Assessing for restoration potential of an ex-agricultural amenity site

Assessing for restoration potential of an ex-agriculturalamenity site

By M. STAMP, D. HEWITT, S. BLACKBIRD, L. CONNOR,H. McALLISTER, A. HENDERSON, T. BARKER, R. MARRS

Applied Vegetation Dynamics Laboratory, School of EnvironmentalSciences, University of Liverpool, Liverpool L69 3GP

Summary

High soil fertility is a common impediment for restoringspecies-rich vegetation on ex-agriculture land. Here, weassessed species abundance and peak-summer biomass of theexisting vegetation, soil chemical composition (four depths) andthe composition of the seed bank. The current vegetation wascomposed mainly of grasses with few forbs. The vegetation wasdescribed as National Classification (NVC) class OV21d, whereasthe target NVC class was either MG1a or MG5d. The seeds bankcontained 4 graminoids and two forbs, all at low densities. Thesoil chemical properties varied spatially and by depth dependingon variable measured, but most were within the ranges ofvariables found in typical semi-natural grasslands elsewhere inGreat Britain. It is suggested that on this site the limitingfactor is not soil fertility but rather a lack of propagules ofthe important dicotyledon species; the propagules are not in thesoil and/or cannot establish in the existing vegetation. Furtherexperiments are needed to test this hypothesis.

Key words: conservation, high soil fertility, agricultural reversion, seed banks, soil nutrients, species composition, NVC,Elleneberg values

Introduction

There is an increasing desire amongst many communities in theurban-rural fringe to improve their environment, and often whenland becomes available there is a desire to create wildflowermeadows to provide wildlife and aesthetic benefits. In somecases, the land has been used for agricultural purposes andthere is a residual high fertility in the soil which impededrestoration attempts (Marrs, 1993). However, there are fewguidelines to help in such restoration attempts. Here, wedescribe a baseline survey of a site in north-west England where

1

local residents have taken over the management of a field withthe aim of restoring a wildflower meadow with a highconservation and amenity value. This case study site it the Cow Field within the “Park

Fields” complex on Wirral (Fig. 1).This field is owned by thelocal authority (now Cheshire West & Chester Council) but leasedto farmers, who used the site for cattle grazing. In 2005, allfarming on the site ceased. Now the land is managed under a 10-year memorandum-of-understanding by the “Friends of ParksFields” with the aim of improving, where possible, its valuefor conservation, and ultimately seeking to create a species-rich, lowland wildflower meadow. Since 2006, it has been managedby hay-cropping with no fertilizer addition. The current aim isto create a grassland classified by the National VegetationClassification as MG1a or MG5c (Rodwell 1992); the Arrhenatheretumeliatoris grassland, Festuca rubra sub-community, and the Centaureo-Cynosuretum, cristati grassland, Danthonia decumbens sub-communityrespectively. Nomenclature follows Stace (1997) throughout. Bothoccur in the moist and mild parts of Britain; MG1 tends to bemown and ungrazed and MG5 tends to be mown and autumn- orwinter-grazed, and is forb-rich (Rodwell, 1992). The aims of this study were, therefore, threefold:

(1) To describe the existing grassland vegetation (speciescomposition and peak summer biomass). These data werecompared using Ellenberg N-values (Hill et al., 1996) withliterature values from typical species-rich grasslandcommunities;

(2) To measure selected soil chemical properties associatedwith fertility spatially across the site and by soil depth.These data were compared with similar soil data for typicalNVC grassland communities. The changes in soil propertieswith depth might inform whether soil stripping or deepploughing might assist in fertility reduction (Marrs 1993).

(3) To assess the composition of the soil seed bank. (4) To provide information about the possibility of restoring

the site using seeds buried in the surface layer.

2

Fig. 1. (a) Aerial photograph of the Cow Field showing the sampling blocks.Photograph courtesy of Cheshire West & Chester Council.

Methods

Baseline surveys

The boundary of Cow Field (Longitude 3.0750W; Latitude =52.2968 N, OS: SJ283785) was delimited using a GPS (GARMIN,etrex), mapped and overlain with a 1m2 grid. The grid was splitinto four blocks to account for topographical variability(Figure 1), and within each Block eight 1m2 positions wereselected randomly from the grid (n = 32). These samplingpositions were located at the end of May and soil samplescollected from each position using an ELE International auger (5cm diameter, 15 cm deep, volume=294 cm3) at four depths (0-15cm,15-30cm, 30-45 cm, 45-60cm). At the end of July, near the point of maximum above-ground

biomass and before hay cutting - these positions were relocated.At each point species composition and cover (%) were determined.Thereafter, the vegetation in the central 0.25m2 area washarvested. Both fresh and dry biomass (80oC for 24h) weredetermined and expressed in gm-2. The species richness (number ofplant species m-2) was then calculated. The NVC class for the field were derived using the cover

scores for the site (mean of all quadrats) using TABLEFIT (Hill,1996). The Ellenberg N-values for each species were extractedfrom Hill et al. (1999) for all species on the site, and the meanEllenberg N-value was calculated in raw form and weighted bycover.

3

Subsequently, the literature was reviewed and data on bothsoil chemical properties (see below) and peak summer biomasswere extracted from potential target grassland sites forcomparison.

4

Chemical description of the Cow Field soil

Soil chemical properties were measured on each soil samples(n=32 (4 blocks x 8 positions) x 4 depths; total= 128) usingtechniques outlined in Allen (1989). Soil pH (1:2.5soil:deionised water) and plant-available concentrations ofnitrogen (2M KCL extract) and phosphorus (Truog’s) were measuredon fresh soils. The soil was then air-dried and passed through a2 mm mesh, and the following properties measured: total C and Nconcentrations, and exchangeable concentrations of K, Na, Ca andMg concentrations after extraction in 1M ammonium acetate. K andNa concentrations were estimated by emission spectrophotometryand Ca and Mg by absorption spectrophotometry (Allen, 1989).Total elemental C and N concentrations were measured directly onair-dried soil samples using a Carlo Erba Instruments NC2500elemental analyser. Elemental concentrations were expressed on aconcentration basis (μg g-1) a volumetric basis (μg cm-3) and anarea basis (g m-2). Only the concentration data are presentedhere.

Seed bank analysis

The seed composition and density from each sampling positionwas estimated in 80g sub-samples of fresh soil in two of thedepth layers: surface (0-15 cm) and sub-surface (30-45 cm) usingmethodology described by Ghorbani et al. (2003). The mixed sub-samples were decanted into trays (20.5 cm x 15.5 cm) lined witha mesh net (1mm), and filled with a layer of washed,horticultural sand, Control trays, containing sand alone, werealso included to identify seeds of any foreign species. Thetrays were set out in an unheated glasshouse, on benches coveredwith plastic sheets and capillary matting, which was wateredfrom below using an underbench watering system (3 x 30 minutewatering cycles per day, supplemented by manual watering asnecessary. All seeds establishing during a 20 week period wereidentified, counted, and grouped into functional types(grasses/forbs/rushes). Where identification was uncertain atthe time of counting, the seedlings were transplanted and grownon for accurate identification at a later date.

Statistical analysis

5

All analyses were implemented in the R software environment (RDevelopment Core Team, 2009). To assess the effects of spatialpositions and soil depth on soil properties Linear Mixed Modelswere used within the NLME package (Pinheiro & Bates, 1996). Eachsoil variable was analysed with spatial block and soil depthincluded as fixed factors in the model, and blocks, positionsand soil depth soil treated as nested random effects. The modeldeletion approach was used (Crawley, 2007) to derive the MinimumAdequate Model starting with the full interaction model andterms deleted sequentially using the Maximum Likelihood methodand analysis of variance. Thereafter selected MAM was re-runusing the Restricted Maximum Likelihood Method (REML) (Crawley2007). For the seed bank analysis, the density of plant groups in

each block was plotted as box plots and the total seed density and total densities of functional groups (grasses, forbs, Juncus spp) in each block analyzed using one-way analysis of variance (ANOVA).

Results

Site survey: vegetation

The vegetation was classified as NVC class OV23d (Lolium perenne-Dactylis glomerata community, Rodwell, 2000), albeit with arelatively low goodness-of-fit of 53%. Species richness wasrelatively low with a mean value of 6.5±0.3 (range=4-10). Themean value for the Ellenberg-N-values for the Cow Field wassimilar to the values for the typical OV2d community (Fig, 2),but higher than either of the potential preferred endpointcommunities, either MG1a, or MG5c, (Fig. 2, Rodwell, 1991).Moreover, although there was great overlap between the Cow Fieldand the three typical communities (Fig. 2), the Cow Field wasthe only one with species without species with an Ellenberg-Nvalue of 2, typical of infertile soils, which were found in theother communities. At the Cow Field all species, except Bromushordaceus, had at least a part-competitive strategy (Grime et al.,1988).The mean peak biomass of the Cow Field was 684±22 gm-2 which is

fairly close to values obtained for an unfertilized Holcus lanatus-Juncus effusus rush pasture (604± 34 gm-2; Tallowin & Smith, 2001),although it is three times that for unfertilized cut grasslandin the Netherlands (289±21 gm-2), and even above that of N-fertilized grasslands (545±47 gm-2; Van Der Woude et. al., 1994)

6

Site survey: soils

Soil chemical properties in relation to NVC typesThe mean values for selected soil chemical properties are

compared with data collected from a national survey ofmesotrophic grasslands (Table 1). The Cow Field soil had a lowersoil, pH loss-on-ignition, and concentrations of total N, andexchangeable K. However, C:N ratios and concentrations ofextractable P and exchangeable Mg were in similar ranges.

Assessment of site heterogeneityThis part of the study assessed spatial variability in both

space (block effect) and depth; essentially five differentstatistical models could be selected: (1) a null model (noeffect), (2) a space effect only (no depth effect), (3) a deptheffect (no space effect), (4) an additive effect where there wasboth a space effect and the same response to depth, and (5) aninteraction effect where there was a space effect but there weredifferent depth responses in each block. A summary of theresponses for each soil variable is outlined in Table 2, and twoexamples are provided (Fig.3). Full data along with allstatistical analyses are available in Hewitt (2010).

There was considerable heterogeneity across the site in bothspace and depth. Only three variables, all associated with soilorganic matter content (loss-on-ignition, total C and N), didnot show spatial differences across the site, but all reducedwith soil depth. All other variables showed a significantspatial effect having significantly different concentrations inthe different blocks. Extractable P only had block effects. Thevariables that were best fitted with the additive model showeddifferential responses: (1) C:N ratio, exchangeable K,extractable NO3-N, and rates of both N-mineralization andnitrification all decreased with depth, (2) exchangeable Caincreased with depth, and (3) extractable NH4-N and total N (NH4-N+NO3-N) showed differences between the depth layers but theydid not show a consistent trend. Finally, soil pH, exchangeableMg and Na showed different responses with depth in the differentblocks; one block showed a consistent trend with depth and theothers did not.

Site survey: seed banks

7

Only seven species (Agrostis capillaris, A. stolonifera, Holcus lanatus,Juncus effusus, L. perenne, Poa pratensis, Poa trivialis, Ranunculus repens andTrifolium repens) were detected in the seed banks, indicating arelatively depauperate flora in the two depth layers assessed(0-15cm; 30-45cm). No significant differences were found betweenspatial block, either in total seed density (per m2), forspecies or functional type (Fig. 4).

8

Table 1. Mean chemical properties of the Cow Field soil compared to values for mesotrophic grassland communities in the National Vegetation Classification (Rodwell 1992). NVC soil properties abstracted from Critchley et al. (2002).

VariableCow Field NVC class

MG1 MG3 MG4 MG5 MG6 MG7 MG8 MG9 MG10 MG11 MG13All-NVC

No of sites 1 23 6 2 29 90 100 28 18 56 31 11 394Loss-on-ignition (%) 5.8±0.1

14.7±4.9

14.6±4.2

19.1±5.1

10.6±4.3

10.4±6.7

14.3±8.0

38.7±17.7

19.4±12.5

26.7±14.3

18.1±11.2

21.5±9.3

17.4±12.7

pH5.3±0.0

37.5±0.

86.4±0.

96.4±0.

66.2±1.

05.9±0.

76.6±0.

9 6±0.69 6±0.476.1±0.

76.4±1.

05.7±0.

86.3±0.

9

Total N (%)0.29±0.

010.84±0.3

0.9±0.42

0.91±0.2

0.67±0.3

0.71±0.3

0.9±0.4

1.96±0.8

1.11±0.6

1.59±0.6

1.22±0.61

1.34±0.4

1.04±0.7

C.N10.2±0.

110.4±2.1

10.1±2.1

12.1±0.6

9.7±3.6

9.1±2.94

9.2±2.33

11.1±3.1

9.7±2.99

9.2±2.45

8.8±3.6

9.1±2.6

9.4±2.8

Extractable P(µgg-

1) 9.4±1.711.7±7.1

7.7±4.5

4.8±1.46

9.6±4.5

13.2±7.9

18.5±13.5

7.8±3.9

10.2±7.1

9.6±6.1

21.7±17.0

10.3±5.7

13.6±10.7

Exchangeable K(µgg-1) 78±4 133±52 128±64

165±109 135±39 164±82

213±120 140±82 143±60 152±89

204±129 160±56 171±97

Exchangeable Mg (µgg-1) 167±14 101±43 96±66

174±101

1423±66 155±78 154±81

167±138 156±71

229±125

234±136 190±56

168±101

Table 2. Summary of the responses of soil chemical properties spatially (blocks) and with soil depth at the CowField; the direction of responses are shown for soil depth, and where it is variable there were fluctuations down

the profile.

Model selected Response in field to depthDecrease with depth Increase with

depthVariable response with depth

Block only Extractable P

9

Depth only Loss-on-IgnitionTotal C Total N

Block+Depth C:N ratioExchangeable KExtractable NO3-N N mineralisation rateNitrification rate

Exchangeable Ca

Extractable NH4-NTotal extractable-N(NH4-N+ NO3-N)

Block*Depth Exchangeable Na (Block4)Exchangeable Mg (Blocks 1,3,4)

pH (Block2) pH (Block 1,2,4)Exchangeable Na (Block 1,2,3)Exchangeable Mg (Block 2 )

10

Fig 2. The mean Ellenberg-N values for the Cow Field compared tothose of potential target NVC sub-communities (Rodwell, 1992, 2000).Mean values ± ranges are shown.

Fig 3. Examples of the distribution on soil propertieswith space and depth at the Cow Field, Parkgate: (a)

11

Extractable P varying with soil depth, and (b) Loss-onIgnition which reduced with soil depth.

Discussion

The aim of this study was to develop a method for assessingthe potential for creating a species-rich wildflower meadow onformer agricultural land. A combination of approaches were usedto assess the fertility of the case study site (the Cow Field,Parkgate) relative to typical communities described within theframework of the National Vegetation Classification scheme(Rodwell, 19922). We used three approaches by assessing: (a) thespecies composition and the biomass, (b) soil chemicalproperties, and here we assessed spatial and depth effects, and(c) the composition of the seed bank.

Assessment of the vegetation

The vegetation was composed mainly of productive grasses, withfew forbs, and all but one species was classified as having apart-competitor strategy (Grime et al., 1988). The vegetation wasclassified as an OV23d, a community typical of managed grassland(Rodwell, 2000). The mean for the Ellenberg N-values for allspecies in the Cow Field was similar to that of the typicalOV23d community and slightly greater than the target MG1a andMG5c communities, implying that it had a greater number ofspecies typical of fertile conditions, and fewer species typicalof infertile conditions. The lack of species with an N-value of2 supports this hypothesis.

Assessment of the soil

Comparison with NVC grassland types The values for most soil chemical variables at the Cow Field

are similar to sites which support typical NVC grasslandcommunities (Critchley et al., 2002), including MG1a and MG5c;indeed the Cow Field had a lower soil pH and much lowerconcentrations of variables associated with soil organic matter.Specifically the available soil phosphorus concentrations arelow and within ranges that can support species-rich vegetation(Gough & Marrs 1990). Thus, the evidence suggests that soilchemical properties will not preclude the establishment of aspecies-rich wildflower meadow.

12

Heterogeneity in space and depthThis was assessed using a stratified random sampling design

analyzed by Mixed-effects models to accommodate the samplingdesign. All soil variables showed significant spatial effects(sampling blocks), soil depth, or both. There are two importantresults. First, within the range of variation sampled across thesite, which is all within or below the ranges of typical NVCcommunities, there are significant differences spatially acrossthe field. This inherent spatial variability might inducevegetation heterogeneity as the field is restored. This mightinduce B-diversity to the developing vegetation. This hypothesisneeds to be tested.The relationship with soil depth showed that some properties

reduced with depth (variables associated with soil organicmatter and N supply), others were variable and exchangeable Caincreased with depth. These results suggest there may be meritin removing the topsoil (0-15 cm depth) or incorporating topsoilthrough the soil profile. This would remove some elements, orreduce their availability, and by exposing more Ca-rich soilmight enhance soil pH, making the soils more similar in pH tothe typical NVC communities (Critchley et al., 2001).

Assessment of the seed bank

The seed banks were depauperate, with few seeds of only sevenspecies, and of these seven only two were forbs. Thus, itappears that the seed banks of Cow Field soils are unlikely toprovide a source of germinating seeds during the restorationprocess. Indeed of the two forbs detected, one Trifolium repens maywell have been sown for agricultural improvement, and

13

(a)

(b)

(c)

(d)

14

Fig 4. Seed density of plant groups within the seed bank of the Cow Field soils: (a) grasses; (b) forbs; (c) rushes (d) total seeddensity.

15

Ranunculus repens can under certain circumstances become adominant weedy species. Therefore, it appears that the CowField soils are seed-limited and to achieve the creation of aspecies-rich wildflower meadow, seeds need to be added.

Conclusions

This site is a paradox; its vegetation is composed ofcompetitive species that are typical of relatively fertilesoils. However, the soils themselves do not appear to be veryfertile, being within ranges typical of semi-natural, species-rich grassland within the UK. The restoration potential forthe site is high in terms of soil fertility but low in termsof species composition (vegetation and seed banks). Wesuggest that this has occurred as a result of long-term haycropping where the hay was cut before seed shed. To create awildflower meadow the main constraint that needs to be reducedappears to be (a) the lack of seed of forbs and species ofmore infertile soils, and (b) ensuring that if seeds are addedthen they are allowed to establish in a relatively competitivevegetation. There may be some merit in removing orincorporating the surface layer, but irrespective of this,there is some spatial heterogeneity in soil properties toperhaps induce variation in the developing vegetation.

Acknowledgements

We thank EnviroLink North-West and the University of Liverpoolfor funding this research, and the Royal Horticultural Societyof Manchester & the Northern Counties for financial support toD Hewitt. The Friends of Parks Fields encouraged this work andallowed access to work on the site. M. Harris, P. Robson and M. Zarzecki gave valuable field support.

References

Allen SE. 1989. Chemical Analysis of Ecological Materials, 2nd ed., WileyBlackwells, Oxford.

Crawley MJ. 2007. The R Book. Wiley, Chichester.Critchley CNR, Chambers BJ, Fowbert JA, Sanderson RA,Bhogal A, Rose SC. 2002. Association between lowland

16

plant communities and soil properties. Biological Conservation105, 199-215.

Ghorbani J, Das PM, Das AB, Hughes JM, McAllister HA, PallaiSK, Pakeman RJ, Marrs RH, Le Duc MG. 2003. Effects ofrestoration treatments on the diaspore bank under densePteridium stands in the UK. Applied Vegetation Science 2, 189-198.

Gough MW, Marrs RH 1990. A comparison of soil fertilitybetween semi-natural and agricultural plant communities:implications for the creation of species-rich grassland onabandoned agricultural land. Biological Conservation 51, 83–96.

Grime JP, Hodgson JG, Hunt R. 1988. Comparative plant ecology: afunctional approach to common British species. Unwin Hyman, London.

Hewitt D. 2010. The value of Biochar application to a grassland site in North-West England. M.Res thesis, University of Liverpool, Liverpool.

Hill MO. 1996. TABLEFIT version 1.0, for identification of vegetation types.Institute of Terrestrial Ecology, Huntingdon.

Hill MO, Mountford JO, Roy DB, Bunce R.H. 1999. Ellenberg’sIndicator values for British plants: Ecofact Volume 2 Technical Annex. CEH,Huntingdon.

Marrs RH 1993. Soil fertility and nature conservation inEurope - theoretical considerations and practical managementsolutions. Advances in Ecological Research 24, 241-300.   

Oksanen J. 2005. Multivariate Analysis of Ecological Communities in R: vegantutorial. URL: www.cc.oulu.fi/ ∼ jarioksa/opetus/metodi/vegantutor.pdf .

Pinheiro J, Bates DM. 1996. Unconstrainedparameterisations for variance-covariance matrices.Statistics & Computing 6, 289-296.

R Development Core Team 2009. R: A Language and Environment forStatistical Computing. R Foundation for Statistical computing,Vienna, Austria. URL: http:// www.r-project.org .

Rodwell JS, ed. 1992. British Plant Communities. Vol. 3. Grasslands andmontane communities. CUP, Cambridge.

Rodwell JS, ed. 2000. British Plant Communities. Vol. 5. Maritimecommunities and vegetation of open habitats. CUP, Cambridge.

Stace C. 1997. New Flora of the British Isles, 2nd ed. CUP, Cambridge.Tallowin JRB, Smith REN. 2001. Restoration of a Cirsio-Molinietumfen meadow on an agriculturally improved pasture. RestorationEcology 9, 167-178.

Van Der Woude BJ, Pegtel DM, Bakker JP. 1994. NutrientLimitation After Long-Term Nitrogen Fertilizer Applicationin Cut Grasslands. Journal of Applied Ecology 31, 405-412.

17