*4.1. Burn Characteristics*

Recreating conditions of a natural burn, whether prescribed fire or wildfire, is one of the biggest challenges of experimental burning. The amount of biomass and type of species can dictate the intensity of the fire; the challenge is that no forest composition is the same. There are other dominating factors such as weather and topography that influence fire behaviour; weather conditions determine whether the fuel load is 'available' to burn. Weather and topography cannot be controlled in this study so therefore the focus was the e ffect of di fferent fuel load and type of fuel.

For this experiment, extensive research was carried out to produce the best experimental burn design as detailed in the Materials and Methods section. Some studies [43,44] that used the same recommendations of fuel load for medium and high intensity burns, 20 and 40 t ha−<sup>1</sup> respectively, found that medium intensity fires burned for ca. 17 min and high intensity fires burned for ca. 37 min [45]. The Damak experiment followed the same fuel load recommendations; however, the mean burn time >100 ◦C for F2 and F3 were 55.67 and 190.67 min, respectively. An explanation for this is could be attributed to the size of burn area; the area for one study was 105 m<sup>2</sup> [44] and another was 80 m<sup>2</sup> [43]. A larger burn area would create a more severe fire and therefore, burn at a quicker rate. These differences are possibly due to different instrumentation, being that these studies used thermocouples that measured temperature directly within the soil. Our experiment used an infrared thermometer that measured the soil surface temperature. The burning material on the soil surface would make both the burn time and temperature a higher reading compared to studies that use thermocouples due to the soils insulating properties.

### *4.2. E*ff*ect of Fire Intensity on Soil Properties*

The pH of noncalcareous soils (such as Haplic Luvisol) should increase after burning, due to the effect of organic acid denaturisation [2] and the release of alkaline cations. Below the threshold of medium fire intensity, burning does not affect soil pH. This is likely to be due to some of the plots for F1 being at a lower pH originally compared to the average pH of all the plots pre-fire. This would mean that the F1 pH would appear to decrease compared to the pH of the plots before burning (F0). pH is not consistent in soil. Substances such as decomposing leaf litter and plant materials can have a higher pH than the underlying soil. Therefore, when F1 was applied to the soil, some of the leaf material is likely to have mixed with the soil. This would produce a higher pH for F1 plots compared to the F0. The fact that the soil sample bags were left for two months before laboratory analysis, might have been enough time for the plant material to change the original soil pH. Studies have found a decrease in pH after low intensity followed by an increase at higher intensity burn sites [32]. It is important to note that the results for F1 pH were not significantly different to F0.

The pH of F2 was higher than F0 but not significantly. However, F3 pH was significantly higher than F0. This shows that the pH after high intensity fires is significantly affected, supported by other studies [2]. Studies have found that significant increases in soil pH only occurred at temperatures >450–500 ◦C [21], this supports our F3 pH data. However, this shows that the maximum temperature for F2 cannot have exceeded 450 ◦C. Many authors report increased pH values after a prescribed fire [46].

The pH of ash above the soil for F1, F2 and F3 were significantly higher than both the original soil pH and each other. This was to be expected as the carbonates and oxides that are produced from combustion are highly alkaline. This is supported by studies that found that topsoil pH (including the ash layer), under a mixed conifer forest, increased three units immediately after the burning [22]. Our high intensity burn plots experienced an increase of six pH units within the ash layer. It is likely that ash leaching, chemicals draining from the ash into the underlying soil by the action of percolating liquid (usually rainwater), could turn the soil from slightly acidic to more alkaline. This would possibly have an effect on tree species, especially in forests like Damak. Common pine *Pinus sylvestris* and nettle would likely cope with these changes to the soil, because the latter prefers high levels of phosphate. However, species like hornbeam can only grow in neutral soils, so in the case the pH becomes too alkaline it will not be suitable for its growth. There would possibly be a profound effect on the native Hungarian oak, as they are specially adapted to heavy acidic soils. Additionally, following a forest fire in the region, invasive species such as black locust would probably pioneer alkaline rich soil: this species has a tolerable pH range of 4.6–8.2, but prefers dry alkaline soils.

The varied results for the OM and C post-fire in our experiment are supported by other studies [47]. The effect of fire on soil OM is highly dependent on a number of factors that are seen in our study including; fire intensity, varied distribution of organic matter in the soil, nature of burned vegetation. Effects have been highly variable in many studies [15,48] with effects ranging from complete combustion of organic matter to increases of up to 30%. This is a prominent discussion in soil research as there is no generalised trend between organic matter and burning, it is variable controlled. For example, soil heated under laboratory conditions have frequently reported significant losses in OM and C of up to 100% [49], whereas after a wildfire there have been records of an increase [10]. Another study suggests this could be due to the addition of necromass in forests (dead wood and organisms) after a fire event which is not as prevalent in a laboratory experiment [50]. One study produced a comprehensive dataset of soils e ffected by wildfire or prescribed fire. Their study concluded that lower C content was found after prescribed fires while higher C content was found after wildfires [14]. This was attributed to the accumulation of charcoal and encroachment of post-fire N-fixing vegetation. We do, however, acknowledge the presence of some limitations due to the application of the LOI method to determine OM and OC. LOI is based on two assumptions: (1) LOI equals OM and (2) the ratio OM/OC is the same for all treatments. LOI incorporated more than just OM, for example hydrated covers of clays, which are expected to be partially lost when a thermal shock occurs, therefore LOI was not exactly equal to OM in this case. In addition, as di fferent temperatures were reached among fire treatments, the OM/OC ratio would not be the same in all of them, likely: OM/OC ratio decreases with the thermal treatment. For these reasons, what we found in our results related to OM and OC can be mainly attributed to using a broad non-specific technique of measuring OM and OC, with its inherent limitations. Further research should imply a more precise method to determine OM and OC, like a modern dichromate oxidation method. This would ensure a precise quantification of OM and OC.

Calcium results sugges<sup>t</sup> that all fire intensities significantly a ffect calcium levels in Haplic Luvisol. This finding is supported by other studies examining the e ffect of fire on Ca levels in soil [2]. As the volatilisation threshold temperature of Ca occurs at 1484 ◦C [15], it is relatively insensitive to fire (temperature of burning woody fuel ca. 1100 ◦C). Therefore, it was not expected in this study to observe any reductions in the amount of Ca. The reason why Ca increases significantly after fire is thought to have come from the mineralisation of the organic matter [32]. Studies sugges<sup>t</sup> that the exchange capacity of surface horizons is depleted after fire which increases the number of cations that are not volatised (Ca and Mg) [51].

The soil samples measured after the burnings would contain Ca released from the overlying biomass. This could be evidence as to why there was a major increase in Ca between F1 and F2/F3. The increased biomass (ca. 2 and 8 kg more for F2 and F3, respectively) would contribute much more Ca cations than 1 kg of litter (F1). Another possibility is that the temperatures in F2 and F3 reached the volatilisation threshold of K and P, ca. 750 ◦C [52].

The potassium (K) level after all fire treatments was significantly lower than the pre-fire levels, ca. 50% reduction. This is not supported by much of the literature measuring K levels post-fire events [2,53]. As the volatilisation of K occurs at ca. 750 ◦C, this reduction would be expected in very high fire intensity events, however, we have observed the reduction in F1 (low intensity) so the reduction in K cannot be attributed to volatilisation. The instrumentation accuracy was high for K (87%), so the reduction cannot be an instrumentation error. There is a possibility that being left in the soil sample bags for 2 months the K cations bound to inorganic substances; however, similar reductions would be observed in F0 if this was the case. A probable reason is that the available potassium was immobilised within mineral structures driven by thermal treatment.

There are very limited studies exploring the e ffect of fire on K levels in Haplic Luvisol. The studies that have been carried out support our findings that K levels are reduced, after a fire event, in Haplic Luvisol [54]. One study measured significantly lower K levels in the 0–5 cm soil layer post-fire for Umbric leptosols, attributing the loss to erosion and leaching [55]. Studies sugges<sup>t</sup> available K should increase after low intensity prescribed fires [46]. However, there were no erosional or leaching processes occurring in our experiment therefore, the reason why there is a reduction of K in Haplic Luvisol is likely to be due to the composition of the soil. Another explanation could be the moisture level of the soil. It has been found that soil moisture significantly a ffects the availability of K [56]. The Damak soil was dry (observed in fieldwork) which could have reduced the level of K recorded. However, dry soil would be present in F0 measurements so the K values should be similar to F1, F2 and F3 K values. The results from this study show that fire was the cause of the reduction in K. The e ffect fire has on K levels in Haplic Luvisol is a therefore a key finding.

Mg levels post-fire were all significantly increased. This increase has been found in many post-fire soil analysis studies [2]. The volatilisation of Mg occurs at ca. 1107 ◦C [15] therefore there should be no reduction in its availability post-fire. As with Ca, the increase in available Mg is likely to come from the mineralisation of the organic matter in the soil.

Following a prescribed fire, we were expecting increased available phosphorus [46]. The fact that available phosphorous increases with fire intensity can be attributed to the conversion of organic P to orthophosphate [57]. This occurs during the combustion of organic matter and is dependent on the type of soil, vegetation species and leaching [58]. We observed an increase in P for Haplic Luvisol underlying a temperate deciduous forest with no leaching (as samples were taken immediately post-fire). This increase confirms other findings in this area of soil science [59]. The general trend is that P increases significantly dependant mainly on the fire intensity [2]. Small amounts of P can be lost through volatilisation. The threshold temperature for P is ca. 774 ◦C [60], therefore, losses would be expected in F2 and F3 as the burn time and temperatures could reach >774 ◦C. However, it is important to note these losses would be very low in comparison to the converted P. In this experiment, there is a clear positive relationship between available P and fire intensity. This result has been found in many levels [2].

In relation to the changes observed in soil OM and C, were somehow inconclusive. This is likely because soil was taken at a depth of 10 cm. Therefore, there would be a dilution e ffect on the impact of fire on the upper soil (organic matter is mainly a ffected in the upper 2 cm layer). It could also depend on the amount of OM in the soil pre-fire and soil type. This is supported by other studies [10]. However, in a forest setting, organic matter has shown to return to pre-fire levels after one year [13]. Therefore, any e ffect of fire on the OM and C would be recoverable in the long term.

One limitation of this study was the instrumentation used to measure soil temperature. This can be observed in the burn time (Figure 4), being much longer than studies that used thermocouples [43,44]. It is likely that the infrared thermometer measured flames from the burning vegetation instead of the soil surface, thus giving the higher temperature (flame). An accurate measurement of the soil surface temperature could not be measured. Therefore, assumptions had to be made in relation to whether certain cations reached their volatilisation threshold. Ideally, a thermocouple would be used that could be inserted into the soil surface and measure the temperature. In addition to that, this study was limited to 0.5 m<sup>2</sup> plots through health and safety restrictions. With accessible funding and Government approval, larger plots could be used, as is the case in Spain, where researchers have had access to 80 m<sup>2</sup> plots [29,31]. A larger scale plot would be a better way to recreate the conditions of a prescribed fire/wildfire. Furthermore, the time between soil sample collection and analysis would ideally be as short as possible, however for our experiment, the gap between collection and analysis was two months. Samples were stored in dark and dry conditions; nevertheless, several processes could still take place within the soil. For example, the exchangeable cations released by the fire treatments could bind to inorganic substances and a ffect the results for that cation. The soil samples contained root systems and small amounts of vegetation. Their decomposition over two months could lower the pH through the release of carbon dioxide. However, in our study, the processes described above would have had negligible e ffect on our samples; pH increased in all samples and the change, for all cations measured, was significantly increased. Lastly, the same prescribed fire experiment could have also been implemented at other locations of Hungary with the same Haplic Luvisol soil type, to provide more conclusive results on the burning e ffect on forest Haplic Luvisol.

The immediate changes to forest soils following burning have been discussed. This study contributes to the concept that burning temperate forests in Central Europe can have a direct impact on soil fertility and forest management, whether that is from prescribed burning or wildfires. The use of small scale experimental fires to measure this e ffect has mirrored larger experimental burning and wildfire studies.
