It matters not how long the light source lasts, assuming the shutter is open for the entire emission from a very short duration light source.
What only matters is how much total light is emitted from a light source during the total exposure time.
Capturing sprites isn't that much different from capturing lightning. In most cases the camera's shutter is open for a few seconds, during which time any particular bolt of lightning may last anywhere from a few to several hundreds of milliseconds. The total exposure time only affects the brightness of light sources that emit light constantly from the beginning of the exposure to the end, as well as the things those constant light sources illuminate.
Imagine we have two identical cameras side by side set to the same aperture and ISO, but one is set to expose for four seconds and the other is set to expose for one second. If both shutters are actuated at the same time and a lighting strike occurs during that first second, both cameras will show the lightning and the things illuminated by the lightning as equally bright. But the camera that is allowed to expose for four seconds will show things illuminated by constant light sources four times brighter than the other camera. Thus the ratio of things illuminated by the lightning to things illuminated by ambient light will be different.
Now imagine that a second bolt of lighting struck during the third second. The first camera didn't record it because its shutter was already closed. In the image from the second camera we have no way of knowing which lighting bolt struck during the first second and which struck during the third (without comparing it to the other image that captured only the first second).
Having done a bit of astro work, I can tell you that the sky was much darker that it appears to be in that photo. For that many stars to be visible the background illumination of the sky would have needed to have been jet black and totally non-perceptible to human eyes. Yet the background brightness of the sky makes it look like the photo was taken at twilight. What appears to be a sunset in the center is, in reality, the accumulation of light from who knows how many bolts of lightning over the course of the exposure.
Think of the very short duration of the sprites this way. Harold 'Doc' Edgerton at MIT practically invented electronic strobe photography in order to take images of bullets going through things such as apples or cards.


The images were taken in a darkened room so that the camera's shutter could be left open for far longer than the instant 'Doc' wished to record. The way that he captured such a short slice of time far beyond the capabilities of the camera's shutter mechanism was to precisely control an extremely short pulse of light set to "pop" a specific number of milliseconds after a microphone attached to the flash detected the sound of the gun being fired.
It doesn't matter how long the shutter is open before and after the flash, the camera records the same amount of light from the flash because all of that light energy happened while the shutter was open.
Digital and film cameras are alike in this respect. Still images record all of the light that falls on the film or sensor when the shutter is open. The resulting image is the grand total of all of the light captured at any point during the exposure.
It's not unlike a water bucket catching rain. If we only measure the amount of water before emptying the bucket once per day, then we can't really tell if that water fell into the bucket slowly during a drizzle over the previous twenty-four hours or if all of the water fell into the bucket during a five-minute cloudburst. We also can't tell if that cloudburst was five-minutes ago, three hours ago, or sixteen hours ago. All we know is that there is an inch and one-half of water in the bucket when we measured it for the first time in twenty-four hours.
Likewise, a still image of these sprites can't tell us how long they lasted. It can't even tell us if all of them happened simultaneously or if they happened at different times over the duration of the full exposure time. It can only tell us how much light each of them emitted in the camera's direction. What the exposure time can tell us is how bright light sources were that were at constant illumination for the entire exposure (assuming they don't fully saturate the sensor, in which case it's analogous to our water bucket overflowing at some point before we measure and empty it so we then have no way of knowing how much more rain fell than what it took to fully fill our bucket).