Package {GreenManureR}

GreenManureR: Green Manure Decomposition, Nutrient Release and Crop Response

Description

Tools for green manure and cover crop experiments (Mucuna, sunhemp, dhaincha and others): litterbag decomposition models, nutrient release, nitrogen balance, residual (carry-over) effects on succeeding crops, a joint analysis linking the three, and standardized reporting.

Main functions

• decomp_fit() fits single-pool, double-pool and asymptotic decay models.

• decomp_t50() estimates the time to 50% (or any other %) mass loss.

• nutrient_release() quantifies N, P, K and other nutrient release.

• greenmanure_n_balance() computes N inputs, crop uptake and balances.

• carryover_effect() quantifies residual effects on kharif and rabi crops.

• gm_joint_analysis() links decomposition, release and crop response.

• greenmanure_report() writes standardized tables, figures and a report.

Example data sets (simulated) are provided as gm_litterbag and gm_agronomy; see also simulate_greenmanure().

Author(s)

Maintainer: Anbukkani maintainer@example.org [copyright holder]

Residual (carry-over) effects of green manure on succeeding crops

Description

Quantifies the effect of each green manure treatment relative to a control on the yield (or any response) of successive crops, for example a kharif crop followed by a rabi crop.

Usage

carryover_effect(
  data,
  response = "grain_yield",
  treatment = "treatment",
  control = "Control",
  season = "season",
  block = "block",
  strata = NULL,
  season_levels = NULL,
  adjust = "holm",
  level = 0.95
)

Arguments

Details

For every stratum and season the model response ~ block + treatment is fitted (randomized complete block design; without a block term if block = NULL). Treatment-versus-control contrasts use the model estimates and standard errors, with p-values adjusted across treatments by adjust. The percentage change is the contrast divided by the control mean; the log response ratio uses the pooled residual variance in a delta-method standard error. retention_vs_first expresses each seasonal effect as a percentage of the effect in the first season.

The pooled analysis fits response ~ season:block + treatment + treatment:season across seasons; a significant treatment x season term indicates that the effect changes between crops. Seasonal variances are assumed homogeneous, which should be checked. The cumulative effect is the sum of seasonal contrasts, with a standard error that assumes independence between seasons and is therefore approximate.

Value

An object of class "gm_carryover": a list with effects (contrasts per stratum, season and treatment), anova (treatment F-tests within season), pooled (pooled analysis, or NULL with one season), cumulative, response and control.

See Also

greenmanure_n_balance(), gm_joint_analysis()

Examples

co <- carryover_effect(gm_agronomy, response = "grain_yield",
                       strata = "fert_N",
                       season_levels = c("Kharif", "Rabi"))
co

Fit litterbag decomposition models

Description

Fits single-pool, double-pool and asymptotic decay models to litterbag mass-loss data, separately for each group (for example each green manure species), and ranks the models by AICc.

Usage

decomp_fit(
  data,
  time = "time",
  mass = "mass",
  group = NULL,
  models = c("single", "double", "asymptotic"),
  mass_scale = c("percent", "fraction")
)

Arguments

Details

With M the percentage of the initial mass remaining and t the time since incorporation, the models are:

• single: M = 100 * exp(-k * t)

• double: M = 100 * (A * exp(-kf * t) + (1 - A) * exp(-ks * t)), fitted with ks = r * kf and ⁠0 < r <= 1⁠, so the fast pool can never decay more slowly than the slow pool.

• asymptotic: M = C + (100 - C) * exp(-k * t), where C is the recalcitrant residue (percent of initial mass).

The initial mass is fixed at 100%, so time-zero observations should be included (or the data expressed relative to the initial mass). Each model is fitted by nonlinear least squares (port algorithm with bounds) from a small grid of starting values, and the best-fitting converged solution is kept.

Value

An object of class "gm_decomp", a list with components table (one row per group and model with parameters k, k_slow, pool_A, C_resistant, fit statistics RSS, RMSE, R2, AIC, AICc, dAICc, Akaike weight, and selected marking the lowest AICc), fits (the nls objects), data, gcols and groups. For the double-pool model k is the fast-pool rate and k_slow the slow-pool rate.

See Also

decomp_t50(), nutrient_release()

Examples

fit <- decomp_fit(gm_litterbag, time = "time", mass = "mass",
                  group = "treatment")
fit
fit$table[fit$table$selected, c("treatment", "model", "k", "R2")]

Time to a given percentage of mass (or nutrient) loss

Description

Solves the fitted decay model for the time at which a proportion p of the initial mass has been lost (p = 0.5 gives t50), with an optional Monte-Carlo confidence interval.

Usage

decomp_t50(
  fit,
  p = 0.5,
  model = "best",
  ci = TRUE,
  nsim = 2000,
  level = 0.95,
  seed = 123
)

Arguments

Details

The interval is obtained by drawing nsim parameter vectors from a multivariate normal distribution centred on the estimates with the covariance matrix of the nls fit, discarding draws outside the admissible parameter space, and taking quantiles of the resulting times. If the model never reaches the target (for example an asymptotic model with a residue above the target), NA is returned.

Value

A data frame with the grouping columns, the model used, p, the time (column named e.g. t50) and the interval limits lwr and upr.

See Also

decomp_fit()

Examples

fit <- decomp_fit(gm_litterbag, group = "treatment")
decomp_t50(fit)
decomp_t50(fit, p = 0.95, ci = FALSE)

Simulated two-season agronomic data

Description

Simulated field data for a control and three green manures at three fertilizer N levels, in four blocks, for a kharif crop followed by a rabi crop. Generated with simulate_greenmanure() (seed 42); the values are synthetic.

Usage

gm_agronomy

Format

A data frame with 96 rows and 10 variables:

treatment

Control or green manure species.

fert_N

Fertilizer N applied (kg/ha).

block

Block (1-4).

season

Crop season (Kharif, Rabi).

gm_biomass

Green manure dry matter (t/ha).

gm_N

Green manure N concentration (%).

grain_yield, straw_yield

Yields (t/ha).

grain_N, straw_N

N concentrations (%).

Source

Simulated; see data-raw/simulate-data.R.

See Also

simulate_greenmanure(), greenmanure_n_balance(), carryover_effect()

Joint analysis of decomposition, nutrient release and crop response

Description

Builds one row per treatment that joins the decomposition rate and t50, the cumulative release of one nutrient by a chosen day, apparent N recovery and the percentage yield response in each season, and reports Spearman rank correlations among these variables.

Usage

gm_joint_analysis(
  decomp,
  nutrient,
  nbal,
  carry,
  key = "treatment",
  nutrient_name = "N",
  at_day = 60,
  anr_season = NULL,
  fert_N_level = NULL
)

Arguments

Details

With only a few treatments (for example three green manure species) the correlations are descriptive only, and a warning is issued when fewer than five rows are available. To draw inference, pool several sites or years.

Value

An object of class "gm_joint": a list with data (the linkage table) and cor (Spearman correlation matrix, or NULL).

See Also

decomp_fit(), nutrient_release(), greenmanure_n_balance(), carryover_effect()

Examples

fit <- decomp_fit(gm_litterbag, group = "treatment")
nr  <- nutrient_release(gm_litterbag, group = "treatment",
                        biomass = "biomass_tha")
nb  <- greenmanure_n_balance(gm_agronomy, by = "season")
co  <- carryover_effect(gm_agronomy, strata = "fert_N")
ja  <- suppressWarnings(
  gm_joint_analysis(fit, nr, nb, co, at_day = 60, fert_N_level = 0))
ja$data

Simulated litterbag decomposition data

Description

Simulated litterbag data for three green manure species, with mass and nutrient concentrations over 112 days. Generated with simulate_greenmanure() (seed 42); the values are synthetic.

Usage

gm_litterbag

Format

A data frame with 108 rows and 8 variables:

treatment

Green manure species (Mucuna, Sunhemp, Dhaincha).

rep

Litterbag replicate (1-4).

time

Days after incorporation (0-112).

mass

Dry mass remaining (% of initial).

N, P, K

Nutrient concentrations (% dry weight).

biomass_tha

Green manure dry matter incorporated (t/ha).

Source

Simulated; see data-raw/simulate-data.R.

See Also

simulate_greenmanure(), decomp_fit(), nutrient_release()

Methods for GreenManureR objects

Description

Print, prediction and plotting methods for the objects returned by decomp_fit(), nutrient_release(), greenmanure_n_balance(), carryover_effect() and gm_joint_analysis(). Plot methods return ggplot objects and require the 'ggplot2' package.

Usage

## S3 method for class 'gm_decomp'
print(x, digits = 4, ...)

## S3 method for class 'gm_decomp'
predict(object, times, model = "best", ...)

## S3 method for class 'gm_nutrient'
print(x, digits = 3, ...)

## S3 method for class 'gm_nbalance'
print(x, digits = 3, ...)

## S3 method for class 'gm_carryover'
print(x, digits = 3, ...)

## S3 method for class 'gm_joint'
print(x, digits = 3, ...)

## S3 method for class 'gm_decomp'
plot(x, y, model = "best", ...)

## S3 method for class 'gm_nutrient'
plot(x, y, ...)

## S3 method for class 'gm_nbalance'
plot(x, y, ...)

## S3 method for class 'gm_carryover'
plot(x, y, ...)

Arguments

Value

print methods return x invisibly. predict.gm_decomp() returns a data frame with the grouping columns, model, time and pred (predicted mass remaining, %). plot methods return a ggplot object.

Examples

fit <- decomp_fit(gm_litterbag, group = "treatment")
predict(fit, times = c(30, 60, 90))
if (requireNamespace("ggplot2", quietly = TRUE)) plot(fit)

Nitrogen inputs, crop uptake and apparent balance

Description

Plot-level nitrogen accounting for green manure experiments, with treatment summaries.

Usage

greenmanure_n_balance(
  data,
  treatment = "treatment",
  control = "Control",
  rep = "block",
  by = NULL,
  gm_biomass = "gm_biomass",
  gm_N = "gm_N",
  fert_N = "fert_N",
  grain_yield = "grain_yield",
  grain_N = "grain_N",
  straw_yield = "straw_yield",
  straw_N = "straw_N",
  nfrv = TRUE
)

Arguments

Details

All quantities are in kg N/ha unless stated otherwise:

• N_gm_input: green manure dry matter (t/ha) x 1000 x N percent / 100.

• N_uptake: grain plus straw N removal.

• N_balance: (green manure N + fertilizer N) minus crop N uptake, an apparent balance (losses and soil N are not measured).

• ANR: apparent N recovery from green manure (percent): (uptake - mean control uptake at the same stratum and fertilizer N) / green manure N input x 100.

• AE_gmN: additional grain (kg) per kg of green manure N.

• NFRV: fertilizer N replacement value, the fertilizer N needed in the control plots (quadratic response fitted per by stratum, requiring at least three fertilizer levels) to match the green manure plot yield, minus the fertilizer N applied. NA when the yield lies outside the fitted response range.

Value

An object of class "gm_nbalance": a list with plot (plot-level results), summary (means, standard deviations and standard errors per stratum, treatment and fertilizer level), by and treatment.

See Also

carryover_effect()

Examples

nb <- greenmanure_n_balance(gm_agronomy, by = "season")
nb
head(nb$summary[, c("season", "treatment", "fert_N", "ANR_mean")])

Standardized tables, diagnostic plots and Word report

Description

Writes tables (CSV), figures (PNG) and, if the 'officer' package is installed, a Word report containing all tables and figures for any combination of analysis objects.

Usage

greenmanure_report(
  decomp = NULL,
  nutrient = NULL,
  nbal = NULL,
  carry = NULL,
  joint = NULL,
  outdir,
  docx = TRUE,
  dpi = 300,
  digits = 3
)

Arguments

Details

Nothing is written unless outdir is supplied; the directory is created if needed. Figures require 'ggplot2' and the Word report requires 'officer'; when either is missing the corresponding output is skipped with a message and the remaining outputs are still produced.

Value

Invisibly, a list with tables (data frames) and figures (file paths).

Examples

fit <- decomp_fit(gm_litterbag, group = "treatment")
out <- file.path(tempdir(), "gm_report")
greenmanure_report(decomp = fit, outdir = out, docx = FALSE)
list.files(out)

Nutrient release from decomposing green manure residues

Description

Calculates the percentage of the initial nutrient content remaining in litterbags, fits decay models to each nutrient and reports the time to 50% release and model-based cumulative release at chosen times.

Usage

nutrient_release(
  data,
  time = "time",
  mass = "mass",
  nutrients = c("N", "P", "K"),
  group = NULL,
  biomass = NULL,
  at = c(7, 14, 30, 60, 90),
  models = c("single", "double", "asymptotic")
)

Arguments

Details

Nutrient remaining (% of initial content) is ⁠mass * concentration / initial concentration⁠, where mass is the percentage of the initial dry mass remaining and the initial concentration is the group mean at the earliest sampling time (which must therefore be present in data). Values above 100% indicate net immobilization. Decay models are fitted with decomp_fit() to the nutrient-remaining percentages.

Value

An object of class "gm_nutrient": a list with data (long table with remaining, released and optionally input_kg_ha, released_kg_ha), fits (a gm_decomp object per nutrient), t50, release_at, gcols and at.

See Also

decomp_fit(), decomp_t50()

Examples

nr <- nutrient_release(gm_litterbag, nutrients = c("N", "P", "K"),
                       group = "treatment", biomass = "biomass_tha",
                       at = c(14, 30, 60))
nr
subset(nr$release_at, nutrient == "N")

Simulate litterbag and two-season agronomic data

Description

Generates synthetic data with the structure expected by the package functions: a litterbag decomposition experiment with three green manure species (Mucuna, sunhemp, dhaincha) and a replicated field experiment with a control, three fertilizer N levels and two successive crops (kharif and rabi). The values are simulated for testing and teaching and do not represent real measurements.

Usage

simulate_greenmanure(seed = 42)

Arguments

Value

A list with data frames litterbag (see gm_litterbag) and agronomy (see gm_agronomy).

Examples

sim <- simulate_greenmanure(seed = 1)
str(sim$litterbag)