Constructing maps of ungulate species Preliminary report for Darwin Initiative Project 17-008 (Cambridge University - WWF)

MsC. Lương Văn Đức - Quang Binh University1 , August 2012

1. Aim of research

The Central Truong Son is considered the crossroads of the South Chinese and Malayan floras with a

fragmented terrain, the influence of the western continental climate system and the east coast

produce a high biodiversity value. Here beside the hoofed mammals found in other places there are

three endemic large mammals: Sao la (Pseudoryx nghetinhensis), large-antlered muntjac (Muntiacus

vuquangensis) and Truong Son muntjac (Muntiacus truongsonensis) (Hoang Ngoc Khanh, 2004). In

particular, the area of Thua Thien Hue and Quang Nam is has the most records of Sao la,

concentrated mainly in the districts of Nam Đông, A Lưới, Hương Thuỷ and Hương Trà in Thừa

Thiên Huế province and the communes of A Nông, B’Halêê, A Vương, Tà Lu and Sông Kôn in

Quảng Nam. However, these species are facing extinction and are in need of urgent conservation

action (Van Ngoc Thinh et al., 2006). Meanwhile, the conservation efforts, especially of endemic

ungulates, face problems including the difficulty of mapping species distributions to aid protected

area management plans. Because these species are rarely seen and their population densities are

very low, information remains limited. Therefore, there is a need to study the current status and

distribution of these species.

As part of the efforts to study the distribution of ungulates, the community mapping method has

been developed and applied to make use of the indigenous knowledge of local people (hunters). In

particular, besides the local distribution of each species, determining the area where people go is an

important source of information to assess the level of knowledge and the reliability of people's

knowledge. People will have a better understanding of the areas where they regularly conduct

activities than of those areas where they rarely, or never, go. Species distribution maps, once

produced, will be an important source of data for the identification of priority conservation areas

with the help of conservation planning tools such as Zonation (Helsinki, Finland). However, when

community maps are made in each village around a study area, the results showed that the areas

used for forest exploitation and the areas where the villagers know the distribution of forest animals

are not the same [for each village]. Therefore, the question is how to combine and analyze the data

1 Mob: 01663 871 696. Email: vanducwwf@gmail.com

(areas where people regularly go [= visit data] and distribution areas of the species) together to

produce maps of the distribution of species in the study area.

This study was done with the purpose of evaluating the differences between the results of the

analyses when combining visit data and species distribution data. This study will have significant

scientific and practical value, is the basis for the next research steps with the conservation planning

tool Zonation.

2. Scope of research

Thừa Thiên Huế province: communes of Thượng Long, Thượng Quảng, Thượng Nhật, Hương Hữu

(Nam Đông district) and A Đớt and A Roàng (A Lưới district); data on the distribution of species in

the areas of some other communes was also collected through community mapping in the above

communes.

Quang Nam: data on the distribution of species in the communes of A Vương, xã B’Halêê, Prao, Tà Lu

and Sông Kôn.

Scope of the study is shown below:

Figure 1. Map of the study area in Thua Thien Hue and Quang Nam.

Figure 2. Protected areas in the study area

3. Research Methodology

Data collection

Community Mapping

- Interview local people and commune officials about the areas of forest exploitation and the species

which are hunted. In each commune select a village (following consultation with village headmen

and FPD staff) which have the most understanding of forest or which go to the forest often, to

conduct interviews for a longer time than other villages, in order to better understand that

commune.

Community mapping:

First, building build a map of streams in the following format (example from Thượng Nhật

commune):

Figure 3. Base map for community mapping in Thuong Nhat

- Community mapping is divided into two stages. In phase 1 construct the map. In phase two

conduct interviews and then use the bean method to map visit and species distribution data. In the

first village in the commune, phase 2 was carried out across two half-day sessions; one for

interviewing and one for ‘beaning’. In the remaining villages in the commune, both these activities

were combined in one half-day session .

- Phase 1: Mapping with the following basic steps:

+ People fill the names of rivers and streams in the correct places on the printed base map.

+ To compare maps between interview groups, the maps are editing and then digitized in ArcGIS.

+ Evaluation of the results in the field with a GPS device in a few places (ground-truthing).

+ Finally edited with the participation of three knowledgeable local people.

Figure 4. Filling in the names of rivers based on community knowledge

- Stage 2: Using the map (bean method).

+ Print the new base map with the names of all rivers and streams which have been identified.

+ Using the beans: ask hunters to use the beans to assess the level of: places where people from the

village most go to the forest and abundance of all ungulate species (Sao la, large-antlered muntjac,

Truong Son muntjac, red muntjac, sambar, wild pig, serow) and the intensity of hunting and

trapping. The method works by placing more beans in areas with high encounter frequency of

species or areas where people go to the forest a lot, and few beans in areas of lower intensity.

The advantages of this approach are that it is rapid, easy to quantify and correct and it is clear.

First, using the local names of the species, ask hunters (12 / village) to classify and describe the

species ( horn shape, hair color, size, habitat where commonly encountered, food), beginning with

the species most easily recognised such as wild pig, serow, etc and then moving on to the species

more difficult to identify such as the different types of muntjacs. These characteristics described by

the hunter are compared with the scientific literature in order to verify the accuracy of the

information people give.

Figure 5. Community-based species-distribution mapping

(ii). GIS methods:

+ Using ArcGIS to calculate the search radius:

The ‘visit’ area data layers need to be combined with the species distribution data layers to create

the density surface for each species from the beans with a search radius (search radius: all points

within the search radius will have the same value as the point where the bean is placed)

r = 0.5 )1/ nA (Equation 1.1)

Where: r is the search radius

A is the area of the minimum convex polygon around the ‘visit’ area of all the

villages (convex polygon is created from the outermost boundary around surrounding the beans

placed on the map to indicate the forest areas people use).

n is the number of beans the local people used to indicate this forest area.

Village 5. Thượng Long

Village 6. Thượng Long

Village 7. Thượng Long

Village 8. Thượng Long

4. The analysis steps

• Step 1: For each of the villages, include all of the visit data for the village (expressed with beans)

and created one convex polygon around them all. Then, based on the calculation of "nearest

neighbor distance" (Expected Nearest Neighbour Distance: ENND) between the beans to create a

buffer zone around the convex polygon. Those are the maps AOC1.

Specifically:

i) building the basic convex polygon:

From paper maps with beans showing the forest area where people usually go (Visit data), create

digital versions using ArcGIS software.

Use the ET Geo Wizards “Build Convex Hull” tool to create a convex polygon. This will create a

polygon around the outermost points (beans)

Figure 6 Convex polygon map layer for the area where people from Huong Son village (A Roang commune) often visit.

ii) Building map layer AOC1

Because each bean represents an area surrounding that bean, it is necessary to calculate the area

that the bean represents. The formula for calculating the ENND can be modified as follows:

)1/(5.0 nAr (1.1)

Where: r: search radius

A: area of polygon (m2)

n: number of beans in the polygon area.

Specifically:

• Count the number of polygons in the convex polygon: n = 28

• Calculate polygon area above (Figure 7): A = 963524m2.

Therefore, the search radius (r) was calculated according to the formula 1.1 is: 298 689. Then, the

buffer tool in ArcGIS 9.3 is used to create a buffer around the basic convex polygon of the distance r

as above. AOC1 results are shown as follows:

Figure 7. AOC1 for Huong Son village (A Roang).

• Step 2: For each of the areas, create a data class2 including the polygon surrounding each bean

representing the places where people go (visit data). Do this by creating a buffer around each bean

of the ENND value3 calculated above. The result will be the class map AOC2.

Figure 8. AOC2 for Huong Son village (A Roang).

• Step 3: Create a map data class buffered in the same way as in step 2, but using the data on the

distribution of each species. This produces the Buffered species files. The data classes are stored

with names like: Saola_hh_v1, Saola_hh_v2, etc; meaning the buffered species file for Saola in

village 1 (Huong Huu), for Saola in village 2 (Huong Huu), etc.

Figure 9. Buffered species file data class

2 i.e. shapefile

3 i.e. ‘r’. ENND = Expected Nearest Neighbour Distance

• Step 4: Create a raster data layer (Naïve species raster). The value of each cell (pixel)

represents the number of pixels of the polygons created in step 3 which overlay on top of that cell.

The size of each cell is 200x200m.

This is carried out through the following steps:

i) Use the Union tool to combine the polygons created in Step 3 for the species, for example:

Saola_hh_v1. Call this data layer: Saola_hh_v1_union.

ii) Create new columns in the attribute table of this data class ( Saola_hh_v1_union), as follows:

"poly_value" (value = 1),and the coordinates of the center of each polygon: “cenX”, and “cenY”

iii) Use the Dissolve tool on the above species data layer, which will create a new column:

SUM_value, this column shows the number of polygon overlapping on top of each other in each grid

cell. The reason is that each polygon in the same cell will have the same value of X Coordinate of the

centroid X and Y Coordinate of the centroid Y are the same.

iv) With this done, produce a raster where the values of the cells are derived from the column

‘SUM_value’

v) Use the Classify tool to assign values to NoData areas a value of 0 (assign NoData with value 0)

and use the raster calculator to sum the rasters for all villages for the species studied.

• Step 5: Create data layer of people's understanding from the polygons produced in step 1 (using

the same convex polygon for all species: AOC1) (Knowledge1): Produce raster data maps with a

resolution of 200x200m (each pixel cell size 200x200m), in which the value of each grid cell (pixel)

represents the number of AOC1 polygons overlay on top of each other in the grid. Perform the same

steps as in step 4.

• Step 6: Create data layer of people's understanding from the polygons produced in step 2 (which

were created with the method of creating buffer polygons around each bean: AOC2) (Knowledge2 ):

raster map layer with resolution 200x200m (each pixel cell size 200x200m Produce raster data maps

with a resolution of 200x200m (each pixel cell size 200x200m), in which the value of each grid cell

(pixel) represents the number of AOC2 polygons overlay on top of each other in the grid. As in step 5

• Step 7: Create a data layer in raster format for each species “Final species raster (verson 1)” for

each of the 7 species. This data layer was produced using the tool Raster Calculator: results of

calculations according to the following formula: FLOAT (Naive species rasters/Knowledge1) (the

value of the grid layer Naïve species rasters (eg Saola) divided by the value of the grid layers map

Knowledge1). Seven layers created will be named after the class Naïve species rasters of that

species.

• Step 8: Final species raster (verson 2): for seven species:. Same as step 7 but use the Knowledge 2

rather than Knowledge 1 raster.

Research outputs

• Output 1: AOC14: A convex polygon around all the beans in the data layer ‘visit’(Hay Đi) (for each

village).

• Output 2: AOC2: A polygon layer formed of circles (buffers) around each individual bean in the

data layer ‘visit’(Hay Đi) (for each village).

• Output 3: Buffered species files: species layer with buffer

• Output 4: Naïve species rasters: raster data layer for the species.

• Output 5: Knowledge1: Raster layers with resolution 200x200m (each pixel of size 200x200m), in

which the value of each pixel represents the number of AOC1 polygons overlapping that pixel.

• Output 6: Knowledge2: Raster layers with resolution 200x200m (each pixel of size 200x200m), in

which the value of each pixel represents the number of AOC1 polygons overlapping that pixel.

• Output 7: Final species raster (verson 1): for all 7 species: Saola, large-antlered muntjac, Truong

Son muntjac, Red muntjac, Sambar, Serow, Wild Pig.

• Product 8: Final species raster (verson 2): for all 7 species: Saola, large-antlered muntjac, Truong

Son muntjac, Red muntjac, Sambar, Serow, Wild Pig.

• Product 9: Report presents detailed scientific arguments, methods, interpretation methods,

analytical methods.

• Product 10: Map shows the result of the combined analysis with the Saola protected areas and

forest compartments

4 AOC stands for ‘Area of Cognizance’