I'd go with whatever the lab ...

I'd go with whatever the lab manager you're running the samples with says with respect to bottles and preservatives. Issues with interference will depend in part on your target reporting limits - at very low analytical limits interference is usually a bigger problem. Here's a link to a poster presentation I found discussing interference between Cl and As at low levels: http://www.cala.ca/nw_oct11_pt_challenge_moe_poster.pdf. Sample collection procedures are something to consider carefully. By total arsenic I assume you mean not filtered for dissolved fraction (as opposed to speciating for the valence state). If you are screening with field test kits, I assume you are not going for ultra-low detection limits so conventional sample collection with latex or nitrile gloves etc. should be fine. USEPA has developed procedures for low-level metals sampling, mostly related to aquatic-life standards and especially for mercury. You could look those up if you want to make sure your sampling protocol is appropriate. Finally, don't forget your quality assurance/quality control (QA/QC) samples. You'll want some duplicates and field blanks to convince yourself what you're measuring is reproducible and real. Also consider equipment rinse blanks for any re-usable equipment that has been decontaminated between samples. For laboratory QA/QC, you'll need some fraction of samples collected at double or triple volume to do matrix spike/matrix spike duplicates. Look up the USEPA Contract Laboratory Program procedures, and their SW-846 analytical protocols for more background.