text
stringlengths 0
6.44k
|
---|
Figure 1A) and with increasing understory plant community evenness (F1,68 = 8.83, p = 0.004; Figure 1B),
|
suggesting the maintenance of higher water levels and even plant understories on tree islands could
|
improve fungal diversity and therefore the variety of ecosystem services they provide (as supported
|
by the functional guild analysis below). Fungal community richness also increased with increasing
|
canopy openness (F1,65= 9.41, p = 0.003; Figure 1C) as well as with increasing understory evenness
|
(F1,65 = 7.66, p = 0.007; Figure 1D).
|
Diversity 2020, 12, x FOR PEER REVIEW 6 of 17
|
supported by the functional guild analysis below). Fungal community richness also increased with
|
increasing canopy openness (F1,65= 9.41, p = 0.003; Figure 1C) as well as with increasing understory
|
evenness (F1,65 = 7.66, p = 0.007; Figure 1D).
|
Figure 1. Factors explaining fungal community diversity (A,B) and richness (C,D). The results from
|
the model selection that indicated fungal Shannon diversity were explained by the relative water level
|
(A; F1,68 = 8.63, p = 0.004) and plant evenness (B; F1,68 =8.83, p = 0.004), and fungal richness was explained
|
by canopy openness (C; F1,65 = 9.41, p = 0.003) and understory plant evenness (D; F1,65 = 7.66, p = 0.007).
|
In all panels, each point represents a site in the experimental tree islands, and lines are based on a
|
linear model relationship between these variables with the shaded area indicating the 95% confidence
|
intervals around the line. In (A), note that relative water level values equal to zero represent sites
|
where the water level was on average at the soil surface. Values greater than zero represent the sites
|
that were on average inundated, and values less than zero represent sites that on average had water
|
below the soil surface.
|
3.3. Fungal Community Composition
|
In addition to its importance in fungal diversity, relative water level also explained a significant
|
amount of variation in fungal community composition (db-RDA, F1,68 = 2.95, p = 0.001; Figure 2). The
|
core type, one of the manipulative restoration treatments implemented during the construction of the
|
experimental tree islands, affected fungal community composition as well (F1,68 = 1.87, p = 0.002). We
|
found that limestone cores resulted in a significantly different community composition than peat
|
Figure 1. Factors explaining fungal community diversity (A,B) and richness (C,D). The results from
|
the model selection that indicated fungal Shannon diversity were explained by the relative water level
|
(A; F1,68 = 8.63, p = 0.004) and plant evenness (B; F1,68 = 8.83, p = 0.004), and fungal richness was
|
explained by canopy openness (C; F1,65 = 9.41, p = 0.003) and understory plant evenness (D; F1,65 = 7.66,
|
p = 0.007). In all panels, each point represents a site in the experimental tree islands, and lines are
|
based on a linear model relationship between these variables with the shaded area indicating the 95%
|
confidence intervals around the line. In (A), note that relative water level values equal to zero represent
|
sites where the water level was on average at the soil surface. Values greater than zero represent the
|
sites that were on average inundated, and values less than zero represent sites that on average had
|
water below the soil surface.
|
Diversity 2020, 12, 0324 7 of 17
|
3.3. Fungal Community Composition
|
In addition to its importance in fungal diversity, relative water level also explained a significant
|
amount of variation in fungal community composition (db-RDA, F1,68 = 2.95, p = 0.001; Figure 2).
|
The core type, one of the manipulative restoration treatments implemented during the construction of
|
the experimental tree islands, affected fungal community composition as well (F1,68 = 1.87, p = 0.002).
|
We found that limestone cores resulted in a significantly different community composition than
|
peat cores on both of the first two axes of fungal community composition (CAPs 1 and 2; Figure 2),
|
with a 218% higher value on CAP 1 (F1,76 = 3.83, p = 0.054) and a 185% smaller value on CAP2
|
(F1,76 = 61.55, p = 2.17 × 10−11) for limestone compared to peat core islands. To gain insight into the
|
difference between the communities in peat and limestone core islands, we performed an indicator
|
taxa analysis. After correcting for multiple comparisons, we determined that Thelephoraceae,
|
Eurotiomycetes, and Inocybe curvipes were indicative of peat communities, while Agaricomycetes and
|
Archaeorhizomyces were indicative of limestone communities (Table S1). We also found that there was a
|
significant relationship between the inter-site variations in the fungal and understory plant community
|
composition (Mantel test: R = 0.11, p = 0.003).
|
Diversity 2020, 12, x FOR PEER REVIEW 7 of 17
|
cores on both of the first two axes of fungal community composition (CAPs 1 and 2; Figure 2), with a
|
218% higher value on CAP 1 (F1,76 = 3.83, p = 0.054) and a 185% smaller value on CAP2 (F1,76 = 61.55, p
|
= 2.17 × 10−11) for limestone compared to peat core islands. To gain insight into the difference between
|
the communities in peat and limestone core islands, we performed an indicator taxa analysis. After
|
correcting for multiple comparisons, we determined that Thelephoraceae, Eurotiomycetes, and
|
Inocybe curvipes were indicative of peat communities, while Agaricomycetes and Archaeorhizomyces
|
were indicative of limestone communities (Table S1). We also found that there was a significant
|
relationship between the inter-site variations in the fungal and understory plant community
|
composition (Mantel test: R = 0.11, p = 0.003).
|
Figure 2. Ordination demonstrating the relationship between fungal community composition and
|
environmental variables. Each point represents the fungal community composition at a site on the
|
experimental tree islands. Points are colored by relative water level (in meters) from high water (light
|
colors) to low water (dark colors). The graphs on either side of the ordination indicate the mean and
|
standard error of tree density and island core type along the first two axes of community composition
|
(CAP1 and CAP2). Different lowercase letters denote significant differences. The table (in upper right
|
corner) details correlations between continuous environmental variables.
|
3.4. Distribution of Fungal Functional Guilds
|
In total, we found 15 fungal functional guilds in our dataset with each of the eight islands hosting
|
an average of approximately 5 guilds (±0.22) (Figure S3A). The most common functional guilds were
|
ectomycorrhizal fungi, dung-wood saprotroph, and plant pathogen-wood saprotroph, which
|
occurredin63%77%and67%ofthesites(respectively)andonallislands(FigureS3A)WhenFigure 2. Ordination demonstrating the relationship between fungal community composition and
|
environmental variables. Each point represents the fungal community composition at a site on the
|
experimental tree islands. Points are colored by relative water level (in meters) from high water
|
(light colors) to low water (dark colors). The graphs on either side of the ordination indicate the
|
mean and standard error of tree density and island core type along the first two axes of community
|
composition (CAP1 and CAP2). Different lowercase letters denote significant differences. The table
|
(in upper right corner) details correlations between continuous environmental variables.
|
Diversity 2020, 12, 0324 8 of 17
|
3.4. Distribution of Fungal Functional Guilds
|
In total, we found 15 fungal functional guilds in our dataset with each of the eight islands
|
hosting an average of approximately 5 guilds (±0.22) (Figure S3A). The most common functional
|
guilds were ectomycorrhizal fungi, dung-wood saprotroph, and plant pathogen-wood saprotroph,
|
which occurred in 63%, 77%, and 67% of the sites (respectively) and on all islands (Figure S3A).
|
When examining factors that influence guild richness, we found that the number of functional guilds
|
among fungi increased in sites that had a more even understory plant community (F1,63 = 10.12,
|
p = 0.0023; Figure 3A) with an approximately 70% increase in guilds in the most even compared to
|
the least even five sites. Core type and relative water level were also included in the best model for
|
the number of functional guilds (core type: F1,63 = 2.92, p = 0.1378; relative water level: F1,63 = 0.87,
|
p = 0.3521), with an 18% greater number of guilds in peat compared to limestone core islands and an
|
approximately 8% decrease in the driest compared to wettest sites. When we examined the factors
|
that influence abundances of the three most common guilds from our dataset, the abundance of only
|
one—plant pathogen-wood saprotroph—could be explained by any of our measured variables. As the
|
relative water level decreased, the abundance of the plant pathogen–wood saprotroph guild increased
|
(F1,68 = 5.47, p = 0.022; Figure 3B).
|
Diversity 2020, 12, x FOR PEER REVIEW 8 of 17
|
fungi increased in sites that had a more even understory plant community (F1,63 = 10.12, p = 0.0023;
|
Figure 3A) with an approximately 70% increase in guilds in the most even compared to the least even
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.