You can set a different I2C address for each of these devices by sending an i2c,<n>
command where <n>
is the address. After that you can address each device on the same bus.
You can probably send this command using I2C, but you need to use the current I2C address of the device.
I guess that all the probes are more or less at the same potential and that there is no need for more isolation than you already have for one probe. If you do need more isolation, you can use an I2C isolator for each device, but you would need as many galvanically separated power sources (isolated DC-DC converter or other method).
Also see https://www.instructables.com/id/UART-AND-I2C-MODE-SWITCHING-FOR-ATLAS-SCIENTIFIC-E/ with more details about different methods.
After reading your comment I understand that you are not using the pH Probe that you mentionned as an example.
You have to know that pH Probes are themselves very high impedance. Therefore if you want to switch from one probe to another, the probe signals already have to be pre-conditionned (amplified/buffered, etc) - and each probe would have its own electronics for that.
One you have an analog signal coming from each probe, you can sum them in a typical OPAMP summing circuit. As you would switch only one probe on at a time, you would get only one of the values. However the circuit to switch probes on and of might be more costly than using an analog switch.
An analog switch allows you to select the signal you are interested in to sample with your I2C circuit.