Author:

Abstract:

The historical atmospheric nitrogen and sulphur emissions in the Western world have acidified forest soils, reduced their pH buffering capacity, and lowered the concentrations of available cations such as calcium (Ca), potassium (K), and magnesium (Mg). Soils of temperate forests of the northern hemisphere are particularly susceptible to acidifying deposition, and the resulting nutrient imbalances currently threaten the forests and their ecosystem services. To ensure forest vitality and future ecosystem services, the restitution of a better buffered belowground system is needed. This work investigates the potential of rock dust (RD) as a soil amendment to restore forest vitality in areas affected by the acidifying deposition. Rock dusts are ground igneous or metamorphic rocks with slow mineral weathering rates. Rock dust can restore soil pH buffering in forests by acting as slow-release fertilisers that raise the pH and increase the exchangeable Ca, Mg, and K in soil. They do not have the drawbacks of quick lime or soluble fertilisers that promote mineralisation of the forest floor, leading to loss of carbon stocks and eutrophication. Rock dusts are, however, ill-defined; there are no tools to predict their rates and extent of weathering and nutrient releases. Against this background, this study was set up to characterise rock dust, identify the mechanisms of their weathering rate in forest soils, and analyse their effects on forest vitality, including understory plant diversity. The responses of soils and forest ecosystems to RD application were assessed at three temporal scales: from short-term in the laboratory (days-months) to middle-long term in mesocosms and field trials (3-7 years) and towards long-term field trials (30 years). The first part was devoted to identifying rapid laboratory-based tools to predict RD's acid neutralising capacities (ANC) by comparing their liming effects in four acidified forest soils. The ANC is the amount of protons neutralised per kg of RD. The dissolution rates of five commercial RDs (two basalts: Eifelgold and Actimin, phonolite Vulkamin, foidite Soilfeed, and trachy-andesite Biolit) were measured in the laboratory. Three accelerated weathering tests were conducted to assess the ANC: aqueous batch renewal (one year), pHstat titration (24 hours), and lime-calibrated agitated soil-RD suspensions (two months). The RD dissolution occurred in two phases, a fast fraction followed by a slower fraction, and their sum was considered the available ANC. This available ANC is 2-25% of that of calcite lime and is significantly smaller than the total ANCs based on total composition (25-50% of that of lime). The pH-dependent dissolution rate has a slope that critically depends on the mineralogy: dissolution rates reduce factor 10 (foidite, phonolite) to 100 (basalts) per unit pH increase, and that, in turn, depends on the average charge of the dominantly dissolving cations. A two-year outdoor mesocosm trial validated that RD dissolution rates (via pHstat titration), soil pH buffering capacity (via soil titration), and ANC-fractions (novel lime-calibrated soil-RD suspension) can accurately predict the temporal pH changes. For future exploration of new RDs, it was concluded that this lime-calibrated soil-RD suspension is a practical method to predict the slow-release liming property of RDs when mixed with soil. Second, RDs were directly mixed in the planting pit, which is recommended for the immediate soil pH elevation needed for increased sapling growth. A field experiment with 960 sycamore maple (Acer pseudoplatanus L.) saplings in two sites in the Campine region (NL) was constructed, the first site was a clearcut of Picea abies L. that has an agricultural legacy (site 1, soil pH= 3.5, 83% sand) and the second site was under the canopy of Pinus sylvestris L. that has a heathland legacy (site 2, soil pH = 3.1, 67% sand). Treatments included six different RDs and four reference treatments with TSP, dolomite lime and KCl, and their combination. Sapling growth was monitored for 40 months. This trial showed that the total ANC and buffering cation content from digestion were poor predictors of product efficacy. Sapling growth was affected by RD depending on the site and RD type. The highest factors volume increase of the saplings, relative to the unamended control, were 2 (site 1) and 8 (site 2). In site 1, the volume growth was larger than in conventional liming and fertilisation treatments. Due to the high sand content of this site, the growth was only explained by RD's water retention, which was superior for a zeolite-containing RD. In the more acidic site 2, growth and foliar nutrient concentrations were best related to liming and nutrient release by RDs inferred from the 2-month laboratory-based soil-RD suspension test. It is concluded that RD amendments in the planting pit are valuable to regenerate acid forest soils and accelerated weathering laboratory tests can indicate their potential. In the third part, the soil chemical changes following broadcast (surface) application were evaluated as a logical approach because mixing rock dust in soils is not practically feasible in existing forests. The effects of broadcast RD applications (5-16 Mg/ha) on soil pH contrasted with that of lime at ten locations spanning 3-34 years. This application on the forest floor (FF) showed a limited effect on the pH of the mineral soil: the slow rainwater-pH-driven dissolution of broadcast-applied RD compared to mixed-in applications highlighted the limitation already known from conventional liming methods. The liming rate in the mineral topsoil was 0.018 pH units/year after 2.5 years, but by year 34, this liming rate had slowed down to 0.002 pH units/year. RD can buffer soils in the long term, but achieving significant improvements in mineral soil pH requires decades. Furthermore, the slow release of base cations like Ca, Mg, and K was expected to shift soils to the exchange buffer domain. But most base cations are likely taken up by plants and trees from the forest floor as the base saturation (BS) in the top mineral soil remained below 30%, even decades after application. The liming effect of RD is driven by the leaching of dissolved organic carbon (DOC), which leaches alkalinity from the humus into the mineral soil layers. Consequently, if more rapid pH and BS restoration are required, higher RD doses up to 15-20 Mg/ha or combinations of RD with dolomite that generate more DOC should be evaluated. However, this work also showed adverse effects of intense liming in the long term: application of 10 Mg/ha of dolomite or wood ash accelerated soil carbon loss from the forest floor. These high lime doses can reduce the FF thickness up to 30 years after application to a factor of 0.25, leading to a carbon loss of 40±13 Mg C/ha; some of this carbon can be translocated to the mineral soil (up to 20±14 Mg C/ha). Also RDs tended to reduce the thickness of the forest floor but to a smaller extent (factor 0.5). Strikingly, the RDs with high Na-release can lead to accelerated dispersion and decomposition of SOM so these RDs should not be used to protect forest soil carbon stocks. Fourth, the effects of broadcast RD application on tree vitality were tested at ten locations, spanning 4-34 years. This was evaluated by tree cores taken of 250 Picea abies L. trees (28-34 years post-application), 62 Quercus robur L. trees (7 years post-application), and 14 Pinus sylvestris L. (4 fours post-application). One location in Austria identified no effect of RD at higher altitudes where plant growth was limited by N; however, where N-deposition was larger, basal area increments were factor 1.3 in RD treated than in untreated plots 23 years after application. In a RD trial in Germany, no effect of RD was found on tree vitality, but other fertiliser tests identified K as a primary limiting nutrient in that stand that could increase growth (factor 1.5). The RD applied has slow K-releasing silicates that did not provide sufficient available K. In the RD trials in the Netherlands, tree radial growth of 40 year-old oaks increased when Ca was the limiting nutrient. This corroborates Liebig's law of the minimum, i.e., RD is only effective when the primary limiting nutrient is alleviated. Moreover, post-application stand biomass increments were generally greater for younger trees. No relative effects could be detected for ages of oaks > 70 years or Norway spruce >200 years. Nutrient availability in the RD can be identified via pHstat titration and ICP-OES analyses. A decision tree was developed to use foliar nutrient concentrations to diagnose the limiting nutrients. Combined with RD titration data, the identification of site-specific optimal RDs or a combination with dolomite or mineral fertilisers is feasible. The fifth part evaluated the understory response and found that broadcast rock dust application initially disturbed the herb layer community and even more the bryophytes, but this disturbance quickly decreased over time as species diversity (effective number of species, ENS) increased. The Ellenberg R (indicating soil reaction) showed a transient increase after RD amendment, but the difference from the control diminished over time. In contrast, the Ellenberg N (indicating fertility) initially showed no apparent change but tended to increase over time. These gradual effects on the Ellenberg indicators of the plant community show that RD application is a safer alternative to dolomite lime. The demonstrated long-term increase in species diversity without a disturbing loss of calcifuge species suggests that RD could be valuable in promoting the ecological resilience of acidified forest ecosystems. In conclusion, rock dust application deserves to be integrated into forest management strategies, as the adverse effects were negligible and the anticipated benefits largely confirmed. Depending on the RD, they can buffer incoming N deposition of 20 kg N/ha/year, i.e., 1.4 kmol H+/ha/year of potential acidifying deposition, for 10-70 years. However, consideration of site conditions, i.e., local nutrient deficiency, and RD composition, are paramount, and future research on long-term high-intensity monitoring sites remains essential to truly operationalise this promising restoration technique.