for resurrection experiments towards the end of the experiment. Plants also have great variability in mating systems, impacting parameters of population genetics as well as the practicability of an experiment. In obligatory selfers such as Arabidopsis thaliana (L.) Heynh. [Brassicaceae], no pollen vector is needed, but there is also no effective sexual recombination, hence the response to selection is a sorting of genotypes among the ones initially present in the starting population of the experiment, making it easier to quantify genotypic changes across generations (Züst et al., 2012). Outcrossing plants, in contrast, need pollen vectors for sexual reproduction, and have effective sexual recombination, higher heterozygosity, less linkage disequilibrium and potentially higher adaptability (Lucek and Willi, 2021). Because animal pollinators as well as abiotic pollen vectors such as wind are strong selective factors in plants (Gervasi and Schiestl, 2017; Tonnabel et al., 2022), this additional layer of selection makes evolution experiments with outcrossing plants fascinating, yet more challenging in their setup. Naturally, variation in pollination and mating system is also a fascinating research topic that has recently successfully been approached with real-time evolution experiments, showing for example, rapid evolution of selfing and sex allocation (Dorken and Pannell, 2009; Roels and Kelly, 2011), as well as divergent evolution after pollinator switch (Figure 1; Gervasi and Schiestl, 2017). In botany, there is wide scope for studying genes, genomes, traits, species, interactions, and even ecosystems in the view of rapid evolution. How plants adapt to environmental parameters, and how trade-offs, constraints, or pleiotropy impact this process is of interest for both basic and more applied research. Importantly, current environmental change is pertinent due to warming climate, intensified agriculture, deforesting, increased fragmentation, declining pollinators, etc. It is safe to assume that all these parameters will impact natural selection and thus adaptive evolution in nature. Therefore, it is timely to address adaptive evolution experimentally, and develop appropriate models for predicting its outcome. Such experiments can be set up in a greenhouse, allowing for great control of environmental factors, and thus great power of recognizing causative factors in evolution. Alternatively, plots can be set up in experimental fields, providing a superior natural setting, yet less easy identification of specific factors causing evolutionary change. A key question for any experimental approach is whether evolution in an artificial or manipulated ecosystem, either in the greenhouse or in the field, effectively mimics evolution in natural ecosystems.

A particularly timely example for the need to address evolutionary questions is the science of plant conservation (Olivieri et al., 2016). Models of adaptability are deeply desired in conservation biology, to predict which populations of rare plants will be able to adapt, and which need measures like assisted migration for survival (Jump and Penuelas, 2005). Rapid adaptation is also of concern in invasive species, that may threaten rare natives (Alexander, 2013). Experimental evolution can act as a kind of test for adaptability or evolutionary rescue (Gonzalez et al., 2013), as done in alpine species of Drosophila, showing a lack of ability to adapt to warmer temperatures (Kinzner et al., 2019). This approach is even more powerful when combined with genomics, in an “evolve and re-sequence” approach. In evolve and re-sequence experiments, the adaptive potential of experimental populations is tested in response to specific factors such as temperature change or drought. Subsequently, plant populations before and after evolution are genotyped and changes in frequencies of specific alleles can be assessed. Such tests of adaptability and assessment of adaptive genetic variation can yield powerful models of variation in specific genes that enable adaptation. This approach has been used in Drosophila to establish that adaptability to harsh environments is correlated to genome-wide variability, rather than patterns of inbreeding (Orsted et al., 2019). Similarly, Stelkens et al. (2014) showed in yeast that hybridization enables evolutionary rescue and adaptation to degraded environments.

Real-time evolution in botany may gain more momentum in the future by the development of new fast-cycling model plant species, analogous to the rapid cycling Brassica [Brassicaceae] species bred by Paul Williams a few decades ago. This would also help to address a greater variety of research questions in a real-time evolution context, because a few “model” species are not enough to represent the diversity of biological functions found in plants. Developing new fast-cycling models would, for example, require artificially selecting for short generation time in an annual plant, yet maintaining genome-wide genetic variability, to allow for sufficient adaptability in subsequent experiments. Besides having an annual life cycle, such plants should be easy to grow, have seeds that can be stored for long time periods, and with easy subsequent germination. Annual species of Mimulus [Phrymaceae], Consolida [Ranunculaceae], Aquilegia [Ranunculaceae], Plantago [Plantaginaceae], Helianthus [Asteraceae], and several members of the Brassicaceae and the Poaceae may be good candidates for such new models, to name just a few. Plants adapted to habitats with naturally short vegetation period, such as desert or alpine annuals, may be promising, too. Selecting for shorter generation time in any plant species may be an interesting experiment in itself, allowing quantitative genetic studies of generation time and its pleiotropic effects. New fast-cycling models would reveal fascinating opportunities for new studies that include the evolutionary dimension, something highly desirable for future botanical research.