What quantities would you need for analysis?

I don't know if the sampling device you need exists. But easy to make.

But, I would use a large-volume Luer Lock syringe with a Luer-to-Tubing Connector on the front. Various medical tubes exist that mount directly to syringes...could skip the fitting. I'd cable tie or tape a piece of bamboo to the syringe barrel and use the stiff bamboo to feed the tubing into the center of the bag. If you need a larger volume the a peristaltic pump would do it.

It strikes me that water turnover would be rapid in a large-mesh bag like that and there may not be a measurable impact of the mussels. But there is really only one way to find out.

Best of luck in your project.

Bill,

Tubing is a good solution, the inlets can be fixed in any location/position, and you can periodically pump with a peristaltic pump. The pumping rate can be similar to that of the oysters themselves, so you can assure the flow conditions are not influenced by active pumping. For example, we have placed tubing around mussel aggregates at fixed distances and orientation ("siphon mimics") to characterize the water quality gradient around the aggregates, and pump at 80 ml/min (e.g. see attached for methods reference, but this can be scaled to single bags). In a pinch, you can pump with syringes at the same rate, if the head is negligible, and just reject the initial tube volume. The tubing can be almost any material, and fixed to a surface buoy (if the bags are submersed). If deployed over a long time period where fouling can be an issue, it is possible to set it up to fill the tubes with air; also I've found that a glass tip (simple tubing or modified Pasteur pipette) coated with a sinalizing agent (e.g. Sigmacote) can be helpful for some conditions.

Depending on your sampling protocol, your setup can be as sophisticated as an autosampling system, or just discrete samples taken by hand. I've had good results with a 'DYI' autosampling system, using a peristaltic pump and Valco selector, run into a single measurement chamber, outfit with sensors. Flush cycles and even calibrations can be programmed. You can run such a setup with a small PC, Rasberry Pi, etc. Otherwise, just discrete pumping and measurement is suitable for most parameters of interest. You might also think about timing the sampling, to control for 'water parcels' passing through the bags; so you may pump all the tubes simultaneously if you have multiple pumps or a multi-channel pump (or syringes on a fixed rack). Anyways, there are some considerations, but in my experience, pumping from different tubes into a single measurement station is preferred for costs and variability between instrumentation.

A full liter sounds like a lot, unless you're interested in short time series under other controlled parameters (e.g. tracer experiment, etc.). Perhaps the instruments are physically large, so they need that amount? For most parameters, smaller volumes (

In sampling water table at various depths, we epoxied a ceramic cup onto plastic pipe. We pressurized system using two hole rubber stopper and tubing to remove water, then used vacuum to withdraw sample, then pressure again to force sample through the longest tube from the ceramic cup. This approach was used by Sopper and Kardos from Penn State in sewage irrigation studies in the 1970s. I am not sure how well our co-authored report in 1975 with Tennyson and Settergren explained this, but most likely we cited the Penn State study.

I think, it will depend of the quantity of water, the number of samples and periodicity you need to sample and of course de resources at hand. I mean, it is not the same a fixed device with tubing and pumping like those used for continuous water sampling (for instance, nutrient autoanalyzer) or a quick hand sampler like a horse syringe or syringe gun for example.