As far as I know, no one uses a zener diode in a analog circuit that is used to measure a voltage. Well, at least no one who is serious and knows what he/she is doing.
The zener diode will start leaking at a lower voltage and it might introduce noise. I never use a zener diode in a analog input circuit that is used to measure something.
What is the maximum positive and negative voltage that you want to protect the Arduino against ?
Suppose using a 5V ATmega chip, with R1 = 100k, R2 = 10k, C1 = 10nF, with 1.1V as reference (and no extra ESD or clamping diodes and no zener diode). Then you are protected up to 160 V.
That is a constant voltage of 160 V that is allowed, as long as R1 can dissapate the heat.
That means without the zener and better choosen values of the resistors, it becomes more accurate and safer and more sturdy.
Calculation: A0 can be up to 5.5V. Then the internal diodes start to conduct, and it is allowed to push 1mA into the pin.
Current R2 = 5.5 / 10k.
Current into pin = 1mA.
Voltage over R1 = 100k * ((5.5/10k) + 1mA).
External signal = 5.5 + voltage over R1.
Suppose a leakage current for a 2.2V at 1V is 0.02mA. That seems a reasonable number from what I can find. That will influence your measurement by maybe 2 percent (I'm not sure, I didn't do the calculation). That would mean that of the 10-bit resolution you dropped to 6-bit resolution (I'm also not sure about this one).
To protect the analog input, you don't needinstead of a 2.2V zener, but a 5V zener would be more appropriate (that is: when going down the wrong path to use a zener). The 1N4733A has a leakage current of 10 µA at 1V, and that will still influence the analog signal and still cause a less accurate voltage measurement.