The research plot Červenec is located in the Western Tatras, Slovakia, in the Jalovecký creek catchment in a climax spruce forest (latitude 49.183617°N, longitude 19.641944°E, altitude 1,420 m a.s.l.) (Figure 1). The plot has a southeastern aspect with a slope of 20°–35° and an area of approximately 0.2 ha. The average tree height is approximately 26.0 m and the average diameter at breast height is 38.0 cm; the stand density index (stocking density) is low, approximately 0.6. The tree density and basal area of conifers at the experimental plot was estimated to be 638 trees.ha^-1 and 61.3 m².ha^-1. The quaternary glacial relief is characteristic of the Jalovecká valley. The geological bedrock is formed by the crystalline and Mesozoic of the Inner Western Carpathians. Granodiorites and gneisses are found on the crystalline and predominantly limestones on the Mesozoic. Cambisol rendzina soil type developed on the limestones, while cambisol podzols and to a lesser extent, lithosols and rankers formed on the crystalline (Bartík et al., 2014; 2016). Based on 30-year data for 1991–2020 (precipitation and air temperature normal), the average annual air temperature in this area is 3.0 °C and the average annual precipitation sum reaches 1454 mm.

Geobotanically are the forest communities of the research plot Červenec classified as blueberry spruce forests (Vaccinio-Piceetea), which belong to the Alpine and Carpathian subalpine spruce forests (Stanová and Valachovič, 2002). According to Hančinský (1972), the research plot falls into the 7th spruce vegetation altitudinal zone, forest type groups: Sorbeto-Piceetum (60%) and Acereto-Piceetum (40%).

The study was conducted in two stands of Norway spruce (P. abies): 1) a 150-year-old living stand (hereinafter referred to as “LIVING”), 2) a dead, standing stand of the same age (“DEAD”), and 3) a deforested open area (“OPEN”), a former clear-cut area created in 2012. In the living stand, the understorey is exclusively formed by the species European blueberry (Vaccinium myrtillus) with reduced density. The understorey of the dead stand included especially species European blueberry (V. myrtillus), with lesser extent of red raspberry (Rubus idaeus), species of the genus Senecio, great wood-rush (Luzula sylvatica), Calamagrostis arundinacea, fireweed (Chamerion angustifolium), rowan (Sorbus aucuparia) and natural regeneration of Norway spruce (Figure 2). Due to the impact of the bark beetle (Ips typographus), part of the stand died in the summer of 2012, and in 2017, the dieback at this locality began to manifest to an increasing extent. The dead stand was largely composed of standing dead trees. In the stand that died in 2012, the spruce branches were densely covered with the lichen Pseudevernia furfuracea. Open area has a vegetation cover consisting mainly of: V. myrtillus, L. sylvatica, C. arundinacea, Avenella flexuosa, R. idaeus, Sorbus aucuparia and natural regeneration of Norway spruce.

Measurements of surface runoff and vertical precipitation were conductedduring the growing seasons of 2018–2020. The collection of experimental data occurred at approximately 2-week intervals. These measurements started at the end of May and concluded at the end of October, depending on the weather conditions, with the exception of the growing season of 2020. In this period, surface runoff measurements were only carried out until the beginning of October (6 October), as a significant cooling accompanied by snowfall occurred on 13 October due to the arrival of a cold front. The cold weather, with a negative average daily temperature, persisted until 19 October. During the measurement on 15 October, we registered approximately 20 cm of freshly fallen snow on the research plot. All measuring devices were covered by snow, which is why the measurements could not be carried out. During the subsequent measurement on 23 October, snow cover was still present on the soil surface; therefore, only precipitation totals in the standard Czechoslovak rain gauges could be measured.

For the measurement of liquid precipitation and evaluate of interception (Equation 2) were utilized standard Czechoslovak rain gauges (METRA) (with an orifice area of 500 cm²). The primary types of measurements conducted include precipitation totals in open area and within the forest stand. Daily precipitation totals, based on which monthly precipitation totals were evaluated and compared, were recorded using a TRwS 504 weighing rain gauge (with an orifice area of 500 cm²). One METRA rain gauge is positioned in an open area, located sufficiently far from the forest stand (41 m) to minimize or eliminate the influence of surrounding trees on precipitation amounts, as a result of the air flow. Within the forest stands (encompassing living and dead stand), data from the METRA rain gauges and rainfall collection troughs (our self-made) located in the dripping zone at the crown periphery were used for the evaluation of interception. Three rain gauges and three rain collection troughs were installed in the studied stands.

The dripping zone at the crown periphery represented precipitation which was caught in the surface parts of the crown and dripped into the sub-canopy space after canopy saturation (Jančo et al., 2021).

After penetrating the forest canopy, throughfall is also intercepted by the herbaceous or shrub layers located beneath the stand. To determine total interception (stand + understorey) (Equation 3), the amount of precipitation reaching the herbaceous layer of the stands was monitored. Throughfall beneath the understorey (in our case European blueberry) were measured using our self-made rainfall collection troughs (Figure 3). The collection trough consists of a plastic gutter with dimensions of 50 cm in length and 10 cm in width (with a capture area of 500 cm²). A plastic tube leads from the collection trough, channeling the precipitation into a collection vessel. The collection vessel is equipped with a stopper to prevent evaporation.

The deluometric method was used to determine surface runoff values; it is one of the methods used in erosion studies (Midriak, 1986). Surface runoff was measured in an open area, in a living stand and in a dead stand (Figure 4). One deluometer was installed at each location. The deluometer is a metal trough with a hinged roof, from which a drainage opening extends, fitted with a plastic tube. This tube is connected to a collection vessel, where the surface runoff water is gathered. The capture area of this device is 10,000 cm², as the runoff area consists of sheet metal plates inserted into the ground, measuring 200 cm in length above the deluometer and 50 cm in width (Figure 5).

Understorey vegetation cover at individual sites (living forest, dead forest and open area) during the growing season of 2020 was evaluated using digital photographs. The photographs were taken by phenocameras (PhenoCam 800, EMS Brno) (Figure 6), with one phenocamera installed at each study site. Each phenocamera assembly consists of a swivel-mount bracket and a protective housing with a metal roof to ensure weatherproofing. The internal components include a Canon digital camera equipped with an 8 GB memory card, a USB output port for data retrieval, and two 1.5 V alkaline batteries. Photographs were assigned to the days of field measurements in the growing season of 2020 (Supplementary Figure S1). All photographs in the jpg. format was processed using ImageJ software. By applying thresholding techniques to filter vegetation from the soil surface and extraneous objects (e.g., measurement device), the percentage of understorey cover was calculated for the living forest, dead forest, and open area.

The surface organic layer (SOL) refers to the uppermost soil horizon dominated by organic material derived from plant and animal residues that have accumulated on the soil surface and undergone various stages of decomposition. The SOL is distinct from the underlying mineral soil because the dominant is the organic matter (e.g., leaves, needles, twigs, mosses, and other litter), with only minor contributions from mineral particles. In soil classification systems, the SOL is typically designated as the O horizon (Deckers et al., 1998) which can be represented by three horizons - the litter horizon (Ol), the fragmented horizon (Of) and the humus horizon (Oh) (Gömöryová et al., 2013; Zanella et al., 2018). We determined the thickness of the SOL in the both forest stands (LIVING and DEAD), as well as on a nearby clear-cut area (OPEN), using methods described by Cools and De Vos (2010), Gömöryová et al. (2013), Pichler et al. (2022). SOL thickness was measured along 10-m linear transects with 1-m spacing in each study area. Samples from the SOL horizon were taken using a 0.12 m × 0.12 m template placed on the soil surface; the layer under the template was cut away with a knife. The thickness of the SOL was recorded on the wall of the soil profile using a ruler.

The recorded totals of throughfall in individual measurements were first expressed as the arithmetic mean for dripping zone at the crown periphery in the both stands.

Surface runoff (S_R) (mm) was calculated as the ratio of the amount of precipitation captured on the soil surface (V) to the catchment area of the deluometer (C_A) using the following equation: S R = V C A   10 (1)

Canopy interception (I_C) (mm) was calculated as the difference between gross precipitation (P_G) and throughfall (P_TH) using the following equation: I C = P G − P T H (2)

Total forest interception (I_F) (mm) was calculated as the sum of understorey interception (I_U) and canopy interception (I_C) I F = I C + I U (3)

All analyses and graph displays were performed using the software programs Statgraphics Centurion 16, Statistica 12, Autodesk Inventor 2012, Grapher 25, ImageJ and Microsoft Excel 2016. When comparing the equality of means for two dependent samples, we first needed to determine if the data (interception in %, surface organic layer in cm) exhibited a normal distribution. To assess normality, we used the Shapiro-Wilk test and Q-Q plots. If the P-value was ≥0.05, the data were considered normally distributed; if the P-value was <0.05, the data were considered non-normally distributed. If the compared samples showed a normal distribution, we used Student’s paired t-test. If the distribution of at least one of the compared samples was not normal, we applied the non-parametric Wilcoxon paired test. The significance level (α) for all tests was set at 95%, meaning that if the P-value was ≤0.05, the difference between the samples was considered statistically significant. Correlation and regression analyses were used to determine the dependence of surface runoff (mm) in both forest stands and open area on gross precipitation.

This relationship is expressed based on the following equation: y   S R = a + b P G (4) where: surface runoff (S_R) is dependent variable and gross precipitation is independent variable (P_G), a is intercept with S_R-axis and denotes the threshold value at which surface runoff begins to generate, b is the surface runoff coefficient and denotes surface runoff response to gross precipitation.

The same relationship was used to determine the dependence of canopy and total interception loss on gross precipitation totals for the studied stands and the dependence of surface runoff (mm) on understorey vegetation cover (%). The statistical significance of the linear relationship between samples was tested using analysis of variance at a 95% confidence level. If the P-value was ≤0.05, the relationship was considered statistically significant.

The growing season of 2018, with a total precipitation of 625.0 mm, was classified as a rainfall-normal period relative to the 1991–2020 normal. The largest amount of precipitation fell in July (302.0 mm), with up to 199.2 mm falling between 17 July and 20 July. The lowest precipitation occurred in August.

The driest precipitation period was the growing season of 2019, with a total precipitation of only 489.2 mm. Specifically, the months of June and July were precipitation-deficient, and no rainfall occurred in the second, above-average warm half of October.

The highest total precipitation was recorded in the growing season of 2020, with significantly higher precipitation totals compared to the normal were registered in June, September and October (Figure 7).

The calculated values of measured surface runoff (Equation 1) in the studied stands and in an open area during the growing seasons of 2018, 2019 and 2020 are displayed in Table 1.

During the growing season of 2018, the lowest surface runoff was recorded in the dead forest at 14.4 mm (2.3%), followed by the open area with a value of 18.3 mm (2.9%) and the highest surface runoff value was registered in the living forest, i.e., 23.4 mm (3.7%). The range of calculated surface runoff values during this period was in the dead forest 0.6%–5.3%, in the living forest 0.6%–7.2% and in an open area 1.1%–6.0%. When comparing the forest stands, the surface runoff in the dead forest reached a higher value than the surface runoff in the living forest only once. In one instance (7 September 2018), we recorded identical values in both evaluated stands. When comparing the surface runoff in the open area and in the living forest, the surface runoff in an open area reached higher values in six cases. Of these, in three cases (29 June, 13 July and 13 August), the surface runoff in an open area only slightly exceeded (by 0.2% and 0.3%, respectively) the values recorded in the living forest. During this growing season, the highest precipitation total for a 2-week measurement period (14 July to 27 July) was recorded on the open area, reaching 232.4 mm, which represents one-third of the total precipitation for the entire growing season of 2018. However, this amount of precipitation did not significantly affect the resulting surface runoff values at the individual sites. Surface runoff reached 3.2 mm (1.4%) in the dead forest, 9.7 mm (4.2%) in the living forest, and 4.4 mm (1.9%) in an open area. The highest surface runoff values were recorded at all monitored sites toward the end of the growing season: on 1 October in the living and dead stands, with values of 7.2% and 5.3%, respectively and on 13 October in an open area, with a value of 6.0%.

During the growing season of 2019, the lowest surface runoff was recorded in the dead forest at 9.2 mm (1.9%), followed by the open area with a value of 15.7 mm (3.2%) and the highest surface runoff value was again reached in the living forest at 18.1 mm (3.7%). The range of calculated surface runoff values during this period was 0%–3.5% in the dead forest, 1.0%–6.1% in the living forest and 1.3%–5.0% in an open area. When comparing the forest stands, the surface runoff in the dead forest reached a higher value than the surface runoff in the living forest only once. When comparing the surface runoff in an open area and in the living forest during this growing season, the surface runoff in an open area reached higher values in three cases and in one case, the values were identical (4 October 2019). The highest recorded precipitation total on the open area during this period was 136 mm (from 23 August to 9 September). This precipitation totals also did not significantly affect the recorded surface runoff values. Surface runoff reached 2.6 mm (1.9%) in the dead forest, 5.0 mm (3.7%) in the living forest and 3.4 mm (2.5%) in an open area. The highest surface runoff values were recorded in the dead forest in June (3.5%) and in the living forest and on the open area in August (6.1% and 5.0%). Since a significant amount of precipitation fell during storm events in August, it follows that the highest surface runoff values were registered in the living forest and in an open area precisely during this period (Figure 8).

Similarly to the two preceding evaluated growing seasons, the lowest surface runoff in the growing season of 2020 was recorded in the following order: dead forest 24.8 mm (3.0%) < open area 26.5 mm (3.2%) < living forest 66.7 mm (8.1%). The range of calculated surface runoff values during this period was 1.5%–7.3% in the dead forest, 0.8%–16.4% in the living forest, and 2.3%–3.9% on the open area, all relative to the open area precipitation total. When comparing the stands and the open area, the surface runoff in the living forest reached higher values throughout the entire growing season, with the exception of one situation when surface runoff was lower. This situation occurred during the period (23 July to 3 August) when the precipitation total on the open area reached only 6.0 mm. Due to the low precipitation total and higher interception in the living forest, the surface runoff was the lowest. In one instance, we also registered identical values in the living forest and in an open area (23 July) (Figure 8). The growing season of 2020 was rich in precipitation, which resulted in higher surface runoff values, particularly in the living forest, where the range of values was 12.1%–16.4% from the beginning of August until the beginning of October. The highest surface runoff values in the living and dead forest (16.4% and 7.3%) were recorded on 19 August. In an open area, the surface runoff reached its highest value (3.9%) on 3 June 2020.

The calculated surface runoff data from each measurement at the individual sites were statistically compared across the evaluated growing seasons. The statistical evaluation of the results is presented in Table 2. Based on the performed paired Student's t-test or the non-parametric Wilcoxon paired test, no statistically significant difference was confirmed between the surface runoff in the living forest and in an open area during the growing seasons 2018 and 2019, and between the surface runoff in the dead forest and in an open area during the growing seasons of 2018 and 2020.

The expressed linear regression of the calculated values of surface runoff (mm) on the amount of gross precipitation (Equation 4), at the studied sites are displayed in Figure 9. The following statistical parameters are displayed above the graphs: the linear regression equation, the Pearson’s coefficient of determination (R ²), the correlation coefficient (R) and the p-value. The p-value indicates whether a statistically significant difference was confirmed in the correlation between surface runoff and gross precipitation. Bold values indicate a statistically significant difference. The Pearson’s correlation coefficient is most significant for surface runoff in an open area, followed by surface runoff in the dead forest and surface runoff in the living forest is slightly less significant. At all three examined localities, the correlation coefficient indicates a moderately strong relationship between the variables. The R-Squared statistic indicates that the model as fitted explains 76.2% of the variability of surface runoff in an open area, 58.4% of the variability of surface runoff in the dead forest, and 52.6% of the variability of surface runoff in the living forest from gross precipitation. A statistically significant relationship between surface runoff and gross precipitation was confirmed at all examined localities at the 95.0% confidence level. It follows that surface runoff in an open area is most affected by the amount of gross precipitation, as this space has no trees above it. Although the surface runoff at this locality has no trees above it, it is influenced by understorey interception.

Surface runoff in the dead and living forest is influenced not only by understorey interception but mainly by canopy interception. The higher variability of surface runoff in the living forest compared to the dead forest may be caused by the effect of occult precipitation, the occurrence of which reduces interception losses in the living forest.

Canopy interception in the dripping zone at the crown periphery of the dead stand generally exceeded the values recorded in the living forest during the growing seasons of 2018 and 2020 (Figure 10). Following the dieback, the tree crowns lost their assimilation organs (needles). However, approximately 3 years post-mortality (August 2012), the lichen P. furfuracea began to colonize the limbs, branches and twigs, expanding progressively in subsequent years. Due to this extensive lichen colonization, the interception of precipitation within the dead canopies increased significantly. In the growing season of 2018, three cases were recorded where the canopy interception in the living forest was higher, while in the growing season of 2020, there were only two such cases. An exception was the very warm growing season of 2019, which was below average in terms of precipitation, when interception reached higher values in the living forest. Since the precipitation captured in the crown space is guided by the structure of the crown to its edge, the canopy interception in the living forest is consequently lower. Another factor influencing canopy interception values during the growing season is the occurrence of occult (horizontal) precipitation. Their interaction at this locality is manifested at this site primarily in the living forest, and to a lesser extent in the dead forest. In the living forest, seven cases were recorded (four of which were in the 2020 period), and in the dead forest, four cases were recorded across all evaluated growing seasons, where the amount of throughfall exceeded the amount of gross precipitation. The colonization of branches by the lichen P. furfuracea results in a reduction of throughfall in the dead stand. This occurs despite the fact that occult precipitation is combed out not only by the needles of the living forest but also by the lichen-covered branches in these elevations.

In the living forest during the growing seasons of 2018–2020, precipitation totals (throughfall) were recorded as follows: 610.4 mm (2018), 391.6 mm (2019), and 834.0 mm (2020). In the dead forest, the totals were 557.2 mm (2018), 409.1 mm (2019), and 739.6 mm (2020). This represents the canopy interception of 2.3% (2018), 20.0% (2019), and −1.1% (2020) in the living forest, and 10.8% (2018), 16.4% (2019), and 9.2% (2020) in the dead forest of the gross precipitation.

Based on a linear regression between the calculated values of the canopy interception loss (I_C) (mm) on the amount of gross precipitation (P_G) in the dead and living stand were determined the linear regression equations, the coefficients of determination (R ²), Pearson´s correlation coefficients (R) and P-values, informing us whether a statistically significant difference was confirmed between the correlation of interception loss and gross precipitation. Bold values represent a statistically significant difference. Dead stand: y (I_C) = 12.0218–0.0643P_G, R ² = 0.0825, R = −0.2873, P = 0.1307. Living stand: y (I_C) = 11.2855–0.1154P_G, R ² = 0.2367, R = −0.4865, P = 0.0074.

During the growing season of 2018, measured values from the collection troughs from 28 July were excluded from the overall results because they had overflowed. The highest total interception was recorded in the dead forest during every evaluated period (Figure 11). In the living forest, total interception in the dripping zone at the crown periphery was lower, because lower canopy interception resulted in higher throughfall; in the growing season of 2020, it reached its lowest value of 33.5%, which is caused by the period being the richest in precipitation and also by the combined effect of occult precipitation. Another important factor was the development of the understorey European blueberry, within the study stands. In the dead forest, the European blueberry was more developed (this is due to higher solar radiation input through the more reduced canopy closure in the dead forest) compared to the living forest. It is also interesting to note that the total interception in the dripping zone at crown periphery in the stands did not reach negative values in any case, whereas the canopy interception showed negative values. This implies that the understorey (in our case, European blueberry) significantly influences the resulting values of total interception.

Total interception in the dead forest reached values of 228.8 mm (58.3%) in 2018, 328.4 mm (67.1%) in 2019, and 365.8 mm (44.3%) in 2020. In the live forest, the corresponding values were 207.0 mm (52.7%) in 2018, 287.7 mm (58.8%) in 2019, and 276.6 mm (33.5%) in 2020.

Consistent with the approach used for canopy interception, the linear regression equations, coefficients of determination (R ²), correlation coefficients (R) and P-values were determined for the relationship between the total interception loss (I_F) (mm) and gross precipitation (P_G) for both stands, based on regression and correlation analysis. Dead stand: y (I_F) = 6.2735 + 0.4378P_G, R ² = 0.8065, R = 0.898, P = 0.0000. Living stand: y (I_F) = 5.2085 + 0.3664P_G, R ² = 0.7059, R = 0.8402, P = 0.0000. Bold values represent a statistically significant difference.

During the growing season of 2020, the living forest exhibited the lowest understorey cover percentages, with values ranging from 43% to 54%. Conversely, the highest understorey cover was observed in the dead forest, where values ranged between 78% and 100%. The open area showed lower coverage levels compared to the dead stand, with a recorded range of 64%–93% (Figure 12). Across all sites, understorey cover values were lower at the beginning of the growing season, subsequently increased, and declined again toward the end of the period. This temporal variation is attributed to the phenological development (phenophases) of the individual species occurring within the living forest, dead forest and an open area.

The percentages of understorey cover at the individual sites during the growing season of 2020 were subsequently analyzed in relation to the recorded surface runoff values (Figure 13). When comparing the study sites, the open area exhibited the highest variability (R ² = 0.7883), which also corresponded to the highest Pearson´s correlation coefficient (R = −0.8879). In the dead stand, the R-squared statistic indicates that the fitted model explains 52.81% of the surface runoff variability. The Pearson´s correlation coefficient equals −0.7267, indicating a moderately strong negative relationship. The lowest variability in relation to understorey cover was recorded in the living stand (R ² = 0.4284), as further supported by the Pearson´s correlation coefficient (R = −0.6546). Statistically significant relationship between understorey cover percentage and surface runoff was not confirmed in the living stand.

The open area has no tree canopy above it, so the understorey plays a dominant role in retaining precipitation. In the dead forest, the dependence of surface runoff on understory cover is also strong, suggesting that the understorey performs a critical hydrological function at this site as well. In contrast, the sparse understorey in the living forest has the least influence on surface runoff volume compared to the dead stand and an open area.

Figure 14 shows the thickness of the surface organic layer. The maximum variation occurred in the living stand (1.5–7.5 cm), followed by the dead stand (2.0–6.5 cm), with the minimum variation found in and open area (1.4–5.7 cm). Based on the performed paired Student's t-test or the non-parametric Wilcoxon paired test (Table 3), no statistically significant difference was confirmed between any of the sites. This implies that the thickness of the surface organic layer at the studied sites does not have a decisive influence on surface runoff formation.