Fire in longleaf pine

Microbial drivers of grass/shrub dynamics in longleaf pine savanna

Anita Simha & Gaurav Kandlikar
slides: https://talks.gklab.org/iite-25

Natural history of the longleaf pine ecosystem

Historic Range

A fire-maintained biodiversity hotspot

image: Outland III 2004. Tapping the pines. LSU Press.

“The beautiful park-like slopes of the pine hills are being converted into a smoking desert of pine trunks on whose blackened soil the cattle seek more vainly every year the few scattered sickly blades of grass whose roots the fire has not killed”

-1850 Mississippi Geological Surveyor

Johnson and Hale, 2002. USDA Northern Research Station 11-23.

3% of the original ecosystem extent remains

Emerging threat: woody encroachment

Soil microbes as potential contributors

Code

# curl::curl_download("https://datadryad.org/downloads/file_stream/2789266", destfile = "../data/jiang2024-dataset.csv")

jiang_dat <- read.csv("../data/jiang2024-dataset.csv")



# jiang_metamod  <- metafor::rma.mv(rr, var, data = jiang_dat,   mods = ~ Life.form-1)
# saveRDS(jiang_metamod, "../data/jiang-metamod.rds")

metamod <- readRDS("../data/jiang-metamod.rds")

metamod_s <- broom::tidy(metamod)

jiang_dat |> 
  ggplot(aes(x = rr, y = Life.form, color = Life.form)) +
  ggbeeswarm::geom_quasirandom(shape = 21) +
  scale_color_manual(values = c("transparent","transparent")) + 
  # annotate("point",
  #          x = metamod_s$estimate[1], y = 1, size = 3, shape = 21, stroke = 1.5) +
  # annotate("point",
  #          x = metamod_s$estimate[2], y = 2, size = 3, shape = 21, stroke = 1.5) +
  ylab("") + 
  xlab("Growth in self-conditioned microbes vs. nonself-conditioned microbes") +
    xlim(-2.5,2.5)+
  geom_vline(xintercept = 0, linetype = 'dashed') + 
  theme(legend.position = "none",
        axis.text.y = element_text(size = 12, color = 'black'))

Data from Jiang et al. (2024)

Soil microbes as potential contributors

Code

jiang_dat |> 
  ggplot(aes(x = rr, y = Life.form, color = Life.form)) +
  ggbeeswarm::geom_quasirandom(shape = 21) +
  scale_color_manual(values = c("#ccbb44","#228833")) + 
  annotate("point",
           x = metamod_s$estimate[1], y = 1, size = 3, shape = 21, stroke = 1.5) +
  annotate("point",
           x = metamod_s$estimate[2], y = 2, size = 3, shape = 21, stroke = 1.5) +
  ylab("") + 
  xlab("Growth in self-conditioned microbes vs. nonself-conditioned microbes") +
    xlim(-2.5,2.5)+
  geom_vline(xintercept = 0, linetype = 'dashed') + 
  theme(legend.position = "none",
        axis.text.y = element_text(size = 12, color = 'black'))

Data from Jiang et al. (2024)

Common shrubs in longleaf pine understories

Common shrubs in longleaf pine understories

Farkleberry
(*Vaccinium arboreum*)
Associates with ericoid mycorrhizal fungi

Most herbaceous plants here associate with arbuscular mycorrhizae

Soil feedbacks vary by mycorrhizal type

Code

# download.file("https://www.science.org/doi/suppl/10.1126/science.aai8212/suppl_file/bennett_aai8212_database-s1.xlsx", destfile = "../data/bennett2017-dataset.xlsx")
readxl::read_xlsx("../data/bennett-2017.xlsx", sheet = 2) |>
  mutate(`Type of Mycorrhiza` = factor(`Type of Mycorrhiza`, c("EM", "AM"))) |> 
  mutate(logratio = log(`Average of Biomass in Conspecific Soil`/`Average of Biomass in Heterospecific Soil`)) |> 
  arrange(logratio) |> 
  mutate(rown = row_number()) |> 
  ggplot(aes(x = logratio, color = `Type of Mycorrhiza`, y = rown, shape = `Type of Mycorrhiza`)) + 
  geom_point(size = 2) + 
  geom_segment(aes(x = 0, xend = logratio), linewidth = 0.125) +  
  xlab("Growth in self-conditioned microbes vs.\n nonself-conditioned microbes") +

  geom_vline(xintercept = 0) + 
  scale_color_manual(values = c("#0077bb", "#cc3311")) +  
  theme(axis.line.y = element_blank(), 
        axis.text.y = element_blank(), 
        axis.title.y = element_blank(), 
        axis.ticks.y = element_blank(),
        legend.position = "inside",
        legend.position.inside = c(0.9, 0.1),
        legend.text = element_text(size = 12),
        legend.title = element_blank())

Data from Bennett et al. (2017)

Motivation

Framework for projecting long-term dynamics of microbially mediated grass-shrub interactions in
fire-prone systems

Model overview

Patch occupancy, with each patch defined by its current plant state and current soil state: \(p_{\text{plant}, \text{soil}}\)

One of three plant states (empty, grass, shrub)
One of three soil states (baseline, grass-conditioned, shrub-conditioned)
Nine patch types:

\[ \text{Empty}~~\text{Grass}~~\text{Shrub} ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\\ P_{00},~~~~~P_{g0},~~~~~P_{s0}, ~~~~ \text{Baseline microbes}\\ ~~~P_{0g},~~~~~P_{gg},~~~~~P_{sg}, ~~~~ \text{Grass-conditioned}~~~\\ ~~P_{0s},~~~~~P_{ss},~~~~~P_{gs}~~~~~~ \text{Shrub-conditioned}~~ \]

Conditioned microbes modify establishment rates of shrubs/grasses (red arrows; symbolized \(m_{ij}\))

Model equations

Patches without plants or soil conditioning

\[\frac{dP_{00}}{dt} = \overbrace{\mu_g P_{g0} +\mu_s P_{s0}}^{\substack{\text{ plant mortality}}\\{\text{in unconditioned soil}}} + \overbrace{d_g P_{0g} + d_s P_{0s}}^{\substack{\text{microbial decay in} \\ \text{empty-but-conditioned patches} }} - \\ \overbrace{r_gP_{00}(P_{{g0}+}P_{gg}+P_{gs})}^{\text{grass establishment}} - \\ \overbrace{r_sP_{00}(P_{{s0}+}P_{ss}+P_{sg})}^{\text{shrub establishment}} \]

Patches with plant present but soil not yet conditioned

\[\frac{dP_{g0}}{dt} = \overbrace{r_gP_{00}(P_{g0} + P_{gg} + P_{gs})}^{\substack{\text{grass establishment in} \\ \text{unconditioned soil}}} - \overbrace{c_g P_{g0}}^{\substack{\text{soil conditioning}\\\text{by grass}}} - \overbrace{\mu_g P_{g0}}^{\text{grass mortality}}\]

\[\frac{dP_{s0}}{dt} = \overbrace{r_sP_{00}(P_{s0} + P_{ss} + P_{sg})}^{\substack{\text{shrub establishment in } \\ \text{empty patch}}} - \overbrace{c_sP_{s0}}^{\substack{\text{soil conditioning}\\\text{by shrub}}} - \overbrace{\mu_sP_{s0}}^{\text{shrub mortality}}\]

Patches with plants and corresponding soil conditioning effects

\[\frac{dP_{gg}}{dt} = \overbrace{c_g (P_{g0} + P_{gs})}^{\text{soil conditioning by grass}} + \overbrace{r_g m_{gg} P_{0g}(P_{g0} + P_{gg} + P_{gs})}^{\text{grass establishment in empty grass-conditioned patch}} - \overbrace{\mu_g P_{gg}}^{\text{grass mortality}}\]

\[\frac{dP_{ss}}{dt} = \overbrace{c_s (P_{s0} + P_{sg})}^{\text{soil conditioning by shrub}} + \overbrace{r_s m_{ss} P_{0S}(P_{s0} + P_{ss} + P_{sg})}^{\text{shrub establishment in empty shrub-conditioned patch}} - \overbrace{\mu_s P_{ss}}^{\text{shrub mortality}}\]

Patches with no plant present but conditioned soil persists

\[\frac{dP_{0g}}{dt} = \overbrace{\mu_g P_{gg}}^{\text{grass mortality}} + \overbrace{\mu_s P_{sg}}^{\text{shrub mortality}} - \overbrace{r_g m_{gg} P_{0g}(P_{g0}+P_{gg}+P_{gs})}^{\text{grass establishment}} - \overbrace{r_s m_{sg} P_{0g}(P_{s0}+P_{ss}+P_{sg})}^{\text{shrub establishment}}\]

\[\frac{dP_{0s}}{dt} = \overbrace{\mu_s P_{ss}}^{\text{shrub mortality}} + \overbrace{\mu_g P_{gs}}^{\text{grass mortality}} - \overbrace{r_s m_{ss} P_{0s}(P_{s0}+P_{ss}+P_{sg})}^{\text{shrub establishment}} - \overbrace{r_g m_{gs} P_{0s}(P_{s0}+P_{ss}+P_{sg})}^{\text{grass establishment}}\]

Patches with plants growing in the other plant’s conditioned soil

\[\frac{dP_{gs}}{dt} = \overbrace{r_g m_{gs} P_{0s} (P_{g0} + P_{gg} + P_{gs})}^{\text{grass establishment into shrub-conditioned soil}} - \overbrace{\mu_g P_{gs}}^{\text{grass mortality}}\]

\[\frac{dP_{sg}}{dt} = \overbrace{r_s m_{sg} P_{0g} (P_{s0} + P_{ss} + P_{sg})}^{\text{shrub establishment into grass-conditioned soil}} - \overbrace{\mu_s P_{sg}}^{\text{shrub mortality}}\]

Assumptions for the following analyses

Grasses have higher intrinsic growth than shrubs \((r_g = 1, r_s < 0)\)

Microbial effects are strictly host-specific \((m_{gs} = m_{sg} = 1)\)
Dynamics summarized in terms of \(p_{\text{grass}} = p_{g0}+p_{gg}+p_{gs}\) and
\(p_{\text{shrub}} = p_{s0}+p_{ss}+p_{sg}\)

Model dynamics

Neutral microbes: grass wins due to \(r_g>r_s\)

Model dynamics

Microbially generated self-limitation in grass and shrubs can enable coexistence

Model dynamics

Microbes can help shrubs overcome grass’ intrinsic growth advantage

See Ke and Levine (2021) for a complete analysis

Impacts of fire

Characterizing the effects of fire

For now, assume that the severity of a fire (\(0<S<1\)) determines its impacts on plant mortality and soil conditioning
- e.g. \(p_{gg, \text{post-fire}} = (1-S)*p_{\text{gg, pre-fire}}\)
Goal is to develop a trait-based framework for fire effects on grass and shrub demography, and on soil microbes
In following analyses, fires are of low severity (\(S = 0.2\))

Research questions

How does variation in fire frequency affect system dynamics under demographic asymmetries?
How does variation in temporal properties of plant-microbe interactions affect vegetation dynamics under fire?
Under what conditions can fire “rescue” savannas that have undergone woody encroachment?

Q1: Fire effects under demographic asymmetry

Question: How does variation fire frequency shape dynamics under different asymmetries in intrinsic growth and self-limitation?

Each cell: mean grass cover across final 100 timesteps (5000 total)

Q1: Fire effects under demographic asymmetry

Question: How does variation fire frequency shape dynamics under different asymmetries in intrinsic growth and self-limitation?

Implication: Fire can maintain grassy savanna across a wide parameter space even when microbes generate self-limitation in grasses

Logic: microbial legacies build up over time, but frequent fire can erase this — thereby reducing self-limitation

Q2: Temporal dynamics of plant–microbe interaction

Question: Assuming grasses cultivate self-limiting microbes, how do the temporal dynamics of plant-microbe interactions affect vegetation dynamics?

Simulations assume self-limitation in grass, self-facilitation in shrub (\(m_{gg} = 0.9, m_{gg} = 1.1\))

Q2: Temporal dynamics of plant–microbe interaction

Question: Assuming grasses cultivate self-limiting microbes, how do the temporal dynamics of plant-microbe interactions affect vegetation dynamics?

Implication: Outcomes of microbially mediated vegetation dynamics depend on rates of microbial conditioning and decay.

Very little empirical knowledge — need to measure! (Ke et al. 2024)

Q3: Can fires “rescue” shrubified savannas?

Question: Assuming that a system has already undergone shrubification, under what conditions can fire “restore” the grassy understory?

Q3: Can fires “rescue” shrubified savannas?

Question: Assuming that a system has already undergone shrubification, under what conditions can fire “restore” the grassy understory?

Q3: Can fires “rescue” shrubified savannas?

Question: Assuming that a system has already undergone shrubification, under what conditions can fire “restore” the grassy understory?

Q3: Can fires “rescue” shrubified savannas?

Question: Assuming that a system has already undergone shrubification, under what conditions can fire “restore” the grassy understory?

Implication: Variation in strength of shrub self-facilitation could explain failure of fire alone to restore grassy understories

Next steps: modeling

Trait based model for variation in fire response (and extend to multiple grass and shrub species)
Stage structure?

Next steps: empirical data

Quantify \(m_{ij}\)s and their dependence on fire with a factorial experiment

Next steps: empirical data

Quantify direct effects of fire on microbes through a fuel manipulation experiment

LSU AgCenter’s Lee Memorial Forest, FranklintonLA

Video Player is loading.

Current Time 0:00

Duration 0:25

Loaded: 0.00%

Stream Type LIVE

Remaining Time 0:25

Cleared all litter from several 5x5m patches of longleaf pine understory (~120kg of litter, mostly pine needles)

Redistributed into the forest to create patches of low, medium, and high standing litter (fuel) to manipulate fire intensity

Video: Edmarie Rivera Sanchez

How does burn intensity affect biotic and abiotic soil properties?
What are the dynamics of recovery?

Acknowledgements

Ameya Kherde (IISER Pune, India)
Joe Nehlig (Lee Memorial Forest, LSU)
Chris Reid (LSU)

Get in touch: gkandlikar@lsu.edu / asimha@lsu.edu

Acknowledgements

Get in touch: gkandlikar@lsu.edu / asimha@lsu.edu

References

Bennett, Jonathan A, Hafiz Maherali, Kurt O Reinhart, Ylva Lekberg, Miranda M Hart, and John Klironomos. 2017. “Plant-Soil Feedbacks and Mycorrhizal Type Influence Temperate Forest Population Dynamics.” Science 355 (6321): 181–84.

Jiang, Feng, Jonathan A Bennett, Kerri M Crawford, Johannes Heinze, Xucai Pu, Ao Luo, and Zhiheng Wang. 2024. “Global Patterns and Drivers of Plant–Soil Microbe Interactions.” Ecology Letters 27 (1): e14364.

Ke, Po-Ju, Gaurav Kandlikar, Suzanne Xianran Ou, Gen-Chang Hsu, Joe Wan, and Meghna Krishnadas. 2024. “Time Will Tell: The Temporal and Demographic Contexts of Plant-Soil Microbe Interactions.”

Ke, Po-Ju, and Jonathan M Levine. 2021. “The Temporal Dimension of Plant-Soil Microbe Interactions: Mechanisms Promoting Feedback Between Generations.” The American Naturalist 198 (3): E80–94.