Loss of native grasslands is fundamentally changing the ecosystem of the arid southwestern United States. These grasslands are threatened by various factors, including the presence of invasive species and environmental perturbations such as declining water tables. In general, plant species are intimately associated with the microbial communities of their environment, with interactions that may range from parasitic to obligate symbiosis. This study investigated the soil fungal communities associated with native grasslands in the southwestern U.S. to determine if community DNA profiling may be indicative of the state of the grassland ecosystem they inhabit. Specifically, the communities associated with Lehmann’s lovegrass (Eragrostis lehmanniana), an invasive grass, and blue grama (Bouteloua gracilis), a sympatric native species, were characterized. In addition, the microbial communities associated with big sacaton (Sporobolus wrightii) growing on sites with varied water tables were also studied.
One of the most endangered native habitats in the American southwest is the semi-desert grassland. Semi-desert grasslands are characterized by average annual rainfall around 17 inches, with the majority of precipitation falling between April and September (Lowe 1964, Judd 1962). These grasslands are generally found at high elevations, between 5,000 and 7,000 feet (Lowe 1964).
Semi-desert grasslands have been vastly reduced in range over the past decades, with only 26% of the native grassland of the entire southwestern United States remaining in pristine condition (Cox et al. 1983). Grasslands have been lost entirely due to agriculture, urban development and suppression of fire (Cox et al. 1983). Additionally, many areas that were previously open grassland are now suffering from shrub encroachment which is gradually changing the community structure. An increase in shrub cover increases competition for available resources and alters erosion patterns (Gori & Schussman 2005).
Las Cienegas National Conservation Area (Las Cienegas NCA) in southern Arizona is a prime example of semi-desert grassland in the United States. The conservation area was established in 2000 and is located approximately 45 miles southeast of Tucson, AZ. The elevation is approximately 4,500 feet above sea level and average annual rainfall is 15 inches, mainly falling during the summer monsoon season. The area is home to a wide variety of plants and animals, including 33 species identified as threatened or endangered (Gori & Schussman 2005). There are over 50 different species of grasses present at the site, including 45 native and eight non-native (invasive) species (Bodner 2009). It is a working cattle ranch and encompasses more than 42,000 acres (Gori & Schussman 2005). Most importantly for this study, over 90% of the area is occupied by native semi-arid grassland (Gori & Schussman 2005).
Lehmann’s lovegrass (Eragrostis lehmanniana) was originally imported into Arizona from South Africa in 1932 as a forage grass for cattle and to stabilize disturbed soils, and has since been shown to generally outcompete and displace native grasses due to its earlier germination and seeding times, longer growing season and faster growth rate (Cox et al. 1988, Trask 2006). It has bunch type growth and is a warm season C4 grass. Its period of active growth includes spring, summer and fall, with seed production during spring and summer (USDA NCRS PLANT Profile). This grass goes not propagate vegetatively and only propagates from seed. The plant currently can be found throughout much of the southwestern United States, including Arizona, California, New Mexico, Oklahoma, Texas and Utah.
At Las Cienegas NCA the invasiveness of Lehmann’s lovegrass is particularly destructive because it tends to displace important native grass species such as blue grama (Bouteloua gracilis). Native to the continental United States and Canada, this grass shares its ecological niche with Lehmann’s lovegrass as it is also a warm season C4 grass with bunch growth. In contrast to Lehmann’s lovegrass, blue grama reproduces both by emergence from the seedbank and, more commonly, by vegetative growth from tillers (USDA NCRS PLANT Profile).
For a species such as blue grama, grazing by cattle also impacts competition because native plants are grazed during the summer growing season whereas Lehmann’s lovegrass, which is not as palatable to livestock, is eaten in fall, winter and spring (Ruyle et al. 1988). In addition, Lehmann’s lovegrass leaves originate at the crown, instead of at the tillers like native grasses, and new leaves elongate horizontally and that protects them from repeated grazing (Cox, Ruyle & Roundy 1990).
Since its introduction, Lehmann’s lovegrass has invaded an estimated 350,000 acres in the southwest, mostly in southern Arizona (Trask 2006). This grass has been used as forage for cattle, but its long-term presence can destroy native diversity. This invasive species also leads to an increased fire risk, due to a larger amount of standing dried vegetative material during drought conditions and has increased recovery following fire (Brooks & Pyke 2000). Many native plants are not as fire tolerant and quickly disappear once invasives establish. Burning can also have serious negative impacts on soil microbial communities, in particular the fungal communities, which are integral for plant health and soil fertility (O’Dea 2007). Importantly, invasive Lehmann’s lovegrass is capable of robust growth in depleted soils containing little or no beneficial rhizosphere fungi (Eom et al. 1999, Neary et al. 1999).
Programs for eradication of Lehmann lovegrass are inefficient, requiring substantial funding and years of work. The two main control methods currently in use are chemical control via application of herbicides and hand pulling of the grasses. Herbicides and their application are expensive and kill non-target plants, and hand-pulling is extremely time and labor intensive. Reintroduction of native species has been long considered unfeasible as seeding is unreliable, with very few seeds successfully sprouting (Ethridge et al. 1997). Environmental conditions that precipitate blue grama recruitment events occur from every 30-50 years on silty soils to once every 5,000 years on sandy soils (Lauenroth et al. 1994). The long periods between recruitment events mean that disturbances to the community affecting blue grama grass can have impacts that last for decades. Thus, it is much more economical and reliable to protect the native grasses that are already present from competition from invasive species.
Protection of the native grasses will improve wildlife habitat, protect local diversity, prevent soil nutrient depletion and lower the risk of wildfires. Control of Lehmann lovegrass will protect other desert species, such as the iconic saguaro cactus, from competition and the deadly consequences of fire. The desert flora and fauna are an integral part of the ecotourism industry of the Southwest, and all are being threatened by these two extremely invasive species. For example, in Tucson, AZ approximately $2 billion in ecotourism dollars per year and almost 40,000 jobs are reliant on preserving the desert landscape (USGS).
Another important plant community at the Las Cienegas Conservation Area are sacaton flats, a rare plant community type found only in the American southwest. They are defined as containing essentially pure stands of big sacaton (Sporobolus wrightii), interspersed with only a few woody plants. These areas create a unique ecological community that provide habitat and forage for a wide variety of grassland species. Certain species such as the Botteri’s sparrow are specially adapted to live in sacaton and fail to thrive in areas where this plant has been displaced (Jones & Bock 2005).
Various factors have led to the decline and disappearance of sacaton throughout its native range, including grazing by cattle, development, agriculture and alterations to natural water systems (Gori & Schussman 2005). Big sacaton grows over a wide range of water depths but is most common at water tables between four and six meters deep and is associated with fine textured soils (Stromberg et al. 1996). Water table has been shown to have a strong influence on vegetation composition (Groeneveld and Griepenttrog 1985, Richter 1993, Busch and Smith 1995). Water tables within 3.5 meters of the surface are considered accessible to sacaton plants directly. Plants growing in areas with deeper water tables generally rely on rainfall or floods for moisture (Tiller 2004, Scott et al. 2006). While sacaton plants can survive in such conditions in the long term, they tend not to thrive and new growth is limited (Bodner 2008).
A previous study investigated the influence of mycorrhizal associations on sacaton seedlings and observed that fungal interactions led to increased growth and survival in both greenhouse and field conditions (Richter & Stutz 2002). Native big sacaton shows significantly improved growth when mycorrhizal fungi are present (Eom et al. 1999, Neary et al. 1999). Conversely, sacaton stand health suffers dramatically in the absence of soil fungi in degraded soils (Eom et al. 1999, Neary et al. 1999, Richter et al. 2002). These data suggest that the health and composition of the members of the soil fungal community may be impacted by the state of the sacaton plant they are associated with. For example, a sacaton plant that is subjected to environmental stresses may not be able to support the same community as an unstressed plant. Conversely, a microbial community that is subjected to environmental stresses may not be able to support a thriving sacaton flat.
Many grassland ecosystems in Southern Arizona are facing environmental stress in the form of a dropping water tables resulting from a prolonged drought and increased pumping of groundwater for agricultural and urban uses. Along with plant species, most components of the soil biome are negatively impacted by decreasing water availability, and water-stress changes in microbial community structure often occur in advance of subsequent declines in plant health (Miller & Bever 1999).
It has been observed that changes in water table depth also negatively affect the fungal species present in association with certain grasses such as maidencane (Panicum hemitomon), suggesting that soil fungi may be less tolerant of changes in soil water profiles as their host plants and may serve as an early indicator of degrading conditions (Miller & Bever 1999). It was found that species composition and amount of colonization varied by water depth, even with a constant host plant. The observed differences may have been due to species preference for certain moisture levels for sporulation. For example, they may be able to form associations but not sporulate at high moisture, and some may only sporulate in high moisture conditions (Miller & Bever 1999).
Stevens and Peterson (1996) also found that the water levels affected the amount and type of mycorrhizal colonization of purple loosestrife. Anderson et al. (1984) also found that fungal species in a mixed plant community were associated with specific environmental conditions, including water level and various soil nutrients. In addition, the same fungal species in different soil conditions formed different associations and varied as to whether these interactions were functional or not.
Plant-fungal interactions are essential for ecological vitality and stability in virtually all ecosystems, and disruption of this interaction has detrimental effect on both plant health and soil fertility (Gange & Brown 1997). For example, rhizosphere fungi are essential for degradation of plant material, especially cellulose and pectin, which most bacteria are incapable of performing (Newman 1985). The products of this degradation are then recycled for subsequent plant use. Fungi are also responsible for modification of soil structure that facilitates both water retention and plant root penetration (Miller & Jastrow 2000). In a reciprocal manner, plant root exudates then contribute to the soil microbial community by stimulation of specific microbiota, creating a dynamic interaction in symbiosis with the host plant. Conversely, exudates from plant roots may also selectively suppress the growth of deleterious rhizosphere fungi (Sullia 1973). Moreover, soil microorganisms may also interact with the plant directly by performing essential tasks in nutrient uptake that have immediate effects on plant growth and health (Allen 1992).
With increasing recognition that soil microbial communities are one of the primary components and indicators of ecosystem health and stability, considerable research has recently focused on examining and describing this feature in numerous environments. Traditionally, soil microbial communities have been studied using culturing techniques. Species identification relied heavily on morphology, and non-culturable organisms were not possible to observe and characterize. More recently, the addition of DNA sequencing to the scientific repertoire has allowed for direct DNA sequencing from environmental samples or sequencing of cultured organisms to expand and refine the identification of species from various natural communities. Both culturing and DNA extraction and sequencing are often used in tandem to create a robust profile of microbial communities. However, with these techniques it is difficult to form direct and efficient comparisons between complex samples containing many different species.
The technique known as Denaturing Gradient Gel Electrophoresis (DGGE) is a well-accepted method for evaluating the diversity of microbial communities (Muzyer & Smalla, 1998). Members of a specific community component can be visualized in a genetic “fingerprint” on a gel, even when the target organisms make up only 1% of the total DNA in the sample (Muzyer et al. 1993). The number of fingerprint bands observed on the gels generally indicates the number of predominant community members (Muzyer & Smalla, 1998). Each band can then be excised, purified and sequenced to identify the exact community member or members it represents.
Equally important determinant of soil microbial communities are the soil properties themselves. Anderson, Liberta and Dickman (1984) found that plant growth and the presence of mycorrhizal spores was negatively correlated to soil pH, available Ca, Mg, P and soil moisture. Total spore counts were also positively correlated with total N and organic matter. Arbuscular mycorrhizal (AM) fungi can affect the outcome of competition by enhancing resource acquisition of some species of plants over others, and this effect varies based on availability of soil nutrients (Caldwell et al. 1985, Allen and Allen 1990). For example, when phosphorus is abundant, fungal symbiotes can be suppressed (Koide and Li 1990, Koide and Schreiner 1992, DeLucia et al. 1997). AM fungi also affect succession, particularly in cases with both mycorrhizal and non-mycorrhizal species, and they also may increase diversity by favoring less competitive species or vice versa, decreasing diversity by favoring a more competitive species (Allen and Allen 1984, Allen and Allen 1988, Allen et al. 1988, Francis and Read 1994). Therefore, in any comprehensive analysis or comparison of soil microbial communities it is important to evaluate and analyze the substrate on which the community is based.
The objectives of this study were two-fold.
The first objective was to examine fungal communities associated with a native and an invasive grass species and to develop DGGE fingerprints of the fungal communities associated with each target plant species and then specifically identify key fungi representing each community.
The second objective was to examine fungal communities associated with big sacaton in Arizona growing on sites with varying water table depths, and to determine if these communities vary by water table, thus providing additional measurable indications as to the stress on these grass communities.
This data was collected through soil sampling at specific sites and DNA analysis of the resident fungal communities, analysis of physical properties of soil from each site, information on water table subsidence at select sites and correlations between fungal community diversity, grass diversity and environmental features. The data collected may provide an additional metric for grassland health as it relates to pathogenic, commensal and symbiotic relationships at the microbial level and the impact of invasive species on soil fungal communities. Furthermore, by understanding the current state of the microbial communities in soil, we may be able to assess the extent of degradation and predict if and when remedial actions are required or the potential success of such actions.
The long-term goal of this research is to develop new means to maintain and manage native grass communities and control the spread of invasive grasses. The incorporation of fungal community profiles as an additional means to measure grassland ecosystem health may result in novel measures to make maintenance or restoration programs more efficient and cost effective.
All sampling was conducted at the Las Cienagas National Conservation Area. For the first objective, three sites were selected for sampling based on the presence of both Lehmann’s lovegrass and blue grama at each site (Fig. 1). At each of the sites, a total of 18 soil samples were taken, and within each site only well-grown plants were selected for sampling. These included three Lehmann’s lovegrass samples from the area including mainly Lehmann’s lovegrass, three blue grama samples from the area of blue grama, and three samples for each grass in the intermingled area (a total of six samples). Six bare ground controls were also collected, three from the Lehmann’s lovegrass-only area and three from the blue grama area (Fig. 2). Individual plants were defined as one solid bunch that consisted of only one individual. Grass bunches that could not be visually isolated from others growing in close proximity were avoided (Fig. 3).
For the second objective, selected sampling sites were separated into three categories based on the depth to the water table under the soil. The three groups included deep (10-14 meters to the water table), mid-range (6-7 meters) and shallow (1.5-3 meters). A total of eight sites were selected based on the water table, as measured by the depth to water in nearby wells (Fig. 20). At each of the sacaton sites, nine samples were collected. Six samples were taken from individual sacaton bunches spread throughout the area, and three controls were taken from bare ground (Fig. 21). A sacaton bunch was defined as a plant that could be identified as consisting of only one individual, either as a solid bunch or as a ring with a small (20 cm or less) open area in the center (Figure 22).
All the sites were level and comprised almost entirely of either Lehmann’s/blue gramma or sacaton in solid stands. In most areas a few mesquite trees were widely interspersed through the site as well. Care was taken to avoid sampling close to any vegetation other than the target plant species, with individuals being selected that were growing at least 7-10 meters from any mesquite. Well-grown plants were selected for sampling and marked with metal stakes for future identification. Sampling was performed in September 2009 and was timed based on the period of most active growth for the grasses, during the summer monsoon season. Samples were taken from the top 7-10 cm of soil, at three points around the circumference of the base of each plant, approximately 5cm from the base of the plant. The three soil cores were placed in a sterile plastic bag and homogenized. The samples were stored in a cooler after collection and during transport and then placed in 4oC storage.
DNA was extracted from 0.5gm of soil using the MP Biomedial FastDNA® Spin for Soil kit. The protocol used was according to manufacturer instructions. DNA extraction was verified by running 2 ?l of the product on a 1% agarose gel. In all cases, a large amount of DNA with minimal evidence of shearing was acquired (Fig. 23). PCRs were performed on the extracted DNA using the protocol described by May et al. (2001). Samples were diluted to a ratio or 1:100 or 1:250 from the stock solution for the PCR reactions.
Several test gels were run and stained in order to determine the optimal denaturant gradient for the samples. It was found that a denaturant gradient of 20-45%, with a uniform 7% acrylamide concentration, worked best for separation and resolution of bands (Fig. 24).
Gels were run at 100V for ten minutes and then 35V for 20 to 24 hours. Gels were stained using SybrGreen for one hour and then visualized and photographed. PCR products were evaluated for DNA concentration using a Nanodrop fluorometer. DGGE gels were then loaded with approximately 2500ng of PCR product per well. On each gel, the DGGE ladder was loaded in wells at both ends and at the center of the gel in order to visualize any variation in the migration of bands throughout the gel.
From the initial test gels for sites four and five, brightly fluorescing, well-defined bands representing a range of positions in the gel were identified and extracted. The bands were excised from the gel using a sterile razor blade, with all excess polyacrylamide trimmed away. The excised band was then placed in 100 ?l of ddH2O in a 1.5mL Eppendorf tube and incubated in a 60oC water bath for one hour. The samples were then spun down at maximum speed in a tabletop centrifuge for 30 seconds, and the supernatant was transferred to fresh tubes using a pipette. The supernatant was stored at 4oC. This undiluted solution was used in PCR reactions following the same protocol as previously described. All of the isolated bands were amplified and run out on DGGE gels, to verify that only a single band had been isolated and that it would amplify well. Five bands that showed the best amplification and which covered a broad range of vertical locations in the gel were selected for use in the ladder. Each of the five bands was PCR amplified separately, the amplification verified on a 1% agarose gel, and the PCR products were then combined to create the ladder solution that was loaded into the DGGE gels for each site.
Analysis of the gels was performed using BioRad QuantityOne software using the DGGE ladder as a standard. For the sites to be compared, all gel images were combined into one large image so that direct comparison of the gels could be performed in order to reduce subjectivity in assigning bands to matching types.
To examine whether soils significantly contributed to microbial diversity, samples from all sites were subjected to physical and chemical analysis. Soils were prepared for analysis first by homogenizing soil samples of the same type from each site. For example, all sacaton samples from the site four were combined in equal amounts. All control samples from each site were combined in the same way. The samples were then dried in an oven and put through a 1mm sieve to remove stones and large pieces of organic matter such as roots. Sample weight was recorded before and after drying to determine the percent soil moisture for each sample.
The dried and sieved samples were submitted to the University of Arizona Water Quality Center for analysis. Characters evaluated included soil texture, total carbon, total nitrogen, pH and available P. Soil content of Mg, Ca, Na and K were determined, and DTPA analysis was performed to evaluate levels of Zn, Cu and Fe.
Initially, cluster analysis was performed using the statistical software PAST (PAlaeontological STatistics) to reveal general patterns of association. Following the cluster analysis, all of the data was subjected to detrended correspondence analysis (DCA) to determine the spread of the data and therefore whether redundancy analysis or canonical correspondence analysis would be the appropriate subsequent test for the data.
Canonical correspondence analysis (CCA) of the data collected from the DGGE gels and the soil analysis was performed using the statistical software PAST. The band type vs. sample data was examined using canonical correspondence analysis to observe any statistically significant patterns based on imposed groupings. For the first objective, the groupings tested for each site included Lehmann’s lovegrass (pure stand), blue grama (pure stand), Lehmann’s lovegrass (intermingled area), blue grama (intermingled area), Lehmann’s lovegrass area bare ground control and blue grama area bare ground control. For each of the sacaton sites, the groupings tested included sacaton plants and bare ground controls. Samples were also grouped and tested by water table depth, separated into three groups: deep (10-14m), mid-range (6-7m) and shallow (1.5-3m) water table.
To examine the possibility of an environmental variable influencing the observed groupings, principal components analysis was performed on the soil physical and chemical properties data. For each PCA analysis, 1,000 bootstrap replicates were performed and a scree plot was created to determine how many factors were likely to be significant for interpretation of the data.
To evaluate the relationships among the samples based on similarity, UPGMA cluster analysis was run on all the samples from three Lehmann’s lovegrass/blue grama sites together (Fig. 1). While no sites showed strict clustering, each site did have a large number of its samples group together in the overall dendrogram.
Subsequent analysis of data generated from DGGE gels was performed by detrended correspondence analysis (DCA). Because the data show a spread of 1.8 s.d. in the DCA analysis, the curve was determined to be monotonic (Fig. 2). A monotonic curve is most effectively analyzed using RDA analysis (Jongman, ter Braak, and Van Tongeren 1995).
The various RDA analyses performed included various combinations of sampling groups and exclusion of groups from the analysis (Table 1). The RDA data revealed that at the site level there was a statistically significant difference between the various groupings. For sites one and three, the P values were very significant at 0.02 and 0.021 respectively. In the case of site two, the statistical significance was 0.085. P values up to and including 0.1 were considered significant. The use of a p-value of 0.1 is precedented in various ecological studies (Huebner et al. 2009, Knops, Naeem & Reich 2007). For all three sites, there was no significant separation between the bare ground controls of the Lehmann’s lovegrass or blue grama areas. The pure Lehmann’s lovegrass was always significantly different from the blue grama controls but only different from the Lehmann’s controls at site three. Interestingly, pure blue grama samples showed the exact same pattern, with significant difference from blue grama controls and significant difference from Lehmann’s bare ground controls only at site three. Furthermore, the Lehmann’s lovegrass from the intermingled area was never significantly different from the Lehmann’s controls but was different from the blue grama controls. Blue grama intermingled samples were significantly different from blue grama controls, with the exception of site three which had a p value of 0.215. The blue grama intermingled samples were different from the Lehmann’s lovegrass bare ground controls only at site two.
Overall, the organic and inorganic nutrients as well as the physical properties of the soils were fairly uniform (Table 2). Moisture content of all samples averaged 3.46% and ranged between 1.3% and 4.72%. In addition, analysis of texture revealed that all samples were comprised of loam soil, ranging from loamy sand to sandy loam across all three sites (Fig. 3). Site one was mainly loamy sand, while sites two and three were primarily or entirely sandy loam. Other soil characteristics were also fairly uniform across samples and sites, with the highest variation in calcium (CV = 1.01) and the lowest variation, and therefore the best fit to the model, with both % sand and copper (CV = 0.056 and 0.146, respectively). Soil characteristics that showed a mode that was very different from the mean and median were considered possibly informative (Webster 2001). In this case, the characters that showed this kind of pattern included available phosphorus, calcium, zinc and iron.
PCA analysis of the soil characteristics revealed an obvious clustering by site, indicating a distinct variation in soil composition by location (Fig. 4). Furthermore, the soil characteristics most correlated with the first axis varied by site. Scree plots of the data with 1,000 bootstrap replicates for all PCA analyses indicated that only the first axis was likely to be significant (Fig. 4). In the PCA of the soil data from all three sites, the first axis accounted for 52% of the variance. Variables that show strong correlation to the same axis are related to each other (Webster 2009). Vectors that lie along the same axis but in opposite directions are negatively correlated. The components that had the greatest correlation to the first axis were pH, calcium, iron and zinc (Table 3). Soil pH and calcium show a strong positive correlation. Iron and zinc are also positively correlated with each other and significantly negatively correlated with pH and calcium (p-value of 0.1). The second axis was most correlated to magnesium and copper which show a weak negative correlation (not significant with p-value of 0.1).
The introduction and establishment of invasive plants is one of the most significant contributors to loss of biological diversity worldwide. As such, the removal of invasive species and restoration of natural communities is a priority in many highly-impacted environments. The objective of this study was to use DNA-based techniques to compare fungal communities associated with an invasive grass, Lehmann’s lovegrass, and a native grass species, blue grama, in a semi-arid grassland located in the southwest U.S. If variation among microbial communities differed significantly among soil samples and the variation was significantly associated with the plant species related to each soil sample, then efforts toward grassland restoration to native states may utilize novel molecular tools to assess restoration progress.
Initial UPGMA cluster analysis of sample-specific DNA fingerprints compared microbial communities across all sample sites, each of which contained a gradient from stands of pure Lehmann’s lovegrass to stands of pure blue grama, but no obvious and well-supported associations between grass species and microbe community were found. Thus, RDA analysis was performed on the same data in an effort to discover hidden groupings among sample sets. This analysis revealed significant separation by site, but clustering patterns within each site was not consistent between sites. Collectively these results indicate that the soil microbial community is more dependent and associated with the location of sample site than by the vegetation of the sample site. RDA analysis also revealed no difference between the bare ground controls sample at each site, which indicated that there is a basal microbial community that encompasses the entire site regardless of the overlying grass community. However, there was a significant difference between the samples from the base of plants compared to bare ground controls. Thus, the basal microbial community is significantly affected differentially by the presence of different plant species.
Interestingly, the bare ground control samples from the Lehmann’s lovegrass sections are apparently more similar to samples from soils actively occupied by grass species, though samples from neither of the two types of grasses tested here were distinct. This may indicate that the Lehmann’s lovegrass is having a much farther reaching area of effect on the microbial community surrounding it than the blue grama grass. The blue grama control samples were very different from samples of soil collected around pure stand plants, indicating that their microbial profile may be more representative of a true “background” profile. Future studies may focus on the abundance of the ubiquitous plant species present in similar soils. In the case of the invasive forb Centuarea maculosa, AM fungi colonize the invasive to a significantly greater level than native bunchgrasses (Marler et al. 1999). For this study, fungal species abundance and degree of root infection were not investigated, only the species richness. It is possible that the Lehmann’s lovegrass forms associations with the same microbial species as blue grama but has a differing association either due to an altered interaction or a different degree of colonization in the roots.
Lehmann’s lovegrass intermingled area samples showed no significant difference from Lehmann’s bare ground controls, supporting the hypothesis that the impact of Lehmann’s lovegrass is more uniform over a large area. Once again, there was a difference observed between the intermingled Lehmann’s samples and the blue grama control samples. This may indicate that there are changes taking place in the intermingled area, perhaps due to the interaction between the Lehmann’s lovegrass and the blue grama communities. It may be informative to investigate and compare these intermingled sites to sites which contain pure stands of each grass.
Previous studies on soil fungal communities in grasses have focused on Glomeromycota, the mycorrhizal fungi. The vast amount of diversity observed in the profiles acquired in this study indicates that there are many disparate species present in these soils. While previous studies have investigated mycorrhizae and focused on specific fungal phyla such as Glomeromycota, this study involved the fungal community of whole soil. The PCR primers also amplify a fairly conserved region of the fungal genome in the 18S rDNA. Duong et al. (2006) felt that these same primers might be less informative due to the limited region of DNA amplified and some apparent specificity for Ascomycetes; however the large number and diversity observed in the bands acquired indicates that these primers are informative enough for a community study.
It must also be considered that the profiles acquired here are based on the whole DNA content of the soil, and it is possible that some species that produced bands in the profiles of the samples were no longer actively growing and reproducing in the soil. Environmental conditions, including plant species, may have changed to the point that a previously extant species could no longer survive, but its DNA still remained in the soil due to hyphal fragments, spores or other long-term survival structures.
The profiles acquired in this study should be evaluated more closely in the future, to determine if there are any single bands that are more specifically associated with one species. The analyses performed here involved the overall pattern, and it is possible that single specific bands were not given enough weight within the analysis. This problem was addressed to an extent by removing bands that had very high frequency from the analysis, but a band-by-band investigation is in order to evaluate the impact of rare bands on sample grouping. Bands that are diagnostic and bands that are found in only a few samples should be identified by DNA sequencing. Based on these identifications, more restrictively targeted studies may be performed, limited to a specific family or genus.
Based on analysis of the soil samples, soil moisture content was similar across sites, and thus we might expect the microbial community to be in similar states of metabolic activity. Overall, the organic and inorganic nutrients as well as the physical properties of the soils were fairly uniform as well. Earlier studies have found that soil texture and soil water potential influence the probability of recruitment of blue grama grass to new sites (Lauenroth et al. 1994). Therefore, reestablishment of the species must be attempted while regarding soil texture and yearly environmental conditions as well. Silt soils have a much greater probability for recruitment than clay or sand, but the greatest sensitivity to soil water potential is also found in silty soils. Reclaimed farmland is often comprised of sandy soil, which means that reestablishment on these sites will be more difficult, and reduced water availability will likely further reduce the frequency of recruitment events. Finally, Lehmann’s lovegrass and blue grama may cause different effects on the wetting/drying cycle of the soils, impacting the microbial community (Frasier and Cox 1994).
Briones et al. (1998) found that weather, specifically precipitation, significantly affects competition. The amount of a resource available (e.g. water) causes variation in intraspecific competition and plant growth. Briske and Wilson (1980) found that blue grama seedlings will die if they do not produce adventitious roots shortly after germination and lack of water severely limits the production of adventitious roots. Furthermore, the younger the seedling the more severe the impact of drought. Lehmann lovegrass does have a minimum moisture requirement for growth, however it has significantly greater recovery after periods of drought than native grasses (Frasier and Cox 1994). This may be due to Lehmann lovegrass utilizing soil moisture during periods when the native grasses are dormant. For this study, the time of sampling (after monsoon storms) may have impacted the profiles that were acquired, as the plants were more competitive. Negative interactions may appear after heavy rains as well. This study should definitely be repeated over multiple years so the effect of precipitation can be evaluated. Furthermore, growth of adventitious roots does not occur until leaf regrowth has been completed following clipping (Briske and Wilson 1980). If cattle are allowed to graze on very young seedlings, they may cause a greater delay in the production of adventitious roots and perhaps a higher mortality rate in the blue grama seedlings at Las Cienegas.
Cluster analysis revealed that although samples from each site mostly clustered together, there was no strict clustering based water table depth (Fig. 5).
Initial analysis of data generated from DGGE gels was performed by detrended correspondence analysis (DCA) using the statistical software PAST (Fig. 6). Because the data points of the samples showed a spread of more than two on the DCA graph, canonical correspondence analysis (CCA) was performed.
The samples from the sacaton field sites showed statistically significant (P < 0.001) separation by site in the CCA analysis (Table 4). When analyzed by water table level, the three groupings also showed statistically significant separation. Various random groupings of sites of different water table depths were tested to determine if this significance was an accurate representation or simply an artifact of the differences between sites. For example, sites four, seven and ten would be a “group” in the analysis, despite being from different water table depth groups. All the random groupings, including pairwise groupings, also showed P values below 0.05, indicating that site-to-site variation was high irrespective of water table. In order to discover if the more diagnostic bands that could be present for a water table depth might have reduced weight in the analysis due to the large number of variables being tested, bands with very high frequency were removed from the analysis. Initially, bands that occurred in 80% or more of the samples were excluded from the analysis. This was repeated additionally without bands found in 60% or more and 40% or more of the samples. In all cases, the P value was well below 0.05 (data not shown).
Moisture content of all samples averaged 5.64% and ranged between 1.92% and 10.61% (Table 5). In addition, analysis of texture revealed that all samples were comprised of loam. Sites were primarily sandy loam, with some having silt loam and loam (Fig. 3). Other soil characteristics were also fairly uniform across samples and sites, with the highest variation in sodium (CV = 0.989) and the lowest with pH (CV = 0.061). Additionally, components such as calcium, potassium, manganese and available phosphorus showed very wide ranges across sites. Soil characteristics that showed a mode that was very different from the mean and median were considered possibly informative. In this case, the characters that showed this kind of pattern included % sand content, % silt content, calcium, sodium and iron.
With PCA analysis of the sacaton data, the first component accounted for 54.88% of the variance, the second axis accounted for 14.73% and finally the third 10.32% (Fig. 7). The scree plot with 1,000 bootstrap replicates showed that only the first axis was likely to have significant meaning. Variables that show strong correlation to the same axis are related to each other (Webster 2009). Vectors that lie along the same axis but in opposite directions are negatively correlated. The first axis was most strongly correlated to the variables of available phosphorus and % sand, with % silt, total nitrogen, total carbon, zinc and copper also having a high correlation (Table 6). Phosphorus and sand also showed the highest contribution to the first axis. In regards to their interaction, phosphorus and % sand content of the soil were strongly negatively correlated.
In arid and semi-arid areas of the world, over-utilization of limited water resources often results in substantially lower ground water tables, which negatively impacts native plant communities and the species diversity they support. What is less clear is the impact of lower water tables on soil microbial communities and the reciprocal effects this has on the supporting plant community. The objective of this study was to examine soil fungal communities associated with a grass species native to the southwest U.S., big sacaton, across a number of sites with varying water table depths. If variation among microbial communities differed significantly among soil samples and the variation was significantly associated with differing water tables, then efforts toward grassland conservation and maintenance of native grass species may utilize novel molecular tools to assess the broader impacts of water management and the success of conservation efforts.
The soil samples collected from the various sacaton sites showed significant clustering of fungal community profiles by site, but did not reveal any substantial clusters based on depth to the water table. RDA analysis was performed on the same data in an effort to discover hidden groupings among sample sets. This analysis revealed significant separation by site, but clustering patterns within each site was not consistent between sites. Collectively, this indicates that the microbial community is impacted more by location than by water supply. Since the sacaton plants do not appear to be losing or gaining a specific set of fungal associations with changes in the water table, then the associated soil microbial communities are likely to be more stable and tolerant of environmental changes.
Changes observed with other plant species indicated that water depth was an important factor in determining mycorrhizal associations (Miller & Bever 1999, Stevens & Peterson 1996). This study did not observe mycorrhizae directly, and this study also worked with water tables not as immediately close to the surface. The patterns observed with other studies may have been due to a more direct impact of water on the soil community than by the host plant. Fungi in those studies may have required specific requirements for oxygen and moisture levels to germinate, possibly with certain species surviving poorly at high water levels. In this study, the soil moisture levels were not high enough to cause saturation of the soil to create such a limiting environment.
Interestingly, the communities of sites in closer proximity to each other were not more similar to each other than other sites. Specifically, sites seven and eight, both at “shallow” water tables, and sites ten and eleven at “deep” water tables, did not cluster more closely. With PCA there are no obvious groupings by geographic location or water table depth that encompass all relevant samples to the exclusion of others. The “deep” water table samples are found in a relatively limited area of the plot, however samples from site eight and site five, shallow and mid-level water table locations respectively, are closely associated with them as well. The most geographically close samples also do not group closely, with sites ten and eleven and also site seven and eight having some distance between them. This difference is more significant for the site seven and eight samples, as their distance falls along axis 1. The site ten and eleven variance mostly falls within axis 2 and is therefore less significant as this axis describes much less of the variance. This indicates that the soil community is highly variable over small distances. Soil characteristics may have accounted for some of this difference. Sites were primarily sandy loam, with some having silt loam and loam. Other soil characteristics were also fairly uniform across samples and sites, with the exception of calcium, potassium, manganese and available phosphorus which varied widely across sites.
Other groups have previously investigated the mycorrhizae of sacaton and have found diversity within more restricted groups such as Glomeromycota (REF). The PCR primers used in this study amplify a fairly conserved region of the fungal genome in the 18S rDNA. Duong et al. (2006) felt that these same primers might be less informative due to the limited region of DNA amplified and some apparent specificity for Ascomycetes; however the large number and diversity observed in the bands acquired indicates that these primers are informative enough for a community study.
It must also be considered that the profiles acquired here are based on the whole DNA content of the soil, and it is possible that some species that produced bands in the profiles of the samples were no longer actively growing and reproducing in the soil. Environmental conditions, including plant species, may have changed to the point that a previously extant species could no longer survive, but its DNA still remained in the soil due to hyphal fragments, spores or other long-term survival structures.
The implications of this for reclamation efforts are interesting. Previous studies in reestablishment of sacaton have used mycorrhizal inocula from environments similar to the test site but not from the site itself (Richter & Stutz 2002). This site-dependent variation in microbial community may be why a decline in mycorrhizal infection was observed at a period of several months after transplanting, followed by a recovery. The introduced fungi may have survived only for the short-term and were then replaced by the local species.
Educational & Outreach Activities
This study will be published in two manuscripts; one focused on the Lehmann’s lovegrass/bluegramma study and one focused on the big sacaton study. Both manuscripts are currently under development.
This study revealed that although at each site differences in microbial communities were apparent between Lehmann’s lovegrass and blue grama, these differences were much smaller than those observed between distantly located sample sites. Thus, it appears that both grasses can establish and grow well with a varied soil microbial community. Moreover, the invasion of Lehmann’s lovegrass into the semi-desert grasslands of southern Arizona does not appear to be fundamentally altering the microbial community. From a restoration standpoint, this indicates that conservationists do not need to undertake restoration of the ecosystem at the level of soil fungi for this environment, and it should be possible to reestablish native species in the same soils without needing to perform re-inoculation of specific fungi.
Similarly, each sacaton site had a rather distinct soil fungal community and did not reveal a close association between water table and a specific fungal community profile. It is not clear what forces are impacting the site-specific fungal community, but as the resident plant community was quite uniform (nearly solid sacaton), it may be that subtle difference in underlying soil characteristics have a greater impact than the overlying plant community. This is consistent with finding from the Leymann’s/blue gramma study. For conservation and restoration efforts, this indicates that the sacaton communities that have been subjected to a drop in water table should recover upon reintroduction of a shallower water table without requiring extraordinary measures to reestablish the microbial community.
No specific economic analyses were conducted in this study.
Farmer adoption was not evaluated in this study, but ongoing discussion with managers at the Empire Ranch may result in modification of grassland management in the near future.
Areas needing additional study
While this study investigated locations where the invasion of Lehmann’s lovegrass into blue grama stands is an active and immediate factor, locations where either plant is well established may indicate if the patterns observed here are artifacts of this dynamic system. The study may also be repeated with a more diverse community sampling, including several different plant species. In this study, the variation in plant species was minimized, but the addition of another species to the analysis would allow for observation of possible gradients. It would also allow comparisons between the plants to be more robust. For example, if other native grasses present at the site such as Rothrock grama or sideoats grama were added to the study, it would allow evaluation of the native community of two related native grasses to the invasive.
Regarding the big sacaton study, for future studies it may also be informative to investigate a temporal element. It has been observed that for Arizona cottonwood trees, a species that is often found in semidesert communities alongside big sacaton, slower water table declines allow time for root growth to keep up with deeper water tables (Scott et al. 1999). In contrast, fast downcuts allow no chance for the plants to react to the change. Perhaps there is a similar effect on the microbiota of the soil. There was very little data on the exact date of water table declines for the sites in this study, but it was likely that some of the sites had been “dry” for only a few years, while others had been for decades. Specifically, site four is located near a fairly recently-created canyon incised by flowing water, causing an abrupt change in water table. In contrast, the other two “deep” sites, sites ten and eleven, were near a much shallower streambed but had likely been “dry” for much longer; the change also having been more gradual.