Catena 147 (2016) 216224

Contents lists available at ScienceDirect

Catena

journal homepage: www.elsevier.com/locate/catena

Soil microbial community proles and functional diversity in limestone

cedar glades
Jennifer Cartwright a,, E. Kudjo Dzantor b, Bahram Momen c
a
U.S. Geological Survey, Lower Mississippi-Gulf Water Science Center, 640 Grassmere Park, Suite 100, Nashville, TN 37211, USA
b
Tennessee State University, 3500 John A Merritt Blvd., Nashville, TN, 37209, USA
c
University of Maryland, College Park, 1425 Animal Sciences/Agricultural Engineering Building, College Park, MD 20742, USA

a r t i c l e i n f o a b s t r a c t

Article history: Rock outcrop ecosystems, such as limestone cedar glades (LCGs), are known for their rare and endemic plant spe-
Received 11 November 2015 cies adapted to high levels of abiotic stress. Soils in LCGs are thin (b25 cm), soil-moisture conditions uctuate sea-
Received in revised form 5 May 2016 sonally between xeric and saturated, and summer soil temperatures commonly exceed 48 C. The effects of these
Accepted 6 July 2016
stressors on soil microbial communities (SMC) remain largely unstudied, despite the importance of SMC-plant
Available online 16 July 2016
interactions in regulating the structure and function of terrestrial ecosystems. SMC proles and functional diver-
Keywords:
sity were characterized in LCGs using community level physiological proling (CLPP) and plate-dilution frequen-
Limestone cedar glades cy assays (PDFA). Most-probable number (MPN) estimates and microbial substrate-utilization diversity (H) were
Rock outcrop ecosystem positively related to soil thickness, soil organic matter (OM), soil water content, and vegetation density, and were
Environmental microbiology diminished in alkaline soil relative to circumneutral soil. Soil nitrate showed no relationship to SMCs, suggesting
Soil microbial community lack of N-limitation. Canonical correlation analysis indicated strong correlations between microbial CLPP patterns
Community level physiological proles and several physical and chemical properties of soil, primarily temperature at the ground surface and at 4-cm
depth, and secondarily soil-water content, enabling differentiation by season. Thus, it was demonstrated that sev-
eral well-described abiotic determinants of plant community structure in this ecosystem are also reected in SMC
proles.
Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction extreme temperatures at the soil surface combined with widely-uctu-

Rock-outcrop ecosystems in the USA contribute to regional and
global plant biodiversity at levels vastly disproportionate to their small
geographic footprints (Quarterman et al., 1993). For example, limestone
cedar glades (LCGs) of the southeastern USA support unique assem-
blages of plant species including rare endemics listed under the U.S. En-
dangered Species Act (Baskin and Baskin, 1999, 1989). Despite their
botanical importance, many LCGs face ongoing degradation (Baskin et
al., 1995; Quarterman et al., 1993). Indeed, roughly 50% of total LCG
area and 90% of ecologically intact LCGs have been lost from Middle Ten-
nessee, rendering LCGs an endangered ecosystem (Noss et al., 1995).
Within their geographic range, LCGs provide a unique physical envi-
ronment, characterized by very thin soils that experience seasonally-
Abbreviations: CCA, Canonical correlation analysis; CF, Canonical function; CLPP,
Community level physiological proling; H, Shannon-Weaver Index representing
microbial substrate-utilization diversity; LCG, Limestone cedar glade; LOI, Loss on
ignition; MPN, Most probable number; OM, Organic matter; PDFA, Plate-dilution
frequency assay; SMC, Soil microbial community.
Corresponding author.
E-mail addresses: jmcart@usgs.gov (J. Cartwright), edzantor@tnstate.edu
(E.K. Dzantor), bmomen@umd.edu (B. Momen).
ating hydrologic conditions. Soil surface conditions are characteristically
hot and xeric in summer and experience prolonged saturation in winter
(Baskin and Baskin, 1989). These stress factors in LCGs effectively ex-
clude most vegetation of the surrounding mesophytic forest and thus
help maintain habitat for endemic plants (Baskin and Baskin, 2003,
1989, 1999). LCGs function as edaphic climax ecosystems where soils
are too thin to support the growth of trees. Instead, the vegetation is
dominated by C3 forbs and C4 summer annual grasses, accompanied
by mosses, lichens, and cyanobacteria in microhabitats of extremely
thin soil or exposed limestone bedrock (Baskin and Baskin, 1999;
Baskin et al., 2007).
Plant communities in LCGs have been systematically described (e.g.
Baskin and Baskin, 2003; Somers et al., 1986; Taylor and Estes, 2012)
but their associated soil microbial communities (SMCs) are largely un-
studied. Indeed, SMCs have received little scientic attention in rock
outcrop ecosystems throughout eastern North America, despite the rec-
ognized importance of these ecosystems to global plant biodiversity
(Quarterman et al., 1993). This is a key knowledge gap, because plant-
SMC interactions inuence a host of ecological processes such as bio-
geochemical cycling, plant-pathogen dynamics, succession, and soil dis-
turbance (de Vries et al., 2012; Reynolds et al., 2003; van der Heijden et
al., 2008).

http://dx.doi.org/10.1016/j.catena.2016.07.010
0341-8162/Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
 J. Cartwright et al. / Catena 147 (2016) 216224
Ongoing efforts to restore glade environments and to understand
and predict ecological changes, such as from climate change or exotic
species invasions (Cofer et al., 2008; Molano-Flores and Bell, 2012),
could be strengthened by studies of SMC dynamics that interact with
these processes (Harris, 2009; Van der Putten, 2010). A number of in-
vestigators have indicated an urgent need for development of sustain-
able management practices for fragile ecosystems such as LCGs
(Baskin et al., 2007; Nordman, 2004; Noss et al., 1995; Quarterman et
al., 1993). Improved understanding of SMCs and their relationships to
abiotic stressors is thus vitally important to the long-term conservation
of plant biodiversity in LCGs and related ecosystems under changing en-
vironmental conditions.
In service of these goals, the objectives of this study were (1) to char-
acterize spatial and temporal patterns of soil microbial numbers and
functional diversity in LCGs, and (2) to relate these patterns to key soil
physical and chemical properties that dene the abiotic stress regime
in this ecosystem. This study provides baseline information useful for
monitoring abiotic and biotic changes in LCG soils in the face of conser-
vation challenges, such as projected land-use and climate changes. This
study also contributes to a holistic ecological understanding of this eco-
system to support scientically-sound management of LCGs.
2. Materials and methods
2.1. Study area
Soil microbial investigations1 were performed in LCGs at Stones
River National Battleeld (Fig. 1) in Rutherford County, Tennessee,
USA (355235N, 862558W). This park is within a karst sinkhole
plain in the Nashville Basin section of the Interior Low Plateaus physio-
graphic province (Fenneman and Johnson, 1946). Bedrock formations
within the park are Middle Ordovician limestones (Thornberry-
Ehrlich, 2012). Two soil associations are present: the Gladeville-Rock
outcrop-Talbott and the Talbott-Bareld-Rock outcrop complex (U.S.
Department of Agriculture Soil Conservation Service, 1977). Several
dozen LCGs are present within the park, ranging from b70 m2 to ap-
proximately 2800 m2. These glades form a roughly ring-shaped cong-
uration surrounding a topographic high underlain by a structural basin
(Cofer et al., 2008; Thornberry-Ehrlich, 2012). LCGs consist of outcrops
of thinly-bedded limestone surrounded by Gladeville soil, a thin, silty
clay loam with limestone ags throughout the soil and forming a layer
at the soil surface (U.S. Department of Agriculture Soil Conservation
Service, 1977). Areas between glades are predominantly Talbott and
Bareld soils, both of which have surface layers of silty clay loam.
Within LCGs at this site, zones of exposed bedrock surrounded by
thin soil support mosses, cyanobacteria, lichens, and sparsely-distribut-
ed herbaceous plants (Fig. S1, supplementary material). In areas with
greater soil thickness, C4 summer annual grasses and C3 forbs are pre-
dominant, with woody species generally conned to pockets of deep
soil in bedrock cracks and to ecotones with the surrounding oak-red
cedar forest (Cofer et al., 2008; Nordman, 2004).
2.2. Sampling design
Twelve LCGs were selected to study spatial and temporal patterns of
SMCs (Fig. 1). Thirty six 0.5-m 0.5-m quadrats were established, with
approximately 80% (28 quadrats) located in areas dominated by moss,
graminoids, and forbs and the remainder within a 3-m buffer of pre-
dominantly Juniperus virginiana forest surrounding glades. Quadrat po-
sition was randomly assigned using ArcGIS v. 9.3 (Esri, Redlands, CA,
USA). Quadrats were sampled according to a rotational schedule over
16 months (February 2012 to May 2013), with each quadrat sampled
4 times at approximately 4-month intervals. For certain soil properties,
1
Data used in the analysis for this study are available from http://dx.doi.org/10.5066/
F7NV9G9C
217
multiple subsamples were averaged to obtain a single measurement for
each quadrat at each sampling event (see Section 2.4).
2.3. Vegetation characterization
The vegetation cover at each quadrat was characterized based on vi-
sual examination. Vascular plants were broadly categorized as
graminoid, forb, vine, or shrub, following Cofer et al. (2008). The per-
centage of each category was estimated for each quadrat and catego-
rized as none, b 30%, 30% to 70%, or N70% covered with scores
assigned from 0 to 3, respectively. An overall vegetation score was cal-
culated for each quadrat as the sum of individual scores for each vegeta-
tion category.
2.4. Soil physical and chemical analysis
Depth to bedrock (soil thickness) was measured by inserting a 1-
cm-diameter metal probe into the soil until it contacted bedrock (4 sub-
samples per quadrat). Soil temperature at 4-cm soil depth (3 subsam-
ples per quadrat) and ground-surface temperature were measured
using a Taylor 9842N digital thermometer and a Taylor 1523 digital
thermometer/hygrometer, respectively (Taylor Inc., Oak Brook, IL, USA).
At each sampling event, soil samples were collected at a depth of 3
4 cm using sterile techniques. Soil samples were sealed in autoclaved
glass jars and transported on ice to the laboratory for analysis. Gravi-
metric soil water content (3 subsamples per soil sample) was measured
using a Mettler Toledo HB43 halogen moisture analyzer (Mettler Tole-
do, Columbus, OH, USA). For chemical analysis, a portion of each soil
sample was air-dried and sieved using a 2-mm screen. Soils were then
analyzed for organic matter (OM), pH, and nitrate concentration. OM
was estimated using the loss-on-ignition (LOI) method (Davies, 1973).
Soil pHH2O (1:1) was measured using a calibrated FieldScout SoilStik
pH Meter (Spectrum Technologies, Inc., Plaineld, IL, USA). Soil nitrate
levels were estimated using a Hach Platinum Series combination nitrate
electrode (Hach Company, Loveland, CO, USA).
2.5. Soil microbiological characterizations
A separate portion of each soil sample was transferred to a refriger-
ator (4 C) within 6 h of leaving the eld and was processed for micro-
bial characterization within 24 h. Soil microbial populations were
characterized using a plate-dilution frequency assay (PDFA), which is
a most probable number (MPN) method. SMC metabolic proles were
assessed using community-level physiological proling (CLPP).
2.5.1. Plate-dilution frequency assay of microbial populations
PDFA (Harris and Sommers, 1968) was used to estimate an MPN of
culturable microbes in soil. Six levels of serial dilutions from each soil
sample (10 2, 10 3, 10 4, 10 5, 10 6, and 107) were inoculated
onto plate count agar (Difco, Benton, Dickinson, and Company, Frank-
lin Lakes, NJ, USA), in petri dishes. The petri dishes were prepared from
dehydrated agar, checked for contamination after 48 h, and pre-marked
with eight circles per dilution level. Ten microliters (10 L) from each di-
lution were inoculated into the center of each circle, going from the least
to the highest concentration in a dilution series. Following incubation in
the dark at 25 C for 7 days, the total number of marked circles with pos-
itive growth (colony development) was referenced to statistical tables
to estimate MPN of culturable microbes per mL of soil suspension at
the 102 dilution level, along with a 95% condence interval on MPN
(Cochran, 1950; Harris and Sommers, 1968). PDFA was performed on
soil samples collected from May 2012 to May 2013.
2.5.2. Microbial community level physiological proling (CLPP)
CLPP based on sole-carbon substrate utilization (Garland and Mills,
1991; Garland, 1996; Preston-Mafham et al., 2002) was performed
using 96-well Biolog EcoPlates (Biolog Inc., Hayward, CA). For each
218 J. Cartwright et al. / Catena 147 (2016) 216224

Fig. 1. Map of sampling locations in 12 selected limestone cedar glades (LCGs) at Stones River National Battleeld. Polygons representing LCGs are courtesy of the National Park Service.
Physiographic divisions are from USDA (2015).
 J. Cartwright et al. / Catena 147 (2016) 216224 219

soil sample, 150 L of the 102 dilution level suspension from the PDFA
procedure (Section 2.5.1) were inoculated into an EcoPlate. EcoPlates
contained 31 carbon substrates replicated 3 times on each plate. Three
wells contained sterilized water to serve as controls. Immediately fol-
lowing inoculation, EcoPlates were covered and incubated in the dark
at 25 C. Color development of a respiration indicator, tetrazolium chlo-
ride, provided a measure of substrate utilization. EcoPlates were read at
intervals of 24, 48, 72, 96, and 120 h after inoculation using a Biolog
MicroStation plate reader (Biolog, Inc., Hayward, CA, USA) at an absor-
bance of 590 nm to monitor changes in respiration over time. CLPP
was performed on soil samples collected from February 2012 to
March 2013.
2.6. Statistical analyses
All statistical analyses were performed using SAS 9.3 (SAS Institute,
Cary, NC, USA). Relationships between vegetation, soil physical and
chemical properties, and MPN and CLPP were analyzed using analyses
of variance (ANOVA), quantile regression analysis, and canonical corre-
lation analysis (CCA). At each sampling event, multiple subsamples
taken per quadrat or soil sample were averaged prior to statistical anal-
ysis to avoid pseudoreplication.
2.6.1. Soil sample categorization based on MPN
Based on PDFA analysis, 95% condence intervals on MPN estimates
were used to assign soil samples to low-, medium-, and high-MPN
groups, such that between-group MPN estimates were signicantly dif-
ferent (P b 0.05), but within-group MPN estimates were not. In this way,
94 out of 99 soil samples were categorized as follows: low: 2.3 105 to
7.8 105; medium: 1.0 106 to 3.1 106; high: 4.3 106 to 13.7 106
organisms per mL of suspension of the 102 dilution level. Estimates for
5 samples were signicantly lower (1) or higher (4) than any of the
MPN groups so these samples were not retained for further analysis.
To determine the effects of various soil physical and chemical properties
on MPN, one-way ANOVA and Tukey's post-hoc tests were performed
using a general linear model with the three MPN groups as categories
(PROC ANOVA and PROC GLM; SAS Institute, 2011).
2.6.2. Transformations for CLPP data
To correct for inoculum-density effects, plate-level transformations
for carbon substrate data followed Garland and Mills (1991):
T RC =AWCD 1 higher water content, and more dense vegetation (Fig. 2). Neither soil
hXn i
AWCD RC =n 2 ature and soil temperature at 4-cm depth) were signicantly related to
i1
Because plots of H against soil physical and chemical properties indi-
cated highly heteroscedastic, wedge-shaped distributions (Section 3.2),
quantile regression was used to investigate relationships between these
soil properties and H (PROC QUANTREG; SAS Institute, 2011).
Heteroscedasticity implies that multiple rates of change (slopes) may
describe the relationships between the response variable and a limiting
factor (Cade and Noon, 2003). Quantile regression is well-suited to such
distributions when ecological responses are constrained by multiple
measured and unmeasured limiting factors (Schmidt et al., 2012).
Quantile process plots show the slope parameter and associated 95%
condence interval at all quantile levels for the quantile regressions.
This approach has advantages over the use of quantile regression at
only one or a few quantile levels because limitations on both the maxi-
mum and minimum values of the biological response variable can be
identied, as can portions of the response variable distribution most
limited by the limiting factor(s) under consideration (Cade and Noon,
2003; Schmidt et al., 2012).
2.6.4. CLPP patterns for soil microbial communities
To investigate relationships between CLPP patterns of SMCs
and various soil physical and chemical properties, CCA was performed
on CLPP data from soils (PROC CANCORR; SAS Institute, 2011). Pairs
of linear combinations of the original variables within each set
(Canonical Functions; CFs) were calculated to maximize the correlation
among the constructed functions (Lattin et al., 2003; Ye and Wright,
2010).
3. Results
3.1. Soil physical and chemical properties in relation to MPN
The soils of the LCGs in this study were generally thin (all b 23 cm
depth to bedrock), neutral to slightly alkaline, with variable levels of
OM (Table 1). Gravimetric soil water content ranged from xeric
(below 5%) to saturated (above 45%). Soil and ground-surface tempera-
tures were often very hot in summer, exceeding 38 C and 48 C
respectively.
MPN estimates of culturable microbes in 94 soil samples ranged
between 2.3 105 and 13.7 106. Higher MPN was signicantly
associated with thicker soils, higher soil OM, lower (more neutral) pH,
nitrate nor either of the temperature variables (ground-surface temper-
MPN.

where T represents transformed substrate-level response values, R is

the mean absorbance of the response wells (3 wells per carbon sub- Table 1
strate), C is the mean absorbance value of the 3 control wells, AWCD is Soil physical and chemical properties across 36 quadrats in 12 limestone cedar glades sam-
average well color development for the EcoPlate, and n is the number pled from February 2012 to May 2013.
of carbon substrates (31 for EcoPlates). To integrate time-series data N R Min Max Mean Standard deviation
from multiple EcoPlate readings (for AWCD and for individual sub-
Depth to bedrock (cm) 36 4 2.40 22.62 8.40 5.43
strates, T), the area under the incubation curvefrom 48 h to 120 h of Soil organic matter 144 0.04 0.27 0.12 0.05
incubationwas calculated and used in subsequent analysis (Preston- (fraction, LOI)
Mafham et al., 2002). Soil pH 144 7.04 8.28 7.80 0.26
Gravimetric soil water 144 3 3.46 49.46 22.63 8.50
content (percent)
2.6.3. Microbial substrate-utilization diversity
Soil nitrate (g g1 NO3-N) 144 1.07 29.84 5.59 4.32
To assess community-level microbial substrate-utilization diversity Vegetation score 36 1 8 2.78 1.67
from CLPP data, the Shannon-Weaver index (H) was calculated as: Ground-surface 144 1.50 48.40 22.61 11.92
temperature (C)
Xn Soil temperature at 144 3 0.53 38.67 16.20 8.94
H i1 pi ln pi 3
4-cm depth (C)

N = number of measurements; depth to bedrock and vegetation were measured once per
where pi is the ratio of the microbial activity on each substrate (T values, quadrat (36 quadrats); all other properties were measured 4 times per quadrat over 16
area under the incubation curve) to the sum of the microbial activities months. For certain properties, each measurement represents an average of R subsamples;
on all substrates for a given EcoPlate. see Section 2.4.
220 J. Cartwright et al. / Catena 147 (2016) 216224

Fig. 2. Distribution by most probable number (MPN) category of A, depth to bedrock; B, soil organic matter; C, soil pH; D, gravimetric soil water content; E, soil nitrate; F, vegetation score;
G, ground-surface temperature and H, soil temperature at 4-cm depth. Differing letters above each boxplot represent signicant differences (P b 0.05).

3.2. Microbial substrate-utilization diversity
Values of the Shannon-Weaver Index (H) for microbial substrate-uti-
lization diversity ranged from 2.00 to 3.38, with 115 samples out of 125
(92%) above 3.0. H was uncorrelated with MPN (r = 0.21; P N 0.05) and
values of H were not different across MPN groups (F = 2.74; P N 0.05),
indicating that microbial substrate-utilization diversity was not merely
a function of the numbers of culturable microbes in soils.
Plots of H versus soil physical and chemical properties indicated
highly heteroscedastic distributions (Fig. S2, supplementary material).
Consequently, these relationships were investigated by means of
quantile process plots based on quantile regression across all quantile
levels, (Fig. 3), with inferences following Schmidt et al. (2012). For
any quantile level , % of H values are the quantile regression func-
tion of the given soil property (Cade and Noon, 2003). Thus, at any given
value (horizontal axes in Fig. 3), the slope parameter estimate of the
regression function can be plotted (vertical axes in Fig. 3) and its 95%
condence interval (shaded area) can be compared to zero to test if
the soil property under consideration affects H. For quantile ranges, ,
where the entire shaded area is above zero on the vertical axis, the effect
of the soil property on H is considered positive (P b 0.05). Conversely,
values for which the entire shaded area is below zero indicate that the
soil property effect on H is negative. Values of for which the shaded
area includes zero indicate no effect of the soil property on H (P N 0.05).
Several soil properties showed signicant associations with H across
most of the intermediate (not extremely high or low) quantile ranges.
The association between soil thickness (depth to bedrock) and H was
positive for 0.15 b b 0.75 (Fig. 3A). The association between soil pH
and H was negative for 0.25 b , indicating that H was higher in
circumneutral soils and decreased with increasing alkalinity (Fig. 3C).
 J. Cartwright et al. / Catena 147 (2016) 216224 221

Fig. 3. Quantile process plots showing slope parameter estimates and 95% condence intervals (shaded areas) for the Shannon-Weaver Index (H) of microbial substrate utilization for A,
depth to bedrock; B, soil organic matter; C, soil pH; D, gravimetric soil water content; E, soil nitrate; F, vegetation score; G, ground-surface temparature and H, soil temperature at 4-cm
depth. All horizontal axes are quantile levels (), meaning that % of H values are the quantile regression function of the given soil property.

Gravimetric soil water content had a positive association with H for
0.15 b b 0.85 (Fig. 3D). The association with vegetation score was pos-
itive for 0.10 b b 0.65 (Fig. 3F).
Other soil properties showed no signicant associations with
H, or their associations were limited to relatively narrow quantile
ranges. Soil nitrate was not associated with H at any quantile
level (Fig. 3E), whereas soil OM had a signicant positive association
with H only for 0.30 b b 0.50 and 0.55 b b 0.80 (Fig. 3B). Both
ground-surface temperature and soil temperature at 4-cm depth had
signicant negative associations with H across limited quantile ranges:
0.25 b b 0.60 and 0.30 b b 0.65, respectively (Fig. 3G and H,
respectively).
222 J. Cartwright et al. / Catena 147 (2016) 216224

Table 2 culturable microbes, and that soil physical and chemical properties ap-
Correlation coefcients between the rst two sets of canonical functions (CFs) and soil pear to inuence these microbial indicators in different ways. However,
physical and chemical properties. CFsoil1 and CFsoil2 were constructed from soil properties;
CFsubs1 and CFsubs2 were constructed from CLPP carbon substrate data.
certain soil properties were prominently related both to microbial
abundance and to substrate utilization diversity, suggesting their eco-
Variable CFsoil1 CFsoil2 CFsubs1 CFsubs2 logical relevance to SMCs.
Soil temperature at 4-cm depth (C) 0.98 0.18 0.72 0.12 In particular, thin and relatively alkaline soils were associated both
Ground-surface temperature (C) 0.93 0.26 0.69 0.17 with reduced MPN and with lower H values. It is noteworthy that the ef-
Gravimetric soil water content (percent) 0.48 0.45 0.36 0.30
fects of soil thickness and pH were so clearly discernible, given that all
Soil organic matter (LOI) 0.37 0.75 0.29 0.49
Soil nitrate (ppm NO3-N) 0.35 0.40 0.27 0.25 soils in this study were very thin and soil pH values were within a rela-
Depth to bedrock (cm) 0.30 0.44 0.22 0.28 tively narrow range (by comparison, soil pH ranged from b4.5 to N8.5
Vegetation score 0.21 0.44 0.18 0.35 across terrestrial ecosystems in North and South America; Lauber et
Soil pH 0.11 0.39 0.11 0.29 al., 2009). In thin-soil areas of LCGs, soil chemistry is inuenced by lime-
indicates signicant correlation at P b 0.05. stone parent material and by relative scarcity of leaf-litter inputs. Thin-
soil areas support distinct vegetation communities (Quarterman et al.,
3.3. CLPP patterns for soil microbial communities 1993; Somers et al., 1986). Our ndings suggest that these thin-soil, rel-
atively alkaline zones support lower numbers of culturable microbes
Of the CFs constructed, the rst and second sets correlated signi- and that SMCs in these zones display reduced substrate-utilization di-
cantly (r = 0.83 and 0.72, respectively, P b 0.01) and together accounted versity compared to zones of thicker, circumneutral soil. This pattern
for 57% of the total variability of the multivariate data. Based on the ab- may account in part for previous ndings of reduced soil respiration in
solute values of the correlations between original variables and the ca- thin-soil areas of LCGs (Cartwright and Hui, 2014).
nonical functions related to soil properties, the rst function (CFsoil1) Not surprisingly, vegetation score was positively associated with
was most strongly related to soil temperature at 4-cm depth and to both MPN and H. Plant community composition inuences SMCs
ground-surface temperature, and secondarily to soil water content through microenvironmental effects (e.g. temperature effects from
(Table 2). The second function (CFsoil2) primarily reected soil OM con- shading, soil moisture effects from transpiration) and nutrient inputs
tent. The carbon substrates that were correlated with CFsubs1 and CFsubs2 from litter and root exudates (Thomson et al., 2010; Zak et al., 2003).
(Table S1, supplementary material) were spread fairly evenly across all Differences in quantity and quality of these inputs are reections of
six of the substrate guilds present on the EcoPlates: carbohydrates, plant community composition that directly affect soil microbial re-
amino acids, esters, carboxylic acids, amines, and polymers (Zak et al., sponses (Cong et al., 2015; Wardle et al., 2004). SMCs also help deter-
1994). All carbon substrates on the EcoPlates that are known compo- mine plant community composition (Wardle et al., 2004), for example
nents of root exudates (Campbell et al., 1997) were correlated with ei- via detrital food webs in which different SMCs generate different nutri-
ther CFsub1 or CFsub2. Seasonal differences in substrate utilization ent pools for plant use (Reynolds et al., 2003).
proles were apparent when warm-season (May through August) sam- If the primary mechanism by which vegetation cover inuences
ples were plotted along with cool-season (November through February) MPN and H is through organic inputs and food-web interactions, one
samples in an ordination plot of CFsubs1 and CFsubs2 (Fig. 4). might expect that soil OM would also be positively associated with
these SMC indicators. We found mixed support for this conclusion. On
the one hand, soil OM was strongly related to MPN, suggesting that
4. Discussion higher rates of carbon additions to soil in more densely-vegetated
zones of LCGs supported greater numbers of culturable microbes. Spa-
4.1. Inuence of soil physical and chemical properties on soil microbial tial variability in soil OM was reected in CLPP data, in that: (1) soil
communities OM was the most strongly-correlated soil property with the second
CF, and (2) all six of the EcoPlate substrates that are components of
Findings from this study indicate that microbial substrate-utilization root exudates were signicantly correlated with either the rst or sec-
diversity in LCG soils is not simply a reection of relative numbers of ond CF.
On the other hand, the association of OM with H was relatively weak,
i.e. signicantly positive only for limited ranges of based on quantile
regression. This is perhaps not surprising given the complexity of factors
regulating microbial metabolism in soil environments (Torsvik and
vres, 2002). Indeed, the full suite of carbon sources available for mi-
crobial metabolism in LCG soils is likely to be vastly more complex
than, and could in fact be qualitatively different from, that represented
by the 31 substrates present on the EcoPlates in this study (Konopka
et al., 1998). Future investigations to quantify expression levels of
genes involved in specic microbial metabolic pathways could help elu-
cidate the role played by various forms of OM in soil food webs of LCGs.
The absence of a relationship between soil nitrate and any of the
SMC indicators in this study may indicate that N was not a limiting nu-
trient in soil at this study site. N-xing cyanobacteria (e.g. Nostoc spe-
cies) are generally prominent constituents of LCG ecosystems (Baskin
and Baskin, 2003; Martin and Sharp, 1983; Quarterman, 1950a), includ-
ing the glades at this study site (Dubois, 1993; Nordman, 2004).
Findings from this study highlight the importance of seasonally-var-
iable soil conditions to SMCs in LCGs. In particular, soil water content
was positively related both to MPN and to H, suggesting that numbers
Fig. 4. Ordination plot of soil samples collected from May through August versus from
November through February based on canonical correlation analysis of microbial
of culturable microbes and microbial substrate-utilization diversity
substrate utilization proles; CFsubs1 and CFsubs2 are the rst two canonical functions were suppressed by seasonally dry soil conditions. Seasonally dry soils
constructed from CLPP carbon substrate data. in LCGs result from intense evaporation during summer and early
 J. Cartwright et al. / Catena 147 (2016) 216224
autumn due to soil thinness combined with full insolation from lack of
canopy cover (Baskin and Baskin, 1999; Quarterman, 1950b). Botanists
have long noted the prevalence of drought-tolerance adaptations
among LCG vegetation, which includes succulents, species using the
crassulacean acid metabolism photosynthetic pathway, and winter-an-
nuals that survive dry summer conditions using dormancy (Baskin et al.,
1995; Quarterman et al., 1993).
This study indicates that patterns of soil water availability in LCGs
are also relevant to soil microbial ecology. Soil moisture limitation can
inhibit microbial function by constraining rates of substrate diffusion
and by cytoplasmic dehydration that reduces enzymatic efciency
(Stark and Firestone, 1995). In a variety of arid and semi-arid ecosys-
tems, soil moisture availability is a strong control on microbe-mediated
processes such as heterotrophic soil respiration (Munson et al., 2009;
Talmon et al., 2011). Martin and Sharp (1983) found no clear relation-
ship between soil moisture availability and species richness for protozoa
in LCGs, possibly because of relatively limited sampling (5 sites each
sampled twice) that did not include xeric conditions (average available
soil moisture always 37%). Because seasonal moisture limitation is so
ecologically important to plant communities in LCGs (Quarterman,
1950a; Quarterman et al., 1993), further research is warranted
concerning its effects on microbial community structure and function,
especially in light of predicted increases in summer drought severity
in the southeastern United States (Mearns et al., 2003).
In contrast to soil-moisture effects, this study found mixed results
for temperature effects on SMCs. Temperatures at the soil surface and
at a depth of 4 cm were both unrelated to MPN estimates and were
weakly related to H, i.e. signicantly negative only for limited ranges
of . A likely explanation is that temperature regimes in the eld were
not represented in the laboratory protocols for this study. Specically,
inoculated agar plates for PDFA analysis and EcoPlates for CLPP were in-
cubated at ~ 25 C, allowing proliferation of culturable microbes that
may have been dormant under extremely cold or hot temperatures in
the eld. However, it is noteworthy that temperatures at the soil surface
and at a depth of 4 cm were both strongly correlated to the rst pair of
CFs based on CCA analysis of CLPP data. Separation of soil samples in or-
dination space based on substrate-utilization proles suggests that
functionally-different SMCs were present in LCG soils on a seasonal
basis (Fig. 4). Future studies could seek to clarify soil-temperature ef-
fects on SMCs in this ecosystem by employing a range of seasonally-rel-
evant incubation temperatures (e.g. ranging from 0 C to 40 C; Table 1)
or by applying culture-independent methods to characterize soil micro-
bial community structure across seasons. Such methods, e.g. phospho-
lipid fatty acid (PLFA) proles and fatty acid methyl esters (FAME)
analysis, have demonstrated temperature-related shifts in soil microbial
community structure (Frey et al., 2008; Zogg et al., 1997).
4.2. Methodological considerations
This study of SMCs in LCG soils employed culture-dependent
methods, the limitations of which are thoroughly documented
(Degens and Harris, 1997; Konopka et al., 1998; Rastogi and Sani,
2011). However, culture-dependent methods continue to generate use-
ful information for characterizing SMCs in a variety of environmental
contexts (Frc et al., 2012; Garland et al., 2010; Le et al., 2011; Sutton,
2010). Although a variety of culture-independent, molecular methods
can circumvent many of the limitations associated with culture depen-
dency, no single molecular technique can comprehensively describe mi-
crobial diversity at a particular site (Dahllf, 2002; Malik et al., 2008).
Because culture-dependent and culture-independent approaches in-
volve multiple, non-overlapping sets of biases, concurrent use of the
two approaches may allow better representation of true microbial di-
versity than either approach alone (Al-Awadhi et al., 2013).
In this study, clear and statistically signicant patterns were
discerned between SMCs and soil properties in LCGs using culture-de-
pendent methods. Although CLPP necessarily restricted our analysis to
223
organisms capable of rapid metabolism of a limited number of carbon
sources under laboratory (not eld) conditions, it nonetheless indicated
clear differences in SMCs between soil samples (Fakruddin and
Mannan, 2013; Preston-Mafham et al., 2002). Furthermore, these differ-
ences were associated with a number of environmental factors (e.g. soil
thickness, pH, and moisture availability) known to be of primary impor-
tance to plant community structure in LCGs (Baskin and Baskin, 2003,
1999; Quarterman et al., 1993; Somers et al., 1986). In future studies,
hybrid approaches could be employedcoupling the methods de-
scribed here with culture-independent proceduresto further elucidate
the role of environmental factors in shaping SMCs in this threatened and
biodiversity-rich ecosystem.
5. Conclusions
Several soil properties that are key components of abiotic stress re-
gimes in limestone cedar glades showed clear relationships with soil
microbial indicators. Most-probable number estimates along with mi-
crobial substrate-utilization diversity were positively related to soil
thickness, soil water content, and vegetation score, and were higher in
circumneutral soil than in alkaline soil. Soil nitrate showed no relation-
ship with SMC indicators, suggesting that N may have been non-limiting
in these soils. Soil temperature, at the surface and at 4-cm depth, was a
prominent variable based on canonical correlation analysis of commu-
nity-level physiological proles, suggesting possible functional differ-
ences in microbial communities based on season. Because soil
microbial analysis has been lacking for rock outcrop ecosystems, this
study provides useful information to begin investigating plant-SMC re-
lationships in these ecosystems. In future studies, a combination of cul-
ture-dependent methods with molecular approaches could enable a
more comprehensive assessment of microbial diversity and function
in this endangered and biodiversity-rich ecosystem.
Acknowledgements
This research was supported by the National Park Service, (Study#
STRI-00025; Wolfe, PI), the U.S. Geological Survey, and Tennessee
State University (USDA/NIFA award 2010-38821-21594; Dzantor, PI).
The authors thank William Wolfe (U.S. Geological Survey) and Gib
Backlund and Troy Morris (National Park Service) for their assistance
with eld logistics. This manuscript was improved based on reviews
by Thomas Byl (U.S. Geological Survey) and anonymous reviewers.
Any use of trade, rm, or product names is for descriptive purposes
only and does not imply endorsement by the U.S. Government.
Appendix A. Supplementary data
Supplementary data associated with this article can be found in on-
line version, at doi:http://dx.doi.org/10.1016/j.catena.2016.07.010.
These data include point locations representing the sampling quadrats
described in this article.
References
Al-Awadhi, H., Dashti, N., Khanafer, M., Al-Mailem, D., Ali, N., Radwan, S., 2013. Bias Prob-
lems in Culture-Independent Analysis of Environmental Bacterial Communities: A
Representative Study on Hydrocarbonoclastic Bacteria. 2. Springerplus, p. 369.
Baskin, J., Baskin, C., 2003. The vascular ora of cedar glades of the southeastern United
States and its phytogeographical relationships. J. Torrey Bot. Soc. 130, 101118.
Baskin, J., Baskin, C., 1999. Cedar Glades of the Southeastern United States. In: Anderson,
R., Fralish, J., Baskin, J. (Eds.), Savannas, Barrens, and Rock Outcrop Plant Communities
of North America. Cambridge University Press, New York, pp. 206219.
Baskin, J., Baskin, C., 1989. Cedar glade endemics in Tennessee, and a review of their aut-
ecology. J. Tennessee Acad. Sci. 64, 6374.
Baskin, J., Baskin, C., Quarterman, E., 2007. Flow diagrams for plant succession in the Mid-
dle Tennessee cedar glades. J. Bot. Res. Inst. Texas 1, 11311140.
Baskin, J., Webb, D., Baskin, C., 1995. A oristic plant ecology study of the limestone glades
of northern Alabama. Bull. Torrey Bot. Club 122, 226242.
Cade, B.S., Noon, B.R., 2003. A gentle introduction to quantile regression for ecologists.
Front. Ecol. Environ. 1, 412420.
224 J. Cartwright et al. / Catena 147 (2016) 216224
Campbell, C.D., Grayston, S.J., Hirst, D.J., 1997. Use of rhizosphere carbon sources in sole
carbon source tests to discriminate soil microbial communities. J. Microbiol. Methods
30, 3341.
Cartwright, J., Hui, D., 2014. Soil respiration patterns and controls in limestone cedar
glades. Plant Soil 389, 157169.
Cochran, W., 1950. Estimation of bacterial densities by means of the most probable num-
ber. Biometrics 6, 105116.
Cofer, M., Walck, J., Hidayati, S., 2008. Species richness and exotic species invasion in Mid-
dle Tennessee cedar glades in relation to abiotic and biotic factors. J. Torrey Bot. Soc.
135, 540553.
Cong, J., Yang, Y., Liu, X., Lu, H., Liu, X., Zhou, J., Li, D., Yin, H., Ding, J., Zhang, Y., 2015. Anal-
yses of soil microbial community compositions and functional genes reveal potential
consequences of natural forest succession. Sci. Report. 5, 10007. http://dx.doi.org/10.
1038/srep10007.
Dahllf, I., 2002. Molecular community analysis of microbial diversity. Curr. Opin.
Biotechnol. 13, 213217.
Davies, B., 1973. Loss-on-ignition as an estimate of soil organic matter. Soil Sci. Soc. Am. J.
38, 150151.
De Vries, F.T., Manning, P., Tallowin, J.R.B., Mortimer, S.R., Pilgrim, E.S., Harrison, K.A.,
Hobbs, P.J., Quirk, H., Shipley, B., Cornelissen, J.H.C., Kattge, J., Bardgett, R.D., 2012. Abi-
otic drivers and plant traits explain landscape-scale patterns in soil microbial com-
munities. Ecol. Lett. 15, 12301239.
Degens, B.P., Harris, J.A., 1997. Development of a physiological approach to measuring the
catabolic diversity of soil microbial communities. Soil Biol. Biochem. 29, 13091320.
Dubois, J., 1993. Biological dinitrogen xation in two cedar glade communities of middle
Tennessee. J. Tennessee Acad. Sci. 68, 101105.
Fakruddin, M., Mannan, K., 2013. Methods for analyzing diversity of microbial communi-
ties in natural environments. Ceylon J. Sci. 42, 1933.
Fenneman, N., Johnson, D., 1946. Physical divisions of the United States: U.S. Geological
Survey map prepared in cooperation with the Physiographic Commission. U.S. Geo-
logical Survey (scale 1:7,000,000).
Frc, M., Oszust, K., Lipiec, J., 2012. Community level physiological proles (CLPP), charac-
terization and microbial activity of soil amended with dairy sewage sludge. Sensors
12, 32533268.
Frey, S.D., Drijber, R., Smith, H., Melillo, J., 2008. Microbial biomass, functional capacity,
and community structure after 12 years of soil warming. Soil Biol. Biochem. 40,
29042907.
Garland, J., Mills, A., 1991. Classication and characterization of heterotrophic microbial
communities on the basis of patterns of community-level sole-carbon-source utiliza-
tion. Appl. Environ. Microbiol. 57, 23512359.
Garland, J.L., 1996. Analytical approaches to the characterization of samples of microbial
communities using patterns of potential C source utilization. Soil Biol. Biochem. 28,
213221.
Garland, J.L., Mackowiak, C.L., Zabaloy, M.C., 2010. Organic waste amendment effects on
soil microbial activity in a cornrye rotation: application of a new approach to com-
munity-level physiological proling. Appl. Soil Ecol. 44, 262269.
Harris, J., 2009. Soil microbial communities and restoration ecology: facilitators or fol-
lowers? Science 325, 573574.
Harris, R., Sommers, L., 1968. Plate-dilution frequency technique for assay of microbial
ecology. Appl. Microbiol. 16, 330334.
Konopka, A., Oliver, L., Turco, R., 1998. The use of carbon substrate utilization patterns in
environmental and ecological microbiology. Microb. Ecol. 35, 103115.
Lattin, J., Carroll, J., Green, P., 2003. Analyzing Multivariate Data. Thomson Learning, Pacif-
ic Grove, CA.
Lauber, C.L., Hamady, M., Knight, R., Fierer, N., 2009. Pyrosequencing-based assessment of
soil pH as a predictor of soil bacterial community structure at the continental scale.
Appl. Environ. Microbiol. 75, 51115120.
Le, X., Hui, D., Dzantor, E.K., 2011. Characterizing rhizodegradation of the insecticide
bifenthrin in two soil types. J. Environ. Prot. 2, 940946.
Malik, S., Beer, M., Megharaj, M., Naidu, R., 2008. The use of molecular techniques to char-
acterize the microbial communities in contaminated soil and water. Environ. Int. 34,
265276.
Martin, E., Sharp, R., 1983. Soil protozoa from a cedar glade in Rutherford County, Tennes-
see. J. Tennessee Acad. Sci. 58, 3136.
Mearns, L.O., Giorgi, F., Mcdaniel, L., Shields, C., 2003. Climate scenarios for the southeast-
ern U.S. based on GCM and regional model simulations. Climacteric 60, 735.
Molano-Flores, B., Bell, T., 2012. Projected population dynamics for a federally endan-
gered plant under different climate change emission scenarios. Biol. Conserv. 145,
130138.
Munson, S., Benton, T., Lauenroth, W., Burke, I., 2009. Soil carbon ux following pulse pre-
cipitation events in the shortgrass steppe. Ecol. Res. 25, 205211.
Nordman, C., 2004. Vascular Plant Community Classication for Stones River National Bat-
tleeld. NatureServe, Durham, NC.
Noss, R., LaRoe, E., Scott, J., 1995. Endangered Ecosystems of the United States: A Prelim-
inary Assessment of Loss and Degradation. U.S. Department of the Interior, Washing-
ton, DC.
Preston-Mafham, J., Boddy, L., Randerson, P.F., 2002. Analysis of microbial community
functional diversity using sole-carbon-source utilisation proles - a critique. FEMS
Microbiol. Ecol. 42, 114.
Quarterman, E., 1950a. Ecology of cedar glades. I. Distribution of glade ora in Tennessee.
Bull. Torrey Bot. Club 77, 19.
Quarterman, E., 1950b. Major plant communities of Tennessee cedar glades. Ecology 31,
234254.
Quarterman, E., Burbanck, M., Shure, D., 1993. Rock outcrop communities: limestone,
sandstone, and granite. In: Martin, W., Boyce, S., Echternacht, A. (Eds.), Biodiversity
of the Southeastern United States: Upland Terrestrial Communities. John Wiley &
Sons, Inc., New York, pp. 3586.
Rastogi, G., Sani, R., 2011. Molecular techniques to assess microbial community structure
function and dynamics in the environment. In: Ahmad, I., Ahmad, F., Pichtel, J. (Eds.),
Microbes and Microbial Technology: Agricultural and Environmental Applications.
Springer, New York, pp. 2957.
Reynolds, H.L., Packer, A., Bever, J.D., Clay, K., 2003. Grassroots ecology: plantmicrobe
soil interactions as drivers of plant community structure and dynamics. Ecology 84,
22812291.
SAS Institute, 2011. SAS/STAT 9.3 User's Guide. SAS Institute Inc., Cary, NC (8621 p).
Schmidt, T.S., Clements, W.H., Cade, B.S., 2012. Estimating risks to aquatic life using
quantile regression. Freshw. Sci. 31, 709723.
Somers, P., Smith, L., Hamel, P., Bridges, E., 1986. Preliminary analyses of plant communi-
ties and seasonal changes in cedar glades of Middle Tennessee. ASB Bull. 33, 178192.
Stark, J.M., Firestone, M.K., 1995. Mechanisms for soil moisture effects on activity of nitri-
fying bacteria. Appl. Environ. Microbiol. 61, 218221.
Sutton, S., 2010. The most probable number method and its uses in enumeration, quali-
cation, and validation. J. Valid. Technol. 16, 3538.
Talmon, Y., Sternberg, M., Grnzweig, J., 2011. Impact of rainfall manipulations and biotic
controls on soil respiration in Mediterranean and desert ecosystems along an aridity
gradient. Glob. Chang. Biol. 17, 11081118.
Taylor, K., Estes, D., 2012. The oristic and community ecology of seasonally wet lime-
stone glade seeps of Tennessee and Kentucky. J. Bot. Res. Inst. Texas 6, 711724.
Thomson, B., Ostle, N., McNamara, N., Bailey, M.J., Whiteley, A.S., Grifths, R.I., 2010. Veg-
etation affects the relative abundances of dominant soil bacterial taxa and soil respi-
ration rates in an upland grassland soil. Microb. Ecol. 59, 335343.
Thornberry-Ehrlich, T., 2012. Stones River National Battleeld: Geological Resources In-
ventory Report. Natural Resource Report NPS/NRSS/GRD/NRR, Fort Collins, CO.
Torsvik, V., vres, L., 2002. Microbial diversity and function in soil: from genes to ecosys-
tems. Curr. Opin. Microbiol. 5, 240245.
U.S. Department of Agriculture Soil Conservation Service, 1977. Soil survey of Rutherford
County, Tennessee.
USDA, 2015. Ecological subregions: sections and subsections for the Conterminous United
States. Vector Digital Data Adapted from Fenneman and Johnson (1946) and
ECOMAP (2007). Forest Service, U.S. Department of Agriculture, Washington, DC.
Van der Heijden, M.G., Bardgett, R.D., van Straalen, N.M., 2008. The unseen majority: soil
microbes as drivers of plant diversity and productivity in terrestrial ecosystems. Ecol.
Lett. 11, 296310.
Van der Putten, W.H., 2010. Impacts of soil microbial communities on exotic plant inva-
sions. Trends Ecol. Evol. 25, 512519.
Wardle, D.A., Bardgett, R.D., Klironomos, J.N., Setl, H., van der Putten, W.H., Wall, D.H.,
2004. Ecological linkages between aboveground and belowground biota. Science
304, 16291633.
Ye, R., Wright, A.L., 2010. Multivariate analysis of chemical and microbial properties in
histosols as inuenced by land-use types. Soil Tillage Res. 110, 94100.
Zak, D.R., Holmes, W.E., White, D.C., Peacock, A.D., Tilman, D., 2003. Plant diversity, soil
microbial communities, and ecosystem function: are there any links? Ecology 84,
20422050.
Zak, J., Willig, M., Moorhead, D., Wildman, H., 1994. Functional diversity of microbial com-
munities: a quantitative approach. Soil Biol. Biochem. 26, 11011108.
Zogg, G.P., Zak, D.R., Ringelberg, D.B., White, D.C., MacDonald, N.W., Pregitzer, K.S., 1997.
Compositional and functional shifts in microbial communities due to soil warming.
Soil Sci. Soc. Am. J. 61, 475.