0005 
China, 
onally, 
present 
asonal 
lles in 
atellite 
NDVI 
ranges 
ther in 
of the 
habitat 
. could 
/1 sifts 
rimary 
quality 
2). It is 
index 
atellite 
nts the 
d parts 
ndeen, 
js and 
trongly 
ere are 
iomass 
enroth, 
sed to 
gazelle 
rounds 
:]les in 
s track 
er they 
ge (Le. 
expert 
et al, 
habitat 
‘tion of 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004 
gazelles are described, analysis between the actual habitat 
selection of gazelles and NDVI in their habitat is possible. Thus, 
we started tracking gazelle movement, which is the first trial of 
satellite tracking on Mongolian gazelles. 
The objectives in the present study are to describe the migration 
routes of Mongolian gazelles by the satellfte tracking, and to 
examine whether their seasonal migrations are influenced by 
change in NDVI in their habitat. 
2. MATERIALS AND METHODS 
We captured two gazelles (ID1 and ID2) in Omnogobi Province 
and another two (ID3 and ID4) in Dornogobi Province in 
southern Mongolia, in late October 2002 (Figure 1). 
Ww ect 
———M MM MÀ —— sata 
Russia 
| Mongolia 
Domeogobi 
Omnogobi 
China 
  
ne 
Figure 1. Study Area 
Each gazelle was collared with a satellite transmitter (also 
termed platform terminal transmitter, or PTT; model ST-18, 
Telonies, Inc., Mesa, AZ, USA). The weight of a PTT with 
collar is 550 g. The PTTs were programmed to transmit radio 
signals for one 8-h period per week, thus providing weekly 
location data. The location data was received through computer 
communications and computer disks sent from the CLS 
(Collecte Localisation Satellites) Service in France. In this 
analysis, we used location data of four gazelles from October 
2002 to October 2003. Location classes (LC) were categorized 
from 0 to 3. The higher the LC, the more accurate the location. 
Less accurate data are also provided as LC A and B. Keating et 
al. (1991) calculated the accuracy of LC 1, 2, and 3 data from 
PTTs. Their one-standard-deviation accuracy results, compared 
to the accuracies reported by Service Argos (1988) were 1188 
m versus 1000 m for LC 1, 1903 m versus 350 m for LC 2, and 
361 m versus 150 m for LC 3. We obtained location data for the 
four gazelles every week for a year and selected the best data in 
each day according to the LC to plot gazelle migration routes. 
When there were several data from the best LC in a day, the 
latest data were selected. About 9696 of the total number of the 
best location data in a day fell into LC 1, 2, or 3 (LC 1: 1894, 
LC 2: 27%, LC 3: 51%), with 4% falling into LC 0. 
We defined three categories of their home ranges with location 
data: annual, summer and winter ranges. These home ranges 
were calculated by applying the kernel method (Worton, 1989) 
using Geographic Information Systems (GIS), ArcInfo/ArcView 
(Environmental System Research Institute Inc.) with the Animal 
Movement Analyst Extension (Hooge and Eichenlaub, 2000). 
Annual range in each province was defined as 95% core area 
with the data from October 2002 to October 2003. Summer and 
winter ranges of each gazelle were defined as 5096 core areas 
419 
with the data from June to August, and from December to 
February, respectively. 
To analyse the environment condition, we used NDVI derived 
from moderate resolution imaging spectrometer (MODIS) 
satellite image. We downloaded vegetation index product 
(MOD13Q1 Product; 16-day composit NDVI, 250m resolution) 
in the period that gazelles were tracked, from NASA's Earth 
Observing System Data Gateway via the Internet 
(http://redhook.gsfc.nasa.gov/- imswww/pub/imswelcome/plain. 
html). Mean NDVI of each range was calculated using 
ERADAS IMAGINE (Leica Geosystems GIS & Mapping, 
LLC.). To get an index of relative quality in the different ranges 
within the annual ranges, we subtracted the overall average 
value from each of the values for the summer and winter ranges 
3. RESULTS 
3.1 Migrations of gazelles 
The two gazelles (ID1 and ID2) in Omnogobi had moved 
toward the west along the mountains in southern side since we 
started to track their locations in October 2002 (Figure 2). IDI 
moved to the west approximately 100 km in liner distance, and 
stayed there from late December 2002 to late March 2003, then 
went back to the area it had been captured and stayed around 
there until October 2003. ID2 moved from the captured point to 
the west approximately 130 km in liner distance, and stayed 
there from late November to late December 2002, then moved 
to the west another 130 km and stayed there from January to 
February 2003. ID2 didn't move after that. We found it had 
been dead during our field survey in the summer 2003. The 
cumulative distances moved over a year were 659 km and 383 
km and the maximum distances moved during one week were 
72 km in late March and 81 km in early January for IDI and 
ID2, respectively. The gazelles used the western areas in winter 
and the eastern area in summer in their annual range (Figure 3). 
  
  
Figure 2. Migration routes of IDI and ID2 in Omnogobi 
from October 2002 to October 2003. 
The double circle is the site of capture.