Executive Summary

This project investigates the potential to use aerial imagery to assess change in important landscape features. The project focuses on 18 sites, seven of which have been part of a long term study on agricultural landscape change, with the other eleven sites being chosen as representative of the diversity of agricultural landscapes present in England. The first part of the project investigated what aerial imagery and supporting geographic information is readily available for the study sites and collated these data into individual project files for each site. The second part of the project identified changes in the counts and patterns of landscape features observable in modern aerial imagery taken before and after the introduction of the Environmental Stewardship Scheme. These measures were compared to the expected change based on data collected for the 1998 and 2007 Countryside Survey

The key findings of this project are that significant resources are required to locate, digitise and analyse changes in landscape features from historic aerial imagery. The spatial and temporal coverage of this aerial imagery is patchy, and the amount of information that can be extracted from historic imagery can be relatively low. In contrast, the modern imagery is significantly easier to work with, having national coverage and being stored in a georeferenced digital format. The results of this study indicate that the diversity of landscape character of the study areas make it difficult to draw generalisations of the observed impact of policy in different landscape types. However, the data collated in this project along with the outputs of the previous ground based survey projects mean that the sites investigated here should make good candidate study areas for future studies relating to drivers of landscape change. Continued collection of digital aerial imagery along with access to very high resolution satellite imagery and imagery from remotely piloted vehicles would provide opportunities to undertake similar studies in the future.

Page 1

Introduction

The aim of this project was to convert aerial photographs, taken over 18 study areas and between the end of World War II and the present day, to a geo-referenced digital format. These were then to be combined with other digital datasets to investigate landscape change and then linked with changes in agricultural policy. The initial project plan intended to collate data across four policy periods which could be broadly defined as 1) post-war agricultural intensification (1947-1973), 2) entry into the EC and implementation of the CAP (1973- 1987), 3) CAP reform and introduction of agri-environment schemes (1987-2003), and 4) introduction of the Single Payment Scheme and Environmental Stewardship (post-2003). Work identifying imagery held by showed that firstly locating imagery for all the study areas was not feasible and secondly that where aerial photographs had been taken over the study areas, only one of the policy periods would be covered. The project was however able to supplement the historic imagery with modern digital imagery taken after 1999. The project therefore produced 18 map files containing the GIS data and modern digital imagery for all 18 study areas and including historic aerial imagery for seven of these areas. The project used a rapid visual comparison (rather than the much more time consuming full manual digitisation) for assessing change in landscape features in order to be able to assess change across all 18 sites. In order to assess all 18 sites, the project was also restricted to using only the modern (post-1999) digital imagery for this assessment of change.

The changes quantified from the imagery were compared to the range of expected change for the site based on Countryside Survey data collected in 1998 and 2007. The sites were also grouped and compared according to their Agricultural Landscape Typology (ALT), with three sites belonging to each of the six typologies. While three sites is not enough to make statistical inferences from, particularly as the sites were chosen to be complementary in describing the range of landscapes within a typology rather than as replicates, some general conclusions regarding changes in the landscapes within the typology have been drawn.

Page 2

Background

The intention of the project was to provide the means to ‘visualise’ the effects of policy change; including the impacts resulting from regulation and agri-environment schemes, on our agricultural landscapes, and thereby inform ongoing policy evaluation and scheme development. This included consideration of both intended and unintended consequences of policy interventions in specific places, from which inferences on the wider agricultural landscape can be drawn. From 2015 onwards there will be new CAP regulations and a new agri-environment scheme which will take over from Environmental Stewardship. The outputs of this project would be able to form the basis for continued monitoring of landscape scale impacts from the changes in agricultural policy and regulation that these represent, as well as other policy drivers such as the Biodiversity 2020 (Defra 2011) targets and planning regulations.

The project followed preliminary work undertaken by Natural England who had started the process of locating and cataloguing historic aerial imagery from Natural England’s archives. The project also ran parallel to a separate Defra project looking into the impact that a reduced coverage of agri-environment schemes would have upon landscape character (Project BD53031). That project mainly reported on data collated for National Character Areas (NCAs) but also involved on the ground surveys of 18 study areas representing the six Agricultural Landscape Types for England (Boatman et al 2010). The work presented in here is based on the same 18 study areas.

Methods

Study Site Locations

Annex 1 contains a description of the key landscape features from the National Character Areas in which the study areas are located. The NCA features were used to guide the selection of the study areas to provide a complementary combination of landscape types to cover the range of landscapes found in each of the six Agricultural Landscape Typologies. These study areas comprise the seven New Agricultural Landscape (NAL) sites, which had been the location for four previous studies into changes to landscape character and features (Countryside Commission 1974, 1983, 1997, Countryside Agency 2006) and 11 other sites, chosen so that there were three study areas for each of the 6 Agricultural Landscape Types (Table 1). While three study areas per Agricultural Landscape Type is insufficient to produce

1 Report available to download from http://randd.defra.gov.uk/Default.aspx?Menu=Menu&Module=More&Location=None&Completed=0&ProjectI D=17454 Page 3 a robust generalisation of the changes in landscape feature metrics within the landscape type, the study areas can be viewed as representative case studies for the six ALTs.

Table 1. Agricultural Landscape Type, National Character Area and total area for the 18 study sites. Descriptions for the NCAs can be accessed at http://www.naturalengland.org.uk/publications/nca/searchpage.aspx

ALT NCA Area of Search or New Agricultural Size Landscape (ha)

Chalk and Limestone East Anglian Chalk Chalk scarp south of Royston 3097 Mixed Dorset Downs and Cranborne Piddlehinton NAL 1573 Chase

Yeovil Scarplands Crewkerne NAL 987

Eastern Arable Vale of York Myton on Swale NAL 1297

The Fens Prickwillow NAL 1202

Bedfordshire and Cambs. Leighton Bromswold NAL 1931 Claylands

South East Mixed Thames Basin Lowlands Eastern Fringe of Guildford 1754 (Wooded) Wealden Greensand Selborne 2204

High Weald Eridge 4633

Upland North Pennines Allendale 3714

Dark Peak Edale and Hope Valley 4403

Dartmoor West of Chagford 5730

Upland Fringe Orton Fells Northern Orton Fells 3299

Manchester Pennine Fringe Oldham Fringe 2336

The Culm Hartland Peninsula and Coast 3823

Western Mixed Dunsmore and Feldon Grandborough NAL 2104

Herefordshire Lowlands Preston on Wye NAL 1360

Marshwood and Powerstock Powerstock Hills 2415 Vales

Page 4

Map overlays

Shapefiles of the study site boundaries were overlaid onto layers containing the 1:10000 tiles and 1:50000 map outlines to identify which flight path maps and map overlays were required.

Aerial photography for the study areas was identified and digitised. Natural England holds an archive of aerial photographs taken in the period between the 1960s and the 1990s. This archive is not centrally catalogued, so appropriate images for each study area needed to be individually identified. There is some variation in how the imagery is referenced across the time periods, but the majority of the images are identified by a combination of a flight number and an image number. For a large part of the archive material that was used in this project, flight paths were recorded on acetate overlays for OS 1:50000 maps. In a smaller number of cases these flight paths were recorded on a photocopied section of a paper map. For two sites (Preston-on-Wye and Selbourne) the images had already been preselected for the study area by Natural England staff.

The boundaries of the study sites were overlaid onto the OS 1:10000 Vectormap product using ArcGIS and the resulting map printed onto an A4 sheet. This allowed the analyst to visually identify the appropriate section of the 1:50000 map for a particular study site, and then to identify whether any of the flights had potentially taken photographs of the study site. The location of the images on the flight paths is only given as the centroid of the image. Flights that did not directly overfly but were close to the boundary of a study site may have taken an image which overlaps with the study site were included in the list to be scanned and georeferenced. All identified imagery was entered into the image catalogue spreadsheet (Supplementary Material: AP_Imagery.xls). The metadata fields included in the catalogue are given in Table 2.

Page 5

Table 2. Metadata recorded in imagery catalogue.

Site Information

10k Tile Ref 1:10000 Tiles in which the study site falls

1:50 000 Sheets 1:50000 (OS Landranger) maps which cover the study site

Centroid GR Study site centroid

County Counties in which study site falls

NCA National Character Areas in which study site falls

ALT Agricultural Landscape Type for the study site

Imagery Information

Flight No. Flight ID for imagery (Potential for multiple runs in 1 flight)

Run Run number (1 run = 1 set of images taken along a single transect)

Flight Date Date of flight

Image Number Image number within flight/run

Image Coordinates Image centroid identified from flight map/map overlay

Direction of Flight Direction in which sequence of images was taken

Scale Scale of imagery

Comments Additional comments related to imagery

Scanned Indicator that image was located and scanned

Georeferenced Indictor that image was georectified against base map

Min X Minimum easting for image

Min Y Minimum northing for image

Max X Maximum easting for image

Max Y Maximum northing for image

In Study Area Indicator to show if image intersects with study site

Resampled to low resolution Indicator to show if image has been resampled to 2m resolution

Page 6

Aerial photography – historic (1960s-1990s)

The historic aerial imagery spans a time frame from the late 1960’s to the mid-1990s. The majority of the imagery was taken at a 1:10000 scale, however a number of the photographs were taken at other scales (for details see scale information in the image catalogue). Photographs were scanned on a flatbed scanner at 600dpi resolution (maximum resolution available on the scanning equipment) and saved as a TIF file. For a 1:10000 image this resulted in a resolution of approximately 40 cm.

The scanned images were georectified against the OS 1:10000 Vectormap Local product using ArcGIS 10 (ESRI 2010). A number of features within the images were identified on the Vectormap product and used as referencing points to georectify the image. There may be a significant time difference between the date the image was taken and the Vectormap Local product, so reference points were chosen that are highly unlikely to have changed location and were easily identified in both the imagery and the basemap (e.g. junctions between major roads, significant buildings, rock outcrops). A minimum of four well-spaced reference points are required to georectify the image, although additional points can help improve the accuracy of the georectification process. The georectified images were then saved as georeferenced TIFs in the OSGB36 projection.

Aerial imagery – modern (2000 onwards)

Post-2000 aerial imagery was obtained for all study sites from the Infoterra Geostore (http://www.geostore.com/geostore4). These files are delivered as 1km2 tiles in the ECW file format. The tiles for each study site were merged and converted to georeferenced TIF files. Images for two time periods were requested for each site, one pre-2005 and the second post-2005. This provides imagery prior to and post the introduction of Environmental Stewardship. In addition to the standard 3 band visible imagery, near-infrared imagery was also requested where this was available for the site.

Additional datasets

In addition to the aerial imagery, GIS data from a number of other datasets was extracted for the study areas. The datasets were included in the project map files and their coverage of the study sites is presented in Annex 2.

Field Parcel Structure

The pattern of fields in the landscape was derived from two sources. The first was the historic pattern of field parcels from the historic Landmark maps for the areas. The second was the current pattern of fields from the Mastermap data. The historic maps comprise raster

Page 7 tiles, scanned from 1:10560 maps created between the mid- 19th and mid-20th centuries. These have been split into four epochs (Table 3). The use of MasterMap to define field parcels rather than digitising field boundaries from aerial imagery was a pragmatic solution to reducing the overheads of digitising.

Table 3. Date ranges for epochs of Landmark mapping datasets.

Epoch Date Range Epoch 1 1846-1899 Epoch 2 1888-1914 Epoch 3 1903-1949 Epoch 4 1922-1969

All the raster tiles for a study site within each epoch were merged into a single raster, and a combination of the earliest epochs that cover the site was used as the base to digitise field boundaries for the pre-Second World War period. OS Mastermap was used to provide the modern field parcel boundaries by filtering out buildings, man-made surfaces, roads and tracks from the OS Mastermap polygon data. The remaining land parcels consist of fields of crops and grassland in addition to parcels of woodland, orchard and scrub. Generally field boundaries followed the boundary of the study area, but where field parcels extended beyond the study area, the whole parcel was included in the analysis. This introduces the potential for the total area of the field parcels to be greater than that of the study area.

The historic and modern datasets were used to calculate mean area and perimeter for field parcels within each of the seven NAL study sites, as well as a count of the number of field parcels and a calculation of the total area represented by the field parcels identified in each of the datasets.

Agri-Environment Agreement Data

For each of the 18 study areas the location and composition of the Environmental Stewardship scheme agreements which were active in 2012, and the location of Countryside Stewardship and Environmentally Sensitive Area agreements for 1998 were also identified for each site. In addition to the 2012/1998 snapshots, the area under agri-environment agreements when the two sets of modern aerial imagery was taken for each site was also extracted. In the cases where the early or later set of modern images extends over a time period greater than one year, the area under agreement for the time of the earliest and latest images was determined.

Page 8

Landcover Maps

Land cover for each site was extracted from the Landcover 2007 (LCM2007, Morton et al 2011) and the Dudley Stamp land utilisation digital map products. The LCM2007 is derived from satellite imagery taken between 2005 and 2008 classified onto a spatial framework of land parcels derived from OS Mastermap and Rural Land Register parcels. The Dudley Stamp Map data is a digital version of the land utilisation mapping undertaken between 1933 and 1944. The paper maps were scanned and georefererenced before a semi-automatic classification routine was applied to produce the final digital product (Baily et al 2011)

Landscape Feature Metrics

The identification of features follows the methodology outlined in the Countryside Survey Field Mapping Handbook (Maskell et al 2008) modified for the application to aerial photography.

Feature Descriptions

In-field Trees

In-field trees are large mature trees that occur within fields and are not associated with linear features such as or water-courses. They are a target of two Entry Level Scheme options (EC1 and EC2) and five Higher Level Scheme options (HC5, HC6, HC12, HC12-14). Potentially, wood pasture can contain a large number of in-field trees, however if the canopies of these trees join up then this would be classed as woodland rather than wood pasture and individual trees would not counted within this class of features.

Visually, in-field trees are relatively easy to identify in colour imagery, as the tree tends to have a significantly different colour profile to the field. In addition, if the angle of the sun is oblique at the time the image was taken, the shadow cast by the tree can be used to indicate its presence. The length of this shadow can also be helpful in distinguishing between a patch of scrub and what would be recognised on the ground as a tree. Each tree within a small patch of trees was individually identified as long as the canopies of the individual trees were distinguishable from each other. Larger patches are classified as a parcel of woodland within the field and therefore individual trees are not identified.

Hedgerow Trees

Hedgerow trees are trees that have been allowed to develop within a hedgerow. In some cases these will be relatively large mature trees, potentially much older than the they are associated with, with canopies which may be significantly larger than the width of the hedge. In other cases they will be individual trees within the hedge which have been Page 9 allowed to develop so that they stand proud of the surrounding hedge. The establishment of hedgerow trees is a voluntary component of ES options EB1 and EB2 (up to 4 trees per 200 metres of hedge) or as a specific option EC23 which was introduced in 2010.

Hedgerow trees are more difficult to distinguish than in-field trees as they occur within a feature which has similar colour characteristics. For the larger, more mature hedgerow trees it is possible to see that the canopy is significantly larger than the width of the hedge in which it is located and identify them by this characteristic. For smaller trees, identification may rely upon the shadow cast by the tree when the sun is at an oblique angle. This allows identification where a tree casts a shadow significantly longer than the hedge in which it is located.

Hedges

Hedges are a contiguous line of shrubs that are one shrub in width and no wider than 5m. They are equivalent to the Countryside Survey Woody Linear Feature of Unnatural Shape. Hedges are generally managed in some way so that the shrubs do not grow into trees, (except in the case of hedgerow trees, as above). All hedges on agricultural land should have at least a 2m field margin associated with them to meet cross-compliance requirements and, provided they meet the requirements of being sited and managed appropriately and contain at least 80% native species, can be entered in the ELS scheme under options EB1- EB3, EB8-EB10 or the HLS scheme under options HB11 and HB12.

In hedges which have recently been trimmed, the hedge can be easily distinguished by a ‘straight’ edge to the hedge feature on the imagery. However if it has been some time since the sides of the hedge were last cut, it can be difficult to tell whether the feature should be classed as a line of trees or a hedge. Where the sun is at an oblique angle at the time the image was taken, looking at the shadow cast by the feature can help determine if all trees are of a uniform height and hence can be considered a hedge, or if they represent a line of trees of differing heights in which the canopies have been allowed to overlap.

While some hedges are easily distinguished as linear features with a significantly different colour profile from the vegetation in the margin, it can sometimes be difficult to tell if a boundary is a hedge or a fence line with scrubby vegetation. Again, comparing shadows between known hedges and the features in question can help determine if the feature is tall enough to be considered a hedge. Hedge-banks are also included as hedges, as it is difficult to identify in aerial imagery where the base of the hedge has been raised above the level of the surrounding land.

Page 10

Line of Trees

The class Line of Trees represents trees along boundary features which have been allowed to develop naturally, and the lines can take two forms. They can be a contiguous line of trees where, unlike hedges, the trees are allowed to take their natural shape, or they can be a line of individual trees that follows another linear feature such as a fence or watercourse. In the first case, hedges and lines of trees can be considered as being on a continuous scale of management where there is some judgement required near the centre of the scale as to whether the feature should be classed as a line of trees or as a hedge. The decision is largely a judgement call on the part of the analyst, but if there are five or more individual trees that can be identified within a 200m length of the feature (in line with Entry Level Scheme guidance for hedge management options), then this should be considered a line of trees rather than a hedge.

The same identification methods used for hedgerow trees and hedges are employed in identifying lines of trees. In many cases a feature will be classified as a line of trees where it does not meet the requirements outlined above for being classed as a hedge. Identification of ditches and fence lines associated with the trees demarcate them as part of a line of trees rather than being included as an in-field tree, even if the trees are relatively sparsely spaced along the boundary feature.

Walls

Walls are brick or stone structures used to delineate field boundaries. Stone walls can be entered into Environmental Stewardship for protection and maintenance under option EB11. For the purposes of this study, brick walls associated with residential land are not included in the feature metrics. Visually these features appear as relatively fine grey linear features in the imagery and are expected to be more prevalent in the upland study areas than in the lowland where hedges are preferred as boundary features.

Farm Buildings

Farm buildings are permanent agricultural structures. These are easy to identify as man- made structures in the aerial imagery and include isolated farm houses, barns, silos and other roofed buildings outside of urban areas. Counts were made on individual complexes of buildings, i.e. if two or more buildings have adjoining roofs, this is considered one building.

Farm buildings were identified both on the historic aerial imagery and from the Mastermap data layer. Where both are available for a NAL site it was noted whether the building exists in both the historical imagery and the current Mastermap data or in just one of these datasets. Page 11

Feeding and Watering Points

These are the locations in grassland fields where feeding/watering stations or supplemental feed for livestock is regularly placed. The outcome of this tends to be that the area surrounding the feeding/watering point becomes poached. It is these poached areas which have been used as the main feature to identify the presence of a feeding/watering point, although it is sometimes possible to see the feeding/watering troughs in the imagery. Feeding points may occur on boundaries as well as in the middle of fields. These feeding points are differentiated from access points, which may also become poached, by being located near to the boundary feature where there is no break visible in the boundary which would indicate that the location is used for access to the field.

Non-Agricultural Use

As economic pressures on farming have increased, land managers have looked to diversify the activities undertaken on their land. There are an increasing number of recreational activities for which land is been converted. The modern imagery was examined to identify locations which had been developed for recreational activities such as equestrian centres, golf courses, tennis courts and swimming pools. While presence and number of these recreational areas was noted, this study has not attempted to quantify their impact on the study site metrics.

Changes in Measures of Landscape Features

Change in features between pre-2005 and post-2005 aerial imagery

For each site, the two sets of modern imagery were compared. The earlier sets of imagery were taken in a period between 1999 and 2005, prior to the introduction of Environmental Stewardship. The second set of images were taken post-2005. The time period between each set of images varies, but is approximately seven years. Images for the two periods may also be taken at slightly different times of year which may impact on some of the features visible in the imagery such as poached areas or in-field agri-environment scheme options.

Differences in agricultural habitats, linear and point features were identified by visual comparison of the imagery. The analysts were looking for a distinct visual change in the imagery. This means that the changes reported for this work are of a different character to those that can be calculated from the digitisation undertaken in the previous section (where all landscape features have been digitised in both sets of imagery).

Page 12

The key parcel level habitat changes looked at were transitions to/from grassland, arable, scrub and woodland. There was also some heathland change observed though this was restricted to a single study site. In addition changes in areas given over to particular forms of land management which provide structure to the landscape were also considered. These included ponds, orchards, coverage by plastic sheeting, field margins and buffer strips, areas managed for nature conservation, and farm buildings and recreational structures.

Changes in the lengths of significant linear features and the counts of point features were also quantified. Linear features included hedges, walls, ditches and access routes. Point feature counts included individual trees, ponds, and poached areas on grassland. Ponds have been reported both in terms of the area change and as a change in the count.

Estimates of expected change in features

For each site, an estimate of the expected change in the measures of the key features was calculated using information on 1km square level changes from the Countryside Survey (CS) performed in 1998 and 2007. The information on features recorded in each CS survey square is available to download in an anonymised form from the Countryside Survey website (www.countysidesurvey.org.uk). Only squares which were surveyed in both 1998 and 2007 were included in the calculations of the expected change. Total change in features for each square were calculated and these were then used to calculate CEH land class (Bunce et al 1996) means and standard deviations for the expected change in features per 1km2. The study site boundaries were intersected with a 1km grid of the CEH land class assignments and the mean (expected) change and standard deviation of the key features for each study site was calculated using the land class areas in each study site.

Reclassification of CS habitat areas

The recording of CS habitats (as broad and priority habitats) does not correspond exactly with the habitat categories that the image analysts were able to assign land use changes to. Arable land was the simplest assignment as it is represented by a single broad habitat in CS. For woodland the two CS broad habitats (Broadleaf and Mixed Woodland, and Coniferous Woodland) and five priority habitats (Lowland Beech and Yew Woodland, Upland Mixed Ashwood, Wet Woodland, Upland Oakland, and Lowland Mixed Deciduous) were amalgamated into a single general woodland habitat class. However, for the grassland and scrub classification additional coding was required. This is because in CS the scrub is split between the Bracken broad habitat and a sub-class of the Neutral Grassland broad habitats. Bracken was therefore recoded directly as scrub, while the appropriate parcels from those recorded as Neutral Grassland were identified by their primary attribute and then combined with the Bracken parcels within the scrub habitat class. The remaining Neutral Grassland Page 13 was combined with the other two broad grassland (Acid, and Improved) and five priority grassland (Lowland/Upland Hay Meadow, Lowland/Upland Calcareous Grassland, and Lowland Acid Grassland) habitats into the general grassland habitat class.

CS Linear and Point Features

For the expected change in linear features information for hedges, walls and access was only extracted from the CS data where a change in those features was observed in the imagery. Hedges in CS correspond to the theme ‘Woody Linear Feature Unnatural Shape’ while access is encompassed by the length of features recorded under the Transport theme. Ditches have not been included as they are contained in a composite theme (Inland Water) within CS that also contains streams and rivers. Expected point feature counts were extracted for points recorded with the individual tree and pond primary attributes.

QA – Comparison from different analysts.

To examine differences between change estimates made by different analysts, changes in features for the Piddlehinton site were assessed separately by two analysts. Each analyst made separate measures of visual change in landscape features and land use between the early and late modern imagery for the Piddlehinton site. The two sets of measures were then compared to give a basic assessment of variability in the change metrics due to analyst differences.

Results

The results for sites are presented grouped by landscape typology. To aid comparison across ALTs, the x-axis scale on the charts is the same across all the ALTs. For each site a brief summary of the changes observed has been included, the full set of data for change in landscape features and land use for each site is included in Annex 3, along with the expected change based on the Countryside Survey square level data. Annex 4 contains a series of charts graphically displaying the estimate of change in feature metrics from the aerial imagery in relation to uncertainty bounds for the expected change feature metrics based on the Countryside Survey square level data.

Page 14

Chalk and Limestone Mixed ALT

Table 4. Imagery dates and area under agri-environment scheme agreements for Chalk and Limestone Mixed sites

Site Imagery Dates Area under AES Agreement

Pre-2005 Post-2005 Pre-2005 Post-2005 Imagery Imagery Imagery Imagery

Crewkerne May 2001 July 2009 0 303.7

May 2010

Piddlehinton September 2001 September 2008 34.0 717.0

May 2002 July 2009

September 2009

Royston June 2000 July 2008 9.3 1495.5

August 2000 November 2008

Table 5. Data Quality assessment of modern imagery for Chalk and Limestone Mixed ALT sites

Site Pre-2005 Imagery Post-2005 Imagery Comparison of Same Collected within 1 Collected Within 1 Time Period (+/-4 year Year Months) Crewkerne Y Y Y Piddlehinton Y Y N Royston Y Y N

Site Descriptions

The following general site descriptions are based on the assessments made in Defra project BD5303 (Defra 2013).

Crewkerne

The Crewkerne study area represents a predominantly grassland landscape with relatively small farms and medium sized fields with boundaries defined by hedges. The study area was also chosen for having a high uptake of agri-environment schemes.

Page 15

In the Crewkerne study area, the only habitat transitions are between grassland and arable land. Some areas of grassland were built upon, but the net change from these transitions was to increase the area under grassland and reduce the area under arable. There was no observed change in the area of woodland or scrub. This pattern for grassland and arable fits very well with the expected change based on the Countryside Survey, and the lack of an observed change in the scrub or woodland areas also falls well within the bounds of the expected change given the land class composition of the study area.

The structure of the landscape has changed relatively little, with only two grassland fields being split to create four new field parcels. This was accompanied by a small loss in the length of hedge within the study area. The landscape management would fit with a land sparing approach. The ES agreements are dominated by options to maintain existing grasslands and restore species rich semi-natural grassland, with relatively little in the way in- field nature conservation options. Crewkerne represents a site with relatively little change occurring between the dates that the images were taken.

Piddlehinton

The Piddlehinton study area is primarily cereal farming. It is a rolling landscape with the arable fields delineated by post and wire fencing and concentrated on the upper slopes. Smaller pasture fields, with boundaries delineated by hedges, are located on the lower slopes and valley floors. The study area has a long history of agri-environment schemes, having once fallen within an Environmentally Sensitive Area. A number of small settlements, including Piddlehinton itself, lie along the eastern boundary of the study area, which extend an urban influence into landscape of the study area.

In the Piddlehinton study area arable the overall trend has been for a decrease in grassland in favour of arable. The conversions between arable and grassland are the major transitions in the study area, with 28 of the 37 identified habitat transitions being between these two habitats. Although there are more arable field parcels moving to grassland than the other way round, the parcels converted from grassland to arable tend to be larger (a mean of 5.7 ha compared to 12.4ha). The areas of the other main habitats (scrub and woodland) have remained relatively constant. This is significantly different to the expected pattern in habitat change given the Countryside Survey, which would suggest an increase in area of grassland, woodland and scrub at the expense of arable land within the study area.

The structure of the landscape has also become slightly more simplified, with an overall reduction in the number of fields by seven. This has mainly been due to the merging of adjacent pairs of arable fields into single fields, only mitigated by a single occurrence of a grassland field being split into two. However there has been an increase in length of hedge Page 16 in the study area. There is relatively little uptake of wildlife conservation options in the Environmental Stewardship agreements for the area, the majority going to options for the maintenance of grassland. This, combined with the merging of arable fields and the loss of access tracks, suggests that the farmers in the study site are interested in maximising the production value of the landscape and that it appears that the main land-management philosophy for this study area is one of land sparing rather than sharing. Without the payments made through ES for grassland, there may well have been a much larger proportion of the grassland converted to arable and it may be that ES is a strong driver in preserving the current landscape character of the study area.

Royston

Royston is a primarily arable landscape with a relatively ordered pattern of fields with straight boundaries. There is a relatively low agri-environment scheme uptake and little habitat of conservation interest. The study area contains a stud farm which is of different character to the rest of the predominantly arable landscape.

In the Royston study area the main change has been a reduction in the area of arable in favour of the three other main habitats. In total 50 areas transitioned out of arable between 2000 and 2008, mainly to grassland (19 areas) or scrub (24 areas), while only 19 areas had transitioned into arable over that time period. There is also a significant increase in the amount of woodland within the study area, which has been planted on both arable and grassland, and includes the establishment of a new orchard. This pattern fits very well with the expected changes from the Countryside Survey where the expected area under arable shows a large decrease in favour of significant increases in grassland and scrub. The change in woodland area observed in the imagery is larger than was expected.

The structure of the landscape has become more complex with a net increase of 64 parcels in the study area. Much of this change is explained by the division of larger grassland fields into multiple paddocks, although there are a significant number of arable fields which are split. This is also accompanied by an increase in the length of hedges in the study area. The primary land management philosophy for the study area appears to be diversification and land sharing, with significant areas of wildlife conservation options, such as nectar flower mixtures and wild bird mixes, in the ES agreements for the area.

ALT Assessment

The Chalk and Limestone Mixed ALT contains some of the highest quality agricultural land in England, and it is therefore not unexpected that arable cropping predominates at two out of

Page 17 the three study areas. Royston represents the landscape which has historically evolved to maximise crop production (simple, relatively large fields and little semi-natural habitat remaining) while Piddlehinton represents a more diverse arable landscape where habitats of cultural and conservation significance have been preserved. Finally Crewkerne is an example of a Chalk and Limestone Mixed landscape where the topography is more suited to dairy and mixed farming.

The three study sites in the Chalk and Limestone Mixed ALT show different patterns of land use change (Figure 1). At Crewkerne the conversion of arable to grassland outweighs the conversion of grassland to arable, while in Piddlehinton the conversions are much more balanced. For both of these sites, the total habitat change is of a similar relative magnitude (Crewkerne 11%, Piddlehinton 14%) and restricted to the arable-grassland conversions. In contrast, Royston shows a smaller proportional change in habitat but also a more diverse range of transitions. In Royston there is an observable increase in area given over to nature conservation measures (Figure 1) and the observed increase in field margins is also proportionally much larger than at the other two sites (Figure 2). This is in spite of the fact that similar proportions of the Piddlehinton (46%) and Royston (48%) study areas are under agri-environment scheme agreements which would be the main drivers for the changes in these landscape features. Royston has also been much more dynamic in terms of parcel changes than the other two sites, with just over 15% of the study area being affected by parcel changes (Figure 3).

Figure 1. Proportional change in habitat area for Chalk and Limestone Mixed sites

Page 18

Figure 2. Proportional change in area given over to field margins for Chalk and Limestone Mixed sites

Figure 3. Changes in number of field parcels and percentage of study area affected for Chalk and Limestone Mixed sites

Page 19

Eastern Arable ALT

Table 6. Imagery dates and area under agri-environment scheme agreements for Eastern Arable sites

Site Imagery Dates Area under AES Agreement

Pre-2005 Post-2005 Pre-2005 Post-2005 Imagery Imagery Imagery Imagery

Leighton July 1999 October 2008 0 (1999) 1275

Bromswold September May 2009 454.2 (2004) 2004

Myton on Swale September May 2009 0 1124.9 2002

July 2009

Prickwillow April 1999 May 2008 0 931

May 2010

Table 7. Data Quality assessment of modern imagery for Eastern Arable ALT sites

Site Pre-2005 Imagery Post-2005 Imagery Comparison of Collected within 1 Collected Within 1 Same Time Period year Year (+/-4 Months)

Leighton Bromswold N Y N

Myton on Swale Y Y N

Prickwillow Y Y Y

Site Descriptions

Leighton Bromswold

The Leighton Bromswold study area was chosen to represent an intensively farmed area dominated by arable cropping. The majority of the study area is farmland with fields defined by a network of hedgerows and ditches. Marginal land has been improved to make it

Page 20 suitable for arable crops and little semi-natural habitat remains, although entry level agri- environment scheme uptake is moderate.

The land use in the Leighton Bromswold study area has moved away from the productive agricultural habitats of grassland and arable to include more scrub and woodland in the landscape. The increase in scrub seems to have mainly been through allowing succession to occur in field corners and margins of both grassland and arable fields, while there have been 17 new woodland areas planted. While there are relatively few fields which have transitioned between arable and grassland, there has been a large increase in the areas of fields given over to field margins and buffers. The decrease in arable area is smaller than expected from the Countryside Survey and this is paired with an observed decrease in grassland area where we would expect an increase based on the Countryside Survey. The change in area of scrub is similar to the expected estimate, while the area of woodland has increased more than expected.

There has been a net increase in the number of parcels in the study area, which involves splitting both arable and grassland fields, with a single instance where two adjacent arable fields have been merged into one. Likewise there has been an increase in the length of hedges observed in the study area. Both of these indicate an increase in the complexity of the landscape. While the mean for the expected change in length of hedges for this study area is negative, there is a large statistical uncertainty associated with this, and the increase in planted hedges observed cannot be considered to be outside the range of expected change derived from the Countryside Survey.

Overall the land management appears to support land sharing, alongside a significant commitment to increasing the woodland area. This is supported by large proportion of the total area of ES options in the study site being associated with field corner management and buffer strips. The woodland planting is supported by the ES option for restoration of woodland. While there are large areas of grassland maintained under ES, options for the creation of grassland represent a relatively small proportion of the total area managed under ES options.

Myton on Swale

Myton on Swale was chosen as an example of an Eastern Arable landscape consisting of smaller arable fields and which still contains many ‘remnant’ patches of valued habitat (eg meadows, copses, etc). It was also chosen to represent a study area with a moderate to high level of agri-environment scheme uptake.

Page 21

The change in land use for Myton on Swale is dominated by an increase in arable land and a smaller but still significant decrease in grassland. Nearly all the loss of grassland comes from conversion to arable land and the majority of the remaining change in arable area comes from converting land used for pig production to arable. The increase in the area of scrub comes from transition from a number of different other land uses and there are four areas of tree planting that contribute of the increase in woodland area. The change in grassland and arable is very different to that expected from the Countryside Survey, where an increase in grassland and decrease in arable land was indicated. The increase in scrub and woodland are both slightly lower than expected but still well within the range of the changes indicated by the Countryside Survey.

The study area has become simplified, with a net reduction of 8 field parcels and a decrease in both hedges and walls/fences. This has mainly occurred through merging arable fields, though occasionally adjacent arable and grassland fields have been merged. The Countryside Survey also indicates that the length of hedge would be expected to decrease, supporting the idea that that the drivers operating in this type of landscape are leading to a simplification of landscape structure. The ES options for this study area are very diverse, including options that support nature conservation activities on arable land and grassland, maintenance of woodland, and management options for upland moorland and grassland. Particularly prevalent are buffer strips of a variety of widths, which may be a consequence of the importance of the River Swale within the study area and that ES option choice is strongly influenced by the desire to reduce impacts on this component of the landscape.

Prickwillow

Prickwillow represents an open, flat intensively farmed landscape in which cereal and vegetable crops dominate the land use. Chosen to represent the fenland landscape within the Eastern Arable typology, the fields are delimited by drainage ditches and trees are mainly present as linear plantings rather than areas of woodland.

The dominant observable land use change in the Prickwillow study area is the loss of arable land the majority of which is attributable to an increase in-field margins and buffers. A much smaller area of grassland has been lost to a variety of land uses, the largest of which is a single parcel which transitioned to arable land. In addition to this, three small areas of grassland had farm buildings and hard-standings constructed on them, one area was converted to an orchard, and another area reverted to scrub. Scrub has also increased, mainly through transition from arable to scrub, but also with some attributable to reversion from grassland. These changes in habitat are somewhat different to the pattern expected from the Countryside Survey where grassland was expected to increase. The decrease in

Page 22 arable land is of similar magnitude to the expected decrease estimated from the Countryside Survey. Likewise the observed direction of change in area of scrub was consistent with the expected change, but the magnitude of the increase was somewhat less than the expected increase based on the Countryside Survey.

There was a net reduction in the number of field parcels, indicating that the landscape was becoming simplified. This is supported by a reduction in both the lengths of hedges and ditches. This reduction in numbers of parcels occurred predominantly from the merging of 26 arable fields to produce eight larger individual arable fields. This change was offset by the splitting up of two arable fields and one grassland field and also by two grassland fields being merged and then split again into seven separate fields. In this study area the land sparing approach to landscape management seems to predominate. The merging of arable fields along with implementation of options to maintain species rich grassland would indicate a land sparing approach where production is maximised on arable land, and higher nature value habitats are preserved. While there is a significant area given over to field buffer strips on cultivated land and field corner management under Environmental Stewardship, relatively little of this is enhanced for wildlife conservation such as floristically enhanced grass margins or use of wild bird seed mixes.

ALT Assessment

While the agricultural activity in all three study areas is dominated by arable cropping, they represent a gradient from the flat, almost monotonous fenland of Prickwillow, through the structured intensive landscape of Leighton Bromswold and onto the more diverse landscape of Myton on Swale. This Agricultural Landscape Type covers much of the most productive and easily farmed agricultural land in England, with the interest being on whether the landscapes might be moving towards the simpler but more productive end of the gradient (towards a fenland like landscape) or increasing in diversity to become more like Myton on Swale.

The three Eastern Arable ALT sites are relatively stable over the observed periods, with relatively small proportions of the study sites involved in habitat transitions. Myton on Swale is the only site in which we can see a dominant land use change (Figure 4), which involves conversion of grassland to arable. Also despite the change in field parcel numbers being relatively small, changes to field structure affects a significant proportion of the land in each of the study sites (Figure 6).

Page 23

For all three sites, changes in conservation measures are observable. However, observed conservation measures in Prickwillow represent a smaller proportion of the AES agreement area compared to at the other two sites. On the other hand Prickwillow has a much larger proportional change in field margins (Figure 5).

Figure 4. Proportional change in habitat area for Eastern Arable sites

Figure 5. Proportional change in area given over to field margins for Eastern Arable sites

Page 24

Figure 6. Change in number of field parcels and proportion of study area affected for Eastern Arable sites

Page 25

South East Mixed ALT

Table 8. Imagery dates and area under agri-environment scheme agreements for South East Mixed sites

Site Imagery Dates Area under AES Agreement

Pre-2005 Post-2005 Pre-2005 Post-2005 Imagery Imagery Imagery Imagery

Eastern Fringe of July 1999 May 2009 0 1133.1

Guildford August 1999

September 1999

September 2000

Eridge August 2003 June 2009 596 1130

July 2004

Selbourne July 1999 April 2007 91.1 346.8 (2007)

September 2000 September 2007 1463.9 (2009)

September 2001 April 2008

August 2009

Table 9. Data Quality assessment of modern imagery for Eastern Arable ALT sites

Site Pre-2005 Imagery Post-2005 Imagery Comparison of Collected within 1 Collected Within 1 Same Time Period year Year (+/-4 Months)

Eastern Fringe of Guildford N Y N

Eridge Y Y Y

Selbourne N N N

Page 26

Site Descriptions

Eastern Fringe of Guildford

The Eastern Fringe of Guildford site is an example of a densely populated rural location within the South East Mixed landscape typology. The study area contains a high diversity of land use, with mature hedgerows and significant areas of woodland and parkland. It also represents a site where there has been a relatively low uptake of agri-environment schemes.

For the Eastern Fringe of Guildford study area there is a large increase in area of grassland and a large decrease in area of arable land. A large proportion of the reduction in arable land is due to conversion of arable to grassland, although there are a few areas where arable has reverted to scrub or had woodland planted upon it. Some of this loss of arable is regained by conversion of grassland and scrub to arable land, but the overall effect is a large net loss of arable land in the study area. The conversion of arable to grassland also contributes the majority of the increase in grassland area, but some areas of scrub are also converted to grassland. A little over half the increase in woodland area is due to conversion of arable land to woodland, approximately a third is due to woodland planting on a golf course, and the rest was planted on grassland. The changes in grassland, arable and woodland fit well with the expected changes in these habitats based on the Countryside Survey. However the loss of scrub is contrary to an expected gain predicted from the Countryside Survey.

There has been a large change in the number of parcels, with 30 additional parcels in 2009 compared to 1999/2000. The main change has happened through the splitting of grassland fields, although some arable fields have also been split, often with one or more of the smaller parcels being converted to grassland. This splitting is offset by some fields which have been merged together, and where these were grassland, they have been converted to arable. Despite there being more parcels of land, there has been a loss of hedges from the study site. This may be reflected by the lack of hedgerow management options in the agreements on the study site. The ES options are dominated by grassland maintenance options, however there are in-field nature conservation options included within the agreements. This pattern of indicates that there hasn’t been a large drive to maximise production in the landscape with aspects of both land sharing and land sparing activities evident in the study area. It appears that land owners are attempting to offset increasing efficiency measures such as the creation of larger arable fields with the inclusion of more in-field areas for wildlife conservation.

Page 27

Eridge

Eridge represents sites in the South East Mixed typology with high woodland cover. This is driven by being part of an AONB which has acted to preserve the historic landscape character. The field pattern is irregular and the agricultural land use is dominated by meadow and pasture, with a moderate uptake of agri-environment schemes.

The Eridge study area has seen a loss of arable land in favour of grassland, as well as a significant increase in woodland. Scrub shows a very small net change in area, with three areas of grassland reverting to scrub, but this has been balanced by the conversion of two areas of scrub into garden. The woodland expansion was done on grassland areas, although there has also been some clearing of woodland. There has been an increase in livestock housing within the study area, with outdoor poultry units being placed on grassland, and outdoor pig housing replacing an area of arable land. The loss of arable and gain in grassland reflects the pattern expected from the Countryside Survey. However the observed increase in woodland is much greater than the expected change although there is a high variability associated with the Countryside Survey.

The field parcel structure is little changed between the two sets of imagery, with only two fields having been split into smaller parcels, but in one of these cases the split has resulted in seven new field parcels. Despite the relatively small amount of field parcel change, there has been a large increase in the length of hedge in the study area and also new access in the form of two new trackways through grassland, although an existing track was removed. The ES agreements have a large area in organic management and for the restoration of species rich grassland. They also contain a number of in-field nature conservation options. All of these together suggest that the maximising of production is not as strong a driver in the study area as it may be elsewhere and that the land managers are open to the concepts of a land sharing approach to nature conservation in the study area.

Selbourne

The Selbourne study area combines areas with regular open field parcels and areas where the parcels are more irregular and bounded by mature hedgerows. In addition there is a significant area of orchard within the study area, which has a very high level of regular structure. Away from the structured area of the landscape, there are blocks of woodland, some of which occurs on . While agri-environment scheme uptake has been low, the study area falls inside an AONB and is part of a National Park.

In Selbourne there has been a large conversion of arable land to grassland, with a much smaller area of grassland and arable being converted to scrub or woodland. Selbourne is

Page 28 also the only study area where fruit production is evident in the imagery and is present in both the 1999/2000 and 2007-2009 imagery. The area under fruit production has increased on land that was originally arable, however there has also been some loss of area from fruit production to both grassland and arable. Diversification and recreation appear to be strong drivers of change in this study area with areas of grassland and woodland sacrificed to build recreational structures, such as tennis courts and an equestrian centre. Changes in grassland, arable land and scrub all conform to the expectations of the Countryside Survey, however the expectation from the Countryside Survey is a net gain in woodland area, but a net loss of woodland was observed.

There was a significant increase in the number of field parcels in the imagery between the 1999/2000 imagery and the 2007-2009 imagery. This was accompanied by increases in both length of hedges and access tracks within the study area. Overall, the drivers in Selbourne seem to favour diversification. Conservation of semi-natural habitats is an important component of the ES agreements with significant areas of options for restoration of heath and semi-natural grassland included in the agreements. There are also in-field wildlife conservation measures such as management of field corners and nectar flower mixtures. These would indicate that the land management is focussed on maintaining semi- natural areas.

ALT Assessment

This ALT contains landscapes which are relatively diverse when compared to other ALTs. Significant amounts of woodland remain in the landscape of all three study areas. The landscape of the East of Guildford site is influenced by having a high population compared to the other study areas, while the characters of the other two study sites are somewhat protected by being within an AONB or National Park. Selbourne represents a landscape in which modern agricultural production methods have been imbedded within the a historic pattern of fields and common land, while the pattern of land use in Eridge appears to be the least changed by modern agricultural practices.

In all three South East Mixed sites the dominant land use change is the conversion of arable to grassland (Figure 7). Eridge and the East of Guildford sites have similar patterns of land use change although Eridge with a smaller proportion of the study site undergoing land use transitions than in the East of Guildford site. This may be due to the relatively short time period between the dates that the pre- and post-2005 imagery was taken over the Eridge site (5 years) in comparison with the other two sites (7 and 9 years).

Compared to many sites in other landscape types, the field structure appears to have changed relatively little, with field parcel changes affecting a very small proportion of the Page 29

Eridge and Selbourne study sites (Figure 9). For all three sites the net change in parcels is positive and differences in the number and area affected mirrors the length of time between compared images, where Eridge has the smallest change and the shortest period between image sets and the East of Guildford site having the largest area and change in number of parcels along with the greatest time period between imagery sets.

Figure 7. Proportional change in habitat area for South East Mixed sites

Figure 8. Proportional change in area given over to field margins for Eastern Arable sites

Page 30

Figure 9. Change in-field parcel numbers and proportion of study area affected for South East Mixed sites

Page 31

Upland ALT

Table 10. Imagery dates and area under agri-environment scheme agreements for Upland sites

Site Imagery Dates Area under AES Agreement Pre-2005 Imagery Post-2005 Imagery Pre-2005 Post-2005 Imagery Imagery Allendale March 2002 April 2010 1597.3 (2002) 2450.7 May 2002 May 2010 1346.0 (2004) May 2004 September 2010 Edale and Hope September 1999 April 2007 2017.5 2509.1 (2007) Valley May 2009 2927.6 (2011) June 2010 April 2011 West of Chagford June 2002 June 2006 3536.2 2937.2 (2006) May 2007 3177.9 (2009)

Table 11. Data Quality assessment of modern imagery for Upland ALT sites

Site Pre-2005 Imagery Post-2005 Imagery Comparison of Collected within 1 Collected Within 1 Same Time Period year Year (+/-4 Months) Allendale N Y N Edale and Hope Valley Y N N West of Chagford Y N Y

Site Descriptions

Allendale

Allendale represents an Upland site centred on a river valley. The study area includes improved grazing in the lower valley and extensive grazing in the upper valley sides. The fields are defined by stone walls and there are a number of large blocks of woodland within the study area. The site has a very high level of agri-environment scheme uptake as well as being part of an AONB.

There are no habitat transitions which particularly dominate in the Allendale study area, although the largest change is the loss of grassland while there were small gains in arable, Page 32 scrub and woodland. Areas of grassland were converted to scrub, arable land and to bare ground, and grassland was also lost to the construction of farm buildings and to a static caravan park. Some of this loss was offset by parcels of arable land that were converted back to grassland, but overall the effect was a net loss. There have been a number of small woodland plantings across the study area but some woodland was also felled, leading to a small net gain in woodland area. While the Countryside Survey predicted that there would be expected losses for all these habitats, it should be noted that the variability associated with the expected changes is very high. This means that we cannot say that the relatively small observed increases are unusual given the land class composition.

There has been little change in the structure of the study area, with just one instance where two field parcels have been merged into one. The length of hedges and ditches has increased, but possibly more significantly for the character of the study area, there has been a loss of length of wall. The ES options for the area are dominated by the moorland management options, and it should be noted that the area of options far exceeds the study area which is likely due to central registration of options on large areas of common land to a location within the study area. Included in these agreements are options for the restoration and maintenance of species rich semi-natural grassland, and management of rough grazing for birds. Although there are relatively few opportunities to include nature conservation measures on more productive land in the study area, there are a number of instances where field corner management has been chosen for inclusion in the agreements. There is also a significant uptake of options to preserve archaeological features in the land under agreement. Overall this would indicate that there is a strong driver to maintain the benefits to wildlife and maintain and enhance the existing cultural value of the landscape in which the study site sits.

Edale and Hope Valley

The Edale and Hope Valley study area contains an upland pastoral landscape associated with a concentration of settlement in the south and east, which merges into open moorland towards the north west of the study area. The study area was chosen to represent upland sites with a low uptake of agri-environment scheme options and are associated with protected areas, as the site is within a National Park and contains significant areas of SSSI land.

There are relatively few changes in habitat observed in the imagery. The largest of these changes is the conversion of grassland to arable. In contrast only one arable field was converted to grassland while a second area of arable land was converted to a pony ring. The loss of grassland to arable in the study area was mitigated in a small way by a

Page 33 construction area and a quarry site being converted to grassland. There was a net increase in the total area of woodland, with the area of woodland created exceeding the area felled. It was also observed that some areas of woodland had been thinned, which would be an indicator of the active management of the woodlands in the study area. Only one area of grassland was allowed to revert to scrub and seven areas of new building were identified. These included two areas of new housing, construction of a farm yard, livestock and poultry housing, commercial units and a tennis court. From the mean expected change for the Countryside Survey we would expect to see reductions in the area of grassland, arable and scrub, and an increase in woodland. However, the high variability associated with these statistics means that the observed changes are well within the expectations for this site.

There has been little change in the structure of the study area, with only two field parcels undergoing modification to create 5 new field parcels in the later imagery. Similarly the loss of hedge is relatively small, however there has been a large increase in the length of walls. The increase in length of wall is due to restoration of stone walls, rather than being associated with the creation of new boundaries. All the observable ES options are for the restoration of species-rich semi-natural grassland (HK7), although the largest components of the ES agreements by area are for the restoration and maintenance of moorland. Maintenance of rough grazing for birds (HL7) is also a relatively common option choice. Options for the restoration of woodland (HC8) and restoration of wood pasture and parkland (HC13) have also been taken up in the study area. However there is little evidence of the impact of these on the study site features. Restoration of wood pasture and parkland would be expected to increase the number of individual trees but only the loss of trees has been observed. The choice of options indicates a combination of management that preserves the existing character of the upland areas while restoring some of the more traditional character features in the other parts of the study area.

West of Chagford

The West of Chagford study area is similar to the Hope Valley site in that it includes a transition between open moorland and enclosed pastoral land, although in this case the open moorland covers a significant proportion of the study area. There is a moderate level of agri-environment scheme uptake within the study area and it also forms part of the Dartmoor National Park and contains protected land in SSSIs. The pastoral land is bounded by mature hedgerows and hedgebanks, and has isolated settlements scattered throughout this part of the study area. There are also blocks of woodland situated on the moorland fringe.

Page 34

There is a net increase in the area of arable land, with 22.6ha of grassland converted to arable and only 2.8ha of arable converted back to grassland. Other loss of grassland is down to one area being converted to a pony ring, and the construction of a new pond and five new houses. Three areas of scrub, totalling 0.5ha, were also cleared and converted to grassland. New areas of poaching on grassland at 37 locations and a reduction in poaching at two locations was noted in the later imagery. Five large (8.5ha) blocks of woodland had been felled along with two smaller (<0.1ha) areas of woodland. While there had been some tree planting in a cleared area of woodland, the overall the loss of woodland was the largest change seen for this study area.

There was little modification of field parcels with only one grassland field being split into two new grassland parcels. There was a small increase in the length of hedge (52m), where restoration of a length of hedge bank slightly exceeded the length of hedge lost from removal. The post-2005 imagery for West of Chagford was taken relatively soon (Table x) after Environmental Stewardship was introduced in 2005. As such many of the options would either be very early in their implementation or not started at the time the imagery was taken. Options for general upland and moorland management (HL9, HL12, UL18, UX3) are the predominant options by area. Significant lengths of hedge management options are also included in the agreements active at the time the imagery was taken. However it should be noted that these options are recorded centrally for a holding, and some or all of the management may be occurring outside of the study area. Overall the chosen options indicate a preference for management that preserves the existing character of the study area.

ALT Assessment

The Upland sites tend to be a combination of extensive low diversity moorland and irregular patchworks of pastoral fields and woodland. Allendale is an example of a very remote upland landscape where the only activities are to do with extensive livestock production. The Hope Valley study area contains a significant concentration of settlement, but the structure of the landscape away from this is similar to that of Allendale. Finally pastoral activities in the West of Chagford study area have a different character to those in the northern sites. The pastoral section of the West of Chagford study area is characterised by small settlements spread throughout the area and mature hedgerows and hedgebanks defining the field structure, rather than the stone walls common to the Allendale and Hope Valley study areas.

For all three Upland study areas the conversion of grassland to arable is greater than the conversion of arable to grassland (Figure 10), however the proportion of the study areas

Page 35 undergoing these transitions is relatively small compared to other ALTs. The impact of conservation measures are visible in all three sites, but the impact of a much larger number and area of conservation measures has been recorded for the Allendale site than the other two study areas. This may be related to a larger increase in area under agri-environment scheme agreements in the Allendale site compared to the other two sites (Table 10). All three study areas show evidence of active woodland management with tree felling and planting, although the observed impact of that is different in the three study areas. There has also been very little change in the parcel structure in any of the Upland ALT study areas (Figure 11).

Figure 10. Proportional change in habitat area for Upland sites

Page 36

Figure 11. Change in-field parcel numbers and proportion of study area affected for Upland sites

Page 37

Upland Fringe ALT

Table 12. Imagery dates and area under agri-environment scheme agreements for Upland Fringe sites

Site Imagery Dates Area under AES Agreement

Pre-2005 Post-2005 Pre-2005 Post-2005 Imagery Imagery Imagery Imagery

Northern Orton July 2003 May 2009 367.8 2319.1

Fells April 2004 September 2009

Oldham Fringe April 2005 June 2009 0 331.7 (2009)

September 2005 May 2011 435.0 (2011)

Hartland May 2001 April 2007 285.0 (2001) 1402.0 (2007)

Peninsula July 2001 August 2007 300.3 (2002) 2018.0 (2009)

Table 13. Data Quality assessment of modern imagery for Upland ALT sites

Site Pre-2005 Imagery Post-2005 Imagery Comparison of Collected within 1 Collected Within 1 Same Time Period year Year (+/-4 Months)

Northern Orton Fells Y Y N

Oldham Fringe Y N N

Hartland Peninsula N N N

Site Descriptions

Hartland Coast

This is a mixed landscape between the coast and moorland which is dominated by mixture of improved and semi-natural grasslands. It consists of relatively large fields enclosed by hedgerows and hedgebanks, with woodland running along stream banks. The study area was included specifically to capture a coastal study area but also represents upland fringe sites with a moderate level of agri-environment scheme uptake

Page 38

The change in habitats is dominated by the loss of grassland, but the Hartland Coast is also has the most diverse and complex set of changes of all the study areas investigated. In addition to changes in grassland, scrub, arable and woodland, there are increases in area of orchard, heath and land under cover. There has also been an increase in livestock housing with additional pig pens having been built. In total 169 areas were identified as having changed land use. The highest proportion of these land use changes was the conversion from grassland to arable (60 areas), followed by the reverse transition of arable to grassland (36 areas). Combined there were 39 areas of arable and grassland which reverted back to scrub, while six areas of scrub were reclaimed back to arable land and one area back to grassland. Thirteen areas of woodland were planted, but four woodland areas were cleared. The loss of grassland was greater than expected from the Countryside Survey, while the increase in scrubland and woodland was less than expected. The observed increase in arable land runs contrary to the expected change, which was a substantial loss of arable land. These figures would suggest that the land management in the Hartland Coast study site seems to have focussed on maintaining and increasing production from the study site, where other similar landscapes in other parts of England have reverted back to more of a semi-natural state.

There has been an increase in the number of field parcels in the study site. There have been a number of grassland fields which have been split and one instance where an arable field was split into two. Offsetting this increase in parcels has been the merging of a small number of both arable and grassland fields accompanied by a loss of hedge within the study site and an increase in the area of poaching by livestock. There would appear to be a land sparing approach to managing the land in this study site, with the increase in heath and woodland contracting to the introduction of covered crops and the merging of some arable fields to increase crop production on suitable land. This land sparing approach is supported by the fact that there is relatively little area given over to in-field nature conservation within the ES agreements, but there is a large component made up of maintenance and restoration of lowland heath and species rich semi-natural grassland.

Northern Orton Fells

The Northern Orton Fells study area is a diverse landscape where the main agricultural activity is dairy farming. The fields are well defined by a combination of walls, hedges and fences, The study site includes some areas of limestone pavement and calcareous grassland which are considered to be of ecological and conservation significance, hence the study area does contain some areas of SSSI. Current agri-environment scheme uptake is high within the study area and was also high under the previous ESA and CSS schemes.

Page 39

The main change in the Orton Fells study area is a loss of grassland. This is accompanied by increases in the area of arable and scrub, and a small reduction in the area of woodland. As with the Hartland Coast study area there are a large number of individual changes in land use. The main component of the loss of grassland is a change to arable with a much smaller proportion changing to scrub or woodland. Many parcels of arable land have also been converted back to grassland, with the transitions between scrub and arable being almost perfectly balanced to give a very small net contribution to the change in these two habitats. The pattern of change fits poorly with the expectations from the Countryside Survey. The expectation is to have small losses in grassland and scrub, and a moderate loss in arable. Woodland was expected to increase in the study area.

There has been a reduction in the number of field parcels in the study site, where 17 arable and grassland fields were merged into seven new parcels, while elsewhere in the study area four grassland fields were split into two. This was accompanied by a relative small loss of length of hedge. The ES options are dominated by the moorland and upland grass options, however there are options for in-field nature conservation such as nectar flow mixtures and floristically enhanced margins. There is also a significant uptake of the option to restore species rich semi-natural grassland and options for haymaking. This would suggest a strong drive to encourage wildlife in the study area and maintain and enhance the existing cultural value of the landscape.

Oldham

The Oldham study area represents the site most strongly affected by an urban population. The study area is bounded by the urban developments of Oldham and the Tame Valley. The agricultural land consists of improved and unimproved grassland used for rearing livestock, but also has areas which are not being managed and are reverting to scrub or semi-natural habitats. Field boundaries are a mixture of stone walls, hedges and fencing, and the study area represents an Upland Fringe site with low uptake of agri-environment schemes.

Compared to the other two upland fringe study areas, the Oldham study area has changed relatively little. There have been small losses of grassland, scrub and woodland while the area of arable land increased slightly. The majority of the loss of grassland is caused by conversion to arable, although some reverts to scrub and some is used for building upon, both for farm buildings and for equestrianism. The expected changes are much larger, with grassland woodland and scrub expected to increase while arable land was expected to decrease.

Page 40

There has been an increase in the number of parcels within the study area, with three grassland fields being split into eight smaller parcels. This is supported by an increase in length of hedges and walls, while access has also increased with two tracks through woodland areas. The ES options on agreements are predominately moorland and grassland management options with very little in the way of wildlife conservation measures. The drivers in this area are probably more influenced by the it’s peri-urban nature, with evidence of a diversification of land use, particularly for recreational activities as evidenced by the development of land for equestrian use and the improved access through woodland.

ALT Assessment The Upland Fringe study areas represent a large diversity of sites, including a coastal site (Hartland) and one that contains a significant area of urban development (Oldham). The Hartland and Orton Fells sites both cover the transition between open upland and enclosed grazing. The zone of agricultural area within the Oldham study area is all enclosed grazing but there additional features, such as a golf course and areas being allowed to revert to scrub , that indicate how the land use of an urban fringe site is affected by the presence of a large population centre nearby.

For two of the three Upland Fringe sites there is significant land use changes observed, with grassland to arable conversion slightly outweighing arable land being converted to grassland (Figure 12). Also in all three sites there has been a slight reduction in woodland, while conservation measures were only observed as having an impact in one of the study areas. There is also no consistent pattern in the change in field margins across the three study areas (Figure 13). The parcel structure of the Upland Fringe Sites are relatively stable, with low proportions of the study areas undergoing structural change (Figure 14).

Figure 12. Proportional change in habitat area for Upland Fringe sites

Page 41

Figure 13. Proportional change in area given over to field margins for Upland Fringe sites

Figure 14. Change in-field parcel numbers and proportion of study area affected for Upland Fringe sites

Page 42

Western Mixed ALT

Table 14. Imagery dates and area under agri-environment scheme agreements for Western Mixed sites

Site Imagery Dates Area under AES Agreement

Pre-2005 Post-2005 Pre-2005 Post-2005 Imagery Imagery Imagery Imagery

Grandbourgh August 1999 June 2006 58.2 (1999) 1317.4 (2006)

October 1999 July 2006 154.6 (2004) 1438.6 (2009)

April 2002 May 2009

September 2004

Preston on July 1999 June 2009 61.1 536.3

Wye May 2001

Powerstock August 2001 August 2008 362.9 873.2

Hills September 2001 September 2008

May 2002

Table 15. Data Quality assessment of modern imagery for Upland ALT sites

Site Pre-2005 Imagery Post-2005 Imagery Comparison of Collected within 1 Collected Within 1 Same Time Period year Year (+/-4 Months)

Grandbourgh N N N

Preston on Wye N Y Y

Powerstock Hills Y Y N

Site Descriptions

Grandborough

The Grandborough study area represents a site of balanced arable and grassland fields. The study area is predominantly agricultural, with a settlement mainly concentrated in a few locations. The study area has a fairly regular field structure in which the field boundaries

Page 43 mainly comprise hedges and with a few small blocks of woodland which have been maintained in the landscape. This study site also has had a low uptake on agri-environment scheme options.

The Grandborough study area has seen a large loss of arable land with a significant increase in grassland. Scrub and woodland have both changed very little. Mainly the observed habitat changes involve transition between the grassland and arable habitats, but the increase in scrub comes from reversion of two arable fields to scrub and another area of arable land has been converted to fishing lakes. There has also been new free-range poultry housing built on grassland. The changes to areas of grassland and arable follow the expected pattern of change, with an increase in grassland area and a decrease in arable land. However a larger increase in scrub and woodland was expected based on the Countryside Survey, although the observed change is still well within the bounds of the variability in the expected change for these two habitats.

There has been a small increase in the number of field parcels caused by one grassland field being split into four paddocks. There has however been a significant loss of length of hedges. Given that there has been no merging of field parcels observed the lost hedges do not appear to have been wholly responsible for defining the boundaries of fields. ES options for wildlife conservation are relatively well represented in the agreements, particularly in the form of management of grassland for waders and wildfowl. In-field wildlife conservation options are also present, with nectar flower mixes, field corner management and floristically enhanced field margins all present. It would appear that a land sharing approach to nature management within the study area has been taken up, sacrificing some productivity in favour of conservation activities.

Powerstock Hills

The agricultural activities in the Powerstock Hills study area are mainly centred around dairy farming and as such there is a predominance of grassland over arable in this Western Mixed landscape. There is a large block of woodland on the higher ground in the northwest of the study area and a few smaller blocks of woodland across the rest of study area. There are a mix of field sizes and shapes within the study area, which are bounded by hedgerows. This study area has a moderate level of agri-environment uptake and is also located in an AONB.

The Powerstock Hills study area has seen a moderate increase in grassland and woodland, with a moderate decrease in arable land and a relatively small decrease in scrub. The majority of the change in area of scrub comes from a single area of scrub becoming woodland, the rest coming from an area of scrub which was reclaimed for arable production. There are many field parcels that have moved from grassland to arable and similarly many Page 44 parcels have moved the other way from arable to grassland. While the numbers moving in each direction are not too different (18 areas move from grassland to arable, 22 areas move from arable land to grassland), the average size of the areas moving from arable to grassland is slightly larger than those fields moving from grassland to arable, leading to the net increase in grassland and decrease in arable land. Some additional losses of grassland are caused by woodland planting and construction of poultry housing, farm buildings and hard-standing. The pattern of change matches well with the expected change from the Countryside Survey where grassland and woodland were both predicted to have increased and arable land to have decreased. Scrub was also expected to have increased, but the small observed loss is well within the variation described for the Countryside Survey.

There has been an increase in parcel numbers, with four fields (three grassland and one arable) being split into 12 smaller grassland parcels. Two arable fields were merged to produce a relatively large (11ha) arable field and two small grassland fields were also merged into a single parcel, giving a net increase of 6 field parcels. This was accompanied by a large increase in the length of hedge and also an increase in the length of access trackways in the study area. A large proportion of the area under agreement is for the creation, restoration or maintenance of species rich semi-natural grassland. There are some in-field nature conservation options included in the agreements, but these are relatively few and do not contribute much to the total area of options under agreement. The study area fits closely with a land-sparing approach to management, with diversification supported by increased access and the introduction of recreational activities such as equestrianism.

Preston on Wye

The Preston on Wye study area contains a diverse landscape with a high proportion of cereal cropping. The agricultural activities are also more diverse than at many of the other study areas, with not only pasture and cereal crops but also orchards, soft fruit and poultry rearing. The landscape is made up of a combination of regular and irregular fields of varying size and it still retains numerous blocks of woodland within the mosaic. Uptake of agri- environment scheme options within this study area has been moderate

In the Preston-on-Wye study area, the main changes are an increase in orchard coupled with loss of grassland and arable land. The majority of the creation of orchard has come from arable land, with only one area of grassland having been converted to orchard. Moving in the other direction, two areas of orchards were lost, one reverting to grassland and the other to arable land. Similarly to the pattern observed in the Powerstock Hills study area, the numbers of fields moving in both directions between arable and grassland are very similar, but the average field size is larger for those parcels moving from grassland to arable land.

Page 45

Three areas of scrub were planted up to become woodland but two areas of woodland were cleared while another two had buildings constructed on them. Some arable was given over to poultry production, both for housing and to provide an outside area for the birds. The observed and expected changes from the Countryside Survey match up quite well, with losses expected for grassland and arable land. The observed change for woodland also fits well with the expected change with a small increase in woodland area, and while it was expected for there also to be a small increase in the area of scrub, the very small decrease is well within the bounds of the variability for the expected change in area of scrub.

There is very little change in the number of field parcels in the study area, with one instance each of merging adjacent areas on arable and grassland fields, and one grassland field which was split into two new field parcels. There was a slight reduction in the total length of hedge in the study area, which comes from an imbalance in the amount of new hedge (540m) created compared the length of hedge lost (720m). A land sharing approach to nature management seems to drive the option choice on ES agreements. While options for the management of productive grassland predominates the area under agreement, options that provide in-field nature conservation are well represented in the agreements. Of these in-field options, management of field corners is the main contributor, but options for nectar flower mixes and floristically enhanced grass margins are also present.

ALT Assessment The Western Mixed ALT study sites show significant diversity in terms of the pattern of fields and the types of land use found within them. They can range from being dominated by arable cropping (Preston on Wye), through a near equal share between pastoral and arable fields (Grandborough), or be predominantly improved grazing (Powerstock Hills). Two of the Western Mixed study areas have maintained significant blocks of woodland within the landscape, while in the most intensive site (Grandborough) only a few small blocks of woodland remain.

The total proportion of the study area undergoing the four habitat changes in each Western Mixed study area is similar (Figure 15). The transitions between arable and grassland dominate in all three sites, with Preston on Wye and Powerstock hills being relatively balanced between conversion of arable to grassland and the conversion of grassland to arable as would be expected under normal crop rotation. However, Grandborough has a larger proportion of arable land converted to grass than the other two sites and correspondingly a smaller proportion of the grassland being converted to arable land, leading to a significant increase in the area of the study site under grass.

Page 46

Figure 15. Proportional change in habitat area for Western Mixed sites

Figure 16. Proportional change in area given over to field margins for Upland Fringe sites

Page 47

Figure 17. Change in-field parcel numbers and proportion of study area affected for Western Mixed sites

Assessment of Variability between Analysts

Table 16 details the differences in the results recorded by the two image analysts for the change in features observed in the Piddlehinton study area. Relative differences in the estimates of change in area of habitat were much greater than those for length of features. The largest discrepancy was in the area converted from grassland to arable, however there was only a difference of one (9 vs 10 fields) in the total count of fields converting from grassland to arable recorded by the two analysts. The difference between the estimates from the two analysts for arable conversion to grassland is much smaller than for grassland to arable, but in this case one analyst identified a larger number of smaller parcels (18 parcels, average size 5.7ha) than the other analyst (13 parcels, average size 7.3ha).

Page 48

Table 16. Total difference and difference as a percentage of the mean of the landscape feature changes recorded by the two analysts for Piddlehinton

Difference as percentage Total difference of the mean of the two between measures measures

Arable converted to grassland 7.3 ha 7.4

Grassland to arable 17.4ha 15.2

Total land use change 14.6ha 6.6

Field parcel changes

(eg merging of fields) 3.6 ha 3.1

Hedgerow removal 14.2 m 4.4

New hedgerow 31.0m 3.7

Tree removal 0 0

Discussion and Conclusions Data issues

The initial aim of the project was to produce a time series of imagery for each site containing sets of images captured during four key policy periods between 1940 and 2010. However in performing the collation task, it became evident that historic imagery for many of the sites was not available within the Natural England archives, and there were no sites where historic imagery was available for more than one of the historic periods identified in the project. While other sources of historic aerial imagery exist, such as national archives and commercial suppliers, we did not attempt to locate imagery from these sources in this project as the project focussed on exploiting what was available within the Natural England archives.

The amount of information available to aid interpretation of the imagery also increases as it becomes more recent. Prior to the 1990’s the imagery is usually black and white, for which the analysts primarily use shape and texture to distinguish habitats and features in the Page 49 images. Colour photographs from the 1990s add the ability to use colour to help distinguish features, but the resolution for the scanned imagery, both colour and black and white, is approximately 40cm. Spatial resolution improves to 25cm in the pre-2005 digital imagery and finally a near-infrared band is added to the red-blue-green visible bands in the post-2005 imagery. The addition of the near-infrared band provides additional discrimination information which helps differentiate vegetation types, for example helping to differentiate between hedges and the vegetation of field margins.

Methodology

The project was split into two phases with a change in methodology for assessing change in landscape features between the early and later phases. The initial assessment of landscape features involved capturing the location of all the features of interest observable in imagery on the GIS. In the second phase, the methodology was changed to a direct visual comparison of pairs of images, capturing only the changes in features, not the underlying stock estimate. This change was based on pragmatic concerns over the length of time taken to perform a full digitisation in the earlier part of the study. This limits the inferences that can be drawn, as changes cannot be considered in relation to the total stock of that habitat (e.g. grassland converted to arable as a percentage of the total grassland), but it is still possible to gather some interesting insights into the types of changes occurring and what this means for the landscape as a whole.

The first phase addressed the first two tasks which involved identifying and digitising the historic aerial imagery, compiling ArcGIS project files containing the digital data for each study site, and a digitisation of all features in the two sets of modern imagery for the NAL sites. By the end of this phase Task 1 (digitisation of historic imagery and compilation of project files) was completed along with the majority of Task 2 (quantification of features) for the NAL sites.

The second phase addressed Task 3, a qualitative assessment of the changes at each of the 18 sites. Based on the time taken to undertake digitisation of features in the first phase of the project, the method to identify changes in feature metrics was altered to a more rapid assessment based on direct comparison of the imagery rather than performing a full digitisation of features in the two sets of modern imagery and then comparing the totals. This method produces a very different set of values for the change in features when compared to the difference measures derived from the full digitisation of the NAL study areas produced in the first phase of the project. The rapid assessment is more akin to an assessment of the visual impact of changes to features in the study areas and is therefore of limited use in assessing stock and change in landscape features, however it still has value

Page 50 as a holistic landscape character assessment of visual impact and landscape experience. The full digitisation provides a better quantification for assessing stock and change, but takes significantly longer to perform and is still sensitive to subjectivity on the part of the analyst.

The study also produced a set of expected changes based on Countryside Survey square data and the CEH Land Class composition of the study areas. The analysis of the Countryside Survey data provides estimates with a large amount of variability associated with the expected value. This variability means that although the expected and observed values can be of different magnitudes, the observed change values for a site are often well within the range provided by the Countryside Survey.

Landscape changes

In all cases, there were more parcels in agri-environment agreement at the time of the second set of images than the first, and many apparently positive changes in environmental terms were observed in the landscapes studied between the two sets of images. These included, in some cases, reductions in field size coupled with increases in the length of hedgerow, increases in areas of woodland and/or scrub etc. It should however be borne in mind that the impact of changes always needs to be considered in the context of the landscape in which it occurs; thus in traditionally open landscapes, the planting of woodland may not always result in a net environmental benefit. Likewise, conversion of arable to grassland may or may not be positive. Where the landscape is already predominantly grassland, it may be more environmentally beneficial to retain the remaining arable area (Robinson et al., 2001).

Changes from arable to grassland and vice versa were often the largest impacts, at least in area terms, within the lowland and upland fringe landscapes. In many cases, the areas converted were similar (e.g. Piddlehinton, Northern Orton Fells, Preston on Wye, which may indicate the use of short term grassland within mixed grassland/arable rotations rather than long-term conversions from one to the other. In other cases, the conversion in one direction is much greater, suggesting a trend towards longer-term change (e.g Crewkerne, Selborne, Grandborough: arable to grassland; Myton on Swale: grassland to arable.

Despite evidence of many apparently positive changes, negative changes were also evident. For example 26 arable fields were merged into eight larger fields, with an associated loss of hedgerows and ditches. Although the rate of change in hedgerows is much reduced since 1990 compared to earlier decades, and recent hedgerow losses appear to be mainly a result of lack of management rather than removal (Carey et al., 2008), removal of hedgerows to allow increases in field size is clearly still occurring in some areas. This is despite the Page 51 requirement to obtain permission from the local authority before hedgerow removal can take place2. Nevertheless, mean change in hedgerow length was always at the positive end of the range expected from the Countryside Survey statistics (see Annex 4).

The upland landscapes contrasted with the other landscape types in several ways: (i) there was greater consistency between sites; (ii) there was much less change; (iii) there was a much higher area in agri-environment schemes at the start of the period;. The slower rate to change may be explained partly a result of the metrics used (uplands tend to be mainly open moorland and rough grass rather than enclosed grassland and arable fields), and partly because the rate of change is inherently slower. The higher areas in agri-environment schemes are to be expected because many uplands were designated as Environmentally Sensitive Areas and take-up of this scheme was high for economic reasons.

Within the other landscape types there were often contrasting trends between the different sites. These may result from different approaches among landowners; many of the sites were relatively small and could easily be affected at the landscape level by the policies of one or two large landowners. Differences between landowners could also explain contrasting trends within individual sites, which were evident in some cases (e.g. Prickwillow, Eastern Fringe of Guildford, and others where both field amalgamation and splitting occurred).

Recommendations This project has shown that there is some potential going forward to use modern remotely sensed imagery to undertake quantitative assessments of landscape features. The future of this work is likely to involve the use of automated classification to produce habitat classifications, to perform automatic segmentation of images to detect parcel changes and object-based classification to identify linear and point features. While this may require initial investment in the software and ruleset development, it would significantly reduce time and effort if either a larger number of study areas were assessed or if a rolling study gathering annual time series of additional imagery was performed in the future.

The cost of collecting good quality aerial imagery is coming down, particularly with the maturation of Remote Piloted Aircraft Systems (also known as Unmanned Aerial Vehicles) for collecting high resolution imagery. Not only does this provide the standard aerial imagery but can also be used to produce high resolution surface models which for instance can be used to help identify ditches and walls, hedgerow trees or ridge and furrow in fields. Reduced cost is also important as it provides the flexibility to gather imagery more frequently and with more control over the date of capture to minimise intra-annual variation in the

2 https://www.gov.uk/countryside-hedgerows-regulation-and-management Page 52 imagery within sites, and to standardise the time period between sets of imagery across all sites.

The initial aim for this project was to be able to trace the development of the study sites through use of historic imagery. There is certainly more historic aerial imagery available, but this may lie in commercial catalogues or with organisations other than Natural England. A significant resource commitment would be needed to locate, digitise and georeference this imagery. Analysis of this imagery would also need to be supported with well referenced land management data for the period in which the images were taken. Potentially this may be best run as a joint project between Natural England and English Heritage.

References Baily, B., Riley, M., Aucott, P., & Southall, H. (2011). Extracting digital data from the First Land Utilisation Survey of Great Britain–Methods, issues and potential. Applied Geography, 31(3), 959-968.

Boatman, N., Willis, K., Garrod, G., and Powe N. (2010). Estimating the wildlife and landscape benefits of environmental stewardship. Prepared for DEFRA by Food and Environment Research Agency and Newcastle University (2010). Available at http://archive.defra.gov.uk/evidence/economics/foodfarm/reports/documents/estimatingthewil dlife.pdf

Carey, P.D.; Wallis, S.; Chamberlain, P.M.; Cooper, A.; Emmett, B.A.; Maskell, L.C.; McCann, T.; Murphy, J.; Norton, L.R.; Reynolds, B.; Scott, W.A.; Simpson, I.C.; Smart, S.M.; Ullyett, J.M. (2008) Countryside Survey: UK Results from 2007. NERC/Centre for Ecology & Hydrology, 105pp. (CEH Project Number: C03259).

Countryside Agency, (2006). Agricultural Landscapes: 33 Years of Change. 116pp

Countryside Commission, (1974). New Agricultural Landscapes. 98pp

Countryside Commission, (1983). Agricultural Landscapes: A Second Look. 80pp

Countryside Commission, (1997). Agricultural Landscapes: A Third Look. 126pp

Defra (2011) Biodiversity 2020: A strategy for England’s wildlife and ecosystem services.

Defra (2013) Monitoring the effects of Environmental Stewardship on Landscape Character and Quality. Report prepared by LUC in association with Countryscape and Professor Carys Swanwick.http://randd.defra.gov.uk/Default.aspx?Menu=Menu&Module=More&Location=No ne&Completed=0&ProjectID=17454 (Accessed 05/01/2015)

Page 53

Maskell, L. C., L. R. Norton, S. M. Smart, P. D. Carey, J. Murphy, P. M. Chamberlain, C. M. Wood, R. G. H. Bunce, and C. J. Barr. (2008). CS Technical Report No. 1/07 Field Mapping Handbook. NERC/Centre for Ecology & Hydrology

Morton, D., C. Rowland, C. Wood, L. Meek, C. Marston, G. Smith, R. Wadsworth, and I. Simpson. (2011) Final Report for LCM2007-the new UK land cover map. Countryside Survey Technical Report No 11/07. NERC/Centre for Ecology & Hydrology

Robinson, R. A., Wilson, J D Crick, H Q P. (2001). "The importance of arable habitat for farmland birds in grassland landscapes." Journal of Applied Ecology 38: 1059-1069.

Page 54