Refer to the following paper for complete methodology:
Meyerhof, M. et al. 2016. Microbial community diversity, structure and assembly across oxygen gradients in meromictic marine lakes, Palau. Environmental Microbiology. doi:10.1111/1462-2920.13416

In summary (extracted from above paper):
We studied five meromictic lakes in Palau: Spooky Lake (SLM), Goby Lake (GLK), Ongeim’l Tketau Lake (OTM, known colloquially as Jellyfish Lake), Clear Lake (CLM) and Ngermeuangel Lake (NLK). One holomictic lake (Mekeald Lake; MLN) and an ocean site at the German Channel on the southwestern side of the islands (OS-GC) were also sampled for comparison. We vertically profiled dissolved oxygen (DO), temperature, pH, chlorophyll fluorescence and salinity/conductivity using a HydroLab DS5 (Hach Company, Loveland, CO, USA) and sampled three depth layers within each meromictic lake: the mixolimnion (0–5 m depth), the monimolimnion (5–20 m depth) and intermediate depths near the chemocline; these intermediate depths ranged from 1 to 15 m, depending on the depth of the chemocline within the individual lakes. We sampled comparable depths at MLN (5–20 m) and OS-GC (5–30 m). For QPCR analysis of functional genes (dsrA, amoA and nirS) and specific functional groups, as well as analysis dissolved nutrient concentrations, we collected additional samples above and below the chemocline to capture abrupt transitions in biogeochemical conditions across this interface.

Samples were collected from small boats using a horizontal, 2.5 L GoFlo bottle (General Oceanics, Miami, FL, USA), transferred to 1 L polycarbonate bottles, and stored in the dark during transit to the Coral Reef Research Foundation laboratory in Koror, Palau. Water samples were filtered using a peristaltic pump and 0.22 um Durapore PVDF hydrophilic filters (Millipore, Billerica, MA, USA). Filters were immediately frozen in 800 uL Sucrose-Tris-EDTA (STE) lysis buffer (750 mM sucrose, 20 mM EDTA, 400 mM NaCl and 50 mM Tris) in 2 mL Lysing Matrix E tubes (MP Biomedicals, Solon, OH, USA), and stored at -20 C until transport to the United States, where they were stored at -80 C until extraction (Dry ice and liquid nitrogen are not readily available in Palau.)

During sample filtration, 50 mL of filtrate was collected in high density polyethylene bottles for subsequent nutrient analysis at the University of California, Santa Barbara (UCSB) Marine Analytical Laboratory. Samples were analyzed for ammonium (UCSB MAL analytical method for ammonium, see below; Diamond and Huberty, 1996), nitrite (Environmental Protection Agency (EPA) Method 353.2; Schroeder, 1997), nitrite+nitrate (EPA Method 353.2; Diamond, 1997) and phosphate (EPA Method 365.1; Huberty and Diamond, 1998), on a Lachat QuikChem 8000 Flow Injection Analyzer (Hach Company, Loveland, CO, USA). A handful of samples containing large concentrations of sulfide were not analyzed for nitrate, as sulfide damages the cadmium reduction column. For ammonium analysis, each sample was injected into a flowing carrier stream through an injection valve, and then merged with an alkaline solution stream; the produced ammonia was diffused through a hydrophobic, gas-permeable membrane into a recipient stream containing a pH indicator. Colour change occurs in the indicator solution due to an increase in pH, and the concentration of ammonia was determined spectrophotometrically based on absorption at 570 nm. For all analyses, a mid-range check standard bracketed every 20 samples to verify the accuracy of the measurements, and samples that were detected outside of the standards’ range were diluted 1:10 and reanalyzed. Detection limits were 0.10 uM for phosphate, 0.10 uM for nitrite, 0.20 uM for nitrite+nitrate and 0.10 uM for ammonium.