Limits and constraints to crop domestication

Abstract

The domestication of plants and animals was one of the most significant changes in human history. A managed cultivation of crops allowed a sedentary lifestyle and the division of work, which freed capacities to develop modern societies. The change from a wild plant to a crop required substantial morphological and physiological adaptation. Crops with similar uses display similar trait changes, which are summarized in the domestication syndrome (Hammer, 1984). For grain crops, loss of seed shattering, increased seed size, and loss of seed dormancy are major domestication traits (Fig. 1). Crops that combine most domestication traits and consequently are well adapted to agroecological environments can be considered fully domesticated, while those that only display a few crop traits may be considered as incompletely domesticated. Although hundreds of grain crops have been cultivated by humans for millennia, most plants show only few of the domestication traits rather than the full syndrome (Meyer et al., 2012). Consequently, only a small fraction of the over 2000 crops that we know today are fully domesticated. Even crops that were of high importance for early cultures display only a minor fraction of the domestication syndrome. Studying the signals of incomplete crop domestication in minor crops could reveal the limits and constraints of crop selection and unlock the potential of novel crops for sustainable food production. Here, I review the evidence of potential genetic limits and constraints that altered the path of crop domestication. I present potential genetic features that might have favored the rapid full domestication of certain plant species and hindered the complete domestication of others. My examples and conclusions are mostly based on annual grain crops because their domestication syndrome is well defined and overlapping. Yet, most of the concepts also hold true for tuber, root, fruit, and vegetable crops, although more domestication traits are based on human preferences (i.e., flavor and color) for these crops. Fully domesticated grain crops include maize, rice, and common bean, but hundreds of other crops, such as buckwheat, teff, and amaranth that have many favorable nutritional properties and are, or historically were, of high importance in their centers of domestication (Meyer et al., 2012) combine only a few domestication traits (Abrouk et al., 2020; Stetter et al., 2020). The loss of seed shattering, which is arguably one of the most important domestication traits for grain crops is lacking in many minor crops (Hammer, 1984). A lack of selection during early domestication is insufficient to explain incomplete trait adaptation because archaeological and genetic evidence suggest that minor crops have received equal selection pressure as fully domesticated crops. As a consequence of incomplete domestication, the production of these crops has strongly declined over the last centuries, as they were replaced by major crops that are fitter in agronomic settings. Today, many crops that are not fully domesticated are referred to as orphan crops because of their declined importance. Traits differ in their complexity. While some domestication traits have been shown to be controlled by only a few genes, other traits are complex and require a polygenic response during domestication (Stetter et al., 2017; Xue et al., 2016). Although numerous traits are highly polygenic, specific genes contribute to the domestication syndrome across different taxa. (Lenser and Theißen, 2013). For instance, mutations in the Shattering1 gene lead to a loss of seed shattering in rice, maize, and sorghum (Lin et al., 2012). This, and other examples of parallelism suggest that certain paths to domestication are particularly efficient or essential for trait adaptation across taxa (Meyer et al., 2012; Lenser and Theißen, 2013). While homologs of most domestication genes can be found in minor crops and their modification has led to advancement in de novo domestication, altering few genes in semi-domesticated crops cannot complete their domestication syndrome (Lemmon et al. 2018). The genetic architecture and the relationship between traits is essential for successful adaptation (Stetter et al., 2018). In less-domesticated crops, domestication traits might be negatively correlated with other essential traits and controlled by highly interconnected regulatory network. The core location of domestication genes in an omnigenic architecture (Boyle et al., 2017) would increase the constraints to successful domestication. Direct or indirect antagonistic pleiotropy might have hindered domestication trait adaptation (Milla et al., 2015). In maize, it has been shown that trait correlations were altered during domestication (Yang et al., 2019), indicating that selection against pleiotropy contributed to successful domestication. Many traits, even those for which major quantitative trait loci have been identified, are controlled by a large number of genes. Highly polygenic traits allow for rapid adaptation, as many mutations can be present as standing variation in the ancestral population. Currently, there is no evidence for extreme deviations in mutation rates between major and minor crops, suggesting that standing genetic variation is more likely to dominate the adaptation of polygenic traits (Purugganan, 2019). Therefore, the amount of standing genetic variation might have limited domestication in crops with low ancestral population size (Stetter et al., 2020). The amount of functional standing genetic variation for a trait is dependent on the effective population size of the ancestral population. In fully domesticated crops, the ancestors showed large effective population sizes at the time of domestication (Purugganan, 2019), but in minor crops, e.g., fonio millet (Abrouk et al., 2020) and grain amaranth (Stetter et al., 2020), ancestral population sizes were lower. The low overall effective population size could be indicative of a lack of functional standing genetic variation for domestication trait controlling loci. While data for functional standing genetic variation is becoming available for major crops (Groen et al., 2020), functional loci will have to be determined for less-domesticated crops to allow comparative studies. Low historic effective population size in the wild ancestor not only reduces standing genetic variation, but also reduces the efficacy of selection due to increased genetic drift. Hence, mildly deleterious mutations can accumulate in such a population. The accumulation could lead to a genetic meltdown, which has been discussed for the extinction of the woolly mammoth and the disappearance of the Neanderthal (Rogers and Slatkin, 2017). During crop domestication, the decline in population size was accompanied by strong artificial selection. Hitchhiking of deleterious alleles with linked positively selected loci and domestication bottlenecks led to an accumulation of genetic load. Such an accumulation, described as the cost of domestication, has been observed in successfully domesticated crops (Moyers et al., 2018). For ancestral populations with initially small population sizes, the cost of domestication might have been too high, and low fitness in early crop populations might have hindered further adaptation. To date, no comparative studies of load between a larger number of crops have been executed because differences in reference genome quality bias the estimation of load in a population. The increasing availability of high-quality genomes, annotations, and genome-wide re-sequencing data for minor crops and their ancestors (Abrouk et al., 2020; Stetter et al., 2020) will give insights into the accumulation of genetic load in minor crops. Continuous selection on domestication traits progressively led to a distinction between crops and their wild ancestors. While in early stages of domestication, admixture between crop populations and wild relatives was frequent, isolation proceeded as crops became more domesticated. Once the distinction between crop and wild ancestor was high, gene flow could still occur and deliver additional variation to the crop without compromising its identity as long as migration rates were low. Such introgression from wild relatives has provided adaptive variants to a number of crops (Janzen et al., 2019). However, continuous strong gene flow from the ancestral species and other wild relatives into the crop might have prevented the isolation of the crop population and hampered the evolution of domestication traits. Another consequence of gene flow that might have constrained the domestication of crops is the reintroduction of wild traits after initial adaptation. Post-domestication gene flow from wild relatives to early-stage crop populations could have decreased the distinction of crops from wild plants. Gene flow from wild populations with larger effective population sizes could have reduced genetic load that accumulated during initial domestication steps and prevented a genetic meltdown of the population. Hence, gene flow potentially led to evolutionary rescue of the crop population by increasing its effective population size and reducing genetic load, but reintroduced wild-plant traits such as seed shattering, seed dormancy, or weedy growth habits. The ecology of the wild plants that were fully domesticated likely played an important role in their success. Multiple Poaceae have been strongly domesticated. Their ability to rapidly inhabit open areas created by early farms might have given them a head start for the domestication process (Milla et al., 2015). Early agroecological systems favored highly competitive species. In addition, the early ability to “recruit” humans for seed dispersal has been suggested as fitness advantage of successfully domesticated crops (Spengler, 2020). The ancestors of several minor crops are less competitive, which slowed down the onset of domestication of these crops. Human-mediated selection pressure was an essential factor in the progression of domestication. While it is still under debate whether early domestication was consciously advanced by humans or whether it resulted from unconscious selection (Purugganan, 2019), the order in which traits were selected might have set the fate of a crop. Early, strong selection on traits with a simple architecture from new mutations diminishes the genetic variation for polygenic traits in the population (Akdemir et al., 2019). Consequently, consecutive selection on polygenic traits would be mutation limited and slower than early selection on polygenic traits followed by selection on simple traits. Genome sequence information from ancient crop samples in combination with advanced population genetic methods could allow dating the onset of selection for traits and elucidate differences in the order of domestication trait selection (Swarts et al., 2017). Crop domestication has been repeated across the globe and left different levels of completeness. Although crops with similar uses show similar domestication traits, their path during domestication was probably as diverse as the species that have been cultivated by humans (Fig. 2). Some crops are still advancing on that path, and modern molecular and quantitative genetic approaches will allow us to bypass some of the detours that constrained adaptation in the past. As rapidly changing environments are threatening cropping systems, it will be essential to understand the domestication history and the genetic control of key domestication traits for a diverse set of crops and the potential limits to domestication. Less-domesticated crops are often well adapted to diverse environmental conditions and are high in nutritional value, but require improvement in key domestication traits to present suitable alternatives to major crops. The evaluation of the outlined potential constraints and the understanding of the distinct paths to domestication will elucidate the possibility to complete the process in many crops and unfold the potential for accelerated de novo domestication of wild plants. I acknowledge funding by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany s Excellence Strategy – EXC-2048/1 – project ID 390686111, in particular the “COVID-19 helping hands” support for early career researchers by the Center of Excellence on Plant Sciences. I thank Gaëlle Caublot for the design of Fig. 2 and the members of the Stetter lab, Julie Jacquemin, Jeffrey Ross-Ibarra, the anonymous reviewer, and Pamela Diggle for feedback on an earlier version of the manuscript.

Topics & Concepts