Study site
The Salt Plains National Wildlife Refuge (SPNWR) resides in Alfalfa County in northwestern Oklahoma where salt flats cover approximately 64 km2 ([14]; Fig. 1). Farm and cattle ranching lands border the flats on three sides, while Great Salt Plains Reservoir is located at the southeastern edge of this system. Tributaries of the Arkansas River intermittently flow over the flats to feed the reservoir. Three permanent high-salinity sampling stations were established on the flats in July of 2000, each characterized by ground water salinities and soil moisture content of ca. 125–250 psu and 15%, respectively. Because this project was designed to specifically target extremophilic algae (i.e., oxygenic phototrophs), site selection was deliberately relegated to high salinity areas devoid of standing water. However, permanent sampling stations were then haphazardly placed within these high salinity areas. Sampling stations depicted in Figure 1 are North Crystal Dig (NCD; N 36°44'18" W 98°16'18"), Clay Creek (CC; N 36°43'51" W 98°15'33") and South Crystal Dig (SCD; N 36°42'26" W 98°15'36").
Experimental design
Temporal and spatial variation in algal biomass (i.e., Chlorophyll-a concentration) and physicochemical characteristics across the Great Salt Plains ecosystem were determined using a nested block design. Each of the three sampling stations consisted of a 20 × 10 m plot with three ground water wells (i.e., lysimeters) placed 10 m apart. Wells were constructed of 1.5-m lengths of PVC, permanently sealed at the bottom with perforations along the length of the pipe to allow for collection of ground water; each well was buried to the water table at a sediment depth of approximately 1.0 m and capped at the sediment surface to avoid rainwater inundation. Water samples were obtained from each of the wells for salinity determination and nutrient analyses (see below). Triplicate sediment samples for chlorophyll and moisture content were always taken at 0, 1 and 10 m distances in a west southwest direction from each well. All samples were collected monthly from July 2000 through July 2001.
Sediment and ground water analyses
Traditional limnological and biochemical assays were used to determine Chlorophyll-a (Chl-a) and nutrient concentrations in ground water and sediment samples. Triplicate ground water samples were collected and analyzed for salt and nutrient content from each well at each station (NCD, CC and SCD; total n = 9 per station). To minimize degradation effects, samples for NH4+ analysis were field-filtered (GF/F), placed on ice and immediately processed upon return to the laboratory as described by Parsons et al. [19]. Samples for NO2- + NO3- and orthophosphate (PO43-) were placed on ice, filtered and frozen upon return to the laboratory for later analysis. Concentrations of NO2- + NO3- and PO43- were determined using standard colorimetric techniques, designed for the chemical analysis of seawater [19] and the ascorbic acid method [9], respectively. Estimates of algal abundance were made using Chl-a concentration as a proxy for relative biomass. Triplicate sediment cores (2-cm deep × 2.5-cm diameter) were obtained at distances of 0, 1 and 10 m from each well (total n = 27 per station) using a 60-mL disposable syringe with its end removed, transferred to 50-mL disposable centrifuge tubes and placed on ice. Samples were frozen upon return to the laboratory for later analysis. Sediment Chl extractions were performed in dim light by adding 8.0 mL N,N-dimethyl formamide (DMF) to each sediment tube. Samples were vortexed daily and allowed to extract for 7–10 d. At the end of the extraction period, samples were centrifuged at 10,000 rpm for 20 min and spectrophotometrically analyzed using the equations of Porra et al. [21]. Sediment samples were also assayed to determine and correct for phaeopigment content using traditional fluorometric methods after Lorenzen [16]. Sediment cores (8-cm deep × 1.3-cm diameter) were obtained using a 10-mL disposable syringe with its end removed, transferred to 15-mL pre-weighed disposable centrifuge tubes, placed on ice and frozen for later analysis. Moisture content of sediment-filled tubes was determined by slowly thawing, weighing (= wet weight) and drying sediment cores to a constant weight (= dry weight) for 7–10 d at 114°C. Percent moisture was then estimated by subtracting sediment dry weight from wet weight to determine the original water weight of each sample and expressing it as a percentage of wet weight.
Algal species isolation and identification
Soil and ephemeral brine pool samples were collected from each of the three stations (i.e., NCD, CC and SCD; Fig. 1). Using the bottom of a sterile polystyrene petri dish (8.5-cm diam), 10 random cores from the top centimeter of salt plains soil were taken at each site and placed in a sterile plastic bag. Brine pool samples were collected in sterile Whirl-Pak® bags. All soil and brine pool samples were placed on ice in the field and transferred to a cold room facility (~ 8°C) until initial isolations were performed the next day. Parallel sub-samples of soil (10 g) were suspended in 75 mL of sterile liquid medium or directly plated (~1 g soil) onto 1% agar plates (Bacto™) made with SP medium [10]. To maximize the diversity of algae isolated, we used three media with salinities of 10, 50 and 100 psu. Liquid and plate media with soil amendments were incubated under cool white light (60 μmol photons m-2 s-1, 14:10 L/D) at a temperature range of 25–28°C. Once algal growth became visible (1–3 days), streak-plating was repeated to obtain unialgal cultures. Filamentous cyanobacteria were isolated using a phototactic purification method [25]. Using a Nikon E400 phase-contrast microscope, all algae were identified from live material to genus and when possible, species, under oil-immersion at 1000 × magnification. Chlorophyte algae were identified using Tomas [26] and Wehr and Sheath [29], while diatoms were identified using Cox [5], Round et al. [23] and Tomas [26]. Cyanobacteria were identified using the taxonomic keys of Komarek and Anagnostidis [15], Anagnostidis and Komarek [2] and Abed et al. [1].
Soil temperature data
A Cox Tracer model CT1ED8 recording thermometer with an external sensor was used to measure soil surface temperature (Tsoil) at SCD from 19 June through 22 July and 31 July through 27 September 2001. Temperatures were automatically recorded at 15-min intervals. Air temperature (Tair, °C) at 1.5-m height, solar radiation (W m-2) and wind speed (m s-1) at 10-m height were obtained from the Cherokee Oklahoma Mesonet Station located approximately 9.5-km northwest of SCD. The raw data were at 5-min intervals, so 15-min means were calculated for comparison to measured Tsoil. Because the latter were available for only part of the summer, Mesonet data were used to model Tsoil over a 4-month period (June – September). A direct multiple regression of the three Mesonet parameters explained 88.4% of the variance in measured Tsoil, and 90% of the residuals were within ± 5.2°C. The relationship was improved by using 1-h sliding means of solar irradiance with a 3-h lag, which resulted in a subjectively acceptable model: predicted Tsoil = 0.924Tair + 0.010lagsolar - 0.383wind + 5.743 (n = 8699, r2 = 0.937, 90% of residuals within ± 3.5°C). Tair, lagged solar and wind speed explained 85.4%, 7.5% and 0.8% of the variance in Tsoil, respectively.
Statistics
Analyses were performed on biotic and abiotic data collected from the three permanent sampling stations (i.e., NCD, CC and SCD). All statistical tests were considered significant at the level of P < 0.05. Sampling dates with missing data were excluded to prevent bias. A one-way ANOVA was performed to test for mean differences among wells for each site. A Tukey pair-wise comparison for each well was conducted for each site between chlorophyll and wells to determine the relative order of variables.
Since samples from each well were potentially intercorrelated, a Principal Components Analysis (PCA) was conducted to summarize the variables into smaller, uncorrelated subsets. Stepwise regressions were conducted on 1) chlorophyll (response variable) versus each predictor variable for each site and 2) chlorophyll versus predictor variables and principal components. Data were log-transformed and a Discriminant Function Analysis was conducted to identify those sites that were most alike and to identify variables most useful for distinguishing among groups.