1. Introduction
Due to the incalculable value of water as an asset for all, both in the present and in the future, and the importance of streamflow as an indicator of long-term hydro-climatic changes, curbing and quantifying changes in streamflow series have become essential for water resource planning and management [
1,
2]. Streamflow integrates the influence of atmospheric variables over a catchment and, presumably, if consistent changes in rainfall and other climate variables (winds, atmosphere/heat fluxes) are observed, these should also be reflected in the catchment scale. However, this is not so obvious, because catchment characteristics (geology, slope, soils) impact rainfall-runoff transformation properties [
3,
4,
5]. Additionally, human activities can greatly influence the river systems and make it difficult for trend detection and attribution [
6,
7]. Against this background, streamflow measures in natural or near-natural rivers become of prime importance in understanding hydrological processes in an area and identifying and attributing emerging trends [
8,
9,
10]. Hydrological predictive modellings also require adequate and suitable local data to represent rainfall-runoff processes accurately. In fact, several studies have highlighted the importance of the hydrological modelling components in the rainfall-runoff modelling chain. Some of these studies investigated uncertainties linked to rainfall and the surface roughness component, and only a few focus on streamflow variations [
5,
11,
12]. In this respect, Bermudez et al. [
11] showed that a better representation of the hydrological processes occurring in the model domain and the availability of more accurate streamflow input data may reproduce a better response and improve model performance.
The term natural-regime stream is not well defined in the literature. Occasionally, it is simply described as rivers where little or no human intervention has taken place, thus rendering them highly unspoilt [
13]. However, setting thresholds to differentiate natural-regime streams is not an easy task [
14]. Various criteria relating to certain features describing the character of the riparian zone and river channel, the discharge regime and land use in the catchment were used to catalog human disturbance. For example, indicators of hydrologic alterations (IHA) are widely used to characterize human alterations of streamflow regimes [
15], but they contain 33 individual parameters. The integrated connectivity status index (CSI) is also used to determine human interference in fluvial connectivity, defining the natural or free-flowing rivers as, i.e., systems largely unaffected by changes to fluvial connectivity, allowing unobstructed movement and exchange of water, energy, material and species within the river system and with surrounding landscapes [
16]. As the impact is extensive, defining the threshold for distinguishing natural rivers is sometimes difficult. For example, Batalla et al. [
17] define the degree of impoundment as an indicator of the extent to which reservoirs could change flow, while Nardi et al. [
18] and Scheel et al. [
19] used river–floodplain disconnectivity to evaluate man-made impacts. In the present study, the term “near-natural-regime stream” is used in the same sense as it has been used in previous research into this topic [
8,
20], meaning that it describes river flow regimes minimally affected by anthropogenic disturbances, such as reservoirs, dams, channelization, water extraction. Therefore, the term near-natural does not necessarily equate to totally undisturbed pristine conditions; but it might include minor disturbances, such as land-use and land management changes.
The request for a reference streamflow dataset for near-natural catchments has been largely recognized worldwide and supported by some international programs, like FRIEND [
9]. In this context, a significant number of studies have been conducted over the last few years with the aim of detecting, measuring and evaluating streamflow trends in near-natural catchments [
6,
7,
8,
20,
21,
22]. Regarding trend studies in Europe, no uniform trend has been observed for streamflow [
5,
6]. Thus, whereas positive trends in annual, monthly and low flows were observed in near-natural catchments in Nordic countries [
23], a generalized pattern of negative trends in annual and seasonal streamflows were reported for southern European rivers [
5].
Most trend studies in the Iberian Peninsula focused on annual, seasonal, monthly and low flows. Lorenzo-Lacruz et al. [
22] detected a generalized reduction in streamflow when studying a large database of Spanish mountain rivers not disrupted by major human interference. These results are in line with the Martínez-Fernández [
20] study that found downward trends in annual and seasonal streamflows in an analysis of 74 near-natural rivers in Spain. Zabaleta et al. [
24] also highlighted dominant-negative trends in various catchments in the Basque Country. Significant downward trends in annual streamflow were also identified in the headwaters of the Ribera Salada [
21], Tagus [
25] and Duero [
26]. However, in the Águeda basin (north-central region of Portugal), Hawtree et al. [
27] found no evidence of significant reductions in streamflow, despite wide afforestation that could, arguably, be explained by the presence of compensatory climate trends over the study period.
To summarize, most of the aforementioned trend studies in the streamflow series carried out in Spain focused on regions with a Mediterranean climate. Thus, these studies do not comprehensively analyze the hydro-climatic changes of streams in the humid region of Spain (NW Spain). There is general agreement that air temperatures have increased over the last century in NW Spain [
28,
29]. However, it is unclear how these changes may have affected streamflow. Evidently, the results obtained for the Mediterranean area are not directly applicable to more humid regions. Hence, it is a challenge to examine stream discharge in humid catchments.
In this paper, a streamflow trend analysis was carried out in a headwater catchment of NW Spain (i.e., under a humid climate) with the aim of contributing to filling a gap in research on a national scale. The analysis will be carried out at different time scales (annual, seasonal and monthly), paying special attention to the duration and severity of the low flow. The choice of the study area (a headwater catchment of the Mero basin) is based on: (i) not being affected by anthropogenic alterations such as reservoirs and dams, therefore qualifying it as an example of a near-natural regimen fluvial system, (ii) its characteristics, which respond to the archetype of the Galician rural environment: small population centers with highly dispersed single-family dwellings, and (iii) the strategic, environmental and social importance of its water resources. The Mero basin is important for the future socio-economic development of the city of A Coruña and its metropolitan area (NW Spain), because it is the main contributor to the Abegondo-Cecebre reservoir, the largest source of water supply for the area. This reservoir also provides high-quality habitats for a large number of species and was cataloged as a Special Area of Conservation of Natura 2000 in 2014.
4. Discussion
The results show that discharge in the Corbeira catchment does not show a marked annual fluctuation. However, the intra-annual variability is much larger, with the maximum variability occurring in autumn.
With respect to trend analysis of climate variables, upward trends were found for the temperature at the annual and seasonal scale, as has been already reported in previous studies in the Iberian Peninsula [
9,
43]. In contrast, no significant trends were detected for rainfall on any time scale, similar to the results of Rodrigo and Trigo [
29] for the La Coruña station (located near the study area), which showed positive (non-significant) trends for annual and autumn rainfall and negative for spring and summer. Similarly, and in the context of long-term evaluation of rainfall in Galicia, Lago et al. [
28] pointed out the absence of a unidirectional trend in annual rainfall. However, at a seasonal scale, these authors indicated a possible change in distribution throughout the year, with wetter autumns and less humid winters, mainly due to the decrease in rainfall in February and springs and summers of sub-dry or dry trends. These facts have a heavy impact because more than two-thirds of the water demand takes place in the period between April and September, in which scarcely one-third of the total annual rainfall is registered. Furthermore, it must be noted that in the autumn-winter semester, the period with the highest water contribution, it is possible to record a marked rainfall deficit. An example would be the winter 2004-05 (169.4 mm in winter 2004-05 vs. 333.4 mean rainfall in winter).
With hydrological variables, the annual discharge trend test revealed that there was no significant trend for stream discharge. However, significant downward trends were observed for autumn and summer. In the present study, the observed trends cannot be explained by changes in rainfall, since Z values, although negative, were close to zero, indicating no trends. The study of low and high flows suggested upward trends in the duration of low flow and severity (volumetric deficit) due to the increase in the number of days with low flow in summer and autumn (
Table 3). In contrast, the duration of the high flow period seemed to decrease, particularly during autumn. The longer low flow duration and severity found in our study are in line with the results from other studies in temperate humid catchments [
24,
27] indicating an extension of the low flow period during autumn, which resulted in a longer period of time with low stream discharge, as well as a temporary increase in the volumetric deficit. Given the length of data, it is difficult to say whether the increase of evaporative demand during these seasons, induced by higher temperatures, has caused the apparent trend. Several studies in the Iberian Peninsula have already reported a reduction in streamflow due to the increase in air temperatures [
8,
9,
20]. For example, Morán-Tejeda [
43] observed an increase in the number of days with low flow and a reduction in the frequency and magnitude of high flows in the Duero basin. These authors argued that this behavior is a consequence of raising temperatures by enhancing evapotranspiration, and changes in the land-cover, as a result of re-growth of vegetation, whose effects are more evident during the growth period (spring and summer), as it is associated with the highest demand for water from the soil, and a greater capacity for rainfall interception by the canopy [
43]. Other authors have also associated re-vegetation and land-cover expansion in headwaters as a primary cause of decreasing stream discharge in the absence of rainfall trends.
The observed discharge trends in the Corbeira catchment and hence the Mero basin may have important implications for water management in the study area. The Abegondo-Cecebre reservoir, which is mainly fed by water from the Mero basin, was built in the mid-1970s to guarantee the availability of water for the city of A Coruña and its metropolitan area during drought periods and to reduce negative effects from floods. The decrease in discharge, as well as increase in the low flow period, enhance future risk and vulnerability especially in summer, which is characterized by the low flow. In recent years, there has been a substantial increase in the population in the area (from 250,000 to 450,000) representing a large rise in water demand. Moreover, tourism is growing in the city of A Coruña and its surroundings, especially in summer when the population may even double, which makes it necessary to satisfy an additional water demand at what is undoubtedly the least rainy time of year. In some cases, this causes the Abegondo-Cecebre reservoir to be insufficient if there has not been enough rain to fill it in the previous months, as occurred in 2010, triggering water restrictions and threatening the ecological environment, society and economy. This highlights the fact that alternative strategies for water management may come to the fore in low flow seasons (the most critical period for water management in the area) in order to reconcile the heavier demand for water for human activities with maintaining functioning riparian ecosystems, as it also has to be borne in mind that the reservoir is a Special Area of Conservation.
5. Summary and Final Remarks
An analysis of trends in annual and seasonal stream discharge (mean, low and high flows) was conducted in the headwaters of the Mero river, a near-natural system representative of the climate and land use characteristics of the northwestern Iberian Peninsula catchments. For this, both the non-parametric Mann–Kendall test and the Sen methods were applied.
The study did not detect statistically significant trends in annual stream discharge. However, significant downward trends in mean discharge were observed for autumn and summer. In addition, a significant upward trend in the number of days with low flow was particularly evident in spring and summer. Additionally, a falling trend in the high flows was observed in autumn. On the contrary, rainfall showed a positive pattern, although it was not significant. The different behavior shown by rainfall and discharge may be explained by the reduction of water resources associated with the increase in temperature in the study area, although this must be interpreted with caution, given the absence of long-term measurements.