Appendix 1 Analysis of plant communities
At Rocks Farm, Kemhide and Little Sprays the composition of the plant communities was recorded within each experimental plot during late May or June each year. From each plot, ten 0.5 x 0.5 m quadrats were randomly positioned at least 1 m within the edge of each plot and the presence or absence of vascular plant species was recorded. The use of frequency data from a large number of small, randomly positioned quadrats was considered a more efficient way of monitoring colonisation events than recording abundance in small number of larger samples. The botanical composition of the donor site was assessed in 2003 using the same methodology in fifteen randomly-positioned 10 × 10 m ‘dummy’ plots. Plots at Dancers End were surveyed using the same methodology in July or August each year, with the donor site assessed in 2004.
A pilot study of the method was carried out in July 2001 at Rocks Farm. For each plot in one of the blocks, 25 quadrats were recorded. The data on the cumulative number of species encountered from different sample sizes was calculated using thirty random permutations of the data. The results are shown below and show clear differentiation of treatment effects on plant species richness for the sample size of 10 quadrats per plot used in subsequent sampling.
At Aston Rowant the composition of the plant communities was recorded for each experimental plot during July or August each year. In each split-plot, ten 0.5 x 0.5 m quadrats were randomly positioned at least 1 m from the edge of the split-plot and the presence of all vascular plant species recorded. Plant species were therefore scored on a scale of zero to ten based on their presence in each of the ten quadrats. The botanical composition of the donor site was assessed in 2004 using the same methodology in 15 randomly positioned 10 × 10 m ‘dummy’ plots in each of the two blocks of vegetation used.
At Brush Hill and both Cold Blow sites, plant community composition was recorded for each experimental plot during in July, August or September from 1999 to 2004. One 1.0 × 1.0 m quadrat was positioned at the centre of each 5 x 5 m plot and the percentage cover of each vascular plant species was assessed. The same methodology was used to asses the donor sites from ‘dummy’ plots in 1998 and 1999 respectively.
Table 1. Plant species abbreviations and full latin name of species used in Appendix 2.
Code / Plant Species Latin Name / Code / Plant Species Latin NameAchimill / Achillea millefolium L. / Lolimult / Lolium multiflorum Lam.
Agrieupa / Agrimonia eupatoria L. / Lotucorn / Lotus corniculatus L
Agrocapi / Agrostis capillaris L. / Luzucamp / Luzula campestris (L.) DC.
Agrostol / Agrostis stolonifera L. / Medilupi / Medicago lupulina L.
Ajugrept / Ajuga reptans L. / Melialti / Melilotus altissima Thuill.
Alopmyos / Alopecurus myosuroides Huds. / Myosarve / Myosotis arvensis (L.) Hill
Anisster / Anisantha sterilis (L.) Nevski / Odonvern / Odontites verna (Bellardi) Dumort.
Anthodor / Anthoxanthum odoratum L. / Origvulg / Origanum vulgare L.
Aphaarve / Aphanes arvensis L. / Phlebert / Phleum bertolonii DC.
Arrhelat / Arrhenatherum elatius (L.) / Phleprat / Phleum pratense L.
Bracpinn / Brachypodium pinnatum (L.) P. Beauv / Pimpsaxi / Pimpinella saxifraga L.
Bracsylv / Brachypodium sylvaticum (Huds.) P.Beauv / Planlanc / Plantago lanceolata L.
Bromhord / Bromus hordeaceus L. / Poaannu / Poa annua L.
Camprotu / Campanula rotundifolia L. / Poatriv / Poa trivialis L.
Careflac / Carex flacca Schreb. / Primveri / Primula veris L.
Caresylv / Carex sylvatica Huds. / Prunvulg / Prunella vulgaris L.
Centnigr / Centaurea nigra L. / Ran r+b / Ranunculus repens L. & R. bulbosus L.
Clinvulg / Clinopodium vulgare L. / Ranuacri / Ranunculus acris L.
Convarve / Convulvulus arvensis L. / Ranubulb / Ranunculus bulbosus L.
Cratmono / Crataegus monogyna Jacq. / Rhinmino / Rhinanthus minor L.
Cynocris / Cynosurus cristatus L. / Rumeacet / Rumex acetosa L.
Dactglom / Dactylis glomerata L. / Rumecris / Rumex crispus L.
Dauccaro / Daucus carota L. / Sangmino / Sangisorba minor Scop.
Epilcili / Epilobium ciliatum Raf. / Scabcolu / Scabiosa columbaria L.
Epilparv / Epilobium parviflorum Schreb. / Senevulg / Senecio vulgaris L.
Euphexig / Euphorbia exigua L. / Soncaspe / Sonchus asper (L.) Hill
Festarun / Festuca arundinacea Schreb. / Stelholo / Stellaria holostea L.
Festrubr / Festuca rubra L. / Taraoffi / Taraxacum officionale Wigg. Group
Galimoll / Galium mollugo L. / Trifdubi / Trifolium dubium Sibth.
Galiveru / Galium verum L. / Trifprat / Trifolium pratense L.
Heliprat / Helictotrichon pratense (L.) Besser / Trifrepe / Trifolium repens L.
Heraspho / Heracleum sphondylium L. / Trisflav / Trisetum flavescens (L.) P. Beauv
Holclana / Holcus lanatus L. / Urtidioi / Urtica dioica L.
Hyporadi / Hypochaeris radicata L. / Verocham / Veronica chamaedrys L.
Koelmacr / Koeleria macrantha (Ledeb.) Schult. / Verserp / Veronica serpyllifolia L.
Leonhisp / Leontodon hispidus L. / Vicisati / Vicia sativa L.
Leonsaxi / Leontodon saxatilis Lam. / Violhirt / Viola hirta L.
Leucvulg / Leucanthemum vulgare Lam. / Vulpbrom / Vulpia bromoides (L.) Gray
Linucath / Linum catharticum L.
A1.1Rocks Farm (High Weald grassland enhancement site)
A1.1.1Analysis of changes in plant species richness
Statistical approach
Analysis of variance was used to assess the response of plant species richness (mean number of vascular plant species recorded in each plot). The analysis was based on a repeated measures model with a 2 disturbance (+/- power harrowing) x 4 seed addition (untreated, high and low hay application and brush harvested seed application) factorial design. Within all models the random effect was block and repeated effect year, with individual plots specifying the subject component of the model.
Results
Both the seed addition and disturbance treatments significantly influenced plant species richness in the experimental plots (Table 2). All three seed addition treatments enhanced species richness relative to the control plots, with the high rate of hay application leading to the greatest increase (Figure 1a). Disturbance also increased plant species richness, although no significant interaction between the seed addition treatments and disturbance was found. This may be the result of the recruitment of species from the soil seed bank in the plots which were power harrowed but received no seed addition. There was a general increase in species richness across all plots through the course of the study. The significant interaction between seed addition treatments and time is reflected in the faster rate of increase in species richness in the plots receiving no seed addition (Figure 1b). This may reflect the local colonisation of species introduced in the hay or brush harvested seed from the plots in which they were introduced into the untreated control plots.
Table 2. Results from repeated measures ANOVA of plant species richness (species per plot) for the Rocks Farm grassland enhancement experiment. Where: YEAR, sample year; SEED, four factor treatment of untreated control, hay applied at low rate, hay applied at high rate, brush harvested seed added at high rate; DISTURBANCE, two factor treatment of untreated control and power harrowing prior to application of material from donor site. NS, non-significant with P>0.05; * P<0.05; *** P<0.001.
Model term / Test statisticsSEED / F3,21 = 18.92, P < 0.001***
DISTURBANCE / F1,21 = 22.42, P < 0.001***
SEED*DISTURBANCE / F3,21 = 2.26, NS
YEAR / F3,72 = 41.97, P < 0.001***
YEAR*SEED / F9,72 = 2.76, P = 0.013*
YEAR*DISTURBANCE / F3,72 = 2.64, NS
YEAR*SEED*DISTURBANCE / F9,72 = 1.90, NS
Figure 1. Effects of the treatment on plant species richness (mean species per plot ± one s.e.m.) at the Rocks Farm grassland enhancement experiment, where Control, plots receiving no seed addition; Low Hay, hay added at low rate; High Hay, hay added at high rate; Brush Harvest, brush harvested seed added at high rate; +PH, power harrowing applied.
(a) main seed addition treatment effects
(b) Interaction between time, seed addition and disturbance for the high rate hay addition and controls.
A1.1.2Analysis of effects of treatments on botanical composition
Statistical approach
To test for the effects of the 2 disturbance (control vs. power harrow) * 4 seed addition (control, hay low rate, hay high rate and brush harvested seed high rate) factorial design, constrained multivariate ordination using partial redundancy analysis (RDA) was used. Scaling was centred on species and no standardisation of samples was performed. Note that in this analysis there is no nominal variable for the control treatments, rather it is denoted by its zero-coding for the other variables. The analysis uses combinations of environmental and covariable treatment interactions with year (continuous) within a split plot design, where multiple temporal samples from a single plot represent the split plot. Monte-Carlo permutation tests were carried out on both canonical axis and were permutated freely at the level of the whole plot, but un-permutated at the split plot level (499 permutations). Individual plot identification is retained as a covariable throughout the analysis (Lepš & Šmilauer, 2003) and the block term is included as a covariate. All ordination analyses were performed in CANOCO 4.5. The following null hypotheses were tested:
C1: The species composition showed no directional changes in time that are common to all the treatments or specific for particular treatments;
C2: The temporal trend in species composition is independent of the treatments;
C3 (C4, C5, C6): The disturbance treatment (or low hay or hay high or brush harvest treatments) has no effect on the temporal changes in species composition.
Results
Rejection of all null hypotheses was made indicating that plant community composition showed a directional change in time that was not common to all the treatments or specific for particular treatments (C1). Importantly there were temporal changes in botanical composition that were dependent on the treatments (C2), and all the treatments (power harrow, low rate hay, high rate hay and high rate brush harvested seed) showed significant temporal trends in species composition (C3-C6). Of the specific tests of temporal changes in species composition with time, the high hay treatment explained the greatest proportion of the variance at 19.3 % (Table 3, Figure 2).
Table 3. RDA analysis for the Rocks Farm grassland enhancement experiment showing the response of plant community composition to the interactions between treatments and year. C1- C6 are specific hypothesis defined in the statistical section. % expl. 1st axis provides a measure of explanatory power of the variable and is the percentage of species variability explained by the 1st axis; r 1st axis is the species environment correlation of the first axis; F ratio is the F statistic for the test on the trace; P is the probability obtained by Monte Carlo permutation test (499 permutations). Explanatory variables/covariables are: Yr, Sample year (continuous); PlotID, individual code for each plot; PH, power harrow; LH, Low Hay; HH, High Hay; BH, Brush Harvest.
Analysis / Explanatory variables / Covariables / % expl. 1st axis / r 1st axis / F ratio / PC1 / Yr, Yr*PH, Yr*LH, Yr*HH, Yr*BH / PlotID, Block / 28.5 / 0.397 / 15.531 / 0.002
C2 / Yr*PH, Yr*LH, Yr*HH, Yr*BH / Yr, PlotID, Block / 26.3 / 0.827 / 16.780 / 0.002
C3 / Yr*PH / Yr*PH, Yr*LH, Yr*HH, Yr*BH, Yr, PlotID, Block / 1.9 / 0.408 / 4.249 / 0.034
C4 / Yr*LH / Yr*PH, Yr*LH, Yr*HH, Yr*BH, Yr, PlotID, Block / 10.4 / 0.68 / 23.72 / 0.002
C5 / Yr*HH / Yr*PH, Yr*LH, Yr*HH, Yr*BH, Yr, PlotID, Block / 19.3 / 0.783 / 43.794 / 0.002
C6 / Yr*BH / Yr*PH, Yr*LH, Yr*HH, Yr*BH, Yr, PlotID / 2.6 / 0.538 / 5.847 / 0.018
Figure 2. RDA analysis for the Rocks Farm grassland enhancement experiment showing the response of plant species composition to the interaction between treatments and year. Thick black vectors represent the continuous effect of the interaction between time and treatment. Year, Sample year (continuous); PH, power harrow; LH, low rate hay; HH, high rate hay; BH, brush harvested seed. Species with high axis 1 or 2 scores are displayed by thin black line vectors, species abbreviations are given in Table 1.
A1.1.3Analysis of similarity to donor site
The similarity of the botanical composition of the experimental plots to the plant community composition of the donor site was assessed by calculation of the Czekanowski similarity coefficient. The hay addition treatments showed the greatest similarity to the donor site, with plots receiving the high application rate having the highest values (Figure 3). Pre-application sward disturbance by power harrowing increased the similarity coefficients for all seed addition treatments, but had no effect on the control plots receiving no material from the donor site. Similarity of the botanical composition to that of the donor site increased through the study for all treatment combinations, even the untreated control plots. This is probably a result of the local dispersal of seed from plots receiving hay or brush harvested seed to untreated control plots.
Figure 3. Similarity of plant community composition (Czekanowski similarity coefficients) to that of the donor site for all treatment combinations at the Rocks Farm grassland enhancement experiment.
A1.1.4Success of colonisation from donor site
The success of colonisation of plant species from the donor site in the collected hay or brush harvested seed material can be assessed by examining frequency of occurrence in the treated plots through the course of the study. The frequency in samples in the plots receiving the high rate of hay addition is shown in Figure 4 for the 20 most frequent species at the Coach Road Field donor site. The treatment led to the successful colonisation of the experimental site by a wide range of species, many of which showed consisted increases in frequency through the 4 year period after application of the treatments. These include Centaurea nigra, Festuca rubra, Hypochaeris radicata, Leontodon hispidus, Leucanthemum vulgare, Lotus corniculatus, Plantago lanceolata, Prunella vulgaris and Ranunculus acris. In addition, a number of species of low frequency at the donor site also colonised the plots, notably Trifolium dubium and Vulpia bromoides. Some species with high frequency at the donor site which occurred on the experimental site before imposition of the treatments increased in frequency in response to the addition of hay, for example Rumex acetosa and Trifolium pratense. Rhinanthus minor colonised the plots as a result of the hay addition, but frequency remained low, and the species was not recorded in 2004. The lack of successful colonisation by this species may have been the result of the grazing of the site in the summer of 2001 because of restrictions in the movement of animals during the foot and mouth disease outbreak. R. minor is an annual species with no persistent seed bank, so would be adversely affected by grazing in the year following imposition of the treatments. Ajuga reptans was the most frequent species at the donor site which failed to colonise, probably as a result of its early phenology and low growth habit. This species flowers early in the season before significant shoot growth of other components of the plant community
Figure 4. Percentage frequency (means ± 1 s.e.m.) of plant species in the plots receiving the high rate of hay addition at the Rocks Farm grassland enhancement experiment for the years 2001-04. Species are arranged left to right in decreasing order of frequency at the donor site, Coach Road Field. Only the 20 most frequent species at the donor site are shown. Species marked with an asterix were present at the experimental site before addition of the hay in August 2000. Species codes are shown in Table 1.
Botanical sampling in this project focussed on recording colonisation of the plots in response to the experimental treatments, so frequency data was collected from randomly positioned quadrats. In the final year of the study, cover abundance data was also collected for the species which had successfully colonised the experimental plots as a result of the application of hay or brush harvested seed. This data suggest that Plantago lanceolata, Lotus corniculatus and Hypochaeris radicata were the most successful colonists, expressed in terms of their cover abundance after 4 years (Table 4).
Table 4. Effects of seed addition treatments on cover abundance of selected plant species at the Rocks Farm grassland enhancement experiment in 2004. Significant differences were identified using ANOVA. F ratio and P values are shown in the test statistic column (*** P<0.001). Control, untreated; Low Hay, hay applied at the low rate; High Hay, hay applied at the high rate; Brush Harvest, brush harvested seed applied at the high rate. Values are means ± 1 s.e.m., treatments sharing the same letter code are not significantly different (P>0.05 using Tukey LSD tests).
Species
/Test statistic
/Control
/Low Hay
/High Hay
/Brush Harvest
Centaurea nigra / F3,21=6.83P= 0.002 *** / 0.34(±0.271)a / 2.35(±0.72)b / 2.53(±0.554)b / 1.56(±0.568)a
Hypochaeris radicata / F3,21=19.38
P= 0.000 *** / 0.13(±0.082)a / 3.58(±1.26)b / 5.48(±1.29)b / 0.91(±0.217)a
Leontodon hispidus / F3,21=7.40
P= 0.001 *** / 0.00(±0.00)a / 1.03(±0.57)b / 0.69(±0.220)b / 0.20(±0.131)a
Leucanthemum vulgare / F3,21=13.03
P= 0.000 *** / 0.26(±0.118)a / 2.21(±0.802)b / 2.92(±0.596)b / 2.06(±0.667)b
Lotus corniculatus / F3,21=18.09
P= 0.000 *** / 0.29 (±0.16)a / 6.29(±2.13)b / 9.21(±1.72)b / 0.89(±0.367)a
Plantago lanceolata / F3,21=12.28
P= 0.000 *** / 1.48(±0.624)a / 10.96(±3.42)b / 19.55(±3.12)c / 14.12(±2.98)b
Trifolium pratense / F3,21=25.57
P= 0.000 *** / 0.17(±0.134)a / 1.14(±0.357)b / 3.97(±0.552)c / 1.45(±0.393)b
A1.1.5Colonisation by undesirable species
The most frequent pernicious weed at the experimental site was Cirsium arvense, frequency of occurrence was low, with the species found in fewer than 5% of samples. Repeated measures ANOVA showed no differences in abundance between treatment plots.
A1.2Dancers End (Chilterns grassland enhancement site)
A1.2.1Analysis of changes in plant species richness
Statistical approach
Analysis of variance was used to assess the response of plant species richness (mean number of vascular plant species recorded in each plot). The analysis was based on a repeated measures model with a 3 disturbance (untreated controls, power harrowing, turf stripping) x 4 seed addition (untreated control, hay collected and applied in August, hay collected and applied in October, brush harvested seed collected in August and October) factorial design. Within all models the random effect was block and repeated effect year, with individual plots specifying the subject component of the model.
Results
Both the seed addition and disturbance treatments significantly influenced plant species richness in the experimental plots (Table 5). All three seed addition treatments enhanced species richness relative to the control plots, with the early hay application leading to the greatest increase (Figure 5). Disturbance increased plant species richness and there was a significant interaction of seed addition and disturbance treatments, with the effects of seed addition being enhanced with increasing disturbance levels. There was a general increase in species richness across all plots through the course of the study.