Journal list menu
The effects of flooding lowland wet grassland on soil macroinvertebrate prey of breeding wading birds
Summary
- 1
Lowland wet grassland in western Europe is often managed for breeding wading birds, especially lapwing Vanellus vanellus, redshank Tringa totanus, snipe Gallinago gallinago and black-tailed godwit Limosa limosa. Recommended conservation management often entails introducing winter flooding, and in Britain there is government funding to encourage this through the Environmentally Sensitive Area scheme.
- 2
Soil macroinvertebrates are important prey for breeding wading birds on lowland wet grassland. This study quantified the response of soil macroinvertebrates to flooding, their ability to survive in flooded grassland, and changes in the abundance and physical availability of soil macroinvertebrates for feeding wading birds as flood water subsides.
- 3
Unflooded grasslands contained high biomasses of soil macroinvertebrates, comprising mainly Tipulidae larvae and earthworm species that are widespread in pastures. Grasslands with a long history of winter flooding contained much lower biomasses of soil macroinvertebrates, comprising mainly a limited range of semi-aquatic earthworm species.
- 4
Introducing winter flooding to previously unflooded grassland greatly reduced soil macroinvertebrate biomass. This was mainly due to the majority of earthworms vacating the soil soon after the onset of flooding. However, when earthworms were artificially confined in flooded soils, most species were capable of surviving periods of at least 120 days continual submergence. Winter flooding also expelled large numbers of overwintering arthropods from the soil.
- 5
Soil macroinvertebrates were slow to recolonize winter-flooded grassland when it was re-immersed in spring. Consequently, prey biomass for breeding wading birds remained low in areas that had been flooded during the preceding winter. However, winter flooding probably benefited breeding snipe by helping keep the soil soft enough for them to probe for prey. It also probably benefited breeding lapwings and redshank by helping keep the sward short and open enough for them to feed in during the latter part of their breeding season. Pools of winter flood water that remained in spring and early summer also provided a source of aquatic invertebrate prey for breeding wading birds.
- 6
We suggest that the best feeding conditions for breeding snipe will be provided by keeping the upper soil soft enough for them to probe in but without reducing soil macroinvertebrate biomass by flooding it beforehand. Optimal conditions for breeding lapwings and redshank will probably be provided by creating a mosaic of unflooded grassland, winter-flooded grassland and shallow pools.
Introduction
Lowland wet grassland, defined as grassland below 200 m that is subject to periodic freshwater flooding or waterlogging, supports a distinctive assemblage of breeding wading birds in western Europe, including lapwing Vanellus vanellus (L.), redshank Tringa totanus (L.), snipe Gallinago gallinago (L.), black-tailed godwit Limosa limosa (L.), ruff Philomachus pugnax (L.), oystercatcher Haematopus ostralegus (L.) and curlew Numenius arquata (L.) (Beintema 1983; Smith 1983). Severe declines have been noted in the numbers of breeding wading birds on many areas of lowland wet grassland in recent years, and the plight of these species has attracted considerable concern from conservation organizations in Britain and the Netherlands. Overall, breeding lapwings have declined significantly, by 38%, between 1982 and 1989 on lowland wet grassland in England (O’Brien & Smith 1992), while the number of 10-km squares occupied by breeding snipe and redshank in the British Isles has declined by 22% and 12%, respectively, between 1968 and 1972 and 1988 and 1991, almost certainly due largely to the loss of suitable wet grassland (Gibbons, Reid & Chapman 1993). Black-tailed godwits and redshank also have an unfavourable conservation status in the rest of Europe (Tucker & Heath 1994).
Recommended conservation management for lowland wet grassland often entails the introduction of winter flooding. For the purpose of this paper, winter flooding is defined as the presence of surface water during periods between October and April inclusive. In Britain there is government funding to encourage patchy winter flooding through the Environmentally Sensitive Area (ESA) scheme. This is a system whereby landowners within designated areas receive incentive payments for entering voluntary management agreements. These agreements are intended to maintain or enhance the biodiversity value of each of the habitats for which the area is particularly noted, the quality of the landscape and its archaeological and historic resource (MAFF 1994). Similar schemes exist elsewhere in Europe. Of the eight English ESA that contain substantial areas of lowland wet grassland, six have management prescriptions intended to benefit breeding wading birds through the introduction of patchy flooding between the beginning of December or January and the end of April. These six ESA contain 46 000 ha of lowland wet grassland, representing about 23% of the English resource of this habitat (Dargie 1993; Glaves 1998).
Winter flooding attracts feeding and roosting wildfowl (Thomas 1982) and, by leaving shallow pools in its aftermath, provides suitable feeding habitat for breeding wildfowl (Thomas 1980) and wading birds (Green 1986; Ausden 1996a; Sanders 2000). Retention of winter flood water also enables field water levels to be kept high in spring and early summer. On peat soils, high field water levels are thought to increase the physical availability of soil macroinvertebrates for feeding snipe by reducing the penetration resistance of the upper soil, and thereby making it easier for them to probe (Green 1986, 1988; Green, Hirons & Cresswell 1990). Despite the introduction of winter flooding to many formerly unflooded grasslands, little is known of the effect that it has on the soil macroinvertebrates that are important prey for breeding wading birds on grassland, especially snipe (Hogstedt 1974; Matter 1982; Cramp & Simmons 1983; Green 1986, 1988; Galbraith 1989; Baines 1990; Green, Hirons & Cresswell 1990; Ausden 1996a).
It has generally been assumed that winter flooding does not adversely affect soil macroinvertebrates, because they are found in grasslands subject to regular and often prolonged periods of flooding (Green 1986, 1988; Green, Hirons & Cresswell 1990). Furthermore, the four most important sites for breeding snipe in England and Wales identified by Smith (1983), the Ouse Washes, Nene Washes, Lower Derwent Valley and the Somerset Levels, are all extensively flooded during winter.
Laboratory experiments have shown that the majority of the most abundant macroinvertebrates found in pastures, the earthworms Allolobophora chlorotica (Savigny 1826) (unpigmented morph), Aporrectodea caliginosa (Savigny 1826), Aporrectodea longa (Ude 1885) and Lumbricus rubellus Hoffmeister 1843, are capable of surviving for long periods (100–166 days) in aerated water (Roots 1956). However, the main physical stresses on earthworms in flooded soils are thought to be due to the development of anaerobic conditions in the soil (Mather & Christensen 1988). Hence the ability of earthworms to survive in aerated water cannot be used to predict their ability to survive in flooded soils.
In this study, we investigated the effects of flooding on the abundance and physical availability of soil macroinvertebrates to feeding wading birds. First, we compared the macroinvertebrate fauna of sites where water levels had recently been raised to that of traditionally winter-flooded grasslands that support important populations of breeding wading birds. Secondly, we carried out two experiments. The first investigated the relative preference of macroinvertebrates in flooded and unflooded soil. The second determined how long macroinvertebrates were able to survive in flooded soils. Finally, we investigated changes in the distribution of soil macroinvertebrates and their physical availability to feeding wading birds as flood waters subsided.
Methods
The soil macroinvertebrate fauna of lowland wet grassland and the effects of introducing winter flooding
The locations of sampling sites are shown in Fig. 1. We selected 12 sites where winter flooding had been introduced in the previous 1–14 years, and which had not been flooded prior to this for at least 20 years (see Table 2 for grid references). These represented most of the main sites where water levels have recently been raised by conservation bodies to benefit breeding waders in England, and are termed ‘recently flooded grasslands’. All of the areas referred to in this paper were flooded with freshwater (salinity < 1‰ as determined using a hydrometer).

Location of sampling sites.
Mean biomass ± SE (gm−2) | |||||||
---|---|---|---|---|---|---|---|
Site | Years | Soil | Veg | Unflooded | Winter-flooded | t | P |
Berney Marshes TG472060 | 5 | CG | I | 82·20 ± 21·2 | 0·68 ± 0·65 | 5·407 | < 0·0001 |
Buckenham TG348052 | 1 | CG | I | 47·15 ± 8·65 | 8·28 ± 5·20 | 4·356 | 0·0004 |
Castle Marshes TM474916 | 1 | CG | I | 109·08 ± 14·83 | 1·08 ± 0·88 | 10·767 | < 0·0001 |
Church Farm Marshes TM464585 | 3 | AG | I | 37·73 ± 7·63 | 0·00 | 4·656 | 0·0002 |
Elmley TQ964674 | 14 | AG | I | 19·70 ± 4·35 | 0·30 ± 0·15 | 4·236 | 0·0005 |
Heigham Holmes TG440205 | 1 | CG | I | 178·20 ± 13·93 | 24·15 ± 6·70 | 9·901 | < 0·0001 |
Hickling, field a TG425204 | 1 | CG | I | 56·35 ± 11·58 | 0·00 | 5·255 | < 0·0001 |
Hickling, field b TG423206 | 1 | CG | U | 69·73 ± 20·63 | 3·45 ± 3·45 | 3·392 | 0·0032 |
Holkham TF878449 | 7 | AG | I | 158·45 ± 24·53 | 33·88 ± 17·73 | 3·841 | 0·0012 |
Old Hall Marshes TL974129 | 1 | AG | U | 28·65 ± 7·03 | 12·53 ± 7·95 | 1·763 | 0·0948 |
Strumpshaw TG342058 | ? | EP | F | 44·73 ± 9·03 | 7·78 ± 2·58 | 3·739 | 0·0015 |
West Sedgemoor, field a ST377265 | 1 | AG | I | 41·55 ± 9·08 | 2·03 ± 1·10 | 5·096 | < 0·0001 |
West Sedgemoor, field b ST348249 | 1 | AG | U | 41·25 ± 11·93 | 1·40 ± 0·53 | 3·925 | 0·0010 |
Whiteslea TG428217 | 3 | EP | U | 34·38 ± 12·70 | 1·63 ± 1·63 | 0·240 | 0·0240 |
Ten of these recently flooded grasslands consisted of areas of uniform, agriculturally improved, grassland (National Vegetation Classification communities MG6 Lolium perenne–Cynosurus cristatus grassland or MG7 Lolium perenne leys and related grasslands; Rodwell 1992). At each of these sites we selected a field that had been partially flooded the previous winter. At two sites, both agriculturally improved and agriculturally unimproved (National Vegetation Classification mire communities or MG5 Cynosurus cristatus–Centaurea nigra grassland; Rodwell 1991, 1992) fields were present. At these two sites we selected an agriculturally improved field and an agriculturally unimproved field, each of which had been partially flooded the previous winter. All the fields sampled had been partially flooded for periods of 20–120 days between October and March. Soil macroinvertebrates were sampled from unflooded and winter-flooded parts of the same field.
A further five sites were selected that had a long tradition (300+ years) of extensive winter flooding. These are termed ‘traditionally flooded grasslands’. They were the four areas of lowland wet grassland identified by Smith (1983) as being the most important for breeding snipe in England: the Lower Derwent Valley, Nene Washes, Ouse Washes and Somerset Levels, and also the Insh Marshes (NH797020), the most important site for breeding snipe in Scotland (RSPB, unpublished figures). These sites also contained important breeding populations of other wading birds.
At the Somerset Levels, sampling was restricted to a hydrologically managed block on the RSPB's West Sedgemoor Reserve (ST360255), as this was the only area of the Somerset Levels that still held high densities of breeding snipe. At the Nene Washes, sampling was restricted to the RSPB Nene Washes Reserve (TF293997), as the rest of the Washes only supported low densities of breeding snipe. At the Ouse Washes, samples were taken from the Ouse Washes RSPB Reserve (TL490877), while at the Lower Derwent Valley samples were taken from Aughton Ings (SE697387) and Wheldrake Ings (SE443702), both parts of the Lower Derwent Valley National Nature Reserve.
Traditionally flooded grassland sites had estimated densities of breeding snipe of between 16·8 and 27·1 pairs km−2. These compared with overall estimated densities of breeding snipe of 0·3, 0·9 and 3·3 pairs km−2 on lowland wet grassland found in three recent surveys of land within ESA (Robins, Smallshire & Street 1992; Weaver 1995; Allwood 1997).
At each traditionally flooded grassland, two flooded fields were selected that were considered typical of the site in terms of their vegetation, hydrology and use by breeding waders. The fields selected at these sites had been continually flooded for between 40 and 270 days during the winter (and autumn) prior to samples being taken.
Soil samples were dug and macroinvertebrates removed by hand-sorting and wet sieving (Gerard 1967; Ausden 1996b). Hand-sorting is considered the most reliable and effective method of sampling earthworms (Heppleston 1971; Edwards & Lofty 1972; Nordstrom & Rundgren 1972). Small and dark coloured worms, however, tend to be under-recorded using this method compared with chemical extraction (Raw 1959, 1960; Nordstrom & Rundgren 1972). However, as macroinvertebrate abundance was described in terms of biomass, rather than density, missing some small worms was unlikely to have a significant effect on the results.
Soil samples were dug with a spade using a quick, levering, action to intercept retreating large earthworms (Sims & Gerard 1985). Each soil sample was 20 × 20 cm in surface area and 10 cm deep. The depth of 10 cm was chosen as this is the approximate length of a snipe's and black-tailed godwit's beak, hence sampling would have removed only those soil macroinvertebrates within reach of feeding birds.
At recently flooded grasslands 12 randomly positioned soil samples were taken from unflooded parts of each field, and an additional eight randomly positioned samples from winter-flooded parts of the same field. At traditionally flooded grasslands, 12 randomly positioned soil samples were taken per field, except at the Lower Derwent Valley where, because of limited time, only eight samples were taken per field.
All but three sets of samples were taken between the first week of March and first week of April 1993–95 in order to measure macroinvertebrate biomass and availability just prior to waders arriving to breed. Samples from the Ouse Washes and West Sedgemoor had to be taken during the second half of May, as these sites were still extensively flooded until late spring. Samples from Insh Marshes were taken in 1992 using a similar methodology (Ausden 1992).
Soil samples were sealed individually in polythene bags, kept at 5 °C to prevent decay of any macroinvertebrates killed during digging, and sorted within a week. Soil macroinvertebrates were removed by first breaking the sample up by hand and removing any soil macroinvertebrates found. Any root mat that could not easily be broken apart by hand was then wet-sieved using a 2-mm gauge sieve and a high pressure water jet.
Soil macroinvertebrates were preserved in 4% formaldehyde solution and left for at least 3 days for their weight to equilibrate (Piearce 1984), blotted dry and weighed on a top pan balance to the nearest 0·01 g. Formaldehyde-preserved weight is about 25% less than fresh weight (Raw 1959). Invertebrates weighing less than 0·01 g were discarded as they were considered too small to be profitable prey for feeding waders and will in any case only have contributed a negligible proportion of the overall earthworm and tipulid biomass.
Soil macroinvertebrates were identified under a × 10–40 binocular microscope. Earthworms were identified to species using the key in Sims & Gerard (1985). Nomenclature follows that of these authors. Although this key is for adult earthworms only, it also proved possible to identify the majority of immature and juvenile earthworms using features such as colour, shape, form of the prostomium, arrangement of setae and presence or absence of conspicuous dorsal pores. It was not possible to identify immature Octolasion spp. and some small immature Lumbricus spp. Also, some fragments of earthworms, cut or damaged during taking samples, could not be identified to species. Unidentified earthworms only amounted to 3·5% of the total weight of earthworms found.
The most abundant Tipulidae larvae occurring on grasslands, Tipula paludosa Meigen 1830 and the widespread grassland species Tipula oleracea Linnaeus 1758, cannot be reliably distinguished using external features. Therefore, all Tipulidae larvae found (all of which resembled T. paludosa or T. oleracea as determined using the key by Brindle 1960) are referred to as Tipula sp.(p.).
Penetration resistance of the soil was determined using a penetrometer (Green 1988; Green, Hirons & Cresswell 1990). This measured the maximum force required to push a 4·5-mm diameter steel probe 10 cm into the ground. The diameter and depth of the probe were chosen to simulate a snipe's beak. Five measurements were taken adjacent to the location of each soil sample.
At six recently flooded sites, water table height was estimated by digging a temporary dipwell using a 2·5-cm diameter soil auger. The depth of the water table below the soil surface was measured at approximately half-hour intervals, until it had stopped rising.
Relative preferences of soil macroinvertebrates for flooded and unflooded soil
The relative preference of soil macroinvertebrates for flooded and unflooded soil was investigated by offering them the choice of flooded and unflooded halves of intact soil samples. Sixteen soil samples with a 20 × 40-cm surface area and a depth of 10 cm were dug from an unflooded agriculturally improved grassland on pelo-calcareous gley soil at Buckenham RSPB Reserve (TG351053) in February 1994. Samples were sealed individually in clear polythene bags and stored in their original vertical orientation at a temperature of 10 °C for 6 days, to allow the soil samples to equilibrate to this temperature and any earthworms disturbed during transportation to re-establish their burrows. Eight of the soil samples then had their lower halves gradually immersed in rain water (also at 10 °C), which was trickled down the inside of the bags containing the soil samples, so that the water level within the bag was raised by 5 cm every hour. The bags were then resealed. The other eight control soil samples were opened and disturbed in the same way, but had no water added to them.
All bags were stored for 48 h at 10 °C and then each soil sample, while still contained within its polythene bag, was cut into upper and lower halves. A 50–100 g soil sample was taken from the centre of each half, sealed in a polythene bag and its available soil moisture content determined by drying to constant weight at 70 °C. The remainder of each soil sample was then also sealed separately in polythene bags prior to sorting. Macroinvertebrates were removed from the remainder of each soil sample, preserved, identified and weighed as described in the previous section.
Ability of macroinvertebrates to survive in floode d soil
The ability of macroinvertebrates to survive in flooded soil was investigated by artificially flooding soil macroinvertebrates in intact soil samples for different lengths of time. The timing and duration of flooding were chosen to mimic those prescribed under ESA management agreements.
Fifty intact soil samples each of 20 × 15-cm surface area and 20 cm deep were dug in December 1994 from the same area of Buckenham RSPB Reserve as described in the previous experiment. Samples were 20 cm deep to allow earthworms to avoid low winter temperatures in the surface soil during the course of the experiment (Gerard 1967). Each soil sample was sealed in a clear polythene bag to contain the invertebrates, and then reburied in the ground with its upper surface flush with the undisturbed soil surface. The soil samples were then left for a month to allow any earthworms disturbed during translocation to re-establish burrows.
During the first week of January the soil samples were flooded rapidly or gradually for up to 120 days using the treatments shown in Table 1. Each treatment was replicated 10 times. ‘Gradual flooding’ was intended to simulate flooding caused by deliberate raising of water levels. This was carried out by slowly pouring 200 cm3 of previously collected rain water down the inside of the bag containing the soil sample each day for 4 successive days until the soil sample became immersed. ‘Rapid flooding’ was intended to simulate flooding by infiltration of water from above, either as a result of heavy precipitation or through flooding caused by sudden large water inputs, for example due to a river overflowing. This was carried out by pouring rain water over the upper surface of each soil sample until, after a few seconds, it became totally immersed. Once each soil sample was immersed, further rain water was immediately added to cover it to a depth of 10 cm, so that the grass on the soil samples was completely submerged.
Treatment | |||||
---|---|---|---|---|---|
0 | 60 (gradual) | 90 (gradual) | 120 (gradual) | 120 | |
Method of flooding | – | Gradual | Gradual | Gradual | Rapid |
Timing of flooding | Left unflooded for 120 days | Flooded for 60 days and then left unflooded for 60 days | Flooded for 90 days and then left unflooded for 30 days | Flooded for 120 days | Flooded for 120 days |
At the end of the flooding period, each soil sample was removed from the surrounding soil and immediately drained by making approximately 50 pin pricks in the bottom of the polythene bag. The unflooded soil samples also had 50 pin pricks made in the bottom of the polythene bag at the beginning of the experiment to allow them to gain or lose water at the same rate as the drained bags.
It was thought that containment of soil samples within polythene bags might cause the water above the soil to heat more than in a natural flooding situation. It was considered that this, together with the prevention of air and water movement within the sealed bags, might result in water in the upper soil becoming more rapidly deoxygenated than during natural flooding. To reduce these effects partially, all bags were opened at 2–5-day intervals for approximately 1 hour (not when raining) during the day to allow fresh air to enter. After the soil samples had been immersed for 4 days (the first day that the bags were checked) it was noticed that some arthropods had emerged from the soil and collected on the surface of the water and the insides of the polythene bags. These were collected and the bags were subsequently opened during the daytime at 2-day intervals to remove any invertebrates on the water's surface or on the bag. Any earthworm seen on the soil surface, in the water or on the insides of the polythene bags was also removed. This was because it was considered that they were not subject to flooding, and in a natural situation would either have migrated elsewhere or have been highly vulnerable to bird predation. Expulsion of arthropods virtually ceased after 10 days of flooding and, after this, bags were only opened approximately every 5 days to allow air to circulate and to collect any additional invertebrates from the water or insides of the polythene bags.
To determine whether conditions in the water in the polythene bags were similar to those under more natural conditions, an additional six soil samples were gradually flooded at the same time as the rest, buried and left with their polythene bags open. The temperature at the surface of the immersed soil samples was measured and compared with that of water above soil samples sealed in polythene bags, under different weather conditions on four occasions each month.
Maximum and minimum air temperature and minimum ground temperature were recorded daily at a weather station within 10 m of the buried soil samples throughout the course of the experiment.
Macroinvertebrates were removed from soil samples, preserved, identified and weighed as described previously.
Changes in soil macroinvertebrate distribution and physical availability to breeding wading birds as flood water subsides
Changes in soil macroinvertebrate distribution and physical conditions were measured along belt transects running perpendicularly across the margins of unflooded and winter-flooded grassland. Single transects were positioned in each of six partially flooded agriculturally improved fields at Buckenham RSPB Reserve (see previous experiments) and on pelo-alluvial gley soil at Church Farm Marshes RSPB Nature Reserve (TM465585). Each of the fields had been partially flooded for 90–120 days during the winter prior to sampling, and had been partially winter-flooded for the previous 1–5 years.
Transects were placed along a relatively straight section of flood margin. Soil macroinvertebrate biomass, vegetation height and physical conditions were measured along the transects in mid-March (immediately after flood water had begun to subside and just prior to waders settling to breed) and in mid-May, by which time lapwings and redshank have generally ceased feeding on soil macroinvertebrates and begun to feed mainly on terrestrial arthropods and/or aquatic invertebrates (Baines 1990; Ausden 1996a). Whittingham, Percival & Brown (2000) showed the importance of considering vegetation height for breeding waders.
Soil macroinvertebrate biomass, water table height, soil moisture, penetration resistance and vegetation height and cover were determined at 0·5-m intervals along the transects.
Soil macroinvertebrates were sampled by digging a 20 × 20-cm surface area and 10-cm deep soil sample dug at 0·5-m intervals along the transect. The vertical distribution of soil macroinvertebrates was investigated by dividing each sample horizontally into depths of 0–3 cm and 3–10 cm using a spade. The resulting two portions of the sample were separately sealed in polythene bags. Macroinvertebrates were removed from soil samples, preserved, identified and weighed as described previously.
Water table height, soil moisture and penetration resistance were determined using the methods described previously. Five measurements of penetration resistance were taken at each sampling point. Vegetation height and cover were measured using a point quadrat. This consisted of 10 1·2-mm diameter vertical graduated metal wires attached in a line at 5-cm intervals from each other. This was lowered vertically until it touched the soil surface. The maximum height at which vegetation touched each wire was recorded. Grass flower and seed heads were ignored, as these were unrepresentatively tall compared with the rest of the vegetation. If no vegetation touched the wire, then that point was recorded as bare ground. Two sets of 10 measurements were made at each sampling point.
Data analysis
Assemblages of soil macroinvertebrates were ordinated in two axes using decorana (Hill 1979). All taxa were considered equally, i.e. there was no weighting for rare species. Data that were not normally distributed were transformed using the log(x + 1) or arcsine transformation to achieve normality prior to parametric tests being carried out (Sokal & Rolf 1969). Means are given ± 1 SE.
Results
The soil macroinvertebrate fauna of lowland wet grassland and the effects of introducing winter flooding
Figure 2 shows the decorana ordination of soil macroinvertebrate assemblages from traditionally flooded grasslands and from unflooded parts of fields at recently flooded grasslands. Axes 1 and 2 of this ordination accounted for 70·8% of the total variance explained by the model. Soil macroinvertebrate assemblages from the unflooded parts of fields showed little, if any, overlap with those from traditionally flooded grasslands.

decorana ordination of lowland wet grassland soil macroinvertebrate assemblages. Data are from traditionally flooded grasslands and unflooded parts of fields at recently flooded grasslands. Squares = gleys; triangles = peat. Black symbols = traditionally flooded grasslands; white symbols = unflooded parts of fields at recently flooded grasslands. The two abbreviations are the most abundant (left) and second most abundant (right) taxa in terms of biomass: Ac, Allolobophora chlorotica; Aca, Aporrectodea caliginosa; Al, Aporrectodea longa; Ar, Aporrectodea rosea; Et, Eiseniella tetraedra; Lr, Lumbricus rubellus; Lt, Lumbricus terrestris; Oc, Octolasion cyaneum; Ot, Octolasion tyrtaeum; T, Tipula sp.(p.).
The majority of the soil macroinvertebrate biomass of traditionally flooded grasslands comprised the earthworms Octolasion tyrtaeum (Savigny 1826), Allolobophora chlorotica (green morph only), Lumbricus rubellus and Eiseniella tetraedra (Savigny 1826). These four species accounted for 95·5 ± 1·9% (range 90·4–100·0%) of the total earthworm biomass and 87·5 ± 3·2% (range 76·8–96·7%) of the total soil macroinvertebrate biomass at these five sites. The majority of soil macroinvertebrate biomass at three of the four peat sites comprised Octolasion tyrtaeum, but this species was not found in samples from the alluvial Lower Derwent Valley. The only specimens of Aporrectodea caliginosa were taken from an area of the Lower Derwent Valley a few metres from unflooded grassland. Their presence therefore may not be typical of traditionally flooded grassland. The only other species of earthworms found in traditionally flooded grasslands were Dendrobaena octaedra, Octolasion cyaneum (Savigny 1826) and Satchellius mammalis (Savigny 1826).
Some earthworm species were obviously able to withstand long durations of continual submergence. Four species, Allolobophora chlorotica (green morph), Eiseniella tetraedra, Lumbricus rubellus and Octolasion tyrtaeum, were found at the Ouse Washes, which had been completely submerged for approximately 270 days (between September and May) prior to sampling in late May. The Ouse Washes are frequently flooded for durations of more than 120 days during the winter and even for short periods during summer. Octolasion cyaneum was present in the hydrologically managed block at West Sedgemoor, which had been flooded for approximately 150 days prior to sampling. Dendrobaena octaedra and Satchellius mammalis were only found in the Lower Derwent Valley and Insh Marshes, respectively, both of which had been continually flooded for periods of approximately 40 days during the winter prior to samples being taken.
In 11 of the unflooded parts of fields the majority of the soil macroinvertebrate biomass comprised the earthworms Allolobophora chlorotica (both green and unpigmented morphs), Aporrectodea caliginosa, Aporrectodea longa, Aporrectodea rosea (Savigny 1826), Lumbricus rubellus and Lumbricus terrestris Linnaeus 1758. These accounted for 93·9 ± 1·6% (range 84·0–100·0%) of the earthworm biomass and 87·6 ± 1·9% (range 74·5–96·4%) of the total soil macroinvertebrate biomass at these 11 sites. At Old Hall Marshes, Whiteslea and Elmley, the majority of the soil macroinvertebrate biomass comprised Tipula sp.(p.), being 72·4 ± 10·6%, 57·0 ± 14·2% and 57·6 ± 11·0% of the total soil macroinvertebrate biomass at these three sites, respectively.
At recently flooded grasslands, winter-flooded parts of fields nearly always contained significantly lower soil macroinvertebrate biomass than unflooded parts of the same field (Table 2). On average, winter-flooded parts of fields contained 9·8 ± 3·5% (range 0·0–44·5%, n = 14) of the soil macroinvertebrate biomass found in unflooded parts of the same field. Mean biomasses of six of the most abundant earthworm species, together with those of Tipula sp.(p.), Elateridae larvae and ‘other Coleoptera larvae’, were significantly lower in winter-flooded parts of fields than in unflooded parts of the same field (Table 3).
Recently flooded grasslands | |||||
---|---|---|---|---|---|
Taxa | Unflooded (n = 14) | P 1 | Winter-flooded (n = 14) | P 2 | Traditionally flooded grasslands (n = 5) |
All taxa | 74·16 ± 13·46 | *** | 6·50 ± 2·74 | * | 13·67 ± 2·96 |
Lumbricidae: | |||||
Allolobophora chlorotica | 10·66 ± 3·95 | ** | 1·63 ± 0·61 | 4·23 ± 1·96 | |
Aporrectodea caliginosa | 17·11 ± 5·16 | *** | 0·44 ± 0·44 | 0·00 | |
Aporrectodea longa | 15·14 ± 6·31 | *** | 0·39 ± 0·27 | 0·00 | |
Aporrectodea rosea | 3·61 ± 1·12 | *** | 0·00 ± 0·00 | 0·00 | |
Dendrobaena octaedra | 0·00 | 0·00 | 0·14 ± 0·14 | ||
Eiseniella tetraedra | 0·70 ± 0·52 | 0·51 ± 0·34 | 1·00 ± 0·72 | ||
Lumbricus castaneus | 1·28 ± 0·65 | * | 0·03 ± 0·03 | 0·00 | |
Lumbricus festivus | 0·40 ± 0·40 | 0·00 | 0·00 | ||
Lumbricus rubellus | 6·49 ± 1·22 | *** | 0·47 ± 0·29 | * | 2·29 ± 1·13 |
Lumbricus terrestris | 1·76 ± 1·12 | 0·04 ± 0·04 | 0·00 | ||
Octolasion cyaneum | 0·25 ± 0·25 | 0·00 ± 0·00 | * | 0·11 ± 0·09 | |
Octolasion tyrtaeum | 0·72 ± 0·42 | 0·00 ± 0·00 | *** | 2·51 ± 0·75 | |
Satchellius mammalis | 0·02 ± 0·02 | 0·00 ± 0·00 | 0·12 ± 0·12 | ||
Immature Diptera: | |||||
Tipula sp.(p.) | 5·51 ± 1·93 | *** | 1·03 ± 0·56 | 0·03 ± 0·03 | |
Others | 0·77 ± 0·20 | 0·48 ± 0·19 | 0·45 ± 0·22 | ||
Coleoptera larvae: | |||||
Elateridae | 0·52 ± 0·13 | *** | 0·01 ± 0·01 | * | 0·16 ± 0·10 |
Others | 0·69 ± 0·35 | * | 0·06 ± 0·04 | 0·17 ± 0·12 |
Total soil macroinvertebrate biomass in recently winter-flooded grasslands was significantly lower than in traditionally flooded grasslands (Table 3), and in the majority of cases was less than 5 gm−2 (Fig. 3). Mean biomasses of Octolasion tyrtaeum, Octolasion cyaneum, Lumbricus rubellus and Elateridae larvae were all significantly lower in winter-flooded parts of fields of recently flooded grasslands than in traditionally flooded grasslands. It is noticeable that even the traditionally flooded grasslands that supported high densities of breeding snipe also had low soil macroinvertebrate biomass compared with unflooded grassland.

Soil macroinvertebrate biomass in fields at traditionally flooded grasslands (black, n = 10) and in winter-flooded parts of fields at recently flooded grasslands (grey, n = 14).
One of the intended benefits of raising water levels on lowland wet grassland is to decrease the penetration resistance of the upper soil so as to make it soft enough for snipe to probe for macroinvertebrates. Overall, there was no significant difference in median penetration resistance between unflooded and winter-flooded parts of fields at the recently flooded grasslands sampled (T = 29·5, n = 13, P = 0·2632). In four of the recently flooded grasslands, mean penetration resistance was significantly lower in winter-flooded parts of the field than in unflooded parts of the same field. In two fields it was significantly higher, and in eight fields there was no significant difference (Table 4).
Mean penetration resistance ± SE (kg) | |||||
---|---|---|---|---|---|
Site | Soil | Unflooded | Winter-flooded | t | P |
Berney Marshes | CG | 10·2 ± 0·5 | 10·1 ± 0·6 | 0·160 | 0·8755 |
Buckenham | CG | 6·5 ± 0·4 | 11·8 ± 0·5 | 9·142 | < 0·0001 |
Castle Marshes | CG | 9·8 ± 0·6 | 8·8 ± 0·4 | 1·330 | 0·2000 |
Church Farm Marshes | AG | 8·9 ± 0·5 | 8·1 ± 0·4 | 1·272 | 0·2195 |
Elmley | AG | 10·7 ± 0·4 | 9·9 ± 0·7 | 1·097 | 0·2870 |
Heigham Holmes | CG | 4·9 ± 0·1 | 5·1 ± 0·3 | 0·666 | 0·5141 |
Hickling, field a | CG | 10·2 ± 0·5 | 10·7 ± 0·8 | 0·578 | 0·5703 |
Hickling, field b | CG | 9·8 ± 0·5 | 8·1 ± 0·4 | 2·683 | 0·0152 |
Holkham | AG | 4·7 ± 0·7 | 4·2 ± 0·5 | 0·544 | 0·5930 |
Old Hall Marshes | AG | > 16·3* | 9·3 ± 0·4 | – | – |
Strumpshaw | EP | 3·8 ± 0·2 | 5·9 ± 0·4 | 5·745 | < 0·0001 |
West Sedgemoor, field a | AG | 11·0 ± 0·3 | 9·8 ± 0·3 | 2·265 | 0·0361 |
West Sedgemoor, field b | AG | 11·3 ± 1·0 | 8·2 ± 0·3 | 2·740 | 0·0134 |
Whiteslea | EP | 7·4 ± 0·2 | 6·5 ± 0·3 | 2·593 | 0·0184 |
Green (1988) found that penetration resistance of the soil tended to increase during the spring and summer, and that snipe stopped nesting once the mean penetration resistance of the soil in the vicinity of the nest exceeded c. 5·8 kg. In many recently flooded grasslands, both unflooded and winter-flooded parts of the field had a mean penetration resistance greatly in excess of 5·8 kg at the beginning of the breeding season, suggesting that they would have been too hard for breeding snipe to feed in.
At the six sites where water table depth was measured, areas of fields that had been flooded during winter had a significantly higher median water table in early spring than unflooded parts of the rest of the field (T = 0, n = 6, P = 0·0277). Despite this, penetration resistance in winter-flooded areas could be lower than, similar to or higher than that in the rest of the field (Fig. 4 and Table 4[link]). At Strumpshaw, visual inspections suggested that surface water only tended to remain on areas of the field that had a higher mineral content, and this was the most likely reason for the higher penetration resistance in the winter-flooded areas. At Buckenham the winter-flooded parts of the field had consolidated during flooding, and this probably increased its penetration resistance.

Relationships between penetration resistance and water table height in different parts of the same field. Black squares = winter-flooded areas; white squares = unflooded areas. For water table height, positive values indicate the depth of water above the soil surface, and negative values the depth of water below the soil surface. (a) West Sedgemoor, site a (pelo-alluvial gley); (b) Whiteslea (earthy eu-fibrous peat); (c) Hickling, site a (pelo-calcareous gley); (d) Holkham (pelo-calcareous gley); (e) Buckenham (pelo-calcareous gley); (f) Strumpshaw (earthy eu-fibrous peat).
Relative preferences of soil macroinvertebrates for flooded and unflooded soil
Earthworms comprised 99·9% of the soil macroinvertebrate biomass in the 16 soil samples. For the four most abundant earthworm species, the percentage of their biomass in the flooded lower halves of the ‘half-flooded’ soil samples was significantly less than that in the lower halves of the unflooded controls (Table 5), indicating that these species strongly avoided flooded soil and would quickly vacate it following the onset of flooding. The flooded halves of the half-flooded soil samples had a significantly higher percentage soil moisture content than their upper halves (lower halves = 78·8 ± 5·2, upper halves = 54·8 ± 2·1, paired t = 4·423, n = 8, P < 0·0001). There was no significant difference in percentage soil moisture between the upper and lower halves of the unflooded controls (lower halves = 55·3 ± 1·8, upper halves = 52·9 ± 1·9, paired t = 1·015, n = 8, P = 0·344).
Percentage of the total soil macroinvertebrate biomass in each soil sample ± SE | ||||
---|---|---|---|---|
Taxa | Unflooded lower halves of unflooded control soil samples | Flooded lower halves of half flooded soil samples | t | P |
All taxa | 48·9 ± 3·6 | 8·4 ± 2·0 | 8·602 | < 0·0001 |
Allolobophora chlorotica | 41·0 ± 9·8 | 12·4 ± 4·0 | 2·934 | 0·0109 |
Aporrectodea caliginosa | 49·7 ± 12·4 | 2·8 ± 2·8 | 4·335 | 0·0007 |
Aporrectodea longa | 54·0 ± 3·1 | 9·1 ± 2·8 | 7·843 | < 0·0001 |
Lumbricus castaneus | 41·9 ± 7·3 | 1·3 ± 0·5 | 6·882 | < 0·0001 |
Ability of macroinvertebrates to survive in flooded soil
An unexpected effect of the experimental flooding of soil samples was the emergence of over-wintering arthropods from them, particularly during the first 10 days of flooding (Fig. 5a). Significantly more arthropods were collected from the water surface and insides of the polythene bags of the flooded soil samples during this 10-day period than from the soil surface and vegetation of the unflooded controls (flooded = 3·08 ± 0·87, unflooded = 0·00, t = 3·377, P = 0·0015). The majority of displaced arthropods were Staphylinidae (69·3%), Coleoptera larvae (12·1%), and Araneae (6·5%). Most adult Diptera collected from the flooded soil samples emerged during April, after between 85 and 115 days of flooding. Fifty-two of the 53 adult Diptera that emerged from the 10 unflooded soil samples during the second half of April and first half of May were Bibio sp.(p.). This represented a mean density of emerging adult Bibio sp.(p.) of 195 ± 87 m−2. No adult Bibio sp.(p.) emerged from any of the soil samples that had been flooded (Table 6).

Mean numbers of arthropods and biomasses of earthworms expelled from flooded (black) and unflooded (hatched) soil samples. (a) Arthropods; (b) earthworms. Sample sizes for flooded soil samples are 40 for days 0–60, 30 for days 64–90 and 20 for days 95–120. Sample size for unflooded soil samples is 10. Bars show ± 1 SE.
Length of time flooded (days) | |||||||
---|---|---|---|---|---|---|---|
Taxa | 0 | 60 (gradual) | 90 (gradual) | 120 (gradual) | 120 | F | P |
Arthropods other than adult Diptera | 0·2 ± 0·1a | 1·9 ± 1·0ab | 4·5 ± 2·4cb | 5·2 ± 2·3cb | 2·4 ± 0·9ab | 4·070 | 0·0080 |
Adult Bibio sp.(p.) | 5·2 ± 2·3d | 0·0e | 0·0e | 0·0e | 0·0e | 5·793 | 0·0010 |
Other adult Diptera | 0·1 ± 0·1 | 1·7 ± 1·3 | 1·0 ± 0·3 | 0·3 ± 0·2 | 1·3 ± 0·9 | 1·321 | 0·2807 |
The only soil macroinvertebrates displaced from the soil by flooding were earthworms, and the majority of these were Lumbricus castaneus (Savigny 1826) (Fig. 5b and Table 7[link]). Although lower biomasses were removed from the unflooded controls, there were no significant differences between the total biomass of soil macroinvertebrates removed from the different treatments.
Length of time flooded (days) | |||||||
---|---|---|---|---|---|---|---|
Taxa | 0 | 60 (gradual) | 90 (gradual) | 120 (gradual) | 120 | F | P |
Removed: | |||||||
All taxa | 0·03 ± 0·01 | 0·15 ± 0·07 | 0·15 ± 0·10 | 0·30 ± 0·13 | 0·08 ± 0·05 | 1·486 | 0·2268 |
Lumbricus castaneus | 0·03 ± 0·01 | 0·12 ± 0·06 | 0·11 ± 0·08 | 0·21 ± 0·10 | 0·06 ± 0·03 | 1·018 | 0·4112 |
Remaining: | |||||||
All taxa | 7·05 ± 0·86ab | 8·04 ± 1·14a | 8·59 ± 0·94c | 4·33 ± 0·87bc | 3·39 ± 0·31cd | 8·917 | < 0·0001 |
Allolobophora chlorotica | 0·72 ± 0·14 | 1·11 ± 0·16 | 1·13 ± 0·18 | 0·89 ± 0·18 | 0·85 ± 0·15 | 0·859 | 0·4980 |
Aporrectodea caliginosa | 2·09 ± 0·40ef | 2·60 ± 0·31e | 1·60 ± 0·30eg | 0·93 ± 0·26fgh | 0·29 ± 0·15h | 12·735 | < 0·0001 |
Aporrectodea longa | 2·66 ± 0·56 | 3·40 ± 0·82 | 4·01 ± 0·86 | 2·06 ± 0·52 | 1·99 ± 0·33 | 1·651 | 0·1828 |
Aporrectodea rosea | 0·40 ± 0·13 | 0·23 ± 0·15 | 0·18 ± 0·08 | 0·23 ± 0·07 | 0·10 ± 0·06 | 1·727 | 0·1654 |
Lumbricus castaneus | 0·72 ± 0·13i | 0·69 ± 0·16i | 1·05 ± 0·15i | 0·12 ± 0·08j | 0·05 ± 0·03j | 19·831 | < 0·0001 |
Flooding resulted in little reduction in soil macroinvertebrate biomass. Only samples flooded for 120 days had a significantly lower biomass than the unflooded controls (Table 7). All of the most abundant earthworm species in the samples survived in flooded soil for 120 days. Two species, Lumbricus castaneus and Aporrectodea caliginosa, showed possible reductions in biomass as a result of flooding, although the latter only occurred at a significantly lower biomass in one of the two treatments that were flooded for 120 days.
Comparisons of the temperature at the surface of the immersed soil samples suggested that the containment of soil samples within polythene bags did not greatly increase the temperature of the flood water compared with under ‘natural’ flooding conditions. The mean minimum and maximum ground temperatures during the experiment were 0·3 °C and 9·0 °C, respectively, for the first 30 days, increasing to 2·7 °C and 17·6 °C, respectively, for the last 30 days. Mean water temperature at the soil surface was only found to be significantly different (P < 0·05) between open and closed polythene bags on two occasions: early morning following a clear night during the first 30-day period (closed bags = 1·8 ± 0·1 °C, open bags = 1·3 ± 0·1 °C, t = 3·411, n = 6, P = 0·0066) and on a sunny afternoon during the last 30-day period (closed bags = 26·6 ± 0·5 °C, open bags = 22·7 ± 0·9 °C, t = 3·680, n = 6, P = 0·0042). This suggested that conditions in the polythene bags were similar to those under natural flooding. However, it was still possible that oxygen diffusion through the surface water from the air above was less than under natural flooding, because of restricted air and water movement within the sealed bags.
Changes in soil macroinvertebrate distribution and physical availability to breeding wading birds as flood water subsides
Figure 6 shows physical conditions and vegetation height and cover along the six transects in March and May. Comparisons between mean physical conditions and vegetation height and cover along the entire lengths of the unflooded and winter-flooded sections of the transects in March and May are presented in Table 8.

Mean environmental variables along six transects running perpendicularly across the margins of unflooded grassland in March (white squares or hatched bars) and May (black squares or black bars). (a) Water table depth relative to the soil surface; (b) soil moisture; (c) penetration resistance; (d) vegetation height; (e) percentage bare ground. Positive and negative values of water table height indicate the water table above and below the soil surface, respectively. Bars show ± 1 SE.
Mean ± SE | |||||
---|---|---|---|---|---|
Variable | March | May | t | P | |
Water table depth below the soil surface (cm) | Unfl. | −21·4 ± 6·0 | −43·0 ± 4·4 | −2·444 | 0·0013 |
W. fl. | −3·3 ± 2·1 | −29·9 ± 3·1 | −6·072 | 0·0018 | |
Soil moisture (% of dry weight) | Unfl. | 109·0 ± 26·8 | 84·5 ± 16·3 | 2·351 | 0·0655 |
W. fl. | 123·9 ± 30·5 | 84·6 ± 14·2 | 2·290 | 0·0706 | |
Penetration resistance (kg) | Unfl. | 5·48 ± 0·84 | 7·31 ± 0·62 | −3·755 | 0·0132 |
W. fl. | 5·51 ± 0·36 | 7·09 ± 0·23 | −2·982 | 0·0307 | |
Vegetation height (cm) | Unfl. | 1·35 ± 0·31 | 3·42 ± 0·54 | −8·278 | 0·0004 |
W. fl. | 1·27 ± 0·36 | 1·63 ± 0·27 | −0·726 | 0·5004 | |
Bare ground (index out of 100) | Unfl. | 1·95 ± 0·80 | 0·83 ± 0·43 | 0·738 | 0·1497 |
W. fl. | 12·22 ± 4·16 | 19·72 ± 5·36 | −0·889 | 0·4147 |
As the flood water subsided, the water table fell significantly at each sampling point along the transects, falling relatively more towards the formerly flooded ends. Mean percentage soil moisture tended to be lower in May than in March, but these differences were not quite significant along each half of the transects. Mean penetration resistance was significantly higher along the unflooded and formerly flooded lengths of the transects during May than it had been during March.
During March the vegetation was consistently short along the whole lengths of the transects. The fields had been heavily grazed by cattle and sheep the previous year to produce suitable conditions for grazing wildfowl, particularly wigeon Anas penelope (L.), which in turn had maintained a short sward over winter. The sward was also relatively open along the transects, there being a high proportion of unvegetated ground, particularly in the areas that had been flooded. By May the vegetation had grown significantly higher along the unflooded lengths of the transects and on the extreme margins of the winter-flooded grassland. These areas were dominated by vigorous, agriculturally productive, grass species, particularly Lolium perenne (L.). The formerly flooded lengths of the transects largely comprised the low-growing grass Agrostis stolonifera (L.). There were no significant changes in the percentage of bare ground between March and May.
Changes in soil macroinvertebrate biomass along the transects between March and May are shown in Fig. 7. Mean total soil macroinvertebrate biomass remained low along the formerly flooded sections of the transects in March and May, although it was slightly higher within a metre of the former flood margin. This distribution was shown by most taxa: Allolobophora chlorotica, Aporrectodea caliginosa, Aporrectodea longa, Aporrectodea rosea, Lumbricus castaneus, Tipula sp.(p.) larvae and ‘other immature Diptera’. The highest biomass of the semi-aquatic Eiseniella tetraedra was around the former flood margin, but this species still only occurred at low biomass within the winter-flooded grassland further than 0·5 m from the former flood margin. Tipula pierrei (Tonnoir in Goetghebuer & Tonnoir 1921) larvae, Eristaline larvae and Chironomidae larvae were only found in the formerly flooded grassland.

Mean soil macroinvertebrate biomass in the upper 10 cm of soil along six transects running perpendicularly across the margins of unflooded grassland in March (hatched) and May (black). (a) All taxa; (b) Allolobophora chlorotica; (c) Aporrectodea caliginosa; (d) Aporrectodea longa; (e) Aporrectodea rosea; (f) Eiseniella tetraedra; (g) Lumbricus castaneus; (h) Tipula sp.(p.) (excluding T. pierrei); (i) main aquatic immature Diptera (Chironomidae larvae, Tipula pierrei larvae and Eristaline larvae and pupae; (j) other immature Diptera. Bars show ± 1 SE.
There were no significant differences in total soil macroinvertebrate biomass between March and May at any sampling point along the transects, or along the unflooded and formerly flooded lengths of the transects as a whole (Table 9). The only taxa to show significant differences in biomass were Eiseniella tetraedra and the main aquatic immature Diptera. The only remaining individuals of the latter found in May were buried Eristaline pupae.
Mean formaldehyde preserved biomass ± SE (gm−2) | |||||
---|---|---|---|---|---|
Taxa | March | May | t | P | |
(a) Upper 3 cm of the soil | |||||
All taxa | Unfl. | 384·6 ± 47·9 | 310·1 ± 27·8 | 1·378 | 0·2266 |
W. fl. | 41·3 ± 15·5 | 74·3 ± 14·0 | −1·644 | 0·1610 | |
Allolobophora chlorotica | Unfl. | 59·6 ± 13·8 | 47·1 ± 15·2 | 0·995 | 0·3653 |
W. fl. | 4·2 ± 1·9 | 9·3 ± 4·2 | −0·392 | 0·8514 | |
Aporrectodea caliginosa | Unfl. | 62·8 ± 21·4 | 36·9 ± 16·3 | 1·744 | 0·1416 |
W. fl. | 1·8 ± 1·1 | 4·2 ± 2·0 | −1·419 | 0·2150 | |
Aporrectodea longa | Unfl. | 5·7 ± 2·0 | 8·8 ± 1·7 | −1·491 | 0·1962 |
W. fl. | 1·2 ± 1·0 | 3·0 ± 1·9 | −1·136 | 0·3076 | |
Aporrectodea rosea | Unfl. | 1·5 ± 1·1 | 0·7 ± 0·7 | 1·346 | 0·2361 |
W. fl. | 0·2 ± 0·2 | 0·0 | 1·000 | 0·3632 | |
Eiseniella tetraedra | Unfl. | 44·9 ± 18·9 | 13·1 ± 5·7 | 4·068 | 0·0097 |
W. fl. | 16·9 ± 14·6 | 9·8 ± 6·5 | −0·312 | 0·7676 | |
Lumbricus castaneus | Unfl. | 36·2 ± 20·3 | 38·7 ± 22·3 | 1·019 | 0·3550 |
W. fl. | 0·0 | 5·5 ± 4·6 | −1·474 | 0·2004 | |
Tipula sp.(p.) larvae | Unfl. | 113·5 ± 59·0 | 98·1 ± 25·1 | −0·886 | 0·4159 |
W. fl. | 8·6 ± 6·3 | 35·3 ± 19·3 | −1·080 | 0·3294 | |
Main aquatic immature | Unfl. | 1·2 ± 1·2 | 0·5 ± 0·5 | 1·000 | 0·3632 |
Diptera | W. fl. | 7·7 ± 3·3 | 0·6 ± 0·6 | 4·599 | 0·0058 |
Other immature | Unfl. | 7·1 ± 2·4 | 4·2 ± 0·7 | 0·925 | 0·3976 |
Diptera | W. fl. | 0·9 ± 0·3 | 2·3 ± 1·0 | −1·123 | 0·3123 |
(b) Upper 10 cm of the soil | |||||
All taxa | Unfl. | 558·7 ± 61·2 | 461·9 ± 30·7 | 1·254 | 0·2652 |
W. fl. | 62·0 ± 16·8 | 84·3 ± 18·6 | −0·842 | 0·4384 | |
Allolobophora chlorotica | Unfl. | 86·3 ± 13·7 | 60·7 ± 19·4 | 2·303 | 0·0696 |
W. fl. | 8·2 ± 2·9 | 10·2 ± 4·7 | 0·197 | 0·8514 | |
Aporrectodea caliginosa | Unfl. | 87·2 ± 24·6 | 96·2 ± 24·4 | −0·494 | 0·6421 |
W. fl. | 1·8 ± 1·1 | 6·9 ± 4·0 | −1·433 | 0·2113 | |
Aporrectodea longa | Unfl. | 19·8 ± 4·6 | 21·5 ± 5·4 | −0·526 | 0·6212 |
W. fl. | 5·2 ± 2·4 | 6·0 ± 4·0 | 0·410 | 0·6985 | |
Aporrectodea rosea | Unfl. | 14·3 ± 6·9 | 5·3 ± 2·6 | 1·929 | 0·1117 |
W. fl. | 1·0 ± 1·0 | 0·2 ± 0·2 | 0·564 | 0·5973 | |
Eiseniella tetraedra | Unfl. | 49·7 ± 21·1 | 13·3 ± 5·8 | 4·066 | 0·0097 |
W. fl. | 17·6 ± 15·1 | 9·8 ± 6·5 | −0·193 | 0·8544 | |
Lumbricus castaneus | Unfl. | 37·3 ± 21·0 | 41·0 ± 22·8 | 0·844 | 0·3550 |
W. fl. | 0·0 | 5·5 ± 4·6 | −1·474 | 0·2004 | |
Tipula sp.(p.) larvae | Unfl. | 120·5 ± 65·7 | 104·6 ± 27·5 | −0·959 | 0·3816 |
W. fl. | 8·6 ± 6·3 | 38·1 ± 21·8 | −1·107 | 0·3816 | |
Main aquatic immature | Unfl. | 1·2 ± 1·2 | 0·5 ± 0·5 | 1·000 | 0·3632 |
Diptera | W. fl. | 8·2 ± 3·2 | 0·6 ± 06 | 4·581 | 0·0059 |
The majority of soil macroinvertebrate biomass in the top 10 cm of soil was within 3 cm of the soil surface, both during March and May (Table 10). Eiseniella tetraedra, Lumbricus castaneus and all immature Diptera including Tipula sp.(p.) were more or less restricted to within 3 cm of the soil surface. Aporrectodea longa and Aporrectodea rosea were mainly 3 cm or more below the soil surface. Allolobophora chlorotica and Aporrectodea caliginosa showed vertical distributions intermediate between these two extremes. Two species, Aporrectodea longa and Eiseniella tetraedra, appeared to retreat from the soil surface as the season progressed. Aporrectodea caliginosa showed a similar, but not significant, trend.
Mean percentage of biomass in the upper 3 cm of the top 10 cm of soil ± SE (gm−2) | ||||
---|---|---|---|---|
Taxa | March | May | t | P |
All taxa | ||||
Unfl. | 69·5 ± 4·9 | 67·5 ± 4·7 | 0·375 | 0·7233 |
W. fl. | 66·3 ± 10·8 | 92·4 ± 4·9 | −1·962 | 0·1070 |
Allolobophora chlorotica | 66·0 ± 7·2 | 79·7 ± 6·4 | −1·857 | 0·1225 |
Aporrectodea caliginosa | 74·6 ± 11·4 | 38·6 ± 12·2 | 2·310 | 0·0820 |
Aporrectodea longa | 24·9 ± 7·6 | 64·1 ± 10·5 | −2·798 | 0·0489 |
Aporrectodea rosea | 16·1 ± 13·9 | 9·6 ± 9·6 | – | – |
Eiseniella tetraedra | 93·1 ± 2·1 | 99·6 ± 0·4 | −3·142 | 0·0348 |
Lumbricus castaneus | 98·7 ± 0·8 | 88·9 ± 7·8 | – | – |
Tipula sp.(p.) larvae | 98·5 ± 1·5 | 95·4 ± 2·2 | 1·520 | 0·1890 |
Main aquatic immature Diptera | 93·0 ± 7·0 | 100·0 | – | – |
Other immature Diptera | 100·0 | 100·0 | – | – |
Discussion
The soil macroinvertebrate fauna of lowland wet grassland and the effects of introducing winter flooding
Grasslands with a long history of winter flooding have a markedly different fauna to that of unflooded grasslands where attempts have been made to raise water levels to benefit breeding wading birds and other waterfowl. Although lowland wet grassland soil macroinvertebrates are capable of surviving long periods of flooding, this ability is restricted to just a few species. The most important of these in terms of biomass are the earthworms Allolobophora chlorotica, Eiseniella tetraedra, Lumbricus rubellus and Octolasion tyrtaeum. Various combinations of these four species have also been found to comprise the majority of the soil macroinvertebrate biomass in other winter-flooded peat and mineral soils (Cotton & Curry 1980; Baker 1983; A.L. Reid & T.G. Piearce, unpublished data).
Of the four most abundant earthworm species capable of surviving long periods of flooding, two species, Allolobophora chlorotica and Lumbricus rubellus, were also both common in unflooded grasslands. Lumbricus rubellus is found in a wide range of habitats, but mainly those with a high organic and soil moisture content (Sims & Gerard 1985). The green morph of A. chlorotica is typical of wet soil and is replaced by its unpigmented form in drier conditions (Satchell 1967a; Sims & Gerard 1985). The other most abundant earthworm species in traditionally flooded grassland, Octolasion tyrtaeum and Eiseniella tetraedra, were less frequently found in samples taken from unflooded grasslands. Eiseniella tetraedra and Octolasion tyrtaeum are capable of surviving for long periods underwater, and are regularly found in rivers (Sims & Gerard 1985). Both species possess quadrangular caudal ends that they move in the water to maintain gas exchange (Bouché 1970). Octolasion tyrtaeum is thought to survive flooding better than other widely distributed soil-inhabiting earthworm species, due to its well-developed subcutaneous net of blood vessels and high concentrations of haemoglobin, which enable it to inhabit badly aerated soils (Perel 1977).
The earthworm-dominated fauna of the unflooded grasslands was similar to that of other lowland pasture (Guild 1951; Gerard 1967) but differed in the presence of the green, rather than unpigmented, morph of Allolobophora chlorotica and in the lower abundance of Lumbricus terrestris.
The biomass of most soil macroinvertebrate taxa was greatly reduced in areas subject to winter flooding, although it was clear that several species of earthworms were capable of surviving for long periods in flooded soils if forced to remain in them. Earthworms are thought to avoid flooded soils because of their lack of oxygen and, once anoxic conditions have developed, the presence of noxious gases such as hydrogen sulphide (Mather & Christensen 1988). The extremely low soil macroinvertebrate biomass in many recently flooded parts of fields compared with most traditionally flooded grasslands could therefore have been due to the former suffering particularly severe oxygen depletion. The rate at which soil oxygen is depleted depends on temperature, the availability of organic matter for microbial respiration, and sometimes on the chemical oxygen demand of reductants in the soil, such as ferrous iron (Gambrell & Patrick 1978). The recently flooded grasslands tended to be flooded for shorter durations than the traditionally flooded grasslands, and in particular for shorter periods during spring and autumn (when temperatures are higher than in winter). Therefore, duration of flooding and soil temperature are unlikely to be the cause of observed differences in biomass. Recently flooded grasslands may have contained a greater volume of organic matter than traditionally flooded grasslands. Large ‘flushes’ of nutrients are often released from decaying vegetation when areas are first flooded (Andersson 1982; Danell & Sjoberg 1982). It was noticeable that many sites that had been flooded for the first time in recent years had a thick, black, anoxic litter layer consisting of dead flooding-intolerant grass species such as Lolium perenne (L.), while traditionally flooded grasslands had no noticeable litter layer. It might be possible to prevent anoxic conditions from developing by removing most of the vegetation prior to flooding (by grazing or mowing), or by allowing water levels to fluctuate so that oxygen supply is replenished during periods of drying out.
Another possible difference between recently flooded and traditionally flooded grasslands was that most of the former were on minerals soils, while the majority of the latter were on peat. It is possible that differences in soil type could at least partially explain differences in abundance of soil macroinvertebrates in flooded grasslands, and this might repay further investigation.
The results showed that earthworms will try to vacate soil as it becomes flooded. The ratio of soil macroinvertebrate biomass in flooded compared with unflooded soil in the experiment (1 : 11·9) was similar to that found in flooded and unflooded parts of fields (1 : 10·2). Thus movement of earthworms out of flooded soils could explain the reduction in biomass found in winter-flooded areas. Earthworms are capable of moving long distances above ground. Mather & Christensen (1992) found that Allolobophora chlorotica, Aporrectodea caliginosa, Aporrectodea longa, Aporrectodea rosea and Lumbricus terrestris routinely made overland forays at night even under normal conditions, and Darwin (1881) recorded overland forays of earthworms of up to 13 m. In many situations, earthworms are likely to be able to vacate grassland as quickly as it floods.
Earthworms displaced by flooding will presumably initially become concentrated in the upper soil of adjacent unflooded grassland. Such concentrations of earthworms would explain the well-known attraction of soil macroinvertebrate-feeding birds, particularly lapwings, golden plovers Pluvialis apricaria (L.), gulls Laridae, starlings Sturnus vulgaris (L.) and thrushes Turdidae, to the margins of flooded grassland. Concentrations of earthworms in adjacent unflooded grassland may be significantly reduced by birds such as wintering lapwings and golden plovers (Bengtson et al. 1976; Barnard & Thompson 1985). The green morph of Allolobophora chlorotica has been found to display a cryptic advantage over its unpigmented pink morph (Satchell 1967a), and this would presumably confer a selective advantage over most other earthworm species during periods of bird predation. Lumbricus castaneus, which was the only earthworm species to leave the soil regularly and move through the floodwater above it, would be highly vulnerable to predation by waterbirds such as gulls, grey herons Ardea cinerea (L.) shoveler Anas clypeata (L.) and mallard Anas platyrhynchos (L.) in natural flood conditions.
The flooding experiment showed that many overwintering arthropods would emerge following the onset of winter flooding. These would probably accumulate on unsubmerged vegetation and also be blown to the margins of flooded areas. Being relatively cold, and therefore slow-moving, they would be easily preyed upon by insectivorous birds, especially meadow pipits Anthus pratensis (L.) and pied wagtails Motacilla alba (L.).
There was no significant recolonization of winter-flooded grassland by soil macroinvertebrates during the spring. This is not surprising, given the slow colonization rates of grassland (2·5–10 m year−1) by species such as Aporrectodea caliginosa and Allolobophora chlorotica (Rhee 1969; Hoogerkamp, Rogaar & Eijsackers 1983). Recolonization of winter-flooded grassland might have been inhibited by adverse changes to the habitat there caused by winter flooding. Flooding can result in soil compaction, consolidation and loss of soil structure, which will impair air and water movement through the soil and impede earthworm movement (Piearce 1984). Recovery of soil structure is largely brought about by the actions of earthworms themselves (Satchell 1967b).
Implications for managing lowland wet grassland for breeding wading birds
This study has shown that winter flooding greatly reduces the soil macroinvertebrate prey of breeding wading birds, but that many of the highest densities of breeding wading birds are found on sites with only low densities of soil macroinvertebrates.
The results of the experiments suggest that the main reason for the decreased biomass of soil macroinvertebrates in flooded soils is simply due to them vacating flooded soils. However, once the majority of earthworms have sought refuge in unflooded soil, prolonging the duration of flooding is unlikely to greatly further reduce earthworm biomass.
For snipe, the overall effects of flooding will be to decrease their soil macroinvertebrate prey. Snipe are thought to require soft soil in which to probe for macroinvertebrates (Green 1988; Green, Hirons & Cresswell 1990). Flooding may in some cases increase the physical availability of the prey to snipe by helping keep the upper soil moist and therefore soft enough for them to probe. However, the wettest and therefore softest soil is likely to be that most recently uncovered by the retreating flood water. As this study has shown, such areas will only have very low soil macroinvertebrate biomass. An alternative way of providing suitable feeding conditions for snipe is to keep the upper soil moist and soft enough for snipe to probe for macroinvertebrates without flooding it beforehand. The ability to do this varies with the hydraulic conductivity of the soil. Undamaged peats tend to have high hydraulic conductivity. On undamaged peats it is possible to keep the upper soil moist and soft through lateral movement of water through the soil from surrounding water-filled ditches, particularly if these ditches are closely spaced (Silsoe College 1989; Youngs, Leeds-Harrison & Chapman 1989; Armstrong 1993). Clays usually have lower hydraulic conductivity, and it is rarely possible to keep the upper soil soft by maintaining high water levels in surrounding ditches (Armstrong 1993).
Flooding in winter and early spring is likely to improve the physical conditions for feeding lapwings and redshank. Lapwings strongly select short vegetation, usually less than 15 cm high (Klomp 1954; Lister 1964; Redfern 1982; Galbraith 1988; Ausden 1996a), and redshank also select shorter swards (Ausden 1996a). Flooding agriculturally improved grassland during winter decreases vegetation height the following spring, largely through the replacement of vigorous agriculturally improved swards with Agrostis stolonifera-dominated inundation grassland (National Vegetation Classification community MG13, Agrostis stolonifera–Alopecurus geniculatus grassland; Rodwell 1992). The height of this MG13 grassland during late spring and early summer is itself negatively correlated with the duration of spring flooding (Ausden 1996a). Therefore, the shortest, most open, vegetation is most likely to be that recently uncovered by the retreating flood water. Such areas will only have a very low biomass of soil macroinvertebrates. Lapwings and redshank also feed on aquatic Diptera larvae (Ausden 1996a; Johansson & Blomqvist 1996), and winter-flooded grassland will provide an additional food source for them in the form of Chironomidae and other aquatic immature Diptera while flood water is still present.
An alternative way of providing suitably short vegetation for feeding lapwings and redshank is through heavy grazing or by reducing soil fertility. Heavy grazing has the disadvantage that stock trample the nests of wading birds and other ground-nesting species (Beintema & Muskens 1987; Green 1988). In the Netherlands trampling of lapwing and black-tailed godwit nests is reduced using nest protectors (Guldemond, Parmentier & Visbeen 1993), although these cannot be used effectively to protect the more cryptic nests of other species such as redshank, snipe and ground-nesting passerines. Cessation of fertilizer use can reduce sward productivity within a few years, although the former unimproved vegetation (which is usually structurally more open) can take very much longer to re-establish (Berendse et al. 1992; Mountford et al. 1994).
In conclusion, the best feeding conditions for breeding snipe are likely to be provided by maintaining a high water table on peat soils without flooding them beforehand. The best feeding conditions for breeding lapwings and redshank will probably be provided by creating a mosaic of unflooded grassland, winter-flooded grassland and shallow pools on peat, clay or other soils. Unflooded grassland can provide a high biomass of soil macroinvertebrates beneath short vegetation in early spring. Winter-flooded grassland can provide short, open, conditions for lapwings and redshank to feed in during the latter part of the breeding season when the vegetation has become too tall for them to forage in elsewhere. However, winter-flooded grassland will only contain low biomasses of soil macroinvertebrates. Shallow pools will provide an alternative source of aquatic invertebrate prey. Such a mosaic of hydrological conditions is also likely to benefit a range of other wetland biodiversity.
Acknowledgements
This work was funded by RSPB and the Broads Authority. We would also like to thank Graham Hirons and Jane Madgwick and are very grateful to English Nature, the National Trust, Norfolk Wildlife Trust and Suffolk Wildlife Trust for allowing access to their land. The referees made useful suggestions.
References
Received 14 January 2000; revision received 29 October 2000