The rise of Vulpia myuros (Poaceae) and the impact of cultivation-timing on plant community structure

Vulpia myuros (L.) C.C. Gmel. has shown a dramatic increase in abundance as a weed of winter wheat since 2000, especially under regimes of no-till husbandry. A long-term experiment on disturbance timing at Silwood Park suggests that this increase is probably not due solely to no-till cultivation or to the plant’s well-known herbicide resistance, but rather to autumn cultivation coupled with warmer winter weather.


Introduction
The study of phenology (the science of the timing of things) has experienced a renaissance since the search for evidence of the impacts of climate change has become so pressing (Fitchett et al, 2015). There are fundamental ecological questions involved in this. For instance, it is important to know how many of the differences between neighbouring plant communities are due to year effects (temporal heterogeneity, e.g. in the timing of soil disturbance) rather than to underlying spatial heterogeneity in soil structure, nutrient concentrations, water regime, herbivore impacts or microclimate. Year-to-year variation in the timing and intensity of inputs such as rainfall, changes in fitness-affecting weather variables such as the earliness of spring, and fluctuation in abiotic drivers such as frosts, fire, or soil disturbance all have the potential to alter species interactions, resulting in long-term consequences for botanical composition and succession (Crawley, 1997). Seedbank emergence dynamics are affected by ecological soil factors and buried weed seed germination and dormancy (Buhler et al., 1997) and by responses to climatic (Forcella et al., 1992), ecological (Benvenuti & Macchia, 1997), and agronomic practice (Froud-Williams et al., 1984).
We know that species differ in their timing of germination (Buhler et al., 1997), and predict that species are likely to increase in abundance when the disturbance regime matches their preferred recruitment time (Crawley, 2004). Thus, spring germinators should increase under repeated spring cultivation, and autumn germinators should increase under autumn cultivation. Some species, however, might be capable of germination at any time. If germination were stimulated by disturbance at whatever time it occurred, then such species would be predicted to increase in abundance under any regular annual disturbance. If, however, germination were to occur independent of disturbance, then such species might decline in abundance if, on average, they did not reach fruiting size by the time of disturbance. It is likely, therefore, that there will be strong selection for species that germinate immediately after disturbance, and selection against species that germinate before disturbance but do not ripen seed before they are killed by cultivation. This was the pattern observed after 10 years of a long-term disturbance timing experiment at Silwood Park in Berkshire (Crawley, 2004): dominance, relative abundance and species composition were all significantly different on replicated plots cultivated in October, March or May each year (see Appendix A).
Since 2000, there has been a pronounced increase in the importance of Vulpia myuros as a weed of winter wheat, particularly on farms practicing no-till cultivation.
This increase is variously attributed to the plant's well-documented herbicide resistance (Büchi et at., 2020) and/or to some factor associated with no-till cultivation (Akhter et al., 2020). The Pound Hill Experiment at Silwood Park has documented a similar increase in the abundance of V. myuros, and therefore provides an indirect opportunity to test the hypothesis that the increase in V. myuros is caused by no-till husbandry coupled with herbicide resistance.

Methods
Pound Hill Field at Silwood Park, Berks, is a long-term disturbance-timing experiment on acid, sandy soils of the Bagshot series (Crawley, 2004). Within the field, an area of 100 m x 40 m was divided into 12 plots of 36 m x 8 m in October 1992. This allowed replication of three timings of cultivation (October, March, or May each year) randomized independently in four blocks (A at the southern end to D at the northern end). The plots were ploughed on their long axis (east to west) using a tractormounted two-furrow plough, and then the ground was rotovated to smooth out the furrows and create a fine seedbed. The plots were then left entirely alone for the plant community to develop, either by recruitment from the seed bank or by immigration of seeds from adjacent plant communities. Seeds ripened between April (for the earliest flowering species such as Myosotis discolor and Arabidopsis thaliana on October-cultivated plots) and August (for the latest flowering species such as Galinsoga parviflora and Solanum nigrum on May-cultivated plots).
The data presented in this paper were collected on 30 June 2021 by visual estimates of percentage cover for every species encountered on two zigzag walks over the long axis of each 36 m x 8 m plot. By this date, the October-cultivated plots had been growing for 8 months, the March-cultivated plots for 3 months and the May-cultivated plots for just 1 month. There was negligible bare ground on the October-or March-cultivated plots by this date, but the May-cultivated plots still had c.45% bare ground. One individual of a species is scored as 0.1 and two as 0.2, so that species-richness data are inclusive.
The data reported in Crawley (2004) were from a one-off test of the impact of 10 years of contrasting cultivation times on the species-composition of the seed bank (1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002). The whole experiment (3 cultivation timings in each of 4 blocks) was split along its east-west axis and, at random, the eastern half was chosen to be cultivated in October 2002, and the western half in May 2003. The data were gathered in late July 2003 from all 24 sub-plots (2 current cultivation timings, 3 long-term cultivation treatments (October, March and May) and 4 blocks). The main findings are summarised in Appendix A & B for comparison with the 2021 data published here.
The data in Appendix C show the botanical composition of May-cultivated plots at the end of their 2018 growing season, for comparison with the early-season results reported here from June 2021.   Table 3 shows the percentage cover of vascular plant species ranked by mean cover on 30 June 2021 on plots cultivated in May each year in blocks A-D. There was still c.40% of bare soil but the wet May and June weather had allowed rapid growth of Agrostis gigantea, Galinsoga parviflora, Spergula arvensis, Rumex acetosella, Fallopia convolvulus, Holcus mollis, Chenopodium album, Erodium cicutarium and Lycopsis arvensis. Species restricted to the May cultivated plots on this date were Galinsoga parviflora, Lamium amplexicaule, Vulpia bromoides, Anagallis arvensis, Hypochaeris radicata, Cerastium glomeratum, Convolvulus arvensis, Geranium molle, Veronica persica and Sonchus oleraceus.
Total species richness across the four replicates was not markedly different with cultivation timing, but ranked as follows: October (36 species) > May (33) > March (31).

Discussion
Three guilds of species were apparent in the Pound Hill Experiment: autumn germinators, spring germinators (seeds with a chilling requirement to break dormancy), and indifferent species (species germinating after disturbance, irrespective of timing). Some species were driven close to local extinction by annual soil disturbance at the 'wrong' time of year; presumably their seed banks were small or rapidly depleted, and recruitment to their populations may have been more reliant on the annual seed rain.
Cultivation in a given month represents extremely strong selection in favour of annual plants that can germinate after that month and set seed before the next cultivation, and of perennial plants that can withstand fragmentation and burial in that particular month (e.g. the dominance of March-cultivated plots by Artemisia vulgaris and of October-cultivated plots by Holcus mollis and Agrostis gigantea).
The ability to survive over winter or through summer drought clearly differs in relative importance for species germinating in October, March or May. Comparison of Tables 1, 2 & 3 highlights clear differences in species composition, with groups of species confined entirely to a single cultivation-timing treatment (there were 19 species restricted to October-cultivated plots, 7 to March-cultivated plots, and 10 to May-cultivated plots).      Cultivation is highly likely to kill seedlings: this is presumably why seedlings of Quercus robur were found on all October cultivated plots, but not on any of those cultivated in March or May. Seedlings of frost-sensitive species may be killed by frosts on October-or March-cultivated plots but never (see Crawley, 2005, p. 31) on May-cultivated plots. Climatic conditions experienced by newly germinated seedlings are markedly different in October (warm and moist), March (cold and moist) and May (warm and dry), and this has consequences for interspecific competition during the rapid growth phase on each timing treatment. Year effects driven by seasonal weather differences are exemplified by a comparison of May-cultivated plots in 2021 after a wet growing period following a very cold dry spring (Table 3) and 2018 (Appendix C) after a long, hot, dry summer. The dominant in 2021 was Agrostis gigantea, but this was a mere trace in 2018. Galinsoga parviflora had more than 3 times the cover in 2021 as in 2018 (16% vs. 5%). The species that performed relatively well in the dry growing season of 2018 were Erodium cicutarium (20%) and Spergula arvensis (15%). The marked increase in the distribution and abundance of Vulpia myuros in Silwood Park matches the recent rises that have been documented in no-till arable cultivation of winter wheat (Akhter et al., 2020), but the causes of the increase in Silwood Park are clearly not the same as those hypothesised in the weed science literature: viz. the use of no-till sowing technology coupled with the welldocumented herbicide resistance of the grass (Büchi et al., 2020). Cultivation in Silwood Park involved traditional inversion ploughing followed by rotovation, and no herbicides have been applied at Pound Hill since 2002 (Crawley, 2004). The timing of cultivation was crucial: V. myuros increased on plots cultivated in October, but not on those cultivated in March or April. It may well be that the increase documented from arable agriculture is driven by a more widespread cause such as increasing winter temperatures (Chapman et al., 2020) interacting with autumn cultivation.
This analysis of the rise of V. myuros is another example of the way that long-term field experiments can be used to address questions that were never envisaged at the time the experiment was set up.