1) The study aimed to develop a grass cover map of Somuncurá Reserve in northern Patagonia, Argentina using Landsat imagery and determine which levels of grass cover guanacos select.
2) Random forest analysis of EVI and MSI bands from Landsat imagery produced a grass cover percentage map with a correlation of 0.776.
3) Analysis of guanaco census data found they significantly selected areas with 20-30% grass cover in most seasons over expected availability, selecting against areas with 10-20% grass cover in some seasons.
1. Guanaco and grassland distribution in
northern Patagonia, Argentina
AAG 2012
February 28,2012
Paul J. Haverkamp, Martin E. Zamero, Jonathan A. Greenberg,
Ricardo Baldi, Susan L. Ustin
2. Study Objectives
Hypothesis: Guanacos will select for areas
with the highest grass cover since grasses
are their primary food source.
• Objectives:
– Objective 1: develop a grass cover map of
Somuncurá Reserve using Landsat ETM+
imagery.
– Objective 2: determine which levels of grass
cover guanacos select in Somuncurá
Reserve.
3. Guanaco and Vegetation
• Grasses
– Preferred vegetation of guanacos (~65% of diet in
spring, ~45% of diet in summer)
– Tend to be sparse and interspersed with large
proportions of soil and shrubs
• Shrubs
– Less preferred in spring (~17% of diet), but greater
use in summer (~37% of diet)
– Varying densities of shrubs
• Forbes
– Varying preference
Baldi 1999. Ph.D. Dissertation.
4. Vegetation Transect Analysis
• Identified species (except grasses) and
grouped transects by percent of grasses,
shrubs, forbes, and soil/rock.
• Added GPS points to ArcGIS and added
transect cover values.
8. Random Forest Results
• Combination of EVI
and MSI gave the
best correlation.
• Pearson’s r = 0.776
– Test data comparing
actual to predicted
grass cover
• Grass cover percent
map
9. Study Objectives
Hypothesis: Guanacos will select for areas
with the highest grass cover since grasses
are their primary food source.
• Objectives:
– Objective 1: develop a grass cover map of
Somuncurá Reserve using Landsat ETM+
imagery.
– Objective 2: determine which levels of grass
cover guanacos select in Somuncurá
Reserve.
10. Guanaco Field Study Methods
• 10 wildlife and domestic animal censuses:
— Oct. 2007 — Dec. 2008
— Nov. 2007 — Aug. 2009
— Jan. 2008 — Oct. 2009
— Mar. 2008 — Dec. 2009
— Jun. 2008 — Feb. 2010
• 95 transects along road system totaling
over 327 km.
– Split based on fences, intersections,
topography changes, length.
– 500m buffer on each side of split.
11. Census Issues
• Not consistent in space or time.
• Censuses done when funding available.
• Measured in different parts of year.
• North and south parts of Somuncurá.
15. Census Analysis
• For each census :
– Guanaco observations per transect.
– Used Distance program to get effective strip
width (ESW) for each transect.
– ESW as buffer to find suitable guanaco
observations per transect, per census.
Thomas et al. 2010. Journ. Appl. Ecol. 5-14.
16. Census Analysis
• Buffered guanaco locations 75m for a 150m
diameter area used resource unit.
• Overlaid onto grass cover map and calculated
mean grass cover of the guanaco resource unit.
– Used resources of four grass cover classes.
• Overlaid ESW buffered survey transects on
grass cover map and retrieved pixel values.
– Available resources of four grass cover classes.
17. Selection Results
• Pearson’s Chi-square goodness of fit.
• Significant selection (using, observed
more or less use, based on available area
of grasses, expected)
– Nov. 2007 – XP2 = 14.249, p = 0.003
– Jan. 2008 – XP2 = 9.714, p = 0.021
– Dec. 2008 – XP2 = 19.322, p = < 0.0005
– Aug. 2009 – XP2 = 14.050, p = 0.003
– Oct. 2009 – XP2 = 39.282, p = < 0.0001
– Dec. 2009 – XP2 = 17.762, p = < 0.0005
19. Guanacos Selecting
• Selecting 20-<30% significantly more than
expected in each significant census except
January 2008.
• Selecting 10-<20% significantly less than
expected in Jan. 2008, Dec. 2008, and Dec.
2009.
• Aug. 2009 selected 40-<53% significantly less
than expected.
• Oct. 2009 selected 30-<40% and 40-<53%
significantly less than expected.
20. Spring Grass Cover Selection, Bi
October 2007 November 2007 October 2009
0.065
0.165 0.191
0.304
0.201
0.330 0.210
0.182
0.189
0.420
0.395
0.350
December 2009 December 2008
0.127 0.105
0.242 0.245 10-<20%
0.349 20-<30%
0.339
0.292 0.300 30-<40%
40-<53%
21. Summer Grass Cover Selection, Bi
January 2008 March 2008 February 2010
0.119
0.144 0.187
0.281 0.275 0.272
0.305 0.294
0.274
0.287 0.268
0.295
10-<20%
20-<30%
30-<40%
40-<53%
22. Fall Grass Cover Selection, Bi
June 2008
0.165
0.292
0.201
0.342
10-<20%
20-<30%
30-<40%
40-<53%
23. Winter Grass Cover Selection, Bi
August 2009
0.046
0.170
0.268
0.516
10-<20%
20-<30%
30-<40%
40-<53%
27. Acknowledgements
• CSTARS – UC Davis
• CONICET – Consejo Nacional de Investigaciones Científicas y
Técnicas CENPAT - Centro Nacional Patagónico
• Geography Graduate Group – UC Davis
• California Space Grant Consortium
• CODEMA – Consejo Provincial de Ecología y Medio Ambiente,
Provincia de Río Negro
• Wildlife Conservation Society
• Hemispheric Institute of the Americas – UC Davis
• Henry A. Jastro Shields Research Scholarship – UC Davis
• Pablo Raposo – Fieldwork
• Ana María Beeskow – Plant identification
Notas del editor
This is part of another study looking at guanaco migration that started in october of 2007
Based on similar areas of Patagonia
Crop 5.5 1.15Size 5.5Pos 2.1 1.95
Crop 5.5 1.15Size 5.5Pos 2.1 1.95
Crop 5.5 1.15Size 5.5Pos 2.1 1.95
As part of the migration study500m on both sides, so at least 1km between transects
Crop 5.5 1.15Size 5.5Pos 2.1 1.95
Crop 5.5 1.15Size 5.5Pos 2.1 1.95
Bi because more intuitive and possible to compare within and between surveysrelative probability of selection for category i – basically a ratio of the used resources to available resourcesForage ratio 𝑤i= oi/𝜋i𝑤i= relative probability of selection for category i.oi = proportion of sample of used resource units of category i.𝜋i = proportion of available resource units of category i.Manly’s standardized selection ratioBi = 𝑤i/(𝑖=1𝐼𝑤𝑖)relative probability of selection for category i – basically a ratio of the used resources to available resourcesForage ratio 𝑤 ̂i= oi/𝜋i𝑤 ̂i= relative probability of selection for category i.oi = proportion of sample of used resource units of category i.𝜋i = proportion of available resource units of category i.Manly’s standardized selection ratioBi = 𝑤 ̂i/(∑_(𝑖=1)^𝐼▒〖𝑤 ̂𝑖〗)
ESW helps with questions of detection probability, and eliminates sightings
Selecting at larger area than just 30m X 30m
When no significant selection based upon available resources, they are using the resources based upon how available they are.2 months (Oct.07 and June08) there were not enough expected values to indicate selection.
Dec is late spring/early summer.Aug is winterOct is spring, could have been late spring that year.This research does not explore reasoning behind the selection, that will come later.When no significant selection based upon available resources, they are using the resources based upon how available they are.So while they may not be selecting for areas with highest grass cover, they are not avoiding them, indicating that they may not be being pushed off by sheep.
Selection ratio based upon used resources/available resources, and Bi is standardized
Crop 5.5 1.15Size 5.5Pos 2.1 1.95
Preferred vegetation of guanacos (~65% of diet in spring, ~45% of diet in summer)Sheep diet consists of ~41% grasses in spring and ~44% in summer.ShrubsLess preferred in spring (~17% of diet), but greater use in summer (~37% of diet)Sheep diet consists of ~28% shrubs in spring and ~34% in summerForbesVarying preferencePlant functional type composition of guanaco and sheep diets overlap by ~65% in spring, and ~82% in summer.
Removed sheep from ranches before 2004 – evidence in other parts of patagonia that suggest rapid recovery depending on grazing history and goats