Participatory GIS for collaborative deer managementPresentation Transcript
Participatory GIS for collaborative deer management Justin Irvine, Althea Davieshttp://firstname.lastname@example.orgAlthea.email@example.com
Structure1. Background to the issue2. Constructing a GIS model3. Participation in GIS4. Validating GIS predictions5. Using GIS to address local NRM conflicts
Background Deer as a case study in conflict and collaboration: Deer are an important rural resource: Income for landowners - Venison - Jobs for stalkers - Enjoyment for hunters/tourists BUT also a source of conflict: Damage to forest/farm crops Road traffic accidents Overgrazing priority habitats
Background Deer are a common pool resource
Background Sporting estates • Red deer regarded partly as an economic resource and partly as a cultural service • Revenue derived from paying clients stalking trophy stags and from venison revenues (only stags generate trophy revenues stags generate twice the venison revenues of hinds) • Costs incurred in stalkers’ wages • management objective: – Sufficient stag densities to ensure sporting success translate to 15- 20/km2 – maximise profit derived from stalking
Background Conservation woodland • managed by conservation authorities SNH, NTS, RSPB • enhance biodiversity • regenerate native woodland • high deer densities prevent regeneration of native tree species deer densitymanagement objective – reduce population density to c. 5 deer per km2 to 20 initiate regeneration of native trees – typically reduce population to this level over 5 year timescale – hold population at reduced level for further 20 5 years to allow regenerating woodland to establish – minimise cost, given these constraints 5 25 time
Background Neighbouring businesses: different objectives 20 deer per km2 5 deer per km2 Sporting Woodland Hinds:stags 1.3:1 Estate Restoration Hinds:stags 0.6:1
Background Neighbouring businesses: conflicting objectives £3004 £6397 -2 -2 km km lower Sporting Conservation higher Estate Woodland profits costs £3537 £3537 £5070 -2 £5070 -2 -2 km -2 km km km Sporting Sporting Conservation Conservation Estate Estate Woodland Woodland
Background 13 21 % % 14 % 15 % 14 27 % 31 % %
Background Increasing deer density: a source of conflict over habitat use: 16 (km2) 14 12 Total deer density 10 8 6 4 2 0 1960 1965 1970 1975 1980 1985 1990 1995 2000 Year Deer Commission for Scotland data redrawn from Clutton-Brock et al 2002
GIS construction Can GIS help? Definitions of GIS • A data input subsystem that collects and processes spatial data from various sources. This subsystem is also largely responsible for the transformation of different types of spatial data (i.e. from isoline symbols on a topographic map to point elevations inside the GIS). • A data storage and retrieval subsystem that organizes the spatial data in a manner that allows retrieval, updating, and editing. • A data manipulation and analysis subsystem that performs tasks on the data, aggregates and disaggregates, estimates parameters and constraints, and performs modeling functions. • A reporting subsystem that displays all or part of the database in a tabular, graphic, or map form.” Michael N. DeMers, 2000
GIS construction Participatory GIS for collaborative deer management Why use a participatory GIS platform for natural resource management? PGIS can facilitate the integration of stakeholders’ and scientific knowledge Collection and integration of knowledge It can facilitate improved understanding and stimulate discussions over the use of resources Analysis and assessment, Modelling/planning Will affect collaboration Facilitate communication of preferences & knowledge exchange Definition of participation:- To take part; to have or possess
GIS construction Participatory GIS for collaborative deer management Can local & scientific knowledge be integrated to create shared knowledge to underpin sustainable management? Capture local practitioner Collate scientific knowledge knowledge. e.g.–habitat maps, topography Integrate to inform conflict – e.g. DeerMap prediction of deer distribution Collaboration tool • Consensus building, negotiation of compromises & development of management innovations
GIS construction Developing the p-GIS….. DeerMAP combines spatial data to produce a preference map. It does this by combining raster maps of: Forage •Shelter •Comfort •Disturbance
GIS construction DeerMAP: - A spatial model of deer habitat preference. Scientific knowledge: Produce a baseline preference map by combining: • Forage - Land Cover of Scotland 1988 with habitats ranked by grazing ecologists • Shelter - Topographic Exposure maps (TOPEX) • Comfort - OS Digital Elevation Model (DEM) • Disturbance – paths and stalking The input maps are ‘multiplied’ together, Preference = Forage x Shelter x Comfort x (1 - Disturbance) Each of the input layers is a continuous value map between 0 and 1, and so the output map is also a continuous value map between 0 and 1
GIS construction DeerMAP idea LCS88 0 3 0 Feeding raster 4 3 5 1 x 1 1 Cover raster 1 0 x Topex 2000 wind direction 0 1 1 0 1 1 Shelter raster = Output raster
GIS construction DeerMAP structure OS Map LCS88 DEM Stalking Paths Cover Habitat Terrain Shelter Shelter Forage Shelter Comfort Disturbance DeerMAP
GIS construction Vegetation map
D ry he 10 12 14 16 0 2 4 6 8 at he rm Sm oo oo r th gr a ss la C n d oa r GIS construction Yo se un gr g as co slSe ni an m fe re d i-n ou at ur s wo al co o dl ni an fe d ro u s w oo dl W an et d he Yo at un he Stags in Winter g rm br oo oa r dl e af wo o dl an M d ix ed w oo C dl on an i fe d r ou M s at pl ur an e ta br t io oa dl n ea f w oo dl an d Br ac ke n Bl an ke tb og Sc ru b Relative forage, shelter and cover scores M for each vegetation type (values set by a group of grazing ecologists then scaled to 0-1) on ta ne
GIS construction GIS Shelter and cover: using DEM & TOPEX TOPEX uses GIS Digital Elevation Maps (DEM) • A measure of Topographic Exposure • It is the sum of angle to skyline in the eight cardinal directions • with the negative angles recorded as zero (Wilson, 1984). • High Topex scores indicate well sheltered locations
Wind Weighted TOPEXWeights the contributions to theTopex score from each cardinaldirection to take account ofprevailing wind. TOPEX Wind Direction Weighting 1.2 Weighting = 1 (cos(angle)+1)/2 0.8 Weighting . 0.6 0.4 0.2 0 0 45 90 135 180 225 270 315 360 Angle (degrees)
GIS construction Wind Weighted TOPEX Weights the contributions to the No Topex score from each cardinal Wind direction to take account of prevailing wind.
Wind Weighted TOPEXWeights the contributions to theTopex score from each cardinaldirection to take account ofprevailing wind.
GIS construction DeerMAP prediction
GIS evaluation Validation and calibration
GIS evaluation DeerMap Validation We have access to 4 datasets which have records of actual deer numbers and locations: 1. Mar Lodge / Invercauld – 9 deer with GPS collars between Apr 1998 and Feb 2000 2. Rum – c.1700 counts (1-2 per month per block) between Mar 1981 and Nov 1999 3. Glen Affric – 25 (monthly) counts between Jun 2003 and Jun 2005 4. Glen Finglas – 46 (bi-monthly) counts between Jul 2004 and May 2007 Plus DCS Deer Census records from 1961 - 2006 Plus the results reported by a paper summarising habitat use by sheep, hinds and stags on Ardtornish estate between Dec 1976 and Oct 1977 Habitat Use By Red Deer (Cervus elaphus L.) and Hill Sheep in the West Highlands B. C. Osborne, The Journal of Applied Ecology, Vol. 21, No. 2 (Aug., 1984), pp. 497-506
GIS evaluation DeerMap evaluation: How good is it? 3. Glen Affric - 464 locations from 25 (monthly) counts 2. Rum - 76,763 locations (5022 between Jun 2003 and Jun 2005 distinct) from c.1700 counts (1-2 per month per block) between Mar 1981 and Nov 1999 1. Mar Lodge - 33,018 locations from 9 deer with GPS collars between Apr 1998 and Feb 2000 Ardtornish - summary of observations of 4. Glen Finglas - 553 locations habitat use by sheep, from 46 (bi-monthly) counts hinds and stags between Jul 2004 and May 2007 between Dec 1976 and Oct 1977
GIS evaluation 1. Validating using Invercauld & Mar Lodge GPS data Comparison of preferred areas with data derived from deer with GPS collars - split the data and use one estate for calibration, the other for validation Filter the location events to remove spatial and temporal auto-correlation (e.g. minimum of 1 day and/or 1 km distance between events) for each filtered event: 1. lookup the weather conditions (i.e. wind direction) 2. generate an appropriate deermap prediction 3. determine the preference score at the event location 4. determine where in the preference scale this value occurred calculate stats on the preference scores The method is VERY time consuming (need to generate a DeerMap map for nearly every filtered event as there is not much overlap in location/season/weather/sex combinations)
GIS evaluation 4. Glen Finglas count data 1. Split location events into sex and season combinations (Hind/Stag + Summer/Winter, ignore the Rut) 2. Determine average wind direction in each season (from wind database) 3. Generate single DeerMap predictions for each sex/season combination 4. Split the prediction into two equal area quantiles (low scores and high scores) determine the proportion of events in the ‘high’ score zone Should be > 50% (!) - and the higher the better (!?) Irvine et al, 2009, J.App.Ecol
GIS evaluation 4. Glen Finglas Count data Glen Finglas:
GIS evaluation 4. Glen Finglas: geo-referenced count data used for evaluation
GIS evaluation 4. Glen Finglas Count data Stags in Winter with original ‘top half’ prediction 51.1% of winter stag locations in top half of the preference scale
GIS evaluation Evaluation before using local knowledge: For each season/sex combination: Percentage of locations in the top half of DeerMap predicted preference areas Summer Hinds 48.1% Summer Stags 25.3% Winter Hinds 55.7% Winter Stags 51.1% Mean 45.1%
AnnotatedHard Copy Map
Kilomet 0 2 4 6 8 10 12 West Sutherland I Estates Roads Footpaths Fenced Areas Revised Priority Habitats Blanket Bog Non-Priority Native Pine Woodland Upland Calcareous Grassla Upland Heathland Upland Mixed Ashwood Upland Oakwood Wet Woodland Added Footpaths,fences, habitat changes
Re-scaling: simple way to deal with non-linearities 10.90.80.70.220.127.116.11 x^20.2 sqrt(x)0.1 one:one 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
GIS evaluation Adding in the local knowledge: Habitat updates Fenced areas Paths and tracks Importance of shelter to deer distribution Preference for higher ground in summer Forage, shelter, comfort and disturbance importance rescaled Then forage, shelter, comfort and disturbance is scaled to reflect emphasis given in interviews
GIS evaluation Evaluation after using local knowledge: Stags in Winter with: shelter rescaled 76.6% of winter stag locations in top half of the preference scale
GIS evaluation Original Deer Map prediction for Stags in Winter (Upper 25%), as used in PGIS interviews with stakeholders
GIS evaluation New Deer Map prediction for Stags in Winter (Upper 25%)
GIS evaluation 3. DeerMAP Validation using Glen Affric data Just looking at high prefernce areas in relation to counts is a bit simple: there ought to be some animals in the ‘low’ half as well – but how many ? As before, split location events into sex and season combinations (Hind/Stag + Summer/Winter and include the Rut) Determine average wind direction in each season (from wind database) Generate single DeerMap predictions for each sex/season combination split the prediction into several equal area quantiles (from low to high scores) determine the proportion of events in each quantile Compare these with the proportion of location events observed in each zone
GIS evaluation DeerMap Validation
GIS evaluation DeerMap Validation All deer count locations in Glen Affric
P-GIS in use Managing wild deer in Scotland: linking science and practice to resolve grazing conflicts Conflicts between:- •Between neighbours •Between livestock and wildlife •Between policy and practice Common thread is that the conflict involves local resource managers Yet these people are not involved in setting policy, regulations or incentives Need inclusive approaches for setting priorities
P-GIS in use DeerMAP as a conflict resolution tool: • To inform conflict between neighbours over deer movement and culling strategies • To communicate and negotiate public and private objectives (local solutions to global issues) • To explore future policy objectives (e.g. woodland expansion) Example 1: Developing a new approach to involving local land managers in achieving biodiversity objectives
chieving biodiversity objectives
Priority habitats Calcareous Upland Oak Grassland Bog Birch, Wet or mosaic Birch, Pine or mosaic Heath
‘ Habitat Tolerance to grazing’ Impact ‘Tolerance’ High High Medium Low Very LowModerate Low Very Low
‘DeerMap Preferences’ DeerMap Preference High High Medium Low Very LowModerate Low Very Low
Stag (crosses, black lines), hind (diamonds, grey lines) and total (circles, dashed lines) deer numbers predicted by the population dynamics model Ballimore Glen Finglas 450 600 400 500 350 300 400Number of deer Number of deer 250 300 200 150 200 100 100 50 0 0 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Year Year
Estate 2 Estate 1 Box 2 Winter Deer Density 2003 counts 2003 counts deer per sq km with full mixingwith no mixing no mixing 0-5 6 - 10 between between between estates 11 - 15 estates estates 16 - 20 0 1 2 3 4 5 Kilometers
Example 3. future scenarios: Woodland expansion1. Land managers have aspirations for deer densities2. These aspirations might not match actual densities3. Aspirations and actual densities might not be consistent with woodland expansion4. Where should trees be planted?
Figure 3. Winter 2010 deer density across CSDMG, represented in the three classes used in the CSDMG aspirationaldeer density map. Data source as for Figure 2.
Figure 2. Current deer density across CSDMG, based on the winter 2010 deer count. Data shown in five deerdensity groups, including zero, to illustrate potential for incorporating more than three classes. Data from Fraser, D.(2010) Red deer counts. East and West Grampians, DCS.
Figure 5. Difference between aspirational deer densities and 2010 count levels, using three density classes(lower/moderate/higher). Symbols ≤ and ≥ are used where an estate has multiple aspiration zones, since 2010 count datarelate to the whole of each estate. Comparisons of aspiration and count per zone may be possible where estates havemore detailed records. Annotations are shown as an example of how text can be added to clarify mapping.
Figure 9. Aspirational deer densities with current woodland cover.
Suitable for woodland but high deer density aspiration Suitable for woodland and low/ moderate deer density aspiration Not suitable for woodland plus high deer density aspirationFigure 12. Current woodland cover and suitability of surrounding ground for woodland growth, based mainly onbiophysical criteria, overlaid with aspirational red deer densities. Woodland suitability data provided by ForestResearch.
Benefits of pGISSpatial focus allows for broad understanding anddetailed discussionCombines local and scientific knowledgeBetter informs management decisionsGreater trust of managers in DeerMap as amanagement toolRespect, trust and understanding built up duringworkshopsIncreases willingness to work towards other solutions
Adaptive managementRecognise that this is a dynamic system – challenges will vary over time in response tochanges in climate, land use and governance.pGIS supports a novel approach to adapt to change:• That integrates across disciplines and• Involves participation of managers and policy makersat the outset
pGIS in adaptive managementManage conflict Identify conflict Evaluate progress Test solutions & Dialogue monitor success PGIS [among Policy, Research & Practitioner communities] Identify alternative solutions [eg technical– incentive–policies–subsidies–markets]