Raster layers: Can anyone help on building a 'climate envelope' please?

[JonButler]

I am building a model with one raster layer defining where a certain plant is growing and several other raster layers with climatic conditions such as rainfall, temperature etc underneath the original. 
What I want to achieve in my model is to be able to define the minimum and maximum value of every climatic condition where my plant currently exists.
Once I have that information defined I then want to predict where my plant can exist in the future by identifying the same conditions in future climates.
Any help would be much appreciated.

[JonButler]

I am building a model with one raster layer defining where a certain plant is growing and several other raster layers with climatic conditions such as rainfall, temperature etc underneath the original. 
What I want to achieve in my model is to be able to define the minimum and maximum value of every climatic condition where my plant currently exists.
Once I have that information defined I then want to predict where my plant can exist in the future by identifying the same conditions in future climates.
Any help would be much appreciated.

I'll try and be more specific with just one question...
Can I from the Raster Calculator interrogate two raster layers by asking: Where this plant is present on one raster, tell me the max and min temperature on the second raster? 
I'd just like to know if the Raster Calculator can handle WHEREs and IFs please?

[JeffreyEvans]

I highly recommend that you research climate envelop and niche modeling before proceeding. There is a huge body of literature. The term "envelope", in the way that you are using it, is not a global min/max but rather a multivariate hypervolume. The original Bioclim envelope approach has really fallen out of favor because of statistical issues regarding distributional assumptions and empirical limitations. 

There are a number of models that achieve what you are after (GARP, MaxEnt, Climate Space Models, Ecological Niche Factor Analysis, Random Forests, the list goes on...) but caution must be used in both transferring and projecting these types of models. For a climate projection to work the model must fix the statistical relationship fit in the contemporary climate space. Models such as MaxEnt must refit the relationship for future climates and, as such, are not projecting contemporary relationships. It is also quite important that, for contemporary and future climates, you use data that is normalized to a specific climate period (i.e., 30 year norms). This accounts for the inherent stochasticity in climate data.   

Here is a starting point that uses Random Forests for climate projections.

Rehfeldt, G.E., N.L. Crookston, M.V. Warwell, J.S. Evans. (2006). Empirical analysis of plant-climate relationships for the western United States. International Journal of Plant Science. 167(6):1123-1150. http://www.treesearch.fs.fed.us/pubs/25706

To answer your actual question, what you want to do is fairly straight froward. Assign the climate values for contemporary climate to your point samples using the "Extract Multi Values to Points" tool. You can then retrieve the min/max values by opening the attribute table for the point sample, right clicking in the appropriate attribute field and selecting statistics in the context menu. You can then use a con statement (using the con tool or raster calculator) to extract a range of values from each raster. You would assign 1 to areas where the conditional ranges for each climate variable are true otherwise 0. You can then just sum the binary rasters and extract areas that are equal to the number of variables.

[JonButler]

Thanks for taking the time to reply.  Much appreciated.