Intro
In general, there are two calibration procedures used with measurement equipment (I come from an acoustics background but this is more generic to metrology than a specific acoustics field).
Temporary calibration
One is to correct possible drifts resulting from temporary environmental changes. For example, when one uses a sensor array to perform Direction-of-Arrival estimation the speed of sound is needed. Alas, the speed of sound is dependent on temperature, which may exhibit spatial and temporal variations (sometimes quite large). For this reason, one has to calibrate their setup before taking a measurement at a specific location and time. This will account for those temporary changes (compared to the laboratory environment where the initial calibration was done).
This "type" of calibration should be performed when the environmental conditions change (for any reason this may happen).
Long-term drifts calibration
The second kind of calibration is done to correct/account for long-term drifts in the parts of the equipment, whether they are mechanical, electrical or of any other type. For example, capacitors may exhibit capacitance drift with time (for various reasons which I am not going to state here). These effects should be corrected or accounted for. Usually, this type of calibration is done from the manufacturer, or some specialised engineer/lab-technician, which in the capacitor example, some values in an embedded algorithm could be changed or the capacitor itself be replaced.
This kind of calibration should be done whenever the manufacturer suggests. It is rather typical for precision measurement equipment to be about every one or two years. For example, I have encountered a Phosphorous Oscilloscope that declared it should be recalibrated once per year by the manufacturer.
Whether this is indeed needed and what the consequences of not calibrating the equipment will be (drifts and the amount of them) strongly depends on the equipment.
Summary
To directly answer your question, calibration should be performed every time prior to taking measurements (one calibration process before a batch of measurements is common practice), unless your equipment does not allow for it (in my field, there are calibrators for sound level metres that should be used before performing measurements on site).
I believe it was also clear that even though your hydropnone/microphone was protected in a nice protective case, it most probably needs recalibration for long-term drifts. As already mentioned, this is very equipment dependent.
Additional remarks
Please note that whether your intended use case is in need of a calibrated piece of equipment is a whole different story. You may be working with relative values and don't care at all about the actual/absolute values of the quantities you are measuring. In this case you may avoid the calibration process altogether.
But, again, this is also equipment dependent. As an extreme example, think of an Analogue-to-Digital Converter that has huge drift at some Phased Locked Loop component and exhibits large jittering values. This can screw the phase (and frequency) content of your measurements which may, or may not, be detrimental for your algorithms, or whatever your use case is.
