The importance of aquifers with karst porosity, which already supply around a quarter of the global needs for drinking water, is gradually increasing [1
]. This proportion even exceeds half of the demand of individual countries in the areas of the Dinaric Karst [2
], the largest contiguous karst region in Europe [3
], and in the Alps. These areas are mostly covered by forest and are important for its rich and unique (underground) ecosystems [4
Due to the dissolution of the host rock, the karst surface is very permeable, the soil cover is usually clayey, very thin or even absent. Characteristic features are the immediate infiltration of rainwater or melted water from snow into the underground and the sinking of streams that drain the adjacent non-karst areas, while surface runoff is often negligible [6
]. Most water in karst aquifers is transported through a network of solution conduits, which evolve along discontinuities in primary fissured aquifer [7
]. Thus, fast diffuse and concentrated infiltration, high heterogeneity and anisotropy, mainly conditioned by the position of conduits and fast transport from inputs to the springs, is characteristic. A specificity of karst hydrology is the great variability in flow and thus transport processes as a function of temporal hydrological conditions [10
], which can have practical implications for groundwater abstraction in terms of adequate quantity and quality.
Clear evidence shows that radiative forcing by anthropogenic greenhouse gases has affected the global water cycle in many ways [12
], especially the alternation in the water balance elements. Observed records and climate simulations of some meteorological variables show increasing trends worldwide (e.g., temperature, evapotranspiration) [13
] and shifts in precipitation patterns, something which affects discharge quantities [15
] and regimes of rivers and springs [18
]. These changes may, therefore, even intensify the high variability of flow and transport in karst. Long-term effects with a decrease in water resource quantities and short-term effects with the increasing frequency of extreme events (e.g., floods, droughts) have already been observed in karst areas in the Mediterranean, China, Texas and elsewhere [20
Despite the important role of karst water resources, there are few studies that thoroughly assess their sensitivity to the predicted effects of global climate change. For example, Long and Mahler [24
] modelled site-specific hydraulic responses to different climate scenarios. Expectations about the consequences of climate stress on the selected karst spring in the Mediterranean region were made by means of trend analyses [25
]. Other studies mainly focus on prediction of future water availability on a global or local scale by hydrological modelling [26
]. The expected effects are primarily attributed to altered recharge conditions (i.e., decreased precipitation aggravated by increased evaporation or decreased glacier melt runoff) and excessive human interventions (i.e., groundwater abstraction). Ravbar et al. [25
] even stress that not only the quantity of spring water, but also its quality, is endangered due to expected climatological changes, especially during flood pulses caused by precipitation events after a long dry period.
Furthermore, changes in weather patterns, especially extreme events, increased CO2
emissions into the atmosphere and the effects of direct anthropogenic activities, such as rapid urbanization and industrialization, can lead to abrupt changes in vegetation cover [31
]. Although some previous studies have shown that the soil and vegetation mantle plays an important role in the infiltration and storage of water in karst systems [37
], hydrological functions of soil and vegetation layers in the hydrological cycle have been only marginally considered or even underestimated in the investigation of future hydrological processes in karst. The main reasons for this should be sought in poorly studied and understood infiltration mechanisms of soil and vegetation layers and their hydrological relationships with the overall groundwater recharge [38
Changes in vegetation cover that, for example, include afforestation or natural overgrowth, deforestation, large-scale forest disturbances due to drought, ice breakage, natural fires, windthrows, and bark beetle infestation, may substantially influence the water balance through altered evapotranspiration and filtration, and thus influence the quantity and quality of water resources and underground habitats [40
]. This may be especially problematic in large parts of the Middle-European karst areas that have been reforested in the last century and a half [37
]. In some cases, the spruce forest has been favored due to its economic values, which is particularly susceptible to windthrow and bark beetle infestation [51
]. In tropical and sub-tropical areas, strong soil erosion and rocky desertification may also occur as a consequence (e.g., [53
This leads to the need to assess the role of vegetation in recharge dynamics and intensify studies on differentiating the effects of climate and land cover changes on hydrological processes in karst. These issues will be addressed in the present study, which: (1) focuses on a catchment scale; (2) was done on a very well-studied test sites with a large amount of historical meteorological and hydrological data of high temporal resolution; (3) aimed to assess the direct impact of changes in climate and vegetation cover on karst water sources. More specifically, the aim of the study was to compare long- and short-term changes in groundwater recharge (i.e., effective precipitation) and in spring discharge, thus assessing the response of a karst aquifer to climate and vegetation cover changes.
3.1. Trend Analysis of the Hydro-Meteorological Parameters
In case of the Unica River, long-term trend analysis in the period 1962–2018 confirmed statistically significant decreasing trend in annual P (28 mm per decade; p < 0.05), increasing trend in annual ETP (28 mm per decade; p < 0.001) in its catchment.
In the 57-year measuring period (1962–2018; Qmean = 21.967 m3/s) of the mean annual discharge of the Unica River at the Hasberg gauging station (Slovenia), in 2014, a record mean yearly discharge of 40.76 m3/s was measured. This is the only value in the entire observation period that is greater than the sum of (1) the mean value in the period (21.967 m3/s) and (2) the value of three standard deviations (1 SD = 5.82 m3/s; 3 SD = 17.45 m3/s) and thus reaches the value of 40.76 m3/s. The year 2014 was exceptionally wet, which is indicated by the mean annual precipitation height from the Postojna meteorological station (2069 mm). However, the precipitation quantity hardly exceeded the sum of (1) mean annual precipitation in the period 1962–2018 (1537.6 mm) and (2) 2 standard deviations (1 SD = 235.3 mm; 2 SD = 470.7 mm). We assume that the measured discharges of the Unica River were quite high in 2014, which influenced the calculation of a trend, although the used Mann–Kendall method and the non-parametric Sen’s slope methods are not as sensitive to outliers. However, the year 2014 is rather at the end of the observed period and an existence of a high positive outlier at the end of the interval resulted in a p-value of only 0.1084. When we experimented and excluded the year 2014 in the trend calculations (the methodology we used allows this), the p-value of the mean annual discharge of the Unica River was lower than 0.05.
Overall, we decided to consider the mean annual discharge trend of the Unica River as statistically significant. It shows a decreasing trend of 820 L/s per decade (Table 2
, Figure 2
). Statistically significant are also negative trends of the absolute minimum annual discharge (170 L/s per decade; p
In case of the Rižana River, long-term trend analysis in the period 1965–2018 confirmed a statistically significant increasing trend in annual ETP of 35 mm per decade (p
< 0.001) in its catchment, while the decreasing trend in annual P (39 mm per decade) is not statistically significant (Table 2
; Figure 3
). Long-term changes in values of P and ETP are reflected in the reduced discharge of the Rižana River, for which a statistically significant decreasing trend in mean annual discharge of 380 L/s per decade (p
< 0.001) is characteristic (Figure 3
). The decreasing trends of the absolute minimum annual discharge (30 L/s per decade; p
< 0.001) and a maximum annual discharge (2.75 m3
/s per decade; p
< 0.05) are statistically significant.
3.2. Evaluation of Changes in Vegetation Cover and Their Influence on Effective Precipitation
Although according to the national database on land use the area of forest in the catchment of the Unica springs in period 2014–2018 remained practically the same, the forestry data show that growing stock decreased for 16.77% due to the large-scale forest disturbances. On the basis of this information, it was calculated that in the assessment of Pef, the share of forests should be gradually reduced from 77.1% before 2014 to 64.2% by the end of the 2014–2018 period. These values were implemented in the soil water balance model in Microsoft Excel and the model was calibrated (calibration parameters: soil water storage capacity and canopy properties) to correspond to the mean discharge of the Unica River in the same period (21.24 m3/s). In this way, the total Pef for the period 2014–2018 was estimated at 4875 mm (annual average 975 mm). Assuming that the large-scale disturbances did not occur, in the model the share of forest was set to 77.1% over the entire period. This would reduce the total Pef in the years 2014–2018 to 4643 mm (annual average 929 mm).
A comparison of the monthly values of Pef
calculated with the two models (Figure 4
) shows that the ratio ranges between 1 and 1.6 and depends on the amount of precipitation and the season. As expected, in dry periods with active vegetation, the increase of Pef
due to the decrease in the share of forests is evidently higher.
In the catchment of the Rižana spring, at present more than 2/3 of the total area is covered with forests. Only about 60 years ago, this share was slightly above 1/3. To evaluate the influence of such intensive natural vegetation overgrowth on the recharge of the karst aquifer, the same calculation mode was used as for the Unica karst spring. The actual share of forest (67.6% of the total catchment area) was taken into account and the model was calibrated to correspond to the mean discharge of the Rižana River in the period 2014–2018 (4.12 m3/s). In this way, the total Pef for the period 2014–2018 was estimated to be 3579 mm (annual average 716 mm). Assuming that natural overgrowth did not occur and the share of forest remains the same as in 1957 (37.4%), this amount would increase by 20% to 4293 mm (annual average 857 mm).
Comparison of the monthly values of Pef
calculated with the two models (Figure 5
) shows that the ratio ranges between 0.5 and 1. The ratio depends on the amount of precipitation and the season. As expected, in dry periods with active vegetation the natural vegetation overgrowth has higher influence and significantly reduces the amount of Pef
Both the Unica and the Rižana springs show a negative long-term trend in mean annual discharge in the studied periods. However, results of the long-term trend analysis show that the magnitude of a decrease in runoff differs between the studied springs. Namely, the relative decrease in mean annual discharge is greater for the Rižana spring. Since an apparent difference in long-term vegetation cover change in the catchments of the studied spring exists, both climate and vegetation factors have been considered for the interpretation of a long-term decrease of mean annual discharges of the springs. The effects of large-scale forest disturbances have been studied and discussed in the catchment of the Unica springs.
4.1. The Unica Case Study
Statistically significant long-term changes in the values of the water balance elements (precipitation and ETP) in the period 1962–2018 reflect in reduced runoff in the Unica spring catchment. From the beginning to the end of the observed period in 2018, the mean annual discharge of the Unica springs dropped from 23.86 to 19.30 m3/s or for 20.8% according to the Sen’s linear regression function. These results are in accordance with the statistically significant increase in annual ETP in the catchment of the Unica springs in the observed period, which, according to the trend function, increased from 624.10 mm in 1962 to 780.35 mm in 2018 or for 22.3%. Using trend function, we also determined 10.4% drop of annual precipitation from 1642.22 to 1483.14 mm in the observed period.
Mean monthly discharges of the Unica River in the cold span of the year show an increasing trend that is most probably the result of an increase in air temperatures in winter months in the region (Table 2
; Figure 2
). Higher temperatures mean less snow precipitation and a shorter snow residence time, which is reflected in increased runoff during the cold period of the year, when, in addition, vegetation is not active; something that has already been observed in catchments with similar relief and climate characteristics elsewhere [83
]. The effect of the described phenomenon on runoff in the Unica catchment is not significantly hindered by the increased ETP. Namely, monthly ETP trends from December to March show a statistically significant increase.
Long-term changes in the vegetation cover in the catchment of the Unica springs are relatively small for the entire study period 1962–2018. As these only account for a few percentage points, changes in the vegetation cover were not attributed any major impact on the springs. The impact of climatological changes (in particular, the increase in ETP) was instead considered more influential in the overall decrease in the Unica River discharge, although some studies indicate that there are significant seasonal variations in ETP, especially in temperate climatic zones where deciduous trees predominate and are not often exposed to drought conditions [40
More significant are the vegetation cover changes caused by ice breakage, bark beetle infestation and windthrow, which occurred in the years 2014–2018, with a gradual decrease in the proportion of forest from 77.1% to 64.2%. We therefore hypothesized consequent changes in the rate of precipitation interception and transpiration. A higher infiltration rates of precipitation into unsaturated zone and subsequent higher recharge of groundwater was expected.
To assess influence of altered forest cover on Pef
, two different scenarios were tested in a model of the soil water balance. Results show that the large-scale disturbances caused an increase in Pef
of 5%; taking into account the catchment size, this would mean an increase in mean discharge of 1 m3
/s. We explain this with the fact that reduced forest vegetation caused diminution of canopy interception, water consumption and soil evaporation (Figure 6
), which are dominant components of evapotranspiration [99
Changes in evapotranspiration processes and consequent groundwater recharge due to altered forest cover have previously been shown by other studies as well [100
]. Although our study provided a rough estimate that cannot take into account the exact natural conditions, the results are consistent with the comparison discussed hereafter. For the 5-year period 2014–2018, compared to the 52-year period 1962–2013, an increase in annual precipitation of 2.7% and annual evapotranspiration ETP of 14.8% and a decrease in annual runoff, calculated as the direct difference between precipitation and ETP, of 6.9% was calculated. Changes in the values of the above-mentioned water balance elements should be reflected in a significant decrease in the mean discharge of the Unica River. However, in the period 2014–2018 its mean discharge decreased by only 0.4% compared to the period 1962–2013. The results obtained support the assumption that the decrease in forest density due to large-scale disturbances causes an increase in Pef
and a relative increase in the discharge of the Unica River. Furthermore, our findings are consistent with virtual experiment conducted by Sarrazin et al. [40
], which is focused on the impact of land cover change.
4.2. The Rižana Case Study
From the beginning to the end of the observed period 1965–2018, the mean annual discharge of the Rižana River dropped from 4.75 to 2.74 m3
/s, or by 51% according to the trend function. For the spring catchment, using the trend function, we determined a 13.7% drop in annual precipitation from 1625.96 to 1416.08 mm in the same period. Since the negative precipitation trend is not statistically significant, we cannot consider it as an important factor of the drop in mean annual discharge of the Rižana River in the observed period, as was already indicated for the Unica River. Furthermore, unlike the Unica River, where we confirmed an increasing trend in mean discharges in winter months, no such phenomenon was determined for the Rižana River. In comparison to the Unica springs catchment, the Rižana spring catchment lies on lower altitudes with less snow precipitation, hence the effect of decreasing snow precipitation and a consequently shorter snow residence time due to a general increase in air temperatures in the observed period was not so evident [25
]. Concerning water availability for the drinking water supply of the region in the warm season, when the water demand is the highest, statistically significant negative monthly trends of Qmean and Qmin of the Rižana spring from April to September are discouraging (Table 2
; Figure 3
Although Ravbar et al. [25
] gave the effects of climatological changes as the reason for a decrease in mean annual discharge of the spring, this study seeks additional causes. The observed significant drop in the mean annual discharge cannot be simply explained by the effects of climate change with increased annual ETP in its catchment that, in the observed period, increased from 756.65 to 942.80 mm or for 22%, which is exactly the same ratio as we calculated for the Unica catchment.
A possible explanation for such a huge decrease in mean annual discharge of the Rižana River is a distinctive change in vegetation cover in its catchment in the last six decades. The percentage of forests increased from 37.4% in 1957 to 67.6% in 2018, which could have a significant impact on changing the recharge characteristics of the entire catchment. As shown in a study by Huxman et al. [103
], woody plant encroachment leads to increased plant transpiration and simultaneously to reduced availability of soil water. Similar anticipations have been reported by previous studies undertaken in karst areas [99
Analysis of evaluation of Pef in the period 2014–2018 for the Rižana spring, assuming that natural overgrowth did not occur and the share of forest remained the same as in 1957, shows that the amount of Pef would be 20% higher and supports the assumption that forest cover expansion leads to a reduction in groundwater recharge. The findings also show that without significant natural vegetation overgrowth the average discharge of the spring would increase from 4.12 to 4.95 m3/s. This value is in accordance with the calculated mean annual discharge according to the trend function of the spring for the year 1965 (4.75 m3/s). If the comparison is focused only on the summer months (June, July, August), the difference in Pef is much higher, reaching up to 49%.
The statistically significant decrease in mean annual discharge of the Rižana River in the period 1965–2018 can thus be considered as a result of interacting impacts of long-term climate and vegetation cover changes (Figure 7
), which is in line with virtual simulations made by Sarrazin et al. [40
Our research is a comprehensive study that evaluates the changes within the entire catchment area in order to investigate individual influences, i.e., of climate and vegetation cover changes, on groundwater recharge in karst regions. It suggests that the vegetation cover change in a catchment of a karst spring can, in addition to climate changes, have a significant impact on the spring hydrology over a short and long timespan. Based on observational data, a further finding of this study is that the methodology applied allowed the quantification of impacts of these changes on karst water sources.
The case of the Unica springs catchment enabled the study of the effects of changes in vegetation cover to karst spring discharge conditions due to large-scale forest disturbances in the period 2014–2018. The findings show a resulting increase in Pef by 5% (due to less canopy interception and evapotranspiration) and a relative increase in the discharge of the Unica River, which were due to changes in the water balance elements, expected to decrease more significantly. The effects of large-scale forest disturbances thus mitigated the effects expected from the trend analysis of hydro-meteorological parameters.
The catchment of the Rižana spring has been impacted by climate and land use changes in the past six decades and enabled studies of their long-term impacts. Comparing the effects of climate change, reflected in the long-term increase in ETP, and of change in percentage of forest, resulting in a long-term decrease in Pef, the study points to interacting impacts of climate and vegetation cover changes. Without significant natural vegetation overgrowth, the mean discharge values of Rižana would be 4.95 m3/s, which is 0.83 m3/s higher than the actual average value. Even more significantly, up to 49% higher Pef and consequently higher discharges would be expected in summer months (June to August), when water scarcity is a serious problem under current conditions.
Although the study is site-specific, the selected karst catchments were nevertheless recognized for the following contributions: (1) well studied complex binary karst aquifers, (2) containing a long history of both climatological and hydrological datasets of high quality and (3) exhibiting strong and clear long-term vegetation cover changes or large-scale forest disturbances. Therefore, the results of this research are transferable to other similar studies aiming to assess the direct impact of various environmental changes on karst water sources. The study is also useful for the further evaluation of ecohydrological processes and the improvement in environmental policies, as it was able to distinguish between effects of climate and land cover changes on hydrological processes in karst. Furthermore, the study has a direct relevance to numerous land management challenges, due to issues of water availability to water-dependent economic sectors in particular.