The measured salinity for La Sal del Rey samples (Figure ) varied from a low of 4 ppt at site A, the water source of the lake, to a high of 420 ppt at site C just along the shoreline of the lake (Table ). Samples from sites G and H were taken after Hurricane Alex deposited very heavy rainfall to the area, greater than 15 cm in some locations from 30 June - 2 July 2010 [31
]. The salinity at site H was nearly half that measured at site C before the hurricane (Table ).
Salinity, EcoPlate™ utilization, and isolate characteristics for samples taken from La Sal del Rey 2010.
The microbial density was measured by viable plate counts on saline media containing 19.5 g L-1 NaCl and other salts. Density was estimated as the number of colony forming units per ml or per gram (cfu ml-1 or cfu g-1) for water and soil samples, respectively. Microbial density for water samples taken on 14 June 2010 (A - C) ranged from 1.2 × 102 to 5.2 × 103 cfu ml-1. In samples A - C, as salinity increased, the population density of the isolated species decreased (Figure ). Soil samples taken at the same time, sites D - F, had much higher microbial densities than the water samples. After the hurricane, the cfu ml-1 in water samples was substantially higher than before suggesting that salinity is a controlling factor of microbial density in the ecosystem.
Comparison of colony forming units per ml (cfu ml-1) and colony forming units per gram (cfu g-1) for water samples versus their counterpart soil samples taken on 14 June 2010.
Assays using EcoPlate™ microplates showed that many carbon sources could be used as substrates by the microbial communities taken from La Sal del Rey samples. The highest number of utilized substrates was observed in the community from sample G, which was positive for all 31 carbon sources tested (Table ). The lowest number of used substrates, 12, was observed in the community in sample E. Each sample's community displayed a different preferred substrate (i.e., the greatest absorbance value) and no single substrate was preferred by all communities. The community from sample A displayed the highest total activity (the sum of all positive absorbance values) and also used a high number of substrates. The environmental conditions at site A are close to freshwater and near the water source for La Sal del Rey. The total activity was generally lower in soil samples (E - F) compared to water samples (A - C, G, H). Among water samples, the total activity was greatest at site A, decreased at higher salinities and after a hurricane event (samples G, H). Interestingly, although site A had the highest total activity, it did not have the highest microbial density whereas sites D - H had higher densities of microorganisms but not as high substrate utilization or activity.
BIOLOG EcoPlate™ assays of samples from La Sal del Rey.
Random isolates were selected from agar plates inoculated on 14 June 2010. A total of 100 isolates were chosen and examined for cell morphology and the presence of the enzyme catalase. Most isolates were gram-negative (76%) and rod-shaped (71%) and virtually all were positive for catalase (Table ). Color morphology of the cultures was typically off-white, orange-red, red, or yellow. Several cultures from site B underwent a series of color changes as the cultures aged. When these isolates were streaked onto fresh medium, the initial color of the colonies was off-white. Within 12 - 18 hours the colonies appeared green and later turned orange or orange-red (Figure ). Further visual inspection revealed the presence of 2 strains; however, it was difficult to isolate them individually as pure cultures. When the cultures were separated, the growth of one was visibly slower and the other had very little growth. It is possible that the organisms have a symbiotic association that enhances the growth of both.
Representative culturable halophile from La Sal del Rey, Texas.
Cultured isolates (n = 49) from sites A, B, and C were screened for biochemical phenotypes using API 20E® test strips. A suite of 21 tests was used to construct a phenotype profile for each isolate. The number of positive tests displayed by the isolates ranged from 1 to 19. Many of the isolates that displayed only 1 positive test came from site C, which had the highest salinity (data not shown). The phenotype profiles were compared against each other and a similarity dendrogram was constructed. The similarity comparison examined both the number of positive tests and which tests were positive between organisms. Each isolate's phenotype profile was compared against the other 48 isolates' phenotype profiles. The similarity comparison is shown in Figure . Overall, all the isolates displayed a minimum similarity coefficient of approximately 0.4 or displayed phenotype profiles that were approximately 40% similar. The isolates' phenotypes were divided into 2 similarity clusters. A group of 10 bacterial isolates from site A, which contained the lowest salinity, formed 1 group that were > 65% similar (Figure , upper part of figure). A few of the isolates in that cluster showed identical phenotype profiles. Isolates A9 and A10 and isolates A11 and A13 were identical to each other. The other cluster was a mixture of isolates from sites A, B, and C. The isolates in this group generally had fewer positive tests than those isolates making up the other cluster. Many of the isolates from site C, which had the highest salinity, had identical phenotype profiles but this was generally due to those isolates displaying few positive tests. Most of the isolates from site C were only positive for a single test. The site C organisms were all positive for the Voges-Proskauer test, which is a qualitative test for the production of acetyl methylcarbinol (acetoin) from glucose. The high number and diversity of cultures from site A combined with the low number of positive tests among cultures from site C suggests that the higher salinity of the environment selects for less diversity in phenotypes. However, the change in diversity was not inversely proportional to the salinity across the entire salinity gradient.
Figure 4 Similarity of cultured microorganisms from La Sal del Rey based on phenotype profiles. A total of 21 different enzymatic reactions were assayed using API 20E® strips. Based on the number of positive tests and which of the 21 tests were positive, (more ...)
A subset (n = 37) of the cultured organisms were presumptively identified by 16S rDNA sequencing (Table ). All isolates were Bacteria; there were no Archaea identified. Isolate D12, although not included in the API assays, showed 95.8% similarity to Bacillus oleronius. Other isolates (A2, A3, A9, A12, and A16) also had the closest match to B. oleronius; however, the matches could not be made to the species level (i.e., a match > 99%). Furthermore, none of these isolates had identical phenotype API® profiles (Figure ). Isolates A9 and A16 were approximately 41% similar to the other isolates and were found in the same cluster on the API® dendogram. Isolates A2, A3, and A12 were located in the other cluster on the dendogram and were approximately 85% similar with regards to their API® profiles. The sequence and phenotype data suggests that the organisms are different species or perhaps different strains or biovars of the same species.
Putative identification of isolated bacteria from La Sal de Rey, Texas, USA.
All 10 isolates from site C, water sampled from the lake shore (salinity 420 ppt), all showed the highest 16S rDNA match to B. firmus (Table ). With the exception of isolate C1, all the cultures from this highly saline site had the same phenotype profile (Figure ). Isolates A15 and B2, from sites A and B, respectively, also showed the closest match to B. firmus. The confidence levels of the matches to B. firmus could only be made to the genus level. Sequencing of the 16S rDNA also elucidated the presence of several other potential members of the genus Bacillus. These included: B. circulans, B. fastidiosis, B. megaterium, B. thuringiensis, and B. horikoshii (Table ).
In addition to Bacillus, 4 isolates (A8, A14, A17, and A20) had the closest match to the Exiguobacterium, which along with Bacillus are grouped within the Firmicutes phylum of low G+C gram-positive bacteria. Other matches to the isolates' 16S rDNA included several gram-negative organisms such as Vibrio alginolyticus, Pseudomonas fulva, and Planococcus citreus. Organisms A11, A18, and B5 were closely matched to the known halophile species Halomonas aquamarina (Table ).
Water and soil samples A - F were taken on the same day prior to the hurricane event; therefore, we compared water samples to soil samples with regards to the cfu ml-1 or g-1, total activity, and number of EcoPlate™ substrates that were utilized. Samples from sites G and H were not considered in the comparison because only water samples were taken on that day. The number of cfu was significantly (F1,12 = 13.48, P = 0.0032) higher in the soil than in the water samples (Figure ). The overall model did not detect a significant location effect (F2,12 = 2.29, P = 0.1435), but this was due to the two orders of magnitude difference between the soil and water values. A secondary analysis by substrate type (ANOVA and Tukey's Post Hoc test) revealed significantly higher colony forming units in the freshwater sample than in either of the two saline samples (F2,6 = 25.76, P = 0.0011). No other significant effects were detected (all P > 0.1437). The amount of total activity was significantly (F1,12 = 196.06, P < 0.0001) higher in the water than in the soil samples and decreased significantly (F2,12 = 121.63, P < 0.0001) when going from fresh to salt water in both substrate types. In the water samples, total activity decreased from 32.858 in the freshwater to an average of 8.243 once the salinity was above 56 ppt. The same pattern was observed in the soil samples, but with lower activities (7.78) in the freshwater decreasing to an average of 0.583 when salinity was above 37 ppt. The number of utilized substrates was significantly (F1,12 = 21.17, P = 0.0006) higher in the water than in the soil samples. The water samples utilized 28.33 EcoPlate™ carbon sources on average whereas the soil samples used an average of 22.0. No other significant effects were detected (all P > 0.0681).