I looked at the numbers from that graph in a spreadsheet and then again in a graphing application. I discovered a few things. First of all, if you change the DoD axis to logarithmic, the graph becomes a straight (linear) line. Second of all, that graph is almost exactly 1/x, with a scaling factor to represent the estimated number of 100% discharge plus recharge cycles.
In other words, if you assume that the general description of Li-ion batteries is correct, in that you get a certain number of full discharge and recharge cycles, regardless of whether you achieve a full cycle by 100% discharge and 100% recharge or two separate sessions of 50% discharge and 50% recharge, et cetera, then the expected number of partial cycle events should be a/x, where 'a' represents the number of full or complete (100%) cycles.
You'll note in my graph that the blue line has a scaling factor of 500, so it lines up with the final figure in the collected data. The green line has a scaling factor of 575, making it line up with the collected data at the 50% DoD point. The orange line is scaled by 775, so it lines up better with the left half of the collected data. In all cases, there is a very close correlation between the collected data and a simple 1/x graph.
This is the long way of saying that there is apparently no effect on the life of the battery based on Depth of Discharge. A given Li-ion technology appears to have a certain number of full, 100% charging cycles. If you fully discharge your battery every time you use it, then you get 500 charging events. If you only use 10% of your charge with each driving session, then you'll have 10 times as many charging events. Or, to put it another way, 10 charging events that bring the battery up by 10% is equivalent to a single charging event that brings an empty battery up by 100%.
The more you use your battery, the quicker it reaches its maximum number of 100% charge cycles. I see no evidence that there is any significant difference how deeply you discharge the battery. In particular, I see no evidence that starting at 80% and using 10% per day with a recharge to 80% is any different than starting at 50% and using 10% per day with a recharge to 50%. There should be exactly the same number of charging events if each one involves an increase in SOC of 10%.
I welcome more data, since I'm obviously working with nothing more than one tidbit of information on Li-ion technology and a single graph of data points taken under incompletely specified test conditions.