How gathering wild rice under human-mediated drought stress could have inadvertently paved the way to Asian rice incipient domestication and cultivation

Biology and ecology of wild rice (Oryza rufipogon Griff.)

Oryza rufipogon Griff. is an Asian AA genome diploid (2n = 24) species considered to be the wild progenitor of domesticated rice, Oryza sativa L. It is a perennial usually defined as a predominantly outcrossing species which is wind-pollinated. It has a running habit, creeping on the ground and rooting at internodes. Its growth exhibits two successive phases: during the vegetative phase, wild rice tends to have a prostrate growth, a short stature with many tillers and a larger tiller angle, while it develops erect panicle-bearing stalks in the reproductive phase. It is a tufted and scrambling herb, 150–400 cm tall. Anthers are usually >3 mm, reaching 7 mm or more, spikelets are usually 8–9 mm and awns are usually 6–10 cm. Oryza rufipogon is an aquatic or semi-aquatic plant growing in water 0.2–4 m deep. It is found in swamps, marshes, open ditches, channels, and along the boundaries of lakes and ponds (Neelam et al., 2018; Vaughan, 1994). Some of these previously described physiological and ecological features had important consequences for early foragers.

A perennial plant

Wild rice is a perennial, and so it means that even though it produces seeds, its main reproductive system is by means of vegetative propagation, through rhizomes and new tillers. This implies that it is possible to gather its seeds season after season without having to store and later to sow some of these seeds. This biological characteristic has important consequences pertaining to what we usually define as cultivation. Fuller et al. (2014: 6147) define “cultivation” as “a group of behaviors aimed at modifying soil environments and the management of the plants that grow in them.” For us it seems important to be more precise about what are the behaviors leading to cultivation compared to those that are associated with proto-cultivation (or the management of wild stands) (see Figure 1). So we consider that the term “cultivation” can be used only when the whole growth cycle – harvesting, storing, and sowing seeds – is human-mediated. If people do not sow some of the seeds they have harvested, then their behaviors are of the proto-cultivation type. Once sowing is included, then cultivation really starts.

OPEN IN VIEWER

Figure 1.

Evolutionary pathway of rice, from foraging to agriculture.

Since wild rice is a perennial, it was not necessary – even though it was possible – for early foragers to sow some seeds, and then not necessary to start cultivation. This observation is quite important because most of the wild progenitors of the cereals domesticated in southwest Asia (emmer, einkorn, barley) where annual species and so their management required explicit cultivation, that is, included sowing. Even though wild rice seeds were not sown, it was nevertheless possible to manage wild stands of wild rice, for instance by controlling – possibly by fire (Fuller and Qin, 2009; Zong et al., 2007: 91) – other plants (weeds) that were competing with them (Brown, 2018).

An aquatic plant

Wild rice is an aquatic plant, so it was possible for early foragers to manage the water level in the environment where wild rice was growing (Fuller and Qin, 2009). For instance, in the case of water-deficiency, early foragers could have tried to channel water close to the plants, and/or to preserve this water around the plants by bunding small plots of land, since bunding improves soil moisture conservation. Wooden as well as bone spades or hoes, which could have been used for building levees around a small field, have been recovered, for example, from the well-known site of Hemudu, a major center of rice exploitation from the very early Holocene (Fuller et al., 2007). This proto-cultivation practice likely led later – between 7000 and 4000 BP – to wet rice cultivation systems and more precisely to the first paddy fields (Fuller et al., 2007; Zheng et al., 2016; Zong et al., 2007). Of course, all these devices – water-channel, water pit, reservoir, dike, bund – built to control the water level can be used in both directions, that is, either to reduce or to induce a water-deficit.

The low efficiency of rice gathering in an aquatic environment

However, these shattered seeds were not necessarily lost for the foragers, since they also collected various parts of plants on the ground, such as acorns, nuts, and seeds. Ground collection of cereal seeds is even easier when these cereals are wild because in most cases their seeds are awned, and awns – which are usually 6–10 cm long for wild rice – provide to collectors an easier grip (Kislev et al., 2004; Svizzero et al., 2019). Then, gathering wild rice seeds can be done by using – separately or not – two methods. On the one hand mature seeds that have not yet shattered are harvested on the plant, for example by the basket beating method. On the other hand mature seeds that have already shattered can be collected on the ground. According to Optimal Foraging Theory (Winterhalder and Kennett, 2006), early Holocene foragers could have maximized the grain gathered by means of both methods.

The main problem for early foragers aiming to ground collect wild rice seeds is that wild rice is an aquatic plant, so when its seeds shatter, they fall in the muddy water (Fuller and Qin, 2009: 91). These muddy seeds, with their long awns, can nevertheless be collected. However they have first to be cleaned, in order to remove the mud, and after to be either consumed on the spot, or dried to reduce their moisture content in order to avoid spoilage during storage. If grain is not dried correctly they get moldy or may germinate, and so become either non-edible or unpalatable. These activities – cleaning and drying seeds – require extra labor expenditure and so reduce the harvest (and post-harvest) efficiency.

Draining rice plots to improve ground collection of shattered grains

One solution to this previous problem faced by early foragers could have been to dry the small plots of land were wild rice was growing, just before seed maturation. If small plots of land where wild rice was growing were organized as paddy-field-like, for example, based on simple artificial bunding, and were used for instance to retain some nutrient rich seasonal floodwater, then they could also have been used to temporarily drain the water surrounding the plants. With such drainage, the seeds that shatter between two successive harvesting passes would have fallen on a dry soil and so could have be collected on the ground without extra labor expenditure. In other words, draining the plots improved the collection efficiency, and even though the drainage required some extra labor, the cost-benefit analysis would have led early foragers to adopt it. This conclusion is reinforced if it is also taken into account that wild rice is a poor grain producer, compared to domesticated rice, with 20–50 seeds per plant, because it mainly relies on vegetative reproduction (Lu, 2006: 143; Mohapatra et al., 2011). Moreover, and according to recent experiments, harvesting mature seeds before they shatter provides a very low efficiency – around 90 kcal per hour of work – compared to the collection of other wild plants and animals (e.g. shellfish, bamboo roots, yam) (Lu, 2006: 143). So, any improvement in the ground collection of shattered seeds, such as one induced by draining, would have been highly valuable for early foragers.

Plant architecture: Transition toward a more erect growth habit

At the vegetative stage, drought stress results in reduced height, tiller number and leaf area (Menguer et al., 2017). It is often observed that the prostrate habit of the perennial O. rufipogon when the plants grow in standing water is modified compared to when it is grown in drier environments. Indeed the tillers then develop into semi-open bushy plants and give the appearance of the annual species (Nayar, 2014: 225). Tillering determines the plant architecture and canopy development for capture of incident light for primary production.

Tiller development is determined by genetic and environmental factors, resulting in high phenotypic plasticity for plants responding to different environmental conditions (Hussien et al., 2014). Under abundant supply of water and nutrients, the extent of tillering might be especially important; on the contrary under water deficits and short growing seasons, activation of tillers should be genetically restrained to accommodate the limited resource supply (Atwell et al., 2014: 53). In other words, under drought stress, plant reproduction through vegetative propagation is hampered and so natural selection leads to the production of more seeds in order to preserve plant survival by means of seed reproduction.

This implies that during the domestication process, the transition from floating to erect habit could be the consequence of phenotype plasticity in a water-deficient environment (Fuller, 2020). In fact the species adapts according to the level of soil moisture available and modifies phenology, biomass production, and grain yield. Such resilience in aberrant climatic conditions led to the origin of new ecotypes, landraces, species, and ultimately cultivated rice (Mohapatra et al., 2011).

Panicle architecture: Presumed transition toward a closed panicle

A simple morphological change in rice panicle shape, namely a shift from an opened to a closed panicle, has recently been identified by QTL analysis (Ishii et al., 2013; Zhu et al., 2013). An open panicle – controlled by the SPR3 locus – is a feature of wild rice while domesticated rice has a closed panicle. The shift from a spreading to a closed panicle could have had two main consequences, both related to the rice domestication process (Fuller, 2020; Ishikawa et al., 2020). On the one hand, since wild rice has long awns, when the panicle is closed they retain grain longer on the plant, that is, seed retention is improved even though the plant still has the wild phenotype pertaining to seed shattering. On the other hand, with a closed panicle pollination is hampered; then the shift to a closed panicle should also have promoted the outcrossing-selfing transition in rice, the latter being crucial for rice domestication since recessive mutants are more easily fixed under selfing. Indeed in wild rice the evolution of the mating system from cross- to self-fertilization resulted from the modification of panicle and floral morphology that was controlled by QTL of small-to-moderate effect (Grillo et al., 2009; Svizzero, 2021b).

What remains a puzzle is which selective pressure could have induced this morphological change from an open to a closed panicle. Our previous narrative, based on the unintended consequences of drainage by early foragers, provides a possible explanation. Indeed, we have previously demonstrated that the wild rice response to a water-deficient environment was a more compact growth habit, that is, tillers are less numerous and more erect. Then, the problem with more erect and compact tillers is that their respective panicles – when they are open – clashed more frequently, especially under windy conditions, and this damaged the flowers or the grains. So natural selection could have led wild rice growing in a water-deficient environment to modify the panicle shape from a spreading to a closed shape; indeed this would have preserved flowers or grains integrity, and thus enhanced plant survival.

Modified plant architecture and grain productivity

Plant architecture is of major agronomic importance as it determines plant survival ability under environmental stress, the suitability of a plant for cultivation, and potential grain yield (Mathan et al., 2016). Rice plant architecture is mainly determined by the tillering pattern, plant height and panicle morphology (Bai et al., 2018; Wang and Li, 2008). The ehrhartoid (e.g. rice) cereal grasses possess many tillers but do not produce axillary branches. These wild grasses and their domesticated forms produce profuse tillers terminating in an inflorescence. Besides, tiller angle, the angle between the main culm and its side tillers, can significantly affect grain yield in Asian rice (Jin et al., 2008; Tan et al., 2008). The vegetative patterns play a major role in determining the overall biomass of the plant (Li et al., 2017) as well as the number of inflorescences produced and finally the amount of seeds borne by the inflorescence.

A more erect growth habit, resulting from artificial drought stress, enhances photosynthesis efficiency, and improves grain yield. Indeed the number of grain-bearing branches per plant increases, and so harvested grains increase by around 100% (Fuller, 2020; Tan et al., 2008: 92–94). Then, an increase in grain production is another response of wild rice to drought stress (Fuller, 2020: 91).

Modified panicle architecture and harvest efficiency

Once rice plants grow erect, the domesticated panicle is more compact, enhancing photosynthesis of the lower leaves (Zhu et al., 2013), and so grain production increases. On its own the shift to a compact panicle increases grains harvested by around 50% because it improves seed retention (Fuller, 2020; Ishii et al., 2013: 92). Both previous effects are combined, that is with a closed panicle, grain productivity increases as well as harvest efficiency, and so both effects are obviously advantageous for foragers.