Fire-derived phosphorus fertilization of African tropical forests

Tropical forests store substantial terrestrial carbon and biodiversity. While nitrogen can often be replenished via biological fixation during succession, phosphorus is frequently the main limiting nutrient in old-growth tropical forests rooted in highly weathered soils. Consequently, these ecosystems depend on trace atmospheric inputs of P (e.g., dust or biomass burning emissions) rather than soil weathering. It has been hypothesized that forest canopies function as traps for atmospheric constituents and that increasing canopy complexity during secondary succession enhances trapping efficiency. Repeated deforestation may reduce atmospheric P inputs and create long-term negative P balances that hinder natural forest recovery. Given Africa’s extensive biomass burning and the rise of secondary forests due to slash-and-burn agriculture, the study asks: What is the magnitude and source of atmospheric P deposition to central African forests, and how does canopy trapping along a successional gradient regulate P delivery to the forest floor?

Prior work indicates nitrogen may limit early succession but is ultimately replenished via fixation, whereas phosphorus is often limiting in old-growth tropical forests on weathered soils. Atmospheric inputs can sustain nutrient budgets over long timescales when weathering inputs are small. The concept of forest canopies as traps for atmospheric particles suggests structural complexity enhances dry deposition. Africa experiences the majority of global burned area, primarily from grassland and savanna fires, which are known sources of nutrient-rich aerosols, including phosphorus. Models and estimates have suggested moderate P deposition in the region, but empirical data are scarce; combustion-related emissions may contribute substantially to global atmospheric P budgets. Repeated deforestation could reduce canopy-mediated deposition, potentially locking systems into nutrient-poor states that impede recovery.

Study site: Post-agriculture forests near Yoko reserve (N00°17′; E25°18′; 435 m a.s.l.), 29–39 km SE of Kisangani, DRC. Semi-deciduous tropical forest (Af climate), annual rainfall 1418–1915 mm, mean temperatures 23.7–26.2 °C. Highly weathered, nutrient-poor Oxisols. Design: Fifteen 40×40 m plots across a successional gradient with three plots per stage: agriculture, and 5-, 12-, 20-, and 60-year-old secondary forests. Time-since-disturbance determined via local interviews; tree heights measured for 20% of individuals per diameter class. Deposition sampling: In each plot, eight collectors (polyethylene funnels with 0.45 μm nylon mesh, 1.5 m high, draining to 5 L PE containers) arranged as two rows of four, ~8 m apart. Open-field collectors captured bulk deposition. Weekly measurements of water volume; equipment replaced and rinsed each visit. Volume-weighted composite sample per plot per week; samples frozen and shipped for analysis. Analyses: Total phosphorus measured by ICP-AES. Throughfall and open-field P loads computed as concentration × rainfall volume; weekly loads aggregated to annual by averaging and ×52 weeks. Wet deposition defined as open-field loads; dry deposition inferred as net throughfall (throughfall − open-field). Dissolved black carbon (DBC) measured as proxy for biomass burning aerosols: samples filtered (0.45 μm), DOC isolated via PPL SPE, and BPCAs (B6CA, B5CA) quantified by HPLC after nitric acid oxidation. DBC calculated from BPCA sum using an established power relationship. Canopy structure: UAV-based structure-from-motion 3D models from surveys on Feb 9, 2020 using DJI Mavic 2 Pro and customized DJI Phantom 3 with GoPro + RTK/PPK GNSS. High overlap flights at 180 m AGL; images processed in Pix4D to derive 3D point clouds and canopy height models (CHM). Canopy complexity quantified as roughness (via aerodynamic roughness length z0 inferred from obstacle height H and obstacle density A) and as rugosity (SD of upper canopy height across transects). z0 linked to particle deposition velocity via published parameterizations. Data analysis: Standardized major axis regressions between DBC and P deposition, and between rainfall volume and P deposition; log–log SMA used where assumptions were violated and back-transformed for visualization. Net canopy effects computed by subtracting open-field from throughfall loads. Satellite and modeling: Backward wind trajectories (HYSPLIT) intersected with MODIS burned area to infer potential aerosol source regions. MERRA-2 black carbon surface mass concentration (BCSMC) used to evaluate temporal variability; MODIS AOD at 550 nm provided regional aerosol context. Uncertainties/controls: Weekly in-field storage without cooling or preservatives; total P used to ensure recovery. DBC analysis on filtered samples excludes particulate BC but serves as tracer for combustion-derived inputs. Open-field collectors are bulk, not wet-only, slightly overestimating wet and underestimating dry deposition. UAV survey occurred ~6 months after monitoring; agricultural plots overgrown by then, handled in visualization adjustments.

• Open-field (bulk) wet deposition at the central African site: 0.43 ± 0.09 kg P ha−1 yr−1.
• Total P deposition (wet + dry) under 60-year-old secondary forest: 3.1 ± 1.4 kg P ha−1 yr−1, substantially higher than previous regional estimates (0.8–1.0 kg P ha−1 yr−1) and comparable to measurements near Lake Victoria (1.8–2.7 kg P ha−1 yr−1).
• Net P throughfall (dry component) scales linearly with DBC deposition along the chronosequence, indicating that excess P input beneath canopies is predominantly fire-derived rather than from canopy leaching.
• Annual P deposition increases with forest age and canopy complexity: approximately 1.6 ± 0.7 (5 yrs), 1.5 ± 0.3 (12 yrs), 1.1 ± 0.2 (20 yrs), and 3.1 ± 1.4 kg P ha−1 yr−1 (60 yrs). Older, more complex canopies (higher roughness and rugosity) trap more P via dry deposition.
• Weekly P throughfall is significantly correlated with weekly rainfall on a log–log basis across forest stages, indicating that P delivery to the forest floor is transport-limited by rainfall-driven wash-off from canopies rather than the timing of aerosol arrival.
• Forests are about eight times more effective at capturing atmospheric P than non-forest (open-field) areas (increase from ~0.4 to ~3.1 kg P ha−1 yr−1 in old secondary forests).
• Estimated total P deposition in older Congo basin forests is roughly three times the amount needed to sustain tropical forest growth in steady state across the tropics (~1.1 kg P ha−1 yr−1).
• Africa’s widespread biomass burning (especially savannas) provides a consistent regional source of P-rich aerosols contributing to deposition at the study site.

The study demonstrates that central African secondary forests receive substantial atmospheric phosphorus inputs largely derived from biomass burning. The strong scaling of net throughfall P with DBC establishes a pyrogenic source for the excess P under forest canopies, supporting the hypothesis that canopies act as efficient traps for atmospheric constituents. Canopy complexity emerges as a key biotic control on dry deposition: older, structurally complex canopies have higher roughness and rugosity, enhancing aerosol capture and P input to soils. Temporally, P delivery to the forest floor is governed by rainfall wash-off, suggesting that canopies function as dynamic reservoirs that accumulate dry deposition between rain events and release nutrients proportionally with precipitation. These findings imply that deforestation and prolonged canopy loss can reduce local nutrient inputs and potentially push ecosystems toward negative P balances that hinder recovery. Conversely, high current P deposition loads—exceeding steady-state requirements—raise questions about the extent of P limitation in old-growth central African forests. Given that biomass burning has been relatively stable to slightly lower than millennial-scale baselines but may increase with anthropogenic activities, future deposition could intensify, with implications for nutrient cycling, productivity, and successional trajectories.

This work provides in situ evidence that biomass burning across Africa drives high atmospheric phosphorus deposition to central African forests and that canopy trapping—mediated by structural complexity—magnifies these inputs along successional recovery. Forest canopies act as dynamic reservoirs, with rainfall controlling P wash-off to the forest floor. Measured P deposition in older secondary forests exceeds model expectations and steady-state requirements, prompting reassessment of the prevalence of P limitation in the region. Future research should (i) expand spatial and temporal monitoring across different forest types and ages to generalize canopy trapping effects, (ii) improve partitioning of wet versus dry deposition using wet-only collectors and characterize particulate versus dissolved phases, (iii) link deposition variability to ecosystem responses (soil pools, plant uptake, productivity), and (iv) integrate fire regime scenarios to project nutrient deposition and forest recovery under changing land use and climate.

• Collectors lacked active cooling and chemical preservatives during weekly in-field storage; although total P measurements mitigate concerns about microbial alteration, some bias is possible.
• Open-field collectors measured bulk rather than wet-only deposition, likely overestimating wet deposition and underestimating inferred dry deposition.
• DBC measurements were performed on filtered samples and exclude particulate black carbon; DBC was used as a tracer of combustion-derived inputs rather than a full mass balance of pyrogenic material.
• UAV canopy surveys occurred ~6 months after the monitoring period; agricultural plots had become overgrown, necessitating visualization adjustments and potentially introducing minor discrepancies in canopy metrics.
• Logistical constraints led to occasional missing weekly samples; annual fluxes were estimated by averaging available weeks and multiplying by 52.