@Sandra,

Try this !

Arcmap - Add point layer - Arctoolbox - Analysis tools - Proximity - Point Distance - Add your point layer into "input and near" boxes - define your output location for saving the the .dbf file - click OK.

You can open your .dbf file using excel.

Hope it helps

Raoof

http://webhelp.esri.com/arcgisdesktop/9.3/index.cfm?id=1353&pid=1347&topicname=Point_Distance_%28Analysis%29

Almost all simil;arity/dissimilarity measures in ecology are pure ad-hockeries, modelless. There is one exception, though very primitive, that of

Raup, David M., and Rex E. Crick. "Measurement of faunal similarity in paleontology." Journal of Paleontology (1979): 1213-1227.

Here is the link to the paper:

http://www.jstor.org/stable/1304099

I would recommend to avoid distances at all, build and test some relevant hypotheses instead. This is a good book about how to do it:

http://www.amazon.com/Doing-Bayesian-Data-Analysis-Tutorial/dp/0123814855/ref=sr_1_6?s=books&ie=UTF8&qid=1398668538&sr=1-6&keywords=Bayes

HTH,

Peter.

@Sandra,

Try this !

Arcmap - Add point layer - Arctoolbox - Analysis tools - Proximity - Point Distance - Add your point layer into "input and near" boxes - define your output location for saving the the .dbf file - click OK.

You can open your .dbf file using excel.

Hope it helps

Raoof

http://webhelp.esri.com/arcgisdesktop/9.3/index.cfm?id=1353&pid=1347&topicname=Point_Distance_%28Analysis%29

Hi,

Really interesting problem. Although many of the terms you've used are not familiar to me it seems that you intend to add the spatial dimension to your gene factors.

My question would be: do you intend to map the differences or only to correlate the phenomena?

Let me expose this example: If I use as base sample the typical gene map (I'm sorry if this expression is poorly chosen) of a population in the arabia peninsula , which we know historically has scattered for a lot of north africa and middle-east, and compare it with the genes from all around the world. I would probably get a map where the areas in north Africa, middle-east, would be greener (closer to the base), and areas in America or far-East would be red (further from the base). In fact this would be a very substantial indicator for a well know historical event.

Other example: If I create classes based on countries or biomes (as you've suggested, although It would be probably more difficult to do this), each with it's own color, and plot it against another dimension, like the gene difference, in a scatter plot, you would probably get some very specific grouping. To represent this specific kind of plot you should consider viewing this: https://www.youtube.com/watch?v=jbkSRLYSojo

It's an amazing use of statistics to actually perceive history by numbers.

Probably an important point to consider is that the person in the video actually separated the color populations in the plot by regions instead of countries (which is far more stable than countries). It's true that they do represent geographic distance in many cases but the color is actually zone oriented because the phenomena (development) that pulls one country is likely to do the same for the neigbours.

Regarding the technical part I usually use the programming language Python to do my studies. If you're going for it you can look at:

http://matplotlib.org/ (to do plots like in the video and also 2D maps, see also: http://matplotlib.org/basemap/users/geography.html)

Best of luck.

Yes both TNC and WWF have validated biome data. It is attached. Change the biomes to line features and use them a boundaries if you require or just make crossing a biome a high weight or even extract you data by biome. I would approach this by converting the biomes to raster, use distance tools in GIS software to make distance pair rasters (in your france, india, sa example) and then create a biome / distance spatial matrix of point pairs (or more than pairs if you desire).

Hi Sandra,

I have used "Hawth's Analysis Tools" in the past . It is a plug-in that works inside ArcGIS with a large number of SPATIAL ECOLOGY functionality (these are listed and described at http://www.spatialecology.com/htools/tooldesc.php).

That original plug-in evolved and was replaced by GME - Geospatial Modelling Environment (http://www.spatialecology.com/htools/tooldesc.php). I haven't used this latest version, but it is said it contains 'most of the functionality of its predecessor with improvements'. It can be downloaded from http://www.spatialecology.com/gme/gmedownload.htm, but requires that you have access to ArcGIS 10 or 10+ to run it.

I hope this can help.

Regards,

Simone

Without having to use GIS, you can calculate distance for lat-long pairs using an excel formula for the haversine method; e.g.: http://bluemm.blogspot.com/2007/01/excel-formula-to-calculate-distance.html

I partly agree with Peter Cejchan that distance can be a problem. It is usually a weak substitute for more complicated processes of dispersal, with factors of history (time, +) and geography. But your approach can provide insights into what to look for. I also recommend the work of Pierre Legendre, once you have your distances.

Hi, try selecting the urbanised areas in a map and use Point-distance in ArcGIS (ArcMap). You can also try to define buffers considering certain moving conditions of the populations, or the tool "Multi-distance spatial cluster analysis" where you can analyse patterns based on distance bands and then apply "distance band from neghbor count". With this tool you can calculate euclidean and Manhatam distances, I'm not sure if this suits your problem, but you can try to apply Point-Distance tool here.

Regards

Teresa

to utilize DMSP/OLS night time light index, I saw some papers doing that kind of work.

Hi Sandra,

This might help: http://www.werc.usgs.gov/productdetails.aspx?id=4017

kind regards,

Deb

http://www.werc.usgs.gov/productdetails.aspx?id=4017