Ecological Archives A020-008-A2

Chad C. Jones, Steven A. Acker, and Charles B. Halpern. 2010. Combining local- and large-scale models to predict the distributions of invasive plant species. Ecological Applications 20:311–326.

Appendix B. Descriptions of habitat layers used in modeling.

TABLE B1. Habitat variables considered in models.

Used in models

Considered but not used

Climate Variables

Climate Variables

   Number of frost days

   Mean temperature of the coldest month

   Annual precipitation

   Mean temperature of the warmest month

   Frequency of precipitation

   Growing degree days

   Humidity

   Temperature seasonality

   Incident solar radiation

   Summer precipitation

 

   Precipitation of the driest month

Topographic Variables

Topographic Variable

   Slope

   Elevation

   Potential solar radiation

 

   Heat load

 

   Topographic moisture index

 

Distance to nearest water

 

Vegetation Variables

 

   Conifer cover

 

   Vegetation cover

 

1We created or adapted GIS layers for the 19 variables listed. Of these, 12 were used in modeling. Seven were not included because of strong correlations (Spearman |r| > 0.8) with variables included in the models. Sources and methods for developing the layers are described below.

 

TABLE B2. Spearman rank correlations among temperature variables at all species datapoints (n = 3774). Because of the high correlations, only the number of frost days was included in final models.

 

Frost days

Warmest month

Coldest month

Elevation

Growing degree days

Warmest month

-0.86

 

 

 

 

Coldest month

-0.94

 0.77

 

 

 

Elevation

 0.88

-0.75

-0.92

 

 

Growing degree days

-0.97

 0.88

 0.97

-0.91

 

Temperature seasonality

 0.79

-0.49

-0.91

 0.82

-0.81

 

TABLE B3. Spearman rank correlations among precipitation variables at all species datapoints (n = 3774). Because of the high correlations, only annual precipitation was included in the models.

 

Precipitation of the driest month

Annual precipitation

Annual precipitation

0.97

 

Summer precipitation

0.99

0.99

 

TABLE B4. Spearman rank correlations among variables used in the model at all species datapoints (n = 3774).

 

Distance to water

Heat load

Ann. precip.

Precip. freq.

Slope

Pot. rad.

Frost days

Incid. rad.

TMI

Hum.

Conifer cover

Heat load

-0.04

 

 

 

 

 

 

 

 

 

 

Annual precipitation

0.04

0.17

 

 

 

 

 

 

 

 

 

Precipitation frequency

0.07

0.11

0.62

 

 

 

 

 

 

 

 

Slope

0.00

0.20

0.46

0.00

 

 

 

 

 

 

 

Potential radiation

-0.04

0.61

0.18

0.09

0.23

 

 

 

 

 

 

Frost days

0.07

0.16

0.78

0.10

0.62

0.17

 

 

 

 

 

Incident solar radiation

-0.03

0.06

-0.14

0.02

-0.14

0.18

-0.25

 

 

 

 

Topographic moisture index (TMI)

-0.09

-0.12

-0.40

-0.11

-0.63

-0.16

-0.45

0.06

 

 

 

Humidity

-0.13

-0.09

-0.49

0.01

-0.44

-0.10

-0.72

0.19

0.33

 

 

Conifer cover

0.19

0.12

0.14

0.14

0.05

-0.13

0.12

-0.02

-0.08

-0.11

 

Vegetation cover

0.05

0.04

0.00

0.12

-0.09

-0.10

-0.09

0.00

0.04

0.06

0.46

 

CLIMATE VARIABLES

All of the following climate layers were developed by the Numerical Terradynamic Simulation Group at the University of Montana using the DAYMET model (www.daymet.org, Thornton et al. 1997). The model is based on a digital elevation model and observations from meteorological stations over an 18-year period (1980-1997). The grids have a 1 km resolution for the contiguous USA. Original projection: Lambert Equal-Area Azimuthal (central meridian -100; central parallel 45).

We clipped all climate layers to a rectangle encompassing the Olympic Peninsula, reset the cell size to 25 m and re-projected them to UTM Zone 10 NAD 1983. Changing the cell size to 25 m prior to projection reduced the distortion caused by projecting a grid.

Variable descriptions

Number of frost days – number of days in the year with average minimum temperatures of 0° C or less.

Annual precipitation – sum of the monthly precipitation over the entire year.

Frequency of precipitation – proportion of days with >0 precipitation.

Humidity – average over the year of the daily partial pressure of water vapor near the surface. Humidity was modeled based on precipitation and temperature values (Thornton et al. 1999, 2000).

Incident solar radiation – annual average of the daily total shortwave radiation flux as estimated based on temperature, precipitation and humidity values (Thornton et al. 1999, Thornton et al. 2000).

Mean temperature of the coldest month – mean daily temperature for the coldest month (December for all grid cells).

Mean temperature of the warmest month – mean daily temperature for the warmest month (August for all grid cells).

Growing degree days – sum of daily mean temperatures for all days in the year with mean temperatures above 0° C.

Temperature seasonality – difference between the mean temperature of the warmest month and the mean temperature of the coldest month, calculated for each cell.

Summer precipitation – sum of the monthly precipitation for July to September.

Precipitation of the driest month – mean precipitation for the month with lowest precipitation (July or August depending on location).

TOPOGRAPHIC VARIABLES

Topographic variables are based on the USGS Digital Elevation Model (DEM), accessed via the University of Washington: http://duff.geology.washington.edu/data/raster/tenmeter/onebytwo10/index.html. Original projection: Zone 10 NAD 1927 (converted to NAD 1983).

Variable descriptions

Elevation – we resampled the original 10 m DEM to 25 m using bilinear interpolation.

Slope – we used ArcView 3.2 to create a slope grid based on the 10 m DEM and resampled to 25 m using linear interpolation.

Topographic moisture index – calculated for each grid cell as ln(A/tan B) where A is the catchment area or the area from which water will flow to the selected grid cell, and B is the slope of the grid cell (Moore et al. 1993). An ArcView extension for calculating this index was obtained from Tim Loesch, Minnesota DNR www.dnr.state.mn.us/mis/gis/tools/arcview/training.html.

Potential solar radiation – calculated from slope, aspect, and latitude (to the nearest 0.001 degree) of each 10 m cell using equations from McCune and Keon (2002). We then resampled to 25 m using bilinear interpolation.

Heat load – similar to potential solar radiation (and developed using the same procedure), but it reaches its maximum on southwest rather than south-facing slopes.

DISTANCE TO NEAREST WATER

This variable was based on the combination of two layers: (1) a wetlands layer from the National Wetland Inventory (www.nwi.fws.gov) via the ONRC Clearinghouse http://www.onrc.washington.edu/clearinghouse/themes/hydro/hydro_theme.html. Original projection: UTM Zone 10 NAD 1927. Wetland polygons were classified by hierarchical wetland type. We removed all upland polygons and all marine areas that are continuously flooded (M1OWL and E10WL); and (2) a layer containing all stream segments on the Olympic Peninsula from Olympic National Forest (http://www.fs.fed.us/r6/data-library/gis/olympic/index.html). Original projection: UTM Zone 10 NAD 1927. Both layers were converted to NAD 1983.

Variable descriptions

Distance to water – the base-10 logarithm of the distance to the nearest wetland or stream. Distance was log transformed because differences over the first 50 m are likely to be the most important.

VEGETATION VARIABLES

We used vegetation data from the Interagency Vegetation Mapping Project (IVMP; http://www.blm.gov/or/gis/index.php). Original Projection: UTM Zone 10 NAD 1927 (converted to NAD 1983). We used two different layers: vegetation cover and conifer cover. About 2% of cells are classified as "unknown" because of topographic shadow, smoke, clouds, or other problems in the images; these were removed from the study. Remote sensing data are Landsat 5 TM images from 1996.

Variable descriptions

Conifer cover – We used the conifer cover layer from IVMP. We converted this layer into 6 classes: 0 = 0% conifer cover (or classified as non-forested), 1 = 1–20% cover, 2 = 21–40% cover, 3 = 41–60% cover, 4 = 61–80% cover and 5 = 81–100% cover. Accuracy for conifer cover using these cover classes was 64.6%.

Vegetation cover – We used the vegetation cover layer from IVMP, which measures all vegetation (herbs, shrubs, deciduous trees, and coniferous trees). We divided cover values into the same categories as used for the conifer cover layer. Accuracy was 87.9%.

LITERATURE CITED

McCune, B., and D. Keon. 2002. Equations for potential annual direct incident radiation and heat load. Journal of Vegetation Science 13:603–606.

Moore, I. D., P. E. Gessler, G. A. Nielsen, and G. A. Peterson. 1993. Soil attribute prediction using terrain analysis. Soil Science Society of America Journal 57:443–452.

Thornton, P. E., H. Hasenauer, and M. A. White. 2000. Simultaneous estimation of daily solar radiation and humidity from observed temperature and precipitation: an application over complex terrain in Austria. Agricultural and Forest Meteorology 104:255–271.

Thornton, P. E., S. W. Running, and M. A. White. 1997. Generating surfaces of daily meteorological variables over large regions of complex terrain. Journal of Hydrology 190:214–251.

Thornton, P. E., and S. W. Running. 1999. An improved algorithm for estimating incident daily solar radiation from measurements of temperature, humidity and precipitation. Agricultural and Forest Meteorology 93:211–228.


[Back to A020-008]