Modeling Species Distributions with Applications to Agriculture

1
Modeling Species Distributions with
Applications to Agriculture

• Víctor Sánchez-Cordero
• Instituto de Biología, UNAM, México
• A. T. Peterson
• The University of Kansas, USA

2
Agroecosystems and Biodiversity

• Systems of interacting species
• Crop organisms
• Pest organisms (rodents, insects, weeds)
• Invasive species
• Pollinators
• Behavior of such systems can be simulated via
  detailed understanding of the ecological
  requirements of each component of the system

3
Case Studies

• Corn ecology, areas of risk for natural areas
  from advancing agricultural frontier, and climate
  change effects
• Risk assessment of potential for invasion by a
  new pest, a vector for Xylella fastidiosa
• Risk assessment for crop damage by rodent pests
  in Veracruz, Mexico

4
Modeling Corn Ecology in Mexico

• Points where maize is planted without irrigation
  are used to create an ecological niche model and
  geographic projection of potential distribution
• Inventario Forestal Nacional is used to locate
  areas actually (2000) planted in maize and areas
  holding natural vegetation
• Comparisons used to assess areas of potential
  expansion of the frontera agricola
• Climate change predictions used to assess how
  this scenario will change over next 50 yr

5
Points Where Maize Planted without Irrigation
6
Ecological Niche Model Based on Points
7
Ecological Niche Model Without Points
8
Independent Data Distribution of Maize
9
Actual Distribution (GREEN) Overlaid on Potential
Distribution (BLUE)
10
Areas Suitable for Maize but Currently with
Natural Vegetation ( Possible Expansion of
Agricultural Frontier)
11
Protected Areas Vulnerable to Expansion
12
Protected Areas Most Vulnerable
ANPs completely within the ecological niche of
maize
13
Protected Areas Least Vulnerable
ANPs outside of the ecological niche of maize, or
mostly outside
14
Vegetation Types Most Vulnerable
15
Maize and Climate Change
16
Maize and Climate Change Difference Maps
DX scenario then AX scenario
Red worsening for maize Blue improving for
maize
17
Maize Plantations that Are Becoming Inviable
18
Case Studies

• Corn ecology, areas of risk for natural areas
  from advancing agricultural frontier, and climate
  change effects
• Risk assessment of potential for invasion by a
  new pest, a vector for Xylella fastidiosa
• Risk assessment for crop damage by rodent pests
  in Veracruz, Mexico

19
Glassy-winged SharpshooterHomolodisca coagulata
20
Homalodisca and Xylella
21
Known Points Native Distribution
22
GARP Model Inferred from Points
23
Projection to California
24
Test Model Predictions
25
Risk Assessment in Brazil
26
Case Studies

• Corn ecology, areas of risk for natural areas
  from advancing agricultural frontier, and climate
  change effects
• Risk assessment of potential for invasion by a
  new pest, a vector for Xylella fastidiosa
• Risk assessment for crop damage by rodent pests
  in Veracruz, Mexico

27
(No Transcript)
28
PREDICTIVE DISTRIBUTION OF RODENT PEST SPECIES
AGRICULTURAL CENSUS DATA
ESTIMATED CROP LOSS IN EACH MUNICIPALITY
PLANTED MINUS HARVESTED AREA
IS CROP LOSS RELATED TO THE PREDICTED
DISTRIBUTION OF RODENT PEST SPECIES?
29
(No Transcript)
30
CROP Total area Area lost (ha) (ha) RICE
21,920 1,359 BEANS 57,988 9,426
SUGARCANE 213,221 69,670 CORN 514,213 49,189 OA
T 1,198 239 COFFEE 175,027
9,427 GRASSES 1322,985 5,315 SORGHUM
5,676 555 WHEAT 1,822 378
31
CROPS NATIVE RODENT PEST RICE
1 SQUIRREL BEANS 13 RATS AND MICE
SUGARCANE 3 POCKET GOPHERS
CORN OAT COFFEE GRASSES SORGHUM WHEAT

32
(No Transcript)
33
(No Transcript)
34
(No Transcript)
35
(No Transcript)
36
Stepwise multiple regression analyses Dependent
variable crop damage in each crop in the 207
municipalities for Veracruz. Independent
variable proportional predicted coverage of that
municipality by each of the 17 rodent species
37
CROP r2 RICE 0.65, P lt 0.0001 BEANS
0.07, P lt 0.05 SUGARCANE 0.04, P lt
0.05 CORN 0.11, P lt 0.05 OAT 0.35, P lt
0.01 COFFEE 0.04, P lt 0.05 GRASSES 0.12, P lt
0.001 SORGHUM 0.07, P lt 0.1 WHEAT 0.22, NS
CONCLUSION Areas of crop damage are not
distributed at random with respect to
distributional areas of pest rodents risk of
crop damage can be predicted based on the
ecological potential of the species causing the
damage .
38
Summary

• Agroecosystem behavior can be predicted based on
  the ecological requirements of component species
• Risk of crop loss, crop inviability, and other
  phenomena can be predicted
• Economic importance of biodiversity informatics
  dead animals in museums can inform human
  economies!

39
MuitoObrigado

• victors_at_ibiologia.unam.mx