Transpiration: Leaf-specific transpiration rates were measured using a LICOR LI-6400XT infrared gas analyzer (IRGA) bi-monthly along marsh transects that contained all species groups. Measurements were made on individual plant leaves at 50-cm intervals from the water surface to the top of the plant canopy. Where plants did not have leaves but only thick stems (Schoenoplectus spp.), we used custom-made extensions of the stock LI-6400XT IRGA sampling chamber to ensure an airtight seal without damaging plant tissue. Gas flux data were collected on one transect at a time from sunrise until as late as possible. These data included measurements of leaf-specific transpiration rate (Tr mmol H₂O m^-2 second^-1), photosynthetically active radiation (PAR, μmol photons), relative humidity (%), and ambient air temperature (degrees celsius), all of which were captured using the IRGA’s default sensors. While the IRGA reports transpiration data in units of leaf surface area, we expressed them in units of dry weight biomass by weighing 8-10 samples of different tissue samples from the IRGA’s chamber and generating relationships between dry weight biomass and surface area for all plant species.

To scale the daily plant-specific transpiration measurements across space, IRGA measurements corrected for dry-weight biomass were combined with whole system live macrophyte biomass estimates as calculated by the bimonthly data collected along the 10 sampling transects (see biomass section for details). To scale the plant-specific measurements in time, micrometeorological data as collected by the IRGA at the time of transpiration sampling were regressed with simultaneous data from an on-site meteorological station operated by the City of Phoenix. In addition, as plant transpiration in Tres Rios is largely driven by temperature, photosynthetically active radiation, and humidity, we regressed plant transpiration rates against these variables to be able to predict transpiration using meteorological data. Combined with the IRGA-meteorological station regressions, these models allowed us to interpolate our plant-specific measurements through time, resulting in total transpiration losses for the whole system (m³ month^-1). See Sanchez et al. 2016 (doi:10.1016/j.ecoleng.2016.01.002) for more details.

Biomass: We used a point intercept transect approach across 10 approximately 50-m transects that were evenly distributed across the wetland cell from inflow to outflow. Every two months, we measured live aboveground biomass in five 0.25m² quadrats that were randomly distributed along each transect. In each quadrat, we measured plant culm and various plant characteristics (stem height, number of leaves, etc.) and converted these measurements to dry weight biomass for each plant using established phenometric biomass models (see Weller et al. 2016 more detail (doi:10.1016/j.ecoleng.2015.05.044)). Plant weights in each quadrat were summed for each quadrat. These measurements were scaled to the whole-system by averaging quadrat-specific estimates across each transect, multiplying by 1/10th of the total vegetated area (21ha), and summing across all transects.

Water Quality: We used a point intercept transect approach across 10 approximately 50m transects that were evenly distributed across the wetland cell from inflow to outflow. These are the same transects used for biomass measurements. Bimonthly triplicate surface water grab samples were collected at the whole-system inflow and outflow using acid-washed 1 L Nalgene bottles, while single grab samples were collected at the beginning and end of each transect. A Lachat QC 8000 Quickchem Flow Injection Analyzer (detection limit 0.85 μg NO₃-N L^-1 and 3.01 μg NH₄-N L^-1) was used to centrifuge and analyze unfiltered samples for inorganic nitrogen (NO₃^-, NO₂^-, and NH₄⁺) and soluble reactive phosphorus (SRP, PO₄^-3). At the sample locations where grab samples were taken, a YSI Pro 2030 meter was used to measure conductance and temperature while a YSI Ecosense ph100 meter was used to measure pH.