Assessing the significance of soil erosion for arable weed seedbank diversity in agro-ecosystems

I Introduction

Soil erosion is a natural process of landform and landscape evolution involving the detachment, transport and deposition of mineragenic and organic sediments. This process is naturally occurring but can be exacerbated by human activity, such as farming. Land-cover change and intensification of farming are widely linked to accelerated soil erosion (Morvan et al., 2008; Zhang et al., 2007). The average annual rate of soil loss by erosion on arable land across Europe is estimated to lie within the range of c. 3–40 t ha⁻¹ yr⁻¹, which locally can exceed 100 t ha⁻¹ yr⁻¹ (Verheijen et al., 2009). This raises concerns about on-site nutrient depletion, decreased soil aggregation and loss of productivity, and off-site impacts through sedimentation and eutrophication of downstream water bodies (Rowan et al., 2012). Climate change is further expected to increase the extent and severity of soil erosion arising from more frequent extreme events (Cerdan et al., 2010). Intensification of farming not only increases soil erosion, but is also associated with declines in the abundance and diversity of arable weeds and may generate increasingly homogeneous plant assemblages (Baessler and Klotz, 2006; Brazier, 2004; Gunton et al., 2011; Hawes et al., 2010).

Weeds provide valuable ecosystem services (e.g. nutrient cycling, pollination and pest regulation) within agro-ecosystems by maintaining diversity of flora that provides stable food and habitat resources for a range of detritivores, herbivores, pollinators, predators and parasitoids (Evans et al., 2011; Gibbons et al., 2006; Hawes et al., 2003; Hyvönen and Huusela-Veistola, 2008). This emerged weed flora is dependent on annual regeneration from the arable weed seedbank. Seedbank diversity is recognized as an important buffer against short-term disturbance events and provides a degree of redundancy facilitating adaption to long-term change (Fried et al., 2009; Loreau et al., 2003; Venable and Brown, 1988). However, there is a need for balance between managing for a healthy arable ecosystem and maximizing crop productivity. To achieve this, weeds need to be managed to achieve densities that are not competitive with the crop, while still occurring in sufficient abundance to maintain viable populations of species with high resource value to arable food webs (Albrecht and Auerswald, 2009).

The inter-relationship of soil erosion with vegetation is well documented as plants intercept direct rain splash, promote infiltration, enhance water retention and dissipate surface runoff (Jiao et al., 2009; Zhongming et al., 2010). However, there is a paucity of field studies exploring the linkages between soil erosion and seedbanks, particularly in arable contexts. Seedbanks are well described in terms of the biological (predation, pathogens and seed death) and agricultural (tillage, chemical treatments and cropping histories) controls on their diversity and abundance patterns (Albrecht and Auerswald, 2009; Buhler et al., 1997), but the potential impact of erosion is rarely addressed. Small-scale studies have focused on seed mobility in laboratory experiments (e.g. Cerdà and García-Fayos, 2002) whilst others have assessed seed movement at the landscape or catchment scales (e.g. Goodson et al., 2003; Gurnell et al., 2006). However, the link between field-scale soil erosion and weed seedbank diversity and abundance within arable fields represents a major gap in our understanding of temperate agro-ecosystems.

This paper assesses the likely impact of soil erosion on the arable weed seedbanks of temperate northern Europe. The potential significance of accelerated soil erosion is assessed as an under-recognized redistributive mechanism influencing the composition and abundance of weed seedbanks in arable agro-ecosystems. A brief review of erosion and field-scale sediment dynamics (entrainment, transport and deposition) is presented. This is followed by a synthesis of the key biological and agronomic factors influencing seedbank characteristics (e.g. field management, dispersal, seed rain and mortality). We then discuss the likely consequences of the redistribution of weed seed assemblages by entrainment, transport and deposition processes associated with soil erosion at the field scale and beyond.

II The extent and impact of soil erosion in arable ecosystems

Soil erosion, by water, wind, tillage or crop harvesting, is a three-stage process involving detachment of soil particles (entrainment), transport of the detached material, and finally deposition (Verheijen et al., 2009). Water erosion is a balance between erosivity (determined by rainfall intensity and runoff shear stress) and erodibility (effective soil strength arising from texture, structure and the binding effects of plant roots) (Brazier, 2004). Water erosion occurs when raindrop impact dislodges material or when overland flow results from rainfall exceeding the infiltration rate by either ‘infiltration-excess’ or ‘saturation-excess mechanisms’ (Morgan, 2005). For temperate environments, intense rainfall (>10 mm hr⁻¹) for a single short-duration event can be as erosive as lower-intensity but longer-duration events depending on antecedent water content and local soil conditions (such as aggregation) and vegetation cover (Bracken and Croke, 2007). In Europe, mean water erosion rates are between 0.1 and 5 t ha⁻¹ yr⁻¹ (Cerdan et al., 2010).

Wind erosion of soil is a widespread phenomenon in agro-ecosystems that can lead to removal and damage of topsoils (Riksen et al., 2003). Wind erosion also causes secondary disturbances by mobilization of dust particles that damage vegetation and transport pathogens, leading to a decrease in vegetation cover, which creates a positive feedback and increases erosion susceptibility (Morgan, 2005). Wind erosion occurs mainly on vulnerable sandy or organic soils (Banwart, 2011; Fullen, 2003). However, Riksen et al. (2003) highlighted that wind erosion is not confined to these areas. Over three million hectares of lowlands in northwestern Europe are prone to wind erosion due to poor management. Wind erosion rates across England and Wales are estimated at 0.1–2 t ha⁻¹ yr⁻¹ (Chappell and Thomas, 2002).

Tillage erosion, involving the systematic downslope displacement of soil by ploughs and associated tillage equipment, has long been recognized but only recently been systematically evaluated. Arable land contributes over 70% of the total soil erosion in Europe with an average soil loss rate of 3.6 t ha⁻¹ yr (Cerdan et al., 2010). The envelope of soil loss rates across European arable landscapes lies within the range of 0.1–10 t ha⁻¹ yr⁻¹ (Van Oost et al., 2006; Verheijen et al., 2009). Soil loss by crop harvesting, either by adhering to cultivation equipment or being co-extracted with the harvest, has also only relatively recently been acknowledged as a contributory ‘erosion’ mechanism, particularly for soil bound to root crops such as potatoes, beets and carrots. An indicative figure up to 2.5 t ha⁻¹ yr⁻¹ for soil loss associated with root and tuber crop harvesting was suggested by Quine et al. (2006), while Verheijen et al. (2009) report a wider range of between 1.3 and 19 t ha yr⁻¹ across mainland Europe.

The cost of soil erosion across the EU is estimated at £0.6–11 billion per year (Jones et al., 2012). In the UK, the impact of accelerated soil erosion was valued up to £38 per hectare per year (Dobbie et al., 2011; Morgan, 2005; Verheijen et al., 2009). Importantly, these costs encompass on-site and off-site environmental and socio-economic impacts, but hitherto such estimates have not sought to account for value loss of ecosystem services provided by the weed seedbank or other soil biota. Dobbie et al. (2011) estimate erosion costs the economy of Scotland £60.5 million annually as a result of loss of organic matter from soils and related loss of ecosystem services.

Erosion processes span scales from the inter-particle scale, controlled by response to rain splash, to the landscape scale where topography, hydrogeology and land management control hydrological pathways (Van Oost et al., 2006). Challenges in reliably estimating erosion rates arise from short-term funding and difficulties in upscaling from detailed plot-scale studies to the field and catchment scale (Boix-Fayos et al., 2006). Thus a practical compromise is typically made between the physical scale over which measurements are made and their temporal resolution. For example, erosion field plots often produce high-quality event-based data over the limited duration of an experimental programme. However, the complexity of routing and sediment delivery found over longer time (e.g. decadal) and large spatial (e.g. field and landscape) scales are not reflected with plot data (Boardman, 2006; Boix-Fayos et al., 2006). Figure 1 illustrates the key elements of a sediment budget representing both erosion and sediment storage within arable landscapes. Quantifying sediment fluxes over larger scales may involve multiple assessment techniques ranging from measurement of rill dimensions, radiometric surveys and sediment fingerprinting to suspended-sediment yield determinations (cf. Rowan et al., 2012; Walling et al., 2006). An important distinction is made between ‘gross erosion’ which is the sum total of all eroded sediment mobilized and ‘net erosion’, which accounts for the proportion of the eroded sediment deposited within fields, e.g. in hollows and footslopes or vegetated buffer strips, ditches and field boundaries. The ratio of gross to net erosion equates to a ‘sediment delivery’ ratio and varies widely within the range c. 20–50% (Brazier, 2004; DEFRA, 2008; Walling et al., 2006) dependent on climate, landscape setting, slope length and shape, soil texture, cultivation practices and slope-channel connectivity (Small et al., 2003).

OPEN IN VIEWER

Figure 1.

Generalized catchment sediment budget within lowland arable landscape settings.

Erosion and downslope sediment delivery typically exhibit a high degree of size selectivity (Table 1). Unconfined sheet flow tends to be the most selective because the finest particles are most readily entrained, while the coarsest fraction tends to be preferentially deposited (Morgan, 2005). The size of particles can be expressed as water stable aggregates (effective particle size) or as ultimate particle size (after chemically dispersing the water stable aggregates). Slattery and Burt (1995) showed that for rills and gullies the detachment of particles is by collapse of side channels in addition to basal scour. The material coming from the collapse does not require dispersion of aggregates and hence the effective particle size in the entrained material is larger than the ultimate particle size. Effective-grain size distributions were also considerably greater than ultimate-grain size distributions because transported sediment were mainly comprised of aggregates containing substantial amounts of clay and silt particles bound by organic matter and, potentially, seeds from the soil matrix. The specific mobility of seeds within the soil is likely to be similar to soil particles of the same size, but will also depend on the nature of local erosion processes, the depth distribution of seeds within the soil profile, and the seed’s hydrodynamic properties (size, shape, mass and other surface characteristics such as mucilage and microsculpture) (cf. Benvenuti, 2007).

OPEN IN VIEWER

Table 1.

Characterization of erosion processes to indicate the impacts on seedbank redistribution in arable ecosystems.

Mode	Processes	Depth (m)	Description	Step-length (m yr−1)	Significance for seedbank redistribution within arable agro-ecosystems
Water erosion	Splash	10−3	Mechanical impact of precipitation on soil surface resulting in net downslope transport.	10−2	Within-field selective detachment and dispersal of small light seeds on soil surface.
	Inter-rill	10−2	Scour and downslope transport of soil by unconfined sheet flow and splash.	10−1–101	Selective entrainment and transport of near-surface seeds, especially those with small, light and simple morphologies without mucilage (feeds rill network).
	Rill	10−2–10−1	Rills are micro-channels formed by flow convergence to a threshold beyond which bed scour and channelized flow occur.	10−1–102	Soil and climate determine rill dimensions and densities. Bed scour and side-wall collapse means seeds lost to full profile depth and non-selective transport of seeds. Loss from field.
	Piping (tunnelling)	10−2–100	Subsurface erosion due to percolating water creating network of pipe-like conduits.	10−1–102	Seed loss patterns depend on depth where pipe gallery is active, i.e. near surface or deeper below ground.
	Gully	10−1–101	Large linear channelized scour features typically with stepped long-profile.	100–104	Deep scour features can extend into the subsoil resulting in complete loss of seedbank (no hydrodynamic sorting) from fields.
	Tramlines	10−2–10−1	Tracks within arable fields used by farm machinery to plant, manage and harvest crops.	10−1–102	Tramlines typically compacted with decreased infiltration. Preferentially capture runoff so locus for rill/gully development.
Wind erosion	Deflation and abrasion	10−2–10−1	Combines abrasion (mechanical impact by wind-entrained debris) and deflation which is entrainment and transport of soil and seeds from the land surface.	100–105	Scour depth depends on soil conditions and wind climate; and seed transport depends on aerodynamic characteristics. Size selective in low-intensity events and physical impacts may change viability of some seeds. Transport typically off-field.
Mass movements	Creep, slides, slips and flows	10−1–101	Range of mechanisms represented from soil creep in surface layers to deep-seated slides and rotational failures.	10−1–104	Depth of active processes depends on geomorphology and landscape setting. Displacement and export of seedbank throughout the depth of the failure – loss into channel network.
Tillage erosion	Ploughing and crop management	10−1–100	Conventional mouldboard tillage operations result in net local downslope soil displacement from the top of the field exposing subsoil at the crest while burying soil at the bottom.	10−1–100	Local topography and slope curvature important, e.g. erosion greatest on upper slopes and ridges with displaced soil accumulating in swales and field base. Seed redistribution may be size selective – effects include mechanical impact, displacement and seed burial. Processes are largely contained within-field.
	Harvesting	10−1–100	Soil loss associated with harvesting, especially root crops such as beets, potatoes and carrots.	102–103	Export of seeds from soil mass bound to root crops and farm vehicles translocated to off-field processing locations.

Recent years have witnessed much greater attention to the off-farm or downstream consequences of soil erosion from arable land in the form of sediment-associated nutrient losses. Much effort has been directed towards the control of diffuse pollution, and in particular the significant role played by runoff derived from farm machinery wheelings (between-plant tramlines or wheel rows). Deasy et al. (2009) found controlled trafficking and tramline disruption decreases soil and nutrient losses more effectively than traditional treatments such as residue incorporation, minimum tillage, contour cultivation and vegetation barriers. Silgram et al. (2010) showed that tramline disruption techniques reduced both soil and sediment-associated phosphorus loss by more than 86%. By comparison, little attention has been paid to assessing the rates and significance of seed mobilization from the seedbank and the consequences, particularly in terms of systemic losses from perennial sources or accumulation within local deposition sites.

III The seedbanks of agro-ecosystems

The weed seedbank is a dynamic reserve of viable seeds on, or incorporated within, the soil (Brenchley, 1918; Brenchley and Warington, 1933; Csontos and Tamás, 2003; Feast and Roberts, 1973; Roberts, 1964). Seedbanks have been described as a botanical ‘memory’ because they preserve genotypes that may be absent from the standing vegetation community (Davis et al., 2005; Harper, 1977). The persistence of seeds in the seedbank make them far less sensitive than the emerged flora to immediate conditions of the field or weather and therefore confer some resilience to the arable weed community (Wilson and Lawson, 1992). The number and diversity of plants comprising the weed flora has declined over the past 50 years, along with other indicators of biodiversity; the main causes of this decline are farming intensification, the increased use of herbicide and the competitive suppression of weeds by autumn-sown crops (Gibson et al., 2006; Marshall et al., 2003; Robinson and Sutherland, 2002). The seedbank is therefore a valuable reference in studies of ecological impact and in the conservation and restoration of the commoner arable flora (Firbank, 1999; Firbank et al., 2003; Heard et al., 2003a, 2003b; Perry et al., 2003).

Within UK agro-ecosystems arable weed seed densities generally lie within a range of between 1,000 and 1,000,000 seeds m⁻², most being concentrated in the top 15 cm of the soil profile (Firbank, 1999; Squire et al., 2003; Thompson et al., 1997). High seed population densities (>10⁵ m⁻²) and species richness (around 40 species per field) were recorded throughout the last century in ploughed land under poor weed control (Brenchley and Warington, 1933; Roberts and Feast, 1973; Squire et al., 2000). Very low seedbank populations of around 100 m⁻² and as low as 10 species in a field were recorded whenever management suppressed weeds for many years, whether by mechanical cultivation (Brenchley, 1918) or the use of herbicides (Marshall and Arnold, 1994). Seedbanks can be highly responsive, readily increasing if control is relaxed over several years, or declining by up to 50% a year if seed return is severely reduced or prevented (Brenchley and Warington, 1933; Roberts, 1962; Roberts and Feast, 1973; Wilson and Lawson, 1992).

Taxa can be separated according to the longevity of their viability within the seedbank (Bekker et al., 2003), with a distinction typically made between transient species, i.e. seeds which remain viable only to the next opportunity to germinate, and persistent species, i.e. seeds which enter secondary dormancy and remain viable in the soil for longer than one year (Hulme, 1998). The key biotic processes determining arable seedbank abundance and composition include augmentation through seed rain and immigration, and losses through dispersal, seed mortality, germination and emigration (Forcella, 2003). These processes are affected by the timing and intensity of crop management (Andreasen and Skovgaard, 2009; Brenchley and Warington, 1933; Hyvönen et al., 2003; Robinson and Sutherland, 2002).

Conventional tillage practices are frequently associated with physico-chemical degradation of soil profiles. This result is due to organic matter losses, compaction and decreased infiltration rates hence increasing soil profile susceptibility to erosion (Brazier, 2004; Fullen, 2003; Morgan, 2005). The style and intensity of erosion within any particular landscape setting will have different consequences in a direct sense through loss or gain in seedbank inventory but also indirectly in terms of the depth distribution and the quality of the resultant seedbed as a growth medium.

IV The impact of soil erosion on arable seedbanks

The spatial relations of weed seedbank assemblages are an important element of biodiversity within agro-ecosystems (Alignier and Petit, 2012; Benvenuti, 2007). Assessing the significance of erosion and sedimentation to redistribute and restructure the seedbank is therefore an important but under-appreciated research challenge. Key to this is combining a better understanding of earth surface processes with specific biological and agronomic controls on seedbank dynamics – involving death, germination, weed control and replenishment through seed rain. Differential mobilities and mortalities depending on seed morphologies, sensitivity to damage during transport and viability following eventual deposition (which could be at depth) are all potentially important and will play out in different ways according to location, time and starting seedbank characteristics.

1 A first-order assessment of the importance of soil erosion to seed transport

Field- and catchment-scale sediment budgets, as represented in Figure 1, provide a valuable analytical framework to assess the spatial and temporal significance of erosion-controlled seedbank dispersal. Sediment budgets focus on sources, pathways and sinks of erosion and on the timescales of delivery (Small et al., 2003). Table 2 provides an estimate of potential seed losses from within different counties (administrative areas) of the UK based on published soil erosion rates (Brazier, 2004). Table 2 also shows arable seedbank densities derived from ‘Farm-Scale Evaluations’ of genetically modified herbicide-tolerant crops data held at The James Hutton Institute in Scotland. Sampling methods have been previously described (Firbank et al., 2003; Heard et al., 2003a). Data at this scale are not available from other northern European countries, so the UK is used here for illustrative purposes. For each county with data available, average and maximum erosion rates and seedbank densities are given. These values are used to calculate rates of potential seed loss through soil erosion, based on the following assumptions: (1) seeds are concentrated in the top 15 cm of the soil profile and this is the ‘active layer’ in relation to surficial erosion processes; (2) soil bulk density approximates to 1 tonne m⁻³; (3) soil erosion processes are not selective for seeds with particular characteristics and so indiscriminately mobilize the seedbank in equal proportions to the bulk soil. Challenges to these assumptions and refinements to the calculations as presented require direct quantification which, to the authors’ knowledge, is currently unavailable.

OPEN IN VIEWER

Table 2.

Observed soil erosion (Brazier, 2004) and seedbank values (Heard et al., 2003a) providing seed loss estimates within UK counties.

UK county	Erosion rates (t/ha/yr−1)		Weed seedbank			Seed flux (seed/ha/yr)		Annual seed export (% of original inventory)
	Average scenario	Maximum scenario	Average seed density (seeds/m2)	Maximum seed density (seeds/m2)	Median of species	Average scenario	Maximum scenario	Average scenario	Maximum scenario
Cambridgeshire	0.36	3.3	463	638	53	1111	10,188	0.02	0.22
Cumbria	0.22	5.0	1422	2124	20	2085	48,049	0.01	0.34
Dorset	1.29	31.0	551	768	60	4739	114,168	0.09	2.07
Herefordshire	0.99	13.2	489	811	42	3227	43,097	0.07	0.88
Kent	4.51	17.8	943	1459	43	28,339	112,223	0.30	1.19
Norfolk	0.92	10.7	505	547	95	3096	36,030	0.06	0.71
Nottinghamshire	1.11	66.1	240	367	28	1773	105,677	0.07	4.41
Shropshire	1.28	49.3	845	1076	66	7208	277,856	0.09	3.29

Maximum erosion rates provide the worst-case scenarios caused by agricultural practice at highly localized points (Brazier, 2004). For example, in Table 2 Kent has the highest average erosion rate but Nottinghamshire has the highest maximum rate. Nottinghamshire also has the lowest arable weed seedbank densities. Differences in erosion rates between counties are likely to reflect regional differences in hydro-climate, topography, soil types and land-management practices (DEFRA, 2008). Combining these data reveals substantial differences in potential seed flux between the average and maximum scenarios. Seedbank losses through germination, death and weed control have been estimated or quantified (Benvenuti, 2007; Davis et al., 2011; Grime et al., 1981; Navntoft et al., 2009; Watson et al., 2003; Westerman et al., 2003, 2011). Additions to the seedbank have also been quanitfied for seed rain (Jones et al., 2003). The additions are exceeded by total losses since Gibson et al. (2006); Marshall et al. (2003) and Robinson and Sutherland (2002) noted that the seedbank number and diversity are in decline. The losses due to movement by erosion are additional to the other loss mechanisms. Annual seed export rates represent only a small fraction of the original seed inventory (averaging approximately 0.1% across the case-study regions); however, the long-term (>10 years) impact may be significant locally, representing seed losses of up to 40%.

Although there is no quantified critical threshold for arable biodiversity, below which ecosystem functions may be compromised, this magnitude of loss from the arable seedbank over decadal timescales is likely to play out through many within-field processes, including nutrient cycling through soil, plant and invertebrate food webs (Marshall et al., 2003). Using available seedbank data with published erosion rates, the contribution of erosion to the seedbank decline can be estimated. The potential significance of erosive redistribution can be made by combining average net erosion rates within the UK of c. 7 t ha⁻¹ yr⁻¹ (cf. Brazier, 2004; DEFRA, 2008; Walling et al., 2006) with average arable weed seed densities of c. 2000 seeds m⁻² within the plough layer (Heard et al., 2003b) which results in an average annual loss rate of c. 0.5% yr⁻¹ of the total seedbank inventory. Thus erosion alters seedbanks by an additional loss of c. 10% over a 20-year period, potentially destabilizing the seedbank. Furthermore, use of net erosion rates at the field scale conceals transient and longer-term sediment storage within swales, footslopes, boundary ditches, buffer strips and hedgerows. The potential for within-field spatial restructuring of seedbank inventories is therefore potentially much greater than whole-field averages, particularly in patches where erosion rates are high (e.g. >10 kg m⁻² yr⁻¹) and seed densities are low (e.g. <100 seeds m⁻²).

Considerable scope exists to refine these preliminary estimates by quantifying the response of different weed seed species to different erosion processes. For example, splash-related dispersal and inter-rill transport have the potential to move small, light seeds on or near the soil surface at a very local scale (in the order of 1–10 cm). The impact of these transport processes is therefore species or phenotype specific, depending on seed characteristics (size, morphology, seed coat and mucilage) (García-Fayos et al., 2010). Selectivity in terms of entrainment and preferential deposition will contribute to a shift in seedbank composition favouring species with large, heavy seeds in eroded areas and those with small, light seeds in depositional areas. Whether these changes in species composition in different parts of the field have any impact on ecosystem service provision will depend on whether there are any correlations between seed morphology and other plant traits, such as germination requirements, competitiveness, shade tolerance, resource value to insect herbivores and timing of flowering and reproductive periods.

At a larger, but still within-field, scale, surface runoff converges into rills and tramlines. There is increased connectivity with channel networks and seeds can be transported greater distances through the field (Bracken and Croke, 2007). In severe cases, rill and tramline erosion can penetrate the full plough-depth, resulting in potential loss of seed below germination depth. The fate and viability of transported seed is likely to be species specific, with different species showing different responses to abrasion damage, burial and growth of viable seedlings within depositional regions (where there is likely to be greater nutrient supply and increased competition) (Davis et al., 2008).

Movement of seeds from fields into a channel network can also occur through subsurface flows, gullies (deep scours extending into the subsoil), mass movement of soil through creep, slides and flows, and on farm machinery at harvest. These larger-scale processes depend on landscape setting, slope and soil types and, because they are non-selective for soil particle size, they also are likely to be non-selective for arable weed seeds of different sizes. Research is needed to quantify the relative importance of each of these processes to the dispersal distance and amount of seed moved within fields and into the channel network.

Tillage erosion can selectively move seeds across fields, particularly from hill crests to the field base (Cousens et al., 2008; Van Oost et al., 2006). Depending on the field topography, eroded material may accumulate in low-lying areas of fields, resulting in seed burial below germination depth. This process could result in increased seed mortality or it could trigger dormancy until germination conditions are suitable. Using beads as proxies for seeds, Westerman et al. (2009) observed resurfacing of beads after rain and wind removed topsoil. Therefore, depositional areas could have germination of resurfaced seeds. Tillage can also change seed viability directly, increasing the germination potential for species requiring scarification and increasing the mortality rates of others. These factors are likely to have significant impacts on both the within-field distribution of seeds and the species composition of the seedbank community.

2 Timing of erosion events

Erosion events (through precipitation, wind, snow melt and tillage) are highly episodic and vary in intensity (Morgan, 2005). In temperate environments, seedbanks are characterized by seasonal patterns of seed rain, dispersal, germination and onset of dormancy. Erosion and seedbank composition are influenced by multiple factors including the amount of light and water available for germination and growth, soil conditions (chemistry and hydraulic conductivity), and the extent of vegetative cover. The susceptibility of the arable weed seedbank to impact from soil erosion is therefore influenced by the interplay of the number and size of runoff-generating storm-events with the local calendar of tillage and harvesting – inclusive of ploughing, seedbed preparation, and periods of bare ground prior to weed and crop emergence (Figure 2).

OPEN IN VIEWER

Figure 2.

Generic crop and weed cover curves estimated from unpublished data gathered as part of the Farm-Scale Evaluations of GMHT crops against 10-year average rainfall (Squire, personal communication, 2012). CC = crop cover; WC = weed cover; BG = bare ground.

Periods of bare ground for spring and root crops in northern Europe occur from late autumn through to early spring, and tillage operations are usually carried out during autumn or spring prior to crop sowing. Without crop cover, soil is exposed to splash, inter-rill (unconfined sheet flow) and rilling (channels formed by convergent flow). Seeds shed during this period are therefore concentrated on the soil surface and exposed to surficial erosional processes prior to incorporation into the seedbank (Westerman et al., 2006). Weed species with a life-cycle characterized by late autumn seeding are therefore likely to be disproportionately affected by soil erosion compared to species that shed seed during periods of dense vegetative cover when soil disturbance is low.

3 Morphology and hydrodynamic behaviour

Seed mass is commonly cited as an important determinant of dispersal distances (Benvenuti, 2007; Smith and Kok, 1984). Cerdà and García-Fayos (2002) demonstrated that below a threshold mass of 50 mg seed size was the main factor affecting mobility, whereas above this threshold seed morphology becomes more important. The relationship between seed mass and size (average axial length) for a representative sample of commonly found arable weeds is shown in Figure 3. What is clear is that while seed size ranges over one order of magnitude (0.4–28 mm), seed mass (reflecting the shape and anatomy of the seeds) varies over three orders of magnitude (0.012–50 mg) translating into a broad spectrum of hydrodynamic behaviours. Differential mobilities are therefore likely at the field scale consistent with the selectivity of erosion and transport as demonstrated by Slattery and Burt (1995). Some species, especially those such as Veronica spp. that have small cup- or boat-shaped seeds, are particularly adapted to secondary dispersal by rain splash (diplochory) (Benvenuti, 2007; Vander Wall and Longland, 2004). The effect and importance of this process is yet to be determined for seed, but there is some evidence for seed movement by splash dispersal (Westerman et al., 2009).

OPEN IN VIEWER

Figure 3.

Seed mass and size of UK arable seeds plotted in relation to Wentworth particle size classes. The sediment size scale is provided for an indication of the relative size of seeds.

Relationships between seed entrainment and seed size are further complicated by the presence of appendages (hairs, wings, awns) and secretion of mucilage upon hydration which increases the resistance to water-borne movement (García-Fayos et al., 2010). Using Capsella bursa-pastoris (shepherd’s purse), Deng et al. (2012) demonstrated that hydration resulted in a six-fold increase in seed volume and a 2.5-fold increase in seed surface area. This mucilage release occurs within five seconds of wetting and its rapid expansion serves to increase the binding of seeds to the soil matrix. This binding also strengthens water-stable aggregates, further inhibiting detachment and lateral transport.

V Conclusions

Arable weed seedbanks are essential for the sustainability of cropping systems because they provide the basis for the above-ground flora that contributes to biodiversity and supports ecosystem services within otherwise simplified agro-ecosystems. Accelerated soil erosion has a negative impact on the physico-chemical characteristics of soil, plant productivity and seedbank abundance, but the linkages between erosion and seedbanks are poorly understood (Jiao et al., 2009). Consequently, there is an important challenge and opportunity to bridge the research gap between ecology and geomorphology in agro-ecosystems.

This paper has described the main drivers of seedbank abundance and composition in the agro-ecosystems of temperate northern Europe and has discussed their likely susceptibility to physical redistribution by soil erosion processes. There are five key questions emerging from this discussion that need to be answered in order to quantify the significance of soil erosion to the abundance, diversity and composition of weed seedbanks in agro-ecosystems.

First, soil erosion can deplete the weed seedbank by removing seed and altering germination capacity, but the extent to which this occurs is unknown. Quantitative data regarding the impact of erosion on seedbanks is limited to laboratory experiments (cf. Cerdà and García-Fayos, 2002) and needs to be expanded with field data. This information is needed to attribute losses to ecosystem services and thus underpin analysis of the true economic cost of erosion.

Second, given that (1) certain erosion processes selectively remove soil particles according to a range of physical characteristics, and (2) weed seeds have diverse morphologies that are designed to protect the seed from physical, chemical and biological impacts, research is required to determine whether erosion selectively removes species with certain seed morphologies or whether seeds are equally impacted by erosion processes.

Third, some weed seed species are affected by disturbance, which aids germination while others lose viability, become dormant (due to displacement and burial below critical emergence depths) or die in response to disturbance. Further work is required to determine the influence of erosion, transport and deposition on the viability and germination of different species in arable systems that are frequently disturbed by tillage practices. As seeds of different species may support different food webs, losses of different species may have different ecological significance.

Fourth, soil erosion studies cover a range of scales from individual particle entrainment to transport and storage at field and landscape scales. Quantifying the relative role of specific processes, such as water-borne versus tillage erosion, to restructure field-scale seedbank assemblages within different landscape settings is therefore an important challenge identified through this review.

Fifth, soil and seeds are mobilized, transported and deposited during individual storm-events the consequences of which are cumulatively expressed over decadal timescales and longer. Understanding the long-term consequences of erosion-mediated seedbank redistribution is therefore an important issue in the context policies geared towards 'sustainable intensification' in arable systems. This analysis confirms the positive contributions that weed flora play in maintaining ecosystem services within agro-ecosystems and highlights the need to better understand the scaling relations between sediment delivery and biodiversity at field, farm and landscape scales.