8/9/2019 DivaGIS-1

1/9

DIVA-GIS --- Exercise 1

Distribution and diversity of wild potatoes

It is assumed that you have gone through the tutorial and at least glancedthrough the manual before you start this exercise.

Wild potatoes (Solanaceae sect. Petota) are relatives of the cultivated potato.There are almost 200 different species, all of which occur in the Americas.Because of their economic importance for potato breeding, they have been thesubject of intensive collection and study.

The geographic distribution of wild potatoes was recently described (Hijmans,

R.J., and D.M. Spooner, 2001. Geographic distribution of wild potato species.American Journal of Botany 88:2101-2112).

In this exercise, we use some of the same data that were used by Hijmans andSpooner to obtain some of the results described in their paper.

A. Import data

1) Create a folder called diva-ex1 and make it the default data folder

(Tools/General Options/Folders)-

2) Download the data for this exercise, save them in the diva-ex1 folderand unzip the files. Open the file wildpot.xls with Excel to convert thedata in degrees and decimal minutes to decimal degrees (see Chapter 2of the manual), and save the file in DBF-4 format. Make sure that you donot lose the decimal numbers of the coordinates as you save the file as

1

8/9/2019 DivaGIS-1

2/9

DBF (see Chapter 2 of the manual).

3) Use this DBF file to make a shapefile called wild potatoes, save it inthe default data folder and add it to the map. Also add thept_countries shapefile to the map. Your map should now look like this:

B. Summarize by country

4) We are first going to summarize thedata by country.Make the potato shapefile the activefile and click on Analysis/Point toPolygon . Select species as the fieldof interest. Add the shapefile ofcountries in the define shape ofpolygon box. Select an outputfilename and press Apply .

The result is a new countries layer.Make this layer visible and the othertwo layers invisible. Double click onthe new layer and change its legendattributes.

2

8/9/2019 DivaGIS-1

3/9

8/9/2019 DivaGIS-1

4/9

The information by country is of interest and can be useful. However, it isnot very helpful for the understanding spatial distribution of wild potatoesbecause the countries have such different sizes and shapes, and becausemost are very large. It is in most cases better to use grids with cells ofequal area, and that is what we will do next.

Save the project to, e.g., exercise1A.

C. Project the data

To be able to use a grid with cells of equal area, the data need to beprojected. If you were to use lat/long data , cells of say 1 square degreewould get smaller as you moved away from the equator: think of themeridians (vertical lines) on the globe getting closer to each other as you gotowards the poles.

For small areas, UTM would be a good projection, but in this case we will

use a projection that can be used for a whole hemisphere: the LambertEqual Area Azimuthal projection. Before you project your data, you mustchoose a map origin for your data. This should be somewhere in the centerof your points, to minimize the distance (and hence distortion) from anypoint to the origin. In this case, the center could be (-79, 0).

6) Remove all layers from the map, except the wild potato and the originalcountry shapefile.

4

8/9/2019 DivaGIS-1

5/9

Project these files using Tools/ Project. Choose the Lambert Equal AreaAzimuthal (equatorial) projection. On the custom tab, change thecentral meridian to -79. Save the files with filenames such as wildpotatoes lambert and countries lambert. Press Apply. Be patient

Make all layers visible. If you zoom to the maximum extent, you see theprojected data. Note that the shape of the countries is much moresimilar to their shape on a globe than before projecting. In the bottomleft corner of the screen you can see that the coordinate system haschanged. You will mostly see very large numbers now if you move thecursor: these show the distance from the origin (-78,0 degrees) inmeters. If you make the projected potato file invisible, you will see thatthe unprojected data is still present on the map, near the origin of theprojected data. Keep zooming in to that area until the unprojectedcountries appear.

Clearly, one cannot combine projected and unprojected data (or data intwo different projections) on a single map.

7) Now go to Map/ Properties and change the projection to Other and theunits to meters. This will allow you to display the correct scale on themap (at the bottom of the map).

8) Save the project to Exercise 1B.

5

8/9/2019 DivaGIS-1

6/9

D Species Richness

9) Lets determine the distribution of species richness using a grid. This canbe done using the point-to-polygon option that we used before, but in

most cases it is more appropriate to use Analysis/Point to grid.

Select Richness and Number of different classes, and select theSpecies Field.

Create a new grid. In the Options window, set the X and Y resolution to100,000 (as the projection is in meters this means that the cells will be100 by 100 km). Use the default option (simple) for the Point-to-gridprocedure.

Choose an output filename (e.g., species richness 100, and press

Apply ).

6

8/9/2019 DivaGIS-1

7/9

10) Now make a grid of the Number of observations instead of speciesrichness. Define the new grid with the option Use parameters fromanother grid and select the species richness grid you just made. In thisway you assure that you use exactly the same grid (cell size, number ofrows and columns, and geographic origin).

11) Compare the two grids with Analysis/Regression. Is the number ofspecies in a cell a function of the number of observations?

7

8/9/2019 DivaGIS-1

8/9

As you can see, there is a relation between the number of observationsand species in a grid cell. Whether the number of species is, therefore,an artifact of collecting bias remains to be seen.

The problem is that this association will always exist. When there are

only few species in an area, collectors will not continue to go there toincrease the number of (redundant) observations.

In this case, the coefficient of determination (r 2) is not very high (0.665)for this type of relationship. Also, there is a clear pattern in speciesrichness maps, they are not characterized by sudden random-seemingchanges in richness. The value of richness in one cell is usually closelyrelated to that in neighbouring cells. This is corroborated by the factthat the data have spatial autocorrelation according to the Geary index,however, not according to the Moran index. See for yourself under

Analysis/Autocorrelation .

Additional ways to look at this collector-bias problem include the useof rarefaction, richness estimators (see manual) and predictivemodeling (see Exercise 2).

This type of analysis, and the effect of bias, is strongly influenced by the(arbitrary) scale of the grid. The larger the grid cells, the smaller thecollector-bias is likely to be. Investigate this by making grids for richnessand the number of observations at 50 km and 250 km resolutions.

12) There are often gradients of species richness across latitudes and

altitudes.

An easy way to investigate a latitudinal gradient in species richness is bymaking grid of a single column.

Use the unprojected data (because we want to summarize the data bydegree latitude). In the grid options window, use adjust with resolutionand set the number of columns to 1. Then use Grid/Plot to create afigure like the one below. The distribution of species richness has twopeaks. What would explain the low species richness between -5 and 15degrees?

8

8/9/2019 DivaGIS-1

9/9

An alternative way to make a graph like this would be to export the datain the grid to a text file ( Layer/Export gridfile ) and import that file intoExcel.

9