Temperature Germination Profile of Celosia argentea

Exploratory Analysis Across 35 Accessions and Five Temperature Regimes

---

Background

This analysis explores the germination behaviour of 35 Celosia argentea accessions (coded Tca-1 to Tca-35) grown under five temperature regimes (15°C, 20°C, 25°C, 30°C, and 35°C). Seeds were sown in controlled-environment chambers at Aarhus University, Flakkebjerg Research Facility (February–April 2026), with germination counts recorded daily from Day 2 to Day 12 After Sowing (DAS).

Three germination metrics are computed and visualised:

• Germination Percentage (GP) — the proportion of seeds that germinated
• Mean Germination Time (MGT) — the average time (in days) to germination, weighted by daily germination counts
• Germination Speed Index (GI) — a speed-weighted sum of daily counts (higher values indicate faster germination)

Note on raw data: The dataset is not included in this repository. It was collected as part of an ongoing, unpublished research collaboration at Aarhus University and cannot be shared publicly without institutional clearance. The script is fully reproducible once a compatible dataset is supplied — see the Data Import section for the expected structure.

---

Setup

Packages

# pacman::p_load installs missing packages automatically before loading
pacman::p_load(tidyverse, readxl, tidytext, here)

Global Theme

A single theme_set() call applies a consistent visual style to all figures, avoiding repetition across individual plot objects.

theme_set(
  theme_bw(base_size = 13) +
    theme(
      axis.title            = element_text(size = 16),
      axis.text             = element_text(size = 14),
      plot.caption          = element_text(hjust = 0, size = 14,
                                           face = "bold", colour = "black"),
      plot.caption.position = "plot",
      legend.text           = element_text(size = 12),
      panel.grid            = element_blank()
    )
)

Colour Palette

A named vector maps each temperature level to a fixed colour. Passing this to scale_fill_manual() and scale_colour_manual() ensures colours are consistent across all eight figures.

temp_colours <- c(
  "15" = "#818cf8",   # cool violet
  "20" = "#60a5fa",   # blue
  "25" = "#4ade80",   # green
  "30" = "#f59e0b",   # amber
  "35" = "#f87171",   # coral
  "40" = "#b91c1c"    # deep red
)

Save Helper

A small wrapper around ggsave() reduces repetition and ensures all output files land in the outputs/ folder with consistent resolution and units.

save_plot <- function(filename, plot, h = 7, w = 12, format = "png") {
  ggsave(
    paste0(filename, ".", format), plot,
    path   = here("outputs"),
    height = h, width = w,
    units  = "in", dpi = 300,
    bg     = "white"
  )
}

---

Data Import and Transformation

Import

The dataset is expected as an Excel file with one row per accession-temperature combination and daily germination counts in columns named DAS2, DAS3, …, DAS12. Temperature 45°C is excluded from all analyses (insufficient germination at that extreme).

# Replace the path below with your own file location
celo <- read_xlsx(here("data/celosia_temperature_stress_data.xlsx"),
                  range = "A1:M246")

Reshape to Long Format

pivot_longer() converts the wide DAS columns into a tidy long format, with one row per accession-temperature-day observation. str_remove() strips the “DAS” prefix so das is numeric and usable on plot axes.

celo_new <- celo %>%
  mutate(
    Acc  = factor(Acc, levels = paste0("Tca- ", 1:35)),
    Temp = factor(Temp)
  ) %>%
  pivot_longer(
    cols      = matches("DAS"),
    names_to  = "das",
    values_to = "count"
  ) %>%
  mutate(das = as.numeric(str_remove(das, "DAS"))) %>%
  filter(Temp != 45)

Derived Datasets

Three summary datasets are computed once here and reused across all figures, avoiding redundant computation inside individual plot chunks.

# ── Germination Percentage (GP) ───────────────────────────────
# Maximum cumulative count divided by total seeds sown (25 per pot)
germination_perc <- celo_new %>%
  group_by(Acc, Temp) %>%
  reframe(
    max_count = max(count, na.rm = TRUE),
    gp        = (max_count / 25) * 100
  )

# ── Mean Germination Time (MGT) ───────────────────────────────
# Weighted mean of DAS, weighted by daily germination increments.
# pmax() floors negative increments (data entry artefacts) to zero.
# NA is returned for accession-temperature combinations with zero
# total germination, avoiding division by zero.
mgt <- celo_new %>%
  arrange(Acc, Temp, das) %>%
  group_by(Acc, Temp) %>%
  mutate(daily = pmax(count - lag(count, default = 0), 0)) %>%
  reframe(
    total_germination = sum(daily, na.rm = TRUE),
    mgt = if_else(
      total_germination > 0,
      sum((daily * das) / total_germination, na.rm = TRUE),
      NA_real_
    )
  )

# ── Germination Speed Index (GI) ─────────────────────────────
# Sum of daily germination counts divided by the DAS at which they
# occurred. Higher values indicate faster, earlier germination.
gi <- celo_new %>%
  arrange(Acc, Temp, das) %>%
  group_by(Acc, Temp) %>%
  mutate(daily = pmax(count - lag(count, default = 0), 0)) %>%
  reframe(gi = sum(daily / das, na.rm = TRUE))

---

Figures

Figure 1 — Germination Percentage Heatmap

The heatmap gives the most complete overview: every accession at every temperature in a single panel. Diverging fill (dark → blue → green) makes the 50% midpoint immediately readable. White text is overlaid for precise values.

fig1_heatmap <- ggplot(germination_perc,
                       aes(Temp, fct_rev(Acc), fill = gp)) +
  geom_tile(colour = "black", linewidth = 0.3) +
  geom_text(aes(label = round(gp, 1)), size = 4.5, colour = "white") +
  scale_fill_gradient2(
    low      = "#1a2332",
    mid      = "#2563eb",
    high     = "#4ade80",
    midpoint = 50,
    limits   = c(0, 100),
    name     = "Germination (%)"
  ) +
  labs(
    x       = "Temperature (\u00b0C)",
    y       = "Accession",
    caption = "Figure 1: Germination Percentage (%) of 35 Celosia argentea Accessions Across Five Temperatures"
  ) +
  theme(panel.border = element_blank())

fig1_heatmap

---

Figure 2 — GP Distribution by Temperature (Boxplot)

Moving from individual accessions to a population summary, this boxplot shows how GP varies across accessions at each temperature. Outliers are suppressed to keep the focus on the bulk distribution.

fig2_gp_boxplot <- ggplot(germination_perc,
                          aes(x = Temp, y = gp, fill = Temp)) +
  geom_boxplot(outliers = FALSE, show.legend = FALSE, width = 0.8) +
  scale_fill_manual(values = temp_colours) +
  labs(
    x       = "Temperature (°C)",
    y       = "Germination Percentage (%)",
    caption = "Figure 2: Distribution of Germination Percentage (%) Across Accessions by Temperature"
  )

fig2_gp_boxplot

save_plot("fig2_gp_boxplot", fig2_gp_boxplot)

---

Figure 3 — Mean Daily Germination by Temperature Over Time

Grouped bar charts show when germination peaks at each temperature. position = "dodge" separates temperatures within each DAS. Error bars are ±1 SE computed over all accessions. This bridges the GP summary (Fig 2) into the time-course dynamics shown in Fig 4.

fig3_daily_germination <- celo_new %>%
  arrange(Acc, Temp, das) %>%
  group_by(Acc, Temp) %>%
  mutate(daily = pmax(count - lag(count, default = 0), 0)) %>%
  group_by(Temp, das) %>%
  reframe(
    mean_daily = mean(daily, na.rm = TRUE),
    # NA-aware denominator: only non-missing observations contribute to SE
    se_daily   = sd(daily, na.rm = TRUE) / sqrt(sum(!is.na(daily)))
  ) %>%
  ggplot(aes(x = das, y = mean_daily, fill = Temp)) +
  geom_col(position = "dodge", colour = "black", linewidth = 0.2) +
  geom_errorbar(
    aes(ymin = mean_daily - se_daily, ymax = mean_daily + se_daily),
    position = position_dodge(0.9), width = 0.3
  ) +
  scale_fill_manual(values = temp_colours, name = "Temperature (°C)") +
  scale_x_continuous(breaks = seq(2, 12, 1)) +
  labs(
    x       = "Day After Sowing (DAS)",
    y       = "Mean Daily Germination (seeds)",
    caption = "Figure 3: Mean Daily Germination Count Across Accessions by Temperature"
  )

fig3_daily_germination

save_plot("fig3_daily_germination", fig3_daily_germination)

---

Figure 4 — Cumulative Germination Curves per Accession

Faceted by accession (7 columns × 5 rows), each panel shows the full time course of cumulative germination across temperatures. This is the most data-dense figure and is best read alongside Fig 1 to connect final GP values to the trajectories that produced them.

fig4_trend <- ggplot(celo_new, aes(das, count, colour = Temp)) +
  geom_line(linewidth = 1, na.rm = TRUE) +
  geom_point(size = 2, na.rm = TRUE) +
  facet_wrap(~Acc, ncol = 7) +
  scale_x_continuous(breaks = seq(2, 12, 2)) +
  scale_colour_manual(values = temp_colours, name = "Temperature (°C)") +
  theme(
    legend.position  = "bottom",
    strip.background = element_rect(fill = "grey92"),
    axis.title.x     = element_text(size = 12),
    axis.title.y     = element_text(size = 12),
    axis.text.x      = element_text(size = 9)
  ) +
  labs(
    x       = "Days After Sowing (DAS)",
    y       = "Cumulative Germinated Seeds",
    caption = "Figure 4: Cumulative Germination Curves for 35 Celosia argentea Accessions Across Five Temperatures"
  )

fig4_trend

save_plot("fig4_trend", fig4_trend, h = 12, w = 14, format = "tiff")

---

Figure 5 — MGT per Accession (Faceted Bar Chart)

Shifting from how many germinated to how fast, this figure mirrors the structure of Fig 1 at the accession level. Shorter bars indicate faster germination. Temperature 40°C is absent where germination was zero (MGT returns NA and is silently dropped by geom_col()).

fig5_mgt_accession <- ggplot(mgt, aes(Temp, mgt, fill = Temp)) +
  geom_col(na.rm = TRUE) +
  scale_fill_manual(values = temp_colours, name = "Temperature (°C)") +
  facet_wrap(~Acc, ncol = 7, scales = "free_x") +
  theme(
    legend.position  = "bottom",
    axis.text.y.left = element_text(size = 11.5)
  ) +
  labs(
    x       = "Temperature (°C)",
    y       = "Mean Germination Time (Days)",
    caption = "Figure 5: Mean Germination Time (MGT) of 35 Celosia argentea Accessions"
  )

fig5_mgt_accession

save_plot("fig5_mgt_accession", fig5_mgt_accession, h = 12, w = 14)

---

Figure 6 — MGT Distribution by Temperature (Violin + Boxplot)

The violin layer reveals the full distribution shape across accessions; the inner boxplot provides median and IQR. Together they show both the central tendency and spread of germination speed at each temperature, summarising Fig 5 in a single panel.

fig6_mgt_violin <- ggplot(mgt, aes(Temp, mgt, fill = Temp)) +
  geom_violin(width = 0.8, alpha = 0.4, show.legend = FALSE) +
  geom_boxplot(width = 0.2, show.legend = FALSE, outliers = FALSE) +
  scale_fill_manual(values = temp_colours) +
  labs(
    x       = "Temperature (°C)",
    y       = "Mean Germination Time (Days)",
    caption = "Figure 6: Distribution of Mean Germination Time (MGT) Across Accessions by Temperature"
  )

fig6_mgt_violin

save_plot("fig6_mgt_violin", fig6_mgt_violin)

---

Figure 7 — GI per Accession (Faceted Bar Chart)

The Germination Speed Index complements MGT: while MGT measures when germination occurs, GI weights seeds that germinate earlier more heavily, giving a single score that captures both speed and volume. This figure mirrors the structure of Fig 5 for direct comparison.

fig7_gi_accession <- ggplot(gi, aes(Temp, gi, fill = Temp)) +
  geom_col() +
  scale_fill_manual(values = temp_colours, name = "Temperature (°C)") +
  facet_wrap(~Acc, ncol = 7) +
  theme(legend.position = "bottom") +
  labs(
    x       = "Temperature (°C)",
    y       = "Germination Speed Index",
    caption = "Figure 7: Germination Speed Index (GI) of 35 Celosia argentea Accessions Across Five Temperatures"
  )

fig7_gi_accession

save_plot("fig7_gi_accession", fig7_gi_accession, h = 12, w = 14, format = "tiff")

---

Figure 8 — GI Distribution by Temperature (Boxplot)

The closing figure summarises GI across accessions per temperature, completing the GP → MGT → GI narrative arc. Higher GI and lower MGT at the same temperature confirm faster, more vigorous germination — the two metrics should be interpreted together.

fig8_gi_boxplot <- ggplot(gi, aes(Temp, gi, fill = Temp)) +
  geom_boxplot(width = 0.8, outliers = FALSE, show.legend = FALSE) +
  scale_fill_manual(values = temp_colours) +
  labs(
    x       = "Temperature (°C)",
    y       = "Germination Speed Index",
    caption = "Figure 8: Distribution of Germination Speed Index (GI) Across Accessions by Temperature"
  )

fig8_gi_boxplot

save_plot("fig8_gi_boxplot", fig8_gi_boxplot)

---

Summary

Table 1: Summary of Germination Percentage by Temperature

Temperature (°C)	Mean GP (%)	SD	Min (%)	Max (%)
15	2.2	3.9	0	12
20	18.2	11.8	0	44
25	24.7	19.3	0	88
30	32.0	19.0	0	80
35	65.7	22.6	4	100
40	48.8	22.0	4	76

Key findings:

• Optimal temperature for germination capacity (GP): 35°C
• Optimal temperature for germination speed (MGT / GI): 30°C
• Germination at 15°C was consistently low across all accessions
• Considerable inter-accession variation is visible at all temperatures, suggesting heritable differences in thermal tolerance worth further investigation