Regional cultural landscape conservation planning using statewide predictive models

1
Regional cultural landscapeconservation planning
using state-wide predictive models
Mal Ridges
Regional Assessment Unit (CHD) Department of
Environment and Conservation, NSW
2
Why do we conserve cultural heritage?

• Cultural heritage (sites) are important to
  Aboriginal people because they provide a key
  connection to country
• Protecting, ensuring access, and managing
  cultural heritage is therefore an investment in
  the capacity for Aboriginal communities to
  maintain those connections into the future
• In doing so, we are also helping to ensure the
  persistence of a cultural landscape
• This means the goal is quite similar to nature
  conservation persistence of biodiversity vrs
  persistence of a cultural landscape
• We can therefore employ similar approaches to
  systematic assessment of cultural heritage

3
Enter conservation planning...

• Cultural heritage lags behind natural heritage in
  its approach to developing regional conservation
  strategies
• AHIMS and site-based approaches, are limited
  because theyre only based on point data
• What we should really be focusing on is the whole
  landscape ie distributions, degree of loss,
  threats vulnerability, and overall conservation
  status
• Central to this issue is addressing
  representativeness systematically and
  quantitatively
• It also requires a change of focus from sites and
  site significance, to landscapes and priority
  areas

AHIMS Aboriginal heritage information
management system (NSW site register)
4
Mapping distributions through predictive modelling

• Predictive modelling can fill in the gaps in site
  records and provide a whole-of-landscape view of
  site potential
• Approach uses international best-practice in
  presence-only modelling
• Models based on AHIMS data (45,000 sites) up to
  21 variables

• Terrain elevation, slope, aspect, curvature,
  ruggedness, topographic position
• Water proximity (cost-distance) to streams,
  rivers, lakes, coastline, weighting for stream
  order, wetlands
• Geology geological map units simplified into key
  rock types
• Vegetation (pre 1750) state-wide communities
  transformed into similarity within landscapes
• Visibility degree to which any location is
  visible within the landscape- separated into
  visibility looking up vrs looking down

5
Liverpool range
Barrington tops
Dubbo
Hunter Valley
Newcastle
Wollemi
Blue Mountains
Sydney
6
AHIMS records for scarred trees
7
Scarred trees model (original extent)
Low Likelihood
High Likelihood
8
State-wide models

• Models have been derived for 12 feature types
• They cover the whole of NSW at a resolution of 1
  Ha
• Models represent relative likelihood (NOT
  probability)
• KEY ASSUMPTION 1 models do not take into account
  detectability (ie archaeological visibility)
• KEY ASSUMPTION 2 models indicate a distribution
  as it was prior to European settlement

9
Stone Artefacts
Low Likelihood
High Likelihood
10
Rock Art
Low Likelihood
High Likelihood
11
Burials
Low Likelihood
High Likelihood
12
Ceremonial rings (Boras)
Low Likelihood
High Likelihood
13
Earth mounds
Low Likelihood
High Likelihood
14
Grinding grooves
Low Likelihood
High Likelihood
15
Hearths
Low Likelihood
High Likelihood
16
Shell middens
Low Likelihood
High Likelihood
17
Stone Arrangements
Low Likelihood
High Likelihood
18
Stone Quarries
Low Likelihood
High Likelihood
19
Scarred Trees
Low Likelihood
High Likelihood
20
Similarity of feature composition
Similarity of colour similarity in the
combination of features
21
Archaeological regions
22
We now have some databut what do we do with it
???
The next step...

• The issue of model accuracy, validation, sample
  bias, locational accuracy etc. etc. Ill leave
  for this talk- ask me afterwards
• What I really want to illustrate is how we use
  this information for mapping priorities...

23
DERIVING A PRIORITY MAP
Original extent
Modelled distribution of AHIMS feature types
Current extent
Original extent filtered using parameters for
land-use
Local status
How much occurs/is protected locally?

Loss
Local loss
How much has been lost locally?

Threat
Potential for land use to change?
How much has been lost/retained in the region?

Vulnerability
How vulnerable are sites to disturbance?

Feature Priority
Regional Status
Combined Priority
Summed across all feature types
X
24
Scarred trees model (original extent)
Low Likelihood
High Likelihood
25
Scarred trees model (current extent)
Low Likelihood
High Likelihood
26
DERIVING A PRIORITY MAP
Original extent
Modelled distribution of AHIMS feature types
Current extent
Original extent filtered using parameters for
land-use
Local status
How much occurs/is protected locally?

Loss
Local loss
How much has been lost locally?

Threat
Potential for land use to change?
How much has been lost/retained in the region?

Vulnerability
How vulnerable are sites to disturbance?

Feature Priority
Regional Status
Combined Priority
Summed across all feature types
X
27
Scarred tree model (degree of loss spatial
context)
Low Loss
High Loss
28
DERIVING A PRIORITY MAP
Original extent
Modelled distribution of AHIMS feature types
Current extent
Original extent filtered using parameters for
land-use
Local status
How much occurs/is protected locally?

Loss
Local loss
How much has been lost locally?

Threat
Potential for land use to change?
How much has been lost/retained in the region?

Vulnerability
How vulnerable are sites to disturbance?

Feature Priority
Regional Status
Combined Priority
Summed across all feature types
X
29
Archaeological regions (classification of data in
previous slide) USED TO CALCULATE
SPATIAL VARIATION IN STATUS
30
DERIVING A PRIORITY MAP
Original extent
Modelled distribution of AHIMS feature types
Current extent
Original extent filtered using parameters for
land-use
Local status
How much occurs/is protected locally?

Loss
Local loss
How much has been lost locally?

Threat
Potential for land use to change?
How much has been lost/retained in the region?

Vulnerability
How vulnerable are sites to disturbance?

Feature Priority
Regional Status
Combined Priority
Summed across all feature types
X
31
Land capability land tenure urban (THREAT)
Low
High
32
DERIVING A PRIORITY MAP
Original extent
Modelled distribution of AHIMS feature types
Current extent
Original extent filtered using parameters for
land-use
Local status
How much occurs/is protected locally?

Loss
Local loss
How much has been lost locally?

Threat
Potential for land use to change?
How much has been lost/retained in the region?

Vulnerability
How vulnerable are sites to disturbance?

Feature Priority
Regional Status
Combined Priority
Summed across all feature types
X
33
Scarred trees model (current extent)
Low Likelihood
High Likelihood
34
Scarred trees priority
Low Priority
High Priority
35
DERIVING A PRIORITY MAP
Original extent
Modelled distribution of AHIMS feature types
Current extent
Original extent filtered using parameters for
land-use
Local status
How much occurs/is protected locally?

Loss
Local loss
How much has been lost locally?

Threat
Potential for land use to change?
How much has been lost/retained in the region?

Vulnerability
How vulnerable are sites to disturbance?

Feature Priority
Regional Status
Combined Priority
Summed across all feature types
X
36
Graphical representation of status
Original extent
Amount left now
Priority amount
37
(No Transcript)
38
(No Transcript)
39
20 loss
40
priority is 57 of current
41
4 of current protected within reserves
42
Regional conservation priority (all features
combined)
Low Priority
High Priority
43
Implications...

• The models provide the first extensive detailed
  view of Aboriginal feature distributions over NSW
  and have research value in their own right (eg
  survey design)
• They provide the first systematic examination of
  representativeness using measures other than
  simply rarity and/or research value
• DEC is currently working with Aboriginal
  communities to combine this information with
  their own assessments of significance to develop
  regional conservation strategies
• This potentially empowers Aboriginal communities
  in the planning process, and helps them to be
  prepared to make the most of conservation
  opportunities
• Still a LOT of validation work to do with the
  models- but this will be an on-going process