Correlation Analyses of Rodents, Precipitation, and NDVI from Six Localities on the Sevilleta Nation

1
Correlation Analyses of Rodents, Precipitation,
and NDVI from Six Localities on the Sevilleta
National Wildlife Refuge, 1989-2000

• Michael T. Friggens, Robert R. Parmenter, Terry
  L. Yates, and James R. Gosz

2
Introduction Bottom-Up Trophic Cascade

• The Trophic Cascade Hypothesis based on the
  assumption that consumers are resource limited.

• Periods of high rainfall are presumed to lead to
  increased plant production and to subsequent
  rodent density increase.
• Ongoing debate as to the strength, applicability
  of the TCH in terrestrial systems and for small
  mammal population dynamics in particular.

3
Introduction Rainfall Plant Production

• WATER the primary limiting resource for small
  mammals in the desert Southwest.
• Moisture availability is an indirect effect
• PPT gtgt Plant Production gtgt Rodent Density
• If TCH at work, predict
• Plant production correlated to Rainfall
• Both correlated to Rodent Density Response (with
  appropriate lag)

4
Introduction Winter/Spring Monsoon

• Annual PPT in the US Southwest characterized as
  bi-modal
• Winter/Spring October through May moisture
  results from westerly, broad scale systems with
  low variability, which may be influenced by the
  El Niño Southern Oscillation
• Monsoon June through September moisture from
  summer convective storms characterized by heavy,
  localized PPT with high spatial variability,
  known as the North American Monsoon.
• Bi-modal nature of PPT inspires
• investigation of environmental
• variation as defined by the North
• American Monsoon period.

5
The Study

• 12 years of semi-annual density records for 20
  rodent species from 6 study sites (5 ecosystem
  types) on Sevilleta NWR.

6
The Study

• To explore trophic relationships, specifically
  addresses correlation between
• A. Semi-annual (spring summer) rodent species
  densities at each of six sites,
• B. Winter/Spring and Monsoon period total PPT
  from proximate meteorological stations,
• C. Winter/Spring and Monsoon period plant
  productivity, using an index of greenness
  derived from Advanced Very High Resolution
  Radiometer Normalized Difference Vegetation Index
  (AVHRR NDVI) remote imagery.

7
Results Precipitation

• Winter/Spring Maxima, All Sites 1992 and 1997
  (1995 at Station 40)
• Monsoon Maxima 1997
• Winter/Spring Minima 1996 (but 1989 at only
  site w/ data)
• Monsoon Minima 1989 or 1995

8
Results NDVI

• Winter/Spring Maxima, All Sites 1998, then 1997
  or 1999
• Monsoon Maxima 1996, 1997 or 1998
• Winter/Spring Minima 1989 and 1994
• Monsoon Minima 1989, 1990, or 1995

9
Results Correlation of PPT and NDVI

• All Sites At least marginally significant
  correlation between same year Monsoon PPT and
  Monsoon NDVI (2nd column).
• 5 of 6 Sites At least marginally significant
  relationship between previous-year Monsoon PPT
  and contemporaneous Winter/Spring greenness (3rd
  column).

10
Results Rodent Spatial and Temporal Variability

• Total effort 165,168 trap nights, 18,871
  captures, and 11,929 individuals over 12 years,
  six sites.
• Results reported based on 11,670 individuals
  representing 20 species.
• Dipodomys merriami, Perognathus flavus, D. ordii,
  Neotoma albigula and Peromyscus truei were the
  species most frequently encountered.

11
The Small Mammals Sciuridae and Heteromyidae
Photos stolen from W B Davis D J Schmidly
Squirrels Chipmunks
Spermophilus spilosoma
Tamias dorsalis
Tamias quadrivittatus
Pocket Mice Kangaroo Rats
Chaetodipus intermedius
Perognathus flavescens
Perognathus flavus
Dipodomys merriami
Dipodomys ordii
Dipodomys spectabilis
12
The Small Mammals Muridae
Photos stolen from W B Davis D J Schmidly
Wood Rats
Grasshopper Mice
Onychomys arenicola
Onychomys leucogaster
Dear Mice
Neotoma albigula
Peromyscus boylii
Peromyscus truei
Neotoma micropus
Peromyscus leucopus
Peromyscus eremicus
13
Rodent Variability Desert Grassland

• Highest overall densities Spring Summer 1997
  during wettest year and after maximum Monsoon
  PPT and NDVI (1996).
• Lowest densities Summer 1996, after extended
  drought in Summer 1995 and Spring 1996.

14
Rodent Variability East Chihuahuan Desert Scrub

• Highest overall densities Spring 1999 during
  high NDVI and following maximum NDVI (1998),
  followed by 1997.
• Lowest densities Spring Summer 1996, after
  extended drought in Summer 1995 and Spring 1996.

15
Rodent Variability West Chihuahuan Desert Scrub

• Highest overall densities Spring Summer 1999
  during high NDVI and following maximum NDVI
  (1998), followed by Summer 1992- Summer 1993.
• Lowest densities Spring Summer 1996, after
  extended drought in Summer 1995 and Spring 1996.

16
Rodent Variability Plains-Mesa Sand Scrub

• Highest overall densities Spring 1993
  directly after maximum Winter/Spring and Monsoon
  PPT and strong NDVI (1992), followed by Summer
  1998, during the greenest year.
• Lowest densities Spring 1996, after extended
  drought in Summer 1995 and Spring 1996.

17
Rodent Variability Pinyon-Juniper Woodland

• Highest overall densities Spring Summer 1998
  after the wettest year and during maximum
  Winter/Spring and Monsoon NDVI , followed by
  Spring 1999.
• Lowest densities Spring 2000, during the driest
  year for the study period.

18
Rodent Variability Juniper Savanna/Arroyo
Riparian

• Highest overall densities Spring 1993 after
  both maximum Winter/Spring and Monsoon PPT and
  NDVI , followed by spring 1998 and 1992.
• Lowest densities Summer 1991, following a very
  dry Winter/Spring and during minimum Monsoon
  period NDVI.

19
Correlation of Rodent Density to PPT and Greenness

• Correlation analysis independently compared
  Spring and Summer density with lagged periodic
  PPT and NDVI
• Of 1080 exploratory correlations, 211 were at
  least marginally significant
• Of the 211, previous-year Winter/Spring and
  Monsoon variables appear to be most important for
  both moisture (51) and greenness (62)
• About a one year lag response for rodents that
  show a relationship.

20
Spermophilus spilosoma at Desert Grassland
endogenous increase

• Persistent population with low coefficient of
  variation
• Significant correlation with 2-years-previous
  Monsoon PPT, then previous-year Monsoon NDVI a
  lagged response
• Spring 1993 and Spring/Summer 1999 peak Density
  coincided with consecutive increases over 3
  seasons
• An endogenous effect which occurred during and
  immediately after the wettest 2 years 1992
  1997

21
Dipodomys merriami at East Chihuahuan Desert
Scrub immediate response

• Dominant species at this site
• Marginally correlated with previous-year Monsoon
  PPT, then previous-year Monsoon NDVI lagged
  response to same-season production
• Yet, peak density occurred in the same years as
  maximum Winter/Spring PPT 1990, 1992, 1995,
  1997, 1999 (during peak NDVI)
• Density response faster than correlation suggests

22
Neotoma albigula at Plains Mesa Sand Scrub
immediate response, endogenously driven peak D

• Dominant species here in terms of biomass
• Significantly correlated with previous-year
  Monsoon PPT and NDVI, then previous year Monsoon
  NDVI lagged response to same season production
• Appeared to respond immediately to PPT, but peaks
  were endogenously driven during favorable
  conditions
• Density response faster than correlation suggests

23
Peromyscus truei at Pinyon-Juniper Woodland
immediate response to delayed seed mast
production?

• Dominant species at this site
• Significant correlation with 2-years-previous
  Monsoon PPT, none with NDVI
• Density response faster than correlation suggests
• Peak densities endogenously driven by increases
  responding to PPT with less lag, but peak in less
  favorable periods
• Dynamics likely associated with complex, delayed
  response of mast Pinus edulus

24
Conclusions

• Individual species show dynamics ranging from
  apparently immediate response to more gradual,
  endogenously driven dynamics over several
  seasons.
• 12 instances of delayed rodent density response
  to PPT, and subsequent delayed response to NDVI
• but results are not conclusive (few overall, many
  marginally significant).
• Others dont appear related to PPT or NDVI,
  suggesting perhaps sub-optimal habitat, the
  importance of resources other than plant
  material, or more complex, non-linear
  relationships to PPT and plants, not to mention
  predation.
• By site, no consistent correlations to PPT or
  NDVI among species present.
• By species, there were no consistent correlations
  to PPT or NDVI among sites.

25
Many thanks to Bob Parmenter, Terry Yates, Jim
Gosz, Kristin Vanderbilt, Greg Shore, and the
SLTER Field Crews