Examensarbete i biologi avseende kandidatexamen, 15 hp VT 2021
Effects of drought on water chemistry in a boreal stream
network
Enrique Gómez de Salazar Martínez
Effects of drought on water chemistry in a boreal stream network
Enrique Gómez de Salazar Martínez
Abstract
Hydrological drought at high latitudes represents a rising environmental hazard induced by global climate change. Yet, we still know little about how drought events influence the biogeochemistry of boreal streams. Here, I used 15 years of data from eight streams within the Krycklan Catchment to test how interannual variability in summer low flows influences stream water chemistry. My analysis focused several key biogeochemical indicators in these streams, including concentrations of dissolved organic carbon (DOC), dissolved organic nitrogen (DON), nitrate (NO3) and ammonium (NH4), as well as the total C/N and NH4/NO3
ratios. Overall, results revealed widespread declines in summer average DOC concentrations and C/N ratios with greater drought severity. These responses likely reflect shifts in the biogeochemical properties of soils that feed streams during high- versus low-flow summers.
By comparison, nitrogen-based parameters were less clearly influenced by drought, except for in mire-dominated headwaters, where NH4 and DON both increased during the lowest flow periods. Overall, the strong effects of flow variability drove a high degree of interannual synchrony for DOC and C/N across all sites in the drainage system. This synchrony was more variable overall for nitrogen-based parameters, with several sites having unique year-to-year changes in concentrations and ratios. However, strong temporal coherence for NH4 across forested streams suggest other broad-scale factors (e.g., related to forest processes) may regulate interannual patterns for this nutrient. Collectively, results provide insight into how increases in drought frequency and severity may alter boreal streams and rivers in the future.
Key words: drought, boreal streams, Krycklan Catchment Study, land cover, biogeochemistry.
Table of contents
1 Introduction
11.1 Background
11.2 Aim of study and scientific questions
22. Methods
32.1 Study sites
32.2 Field sampling
42.2.1 Long term drought assessment 4
2.2.2 Water record sampling 4
2.3 Statistical analyses
53. Results
53.1 Average chemical trends
53.2 Responses to drought
63.3 Interannual synchrony in water chemistry
94. Discussion
104.1 Effects of drought on dissolved organic compounds
104.2 Effects of drought on landscape synchrony
125. Conclusion
126. Acknowledgements
137. References
131
1. Introduction
1.1 Background
Understanding the consequences of climate change is a central goal in environmental science and key for predicting the future of ecosystems globally. In this aspect, drought stands as one of the main rising trends affecting both environment and society (Van Loon et al. 2016).
Hydrological droughts refer to below normal conditions of low or lack of discharge in surface streams and groundwater, which can affect aquatic ecosystem biogeochemistry and
community dynamics, as well as general water quality (Mosely 2015). Drought occurrence is a result of climatic variations affecting atmospheric conditions that ultimately lead to a lower water input to the catchment. These shifts include broad periods of precipitation deficiency or temperature variations that may affect snowmelt water release and/or evapotranspiration processes. While these events occur globally, the consequences of drought for aquatic
ecosystems have mostly been studied in regions with arid or Mediterranean climates, which are frequently exposed to low flow periods in their natural annual cycles, usually intensified by impacts associated with human activities (Lake 2003). However, current climate models suggest that similar events may become more frequent and severe in boreal landscapes of northern Europe (Spinoni et al. 2018), where they have not been a focus of research in the past. In fact, over the last 20 years, multiple severe dry spells have been documented at high latitudes, including an extreme drought that hit northern Europe in the summer of 2018 (Gomez-Gener et al. 2020). Given that the effects of drought have not been well studied at high latitudes, and that such event may become more frequent, further research in this area becomes important to assess this dynamic environmental hazard.
The chemistry of boreal headwater streams is strongly regulated at the land-water interface, from riparian zones rich in organic matter (OM), which are the primary source of solutes to streams (Blackburn et al. 2017, Ledesma et al. 2017). The accumulation of OM in the riparian zones (RZ) over time as a result of peat formation is a defining feature of many boreal
landscapes and is influenced by both topography and climatic conditions, such as shallow water tables and cold temperatures through the year leading to slow decomposition and paludification (Ledesma et al. 2017). Biogeochemical fluxes from soils to surface waters are closely related to the hydrological connectivity through these organic soils (Laudon and Sponseller 2017). In a vertical dimension, the lateral water and solute fluxes from riparian soils are further influenced by the water table depth. Accordingly, soil permeability and conductivity increase towards the surface and shallower water tables have the highest flow contribution to the stream, increasing lateral connectivity along the hillslope (Bishop et al.
2004). Deeper groundwater, on the other hand, often contributes little to this lateral flow because hydraulic conductivity through deep soils is slow. As a consequence, on an annual basis, most of the solute and water contribution to the stream occurs in a narrow range of soil depths specific for each RZ, or the Dominant Source Layer, which tends to dominate land- water exchange (Ledesma et al. 2017).
Shifts in the water table over time have a direct effect on stream water chemistry though the influence of redox state, as well as by changes in riparian soil chemistry with depth.
Typically, deeper RZ groundwater is hypoxic/anoxic due to the very slow movement of water and elevated bacterial oxygen consumption in OM-rich waters (Helton et al. 2015). These conditions promote the accumulation of reduced solutes and gases, including important nutrients like ammonium (NH4+), whereas shallow groundwater is more likely to have oxidized forms (e.g., nitrate; NO3-) (Blackburn et al. 2017). Additionally, decreases in soil organic matter content with depth in riparian zone (e.g., Nyberg 1995) may in turn influence the concentrations and characteristics of DOC delivered to streams. These changes in soil properties are likely to be relevant to understanding how drought events influence water chemistry in boreal landscapes, yet most of what we know about soil-water table interactions comes from studies of flooding effects (e.g., Bishop et al. 2004). However, some research does indicate that periods of drought can result in streams with lower DOC concentrations
2
(e.g., Schindler et al. 1997, Worrall and Burt 2008) and elevated concentrations of reduced solutes (Gomez-Gener et al. 2020). These observations suggest drought effects on boreal stream chemistry are predictable from vertical changes in groundwater chemistry. Yet, no studies have tested the importance of these influences using long-term records that span large gradients in drought severity in northern boreal landscapes.
In addition to the vertical regulation of water chemistry in riparian soils, differences in land cover attributes (e.g., forest vs. mire cover) across boreal catchments can drive spatial patterns in stream biogeochemistry. In this context, mires are well known to be key sources of dissolved organic carbon (DOC) and other solutes to streams (Laudon et al. 2010). High concentrations of DOC contribute to an increase of bacterial productivity by supporting their energy/carbon requirements, while also potentially constraining autotroph communities by reducing light penetration (Berggren et al. 2010). Under normal circumstances, in mixed land-cover boreal streams, mires provide a large proportion of DOC during low flow periods.
By comparison, under higher flow periods (e.g., snowmelt), DOC supply is dominated by near-surface riparian forest soils, which contribute low molecular weight organic compounds with relatively high bioavailability (Berggren et al. 2007). This change in source contribution creates differences in DOC dynamics across catchments and over time (Laudon et al. 2010).
However, much of what know about these landscape controls comes from studies of flood responses, and we know little about how these catchments features influence chemistry during drought. We know that mires can support greater water fluxes to streams during low flow periods (e.g., summer; Karlsen et al. 2016), and represent major sources of solutes under these conditions (Sponseller et al. 2018), but it is not clear whether or how reducing the water table in mires will influence stream chemistry during droughts.
Finally, the majority of the boreal river network length is made up by small streams
(headwaters), which represent up to 90% of the total channel length (Bishop et al. 2008). The abundance of these streams, which are mainly influenced by relative shallow groundwater sources and solute inputs from surrounding RZs, makes these areas especially important for the catchment biogeochemical fluxes. On the other hand, larger streams and rivers are less dependent of the RZ controls as they receive greater water and solute inputs from deep groundwater sources (Tiwari et al. 2017). These deeper groundwater sources are known to increase abruptly their water inputs as drainage area increases, reaching up to 70-80% of contribution to catchments than 10.6 km2 (Peralta-Tapia et al. 2014). Under regular flow conditions, these large groundwater inputs feeding surface streams contribute relatively low DOC concentrations as well as greater concentrations of base cations that reflect weathering processes in deep sediments. From these observations, it is noticeable how heterogeneity of water sources with distinctive properties can regulate the biogeochemistry of stream at network scales (Peralta-Tapia et al. 2014). High flow periods such as flood events modify the patterns of solute inflows further, generally increasing stream DOC concentrations with discharge due to an enhanced accessibility to the organic matter pools resulting in its mobilization (Fork et al. 2020). Yet, as above, it is not clear how these larger streams and rivers respond to low flow periods. It might be that less reliance on near surface soils will make these systems less sensitive to a lowering of the table, but this has not been tested.
1.2 Aim of the study and scientific questions
The objective of this study is to use 15 years of monitoring data to ask how summer drought conditions influence water quality at network scales in the Krycklan Catchment Study (KSC).
The investigation is motivated by recent climate trends, including warmer temperatures and elevated evapotranspiration, which may promote environmentally hazardous conditions like hydrological drought (Laudon et al. 2021). To answer this question, I used several
biogeochemical indicators measured during July and August each year, as these months have the highest drought incidence owing to warmer and dryer climatic conditions. Specifically, I focused on carbon (C) and nitrogen (N), including organic forms like DOC, dissolved organic nitrogen (DON), and the C/N ratio, which all may reflect how changes in lateral flowpaths
3
intersect with near stream soils or peatlands. In addition, I analysed the responses of the dominant inorganic nitrogen forms, including nitrate (NO3-N) and ammonium (NH4-N), which are often are limiting nutrients in boreal stream (Burrows et al. 2021). In addition, these solutes are redox sensitive and therefore the ratio (NH4/NO3) may be useful for exploring how low-flow conditions influence the metabolic of stream and near-stream environments. Specifically, I expected that extreme low flow would promote lower oxygen conditions (Gomez-Gener et al. 2020), which should favour the reduced form of the inorganic N (NH4) over the oxidized form (NO3).
2. Methods
2.1 Study sites
The study used long term (2005-2019) data from Krycklan Catchment Study (KCS), an experimental area stablished for characterizing the hydrological and biogeochemical dynamics of Boreal ecosystems. The KCS is located ca. 50 km northwest of Umeå in Västerbotten County (Figure 1; see Laudon et al. 2021 for complete site description). Land cover in the KCS is dominated by coniferous by forest, as well as headwater mires, and a small number of lakes. In terms of climate, the area has cold humid weather and persistent snow cover on winter, with spring snowmelt events also affecting the catchment. The mean annual temperature is 1.8C (January -9.5C; July +14.7C), and mean annual precipitation is 614 mm (Laudon et al. 2013). The KCS holds a long-term record of hydrological and
biogeochemical data from ca. 18 streams, constituting a unique infrastructure for catchment research at northern latitudes. My study was centred on the headwater streams with distinct land cover attributes, including C2 (12 ha; 99,9% forest cover), C4 (18 ha, 44,1% mire cover), and C5 (65 ha, 6,4% lake cover, 39,5% mire cover). I also analysed drought responses for several larger downstream sites that are dominated by forest cover, including C6 (110 ha), C7 (47 ha), C9 (288 ha), and C13 (700 ha). Finally, I analysed drought responses for the largest stream/river that drains the KCS (C16; 6790 ha). KCS sites affected by artificial structures (for example, recent forestry) were excluded from the analysis.
4
Figure 1. Krycklan Catchment Study area sitemap. This study focuses on the streams: C2 (12 ha), C4 (18 ha), C5 (65), C6 (110 ha), C7 (47 ha), C9 (288 ha), C13 (700 ha) and C16 (6790 ha).
2.2 Field sampling
2.2.1 Long term drought assessment
Drought severity for each July and August period between 2005-2019 was estimated by the number of days where stream flow at the long-term Krycklan gauging station (C7) was below the long-term 10th percentile value, which has been used elsewhere as a proxy for severe drought (e.g., Hosen et al. 2019). This cut-off value was determined by charactering the distribution of all daily discharge values for C7 from 1991- 2020. The 10th percentile value from this distribution was 0.5 L/s. Then, for each July/August period between 2005 and 2019, the total number of ‘low flow’ days, or days with discharge below this threshold was counted (Figure 2). This number was used as a proxy for summer drought and as
independent variable in correlation analyses with average chemical variables for July/August each year.
Figure 2. Summer flow records for the C7 catchment from 2003 to 2019 (July and August). Blue dots represent measured days above or below the 0,5 l/s mark, which was used here to indicate drought conditions.
2.2.2 Water record sampling
Water chemistry data for July and August were obtained from the Krycklan Catchment monitoring database (https://data.krycklan.se/). During this time of year, samples were collected approximately biweekly using pre-washed bottles and filtered on the same day in the lab. Filtered samples were refrigerated prior to analysis of dissolved organic carbon (DOC) and total dissolved nitrogen (TDN), and frozen before analyzing nitrate and
ammonium. DOC and TDN were analyzed using the combustion catalytic oxidation method on a Shimadzu TOCVCPH analyzer (Shimadzu, Duisburg, Germany). Nitrate (NO3-N;
Method G-384-08 Rev. 2), and ammonium (NH4; Method G-171-96 Rev. 12) were analyzed using a SEAL Analytical AutoAnalyzer 3 (SEAL Analytical, Mequon, WI, U.S.A). DON was
5
calculated by subtracting total dissolved inorganic nitrogen (DIN, NO3-N + NH4-N) from TDN. The C:N ratio was calculated as the mass ratio between DOC and TDN.
2.3 Statistical analyses
The raw data were collected and uploaded to an Excel file for the analysis. Average
concentrations of each compound and average values for both ratios (C/N and NH4/NO3) were calculated from data collected during July and August for each year from 2005 to 2019.
Note that the data records for DON, NO3, and NH4 did not begin until 2008.
I used Pearson correlation analysis (Sigma Plot V. 14) to test whether the year-to-year
changes in summer water chemistry were correlated with the severity of drought, represented by the number of days in July/August for which flow was less than 0.5 l/s. I repeated this test for DOC (mg/L), DON (mg/L) C/N (molar), NO3-N (µg/l), NH4-N (µg/l) and NH4/NO3
for each of the studied streams (C2, C4, C5, C6, C7, C9, C13 and C16). Significant correlations were based on a p-values of less than 0.05, but p < 0.1 was also acknowledged as marginally important give the relatively low power of the test.
Finally, to ask whether the year-to-year changes in water chemistry were ‘synchronous’
across these sites for a given variable, I extracted and analysed all of the pairwise correlations for each compound or ratio between the different streams. From this, synchrony across sites is tested by the significance of pairwise correlations for a specific chemical indicator. Here, variables that are changing in the same way from headwaters to the outlet (i.e., are
synchronous) should generally have high pairwise correlation coefficients (r). High
synchrony may be expected, for example, if drought responses by a given solute are similar across all streams in the KCS network. By contrast, ‘asynchrony’ may emerge for chemical variables that are controlled by local and/or potentially unique drivers across sites and these should have low pairwise correlations over the period. Finally, I tested for differences in median synchrony across compounds and ratios using a non-parametric ANOVA test (Kruskal-Wallis).
3. Results
3.1 Average chemical trends
The KSC´s biogeochemical record during the years of the study integrates the interannual changes for each compound and ratio providing the basis for my investigation. Overall, the sites I used in this study varied considerably in terms of average stream chemistry, but particularly for DOC. For this parameter, the highest average concentrations were at sites 4 (37,5 mg/L) and 7 (27,1 mg/L), and the lowest was observed for site 16 (10,2 mg/L). Also, the greatest difference in DOC from year to year was observed for site 4, with 33,2 mg/L of difference between the summer with the highest average DOC and the summer with the lowest. Unsurprisingly, overall trends for DON across sites were similar to DOC. Here, maximum average DON concentrations were also observed at sites 4 (540,5 ug/L) and 7 (441,0 ug/L), and minimum at site 16 (227,0 ug/L). The largest interannual changes in average DON were observed for sites 4, 13 and 5; with 321,4, 299,7 and 249,6 ug/L of difference, respectively. Compared to DOC and DON, the C/N ratio was more consistent across sites, with the highest average value at sites 4 (66,6), 7 (58,5) and 2 (57,3), and the lowest at site 16 (42,6). The largest year-to-year changes in C/N were observed for sites 4 (32,2) and 5 (27,5), with the lowest observed for site 16 (15,5).
Inorganic nutrients (NO3, NH4, and NH4/NO3) also varied across sites, but the range of concentrations were relatively more constrained when compared to DOC, DON, and C/N.
For example, maximum NO3 concentrations were observed at sites 7 (20,6 µg/l), 16 (18,9 µg/l) and 9 (18,7 µg/l), and minimum averages at site 2 (8,9 µg/l). For NO3, interannual ranges were greatest at site 16 (21,8 µg/l) and lowest at site 2 (10,40 µg/l). NH4 had larger
6
variation across sites, with average concentrations reaching up to 56,5 and 40,7 µg/l for sites 4 and 5, and with the lowest averages of 5,5 and 7,6 µg/l at sites 16 and 2, respectively. These patterns also corresponded to elevated interannual ranges found at sites 4 (86,8 µg/l) and 5 (157,7 µg/l), as these represent the streams with much higher NH4 concentrations. Lastly, the average NH4/NO3 ratio was greatest at sites 4 (4,3) and 5 (3,5), as compared with the lowest values at streams 16 (0,3) and 7 (0,6). Finally, the interannual range in the NH4/NO3 ratio was highest at sites 4 (7,3) and 5 (16,7). For the rest of the streams, this variation was much closer to the lowest value (0,8) found at site 16.
Table 1. Averages plus maximum to minimum values for DOC, DON, C/N, NO3, NH4 and NH4/NO3 concentrations and ratios from the 2008-2019 period.
3.2 Responses to drought
Overall, there was variation in the degree to which water chemistry variables correlated with summer drought (Table 2). Specifically, DOC and C/N were the most drought sensitive overall, showing a negative correlation between concentrations or ratios and drought occurrence through the years, for nearly all sites (Figure 3 and 4). Although all sites showed this negative trend, three of them weren’t significant for the analysis due to p-values greater than 0,05. This includes DOC at C2, as well as the C/N ratio at C2 and C16.
By comparison, DON, NO3, NH4, NH4/NO3 were less universally influenced by drought, with only a few sites having significant correlations (Table 2). However, when there were
significant patterns, drought tended to result in higher concentrations of reduced forms of N (DON and NH4). For example, DON concentrations at C4 increased significantly (+0,64, p<
0.05) with greater drought severity, yet no other sites showed significant correlations for this variable. NH4 concentrations also increased significantly with drought severity, but only for the C4 and C5 headwater sites. NH4/NO3 ratio also increased significantly at these two sites, as well as at C6, and marginally (p<0.1) for C2. Finally, for NO3, the oxidized from of
inorganic N, only C16 showed a significant (negative) correlation with drought severity (r = - 0.62).
Table 2. Correlation values (r values) between DOC, C/N, DON, NO3, NH4 and NH4/NO3 parameters and the study sites C2-C16.
C2 C4 C5 C6 C7 C9 C13 C16
DOC -0,38 -0,67** -0,59** -0,76** -0,80** -0,78** -0,63** -0,80**
C/N -0,45 -0,86** -0,83** -0,66** -0,81** -0,73** -0,67** -0,47
DON 0,02 0,65** 0,42 0,15 0,21 -0,17 0,33 -0,28
NO3 -0,31 -0,05 0,35 -0,16 0,40 0,43 -0,24 -0,62**
NH4 0,131 0,82** 0,71** 0,14 0,11 0,16 -0,02 0,04
NH4/NO3 0,50* 0,86** 0,69** 0,61** 0,02 -0,02 0,18 0,44
DOC DON C/N NO3 NH4 NH4/NO3
Site 2 22,4 (28,5 - 16,3) 387,0 (487,4 - 333,1) 57,3 (68,6 - 46,2) 8,9 (15,7 - 5,3) 7,6 (16,6 - 0,8) 0,8 (1,3 - 0,1) Site 4 37,5 (47,6 - 14,4) 540,5 (773,1- 451,8) 66,6 (83,3 - 51,1) 15,0 (24,2 - 9,9) 56,5 (103,4 - 16,6) 4,3 (8,6 - 1,4) Site 5 17,9 (22,8 - 13,2) 351,5 (500,6 - 251,1) 46,7 (54,6 - 27,1) 13,7 (18,5 - 7,9) 40,7 (168,5 - 10,7) 3,5 (17,6 - 0,9) Site 6 16,2 (22,2 - 6,0) 314,2 (375,0 - 262,6) 49,8 (62,4 - 41,7) 16,6 (25,5 - 9,0) 19,6 (35,3 - 8,8) 1,2 (1,6 - 0,8) Site 7 27,1 (32,6 - 17,4) 441,0 (491,6 - 373,7) 58,5 (70,0 - 46,8) 20,6 (29,6 - 12,4) 12,2 (25,8 - 2,0) 0,6 (1,0 - 0,1) Site 9 16,9 (22,0 - 9,9) 318,7 (383,3 - 266,0) 50,3 (59,1 - 42,9) 18,7 (25,6 - 13,0) 13,9 (27,2 - 5,4) 0,7 (1,2 - 0,4) Site 13 23,1 (29,3 - 14,8) 436,8 (673,7 - 373,9) 53,3 (63,1 - 41,3) 12,2 (20,3 - 9,0) 11,0 (25,2 - 4,1) 0,9 (1,3 - 0,3) Site 16 10,2 (16,1 - 2,1) 227,0 (292,7 - 160,9) 42,4 (51,3 - 35,8) 18,9 (28,5 - 6,7) 5,5 (16,6 - 0,5) 0,3 (0,8 - 0,0)
7
Figure 3. DOC to drought (<0,5 l/s) correlation values and trendlines across sites; graph 1 (C2, C4 and C5), graph 2 (C6, C7, C9 and C13) and graph 3 (C13).
8
Figure 4. C/N ratio to drought (<0,5 l/s) correlations and trendlines across sites; graph 1 (C2, C4 and C5), graph 2 (C6, C7, C9 and C13) and graph 3 (C13).
9
3.3 Interannual synchrony in water chemistry
The overall degree of pairwise synchrony across the KCS network was significantly different among the chemical forms addressed in this study (Kruskal Wallis H = 33.1, p < 0.001;
Figure 5). Consistent with widespread responses to summer drought, DOC and C/N were highly synchronous from year to year (median pairwise correlation was 0,738 and 0,776, respectively). The overall strength of synchrony for these two variables is seen both in the high median correlation and by the only minor variation in correlation values across all pairwise combinations. Indeed, the significance test between the different sites for these measures reveals a synchronous behaviour for nearly all streams. The only exceptions to this shared pattern include C2’s pairwise relationship to C4 and C5 for DOC, and C2’s
relationship with C4, C5 and C16 in the case of the C/N ratio.
In contrast to DOC, DON was significantly less synchronous overall (with a median correlation of 0,398), with fewer significant correlations across sites, and a much broader variation between their values, indicating a synchronous response shared only by a few streams. Similarly, NO3 and NH4/NO3 were also relatively less synchronous at interannual scales, with median correlations of 0,616 and 0,548, respectively. However, while NH4 was not strongly influenced by drought, except for the mire-dominated headwaters, it was highly synchronous across the forest-dominated catchments on a year-to-year basis (median correlation of 0,897), although the behaviour of sites C4 and C5 enlarged the range of the synchrony correlations.
Figure 5. Box plot presenting median, average and variations for site-to-site correlation values of DOC, C/N, DON, NO3, NH4 and NH4/NO3.
10
4. Discussion
The goal of this study was to assess how potentially rising trends of summer low flow periods during July and August affect biogeochemical parameters of boreal streams. Overall, results show that episodes of summer drought over the last 15 years have had a direct influence on water quality in streams draining the Krycklan landscape. More specifically, my results indicate relatively large, immediate, and widespread impacts of drought on carbon-based parameters such as DOC and C/N ratio, both of which decreased among summers as function of drought. By comparison, drought effects were less strong, and less widespread for
nitrogen-based parameters, including DON, NO3, NH4 and NH4/NO3, which appeared to have more variable and site-specific responses to low flow conditions. The shifts in water chemistry revealed by this analysis illustrate the potential for global climate change to alter the attributes and dynamics of boreal ecosystems as a direct result of hydrological drought intensification.
4.1 Effects of drought on dissolved organic compounds
As my results show, drier summer conditions at northern latitudes were responsible for the lowering of DOC concentrations of boreal headwater streams. According to this, DOC stands as a very sensitive parameter to drought with essentially all streams showing some sort of decline. However, magnitude of potential variation in summer DOC was greater for some streams than others. For example, at C4, there was a 27,02 mg/l difference between the driest year (2006: 14,43 mg/l) and one of the most humid ones (2017: 41,45 mg/l). By, comparison, site C5 was less variable, with a difference of 3,68 mg/l between 2006 (13,19 mg/l) and other of the wetter years; 2016 (16,87 mg/l). Overall, these observations for DOC are consistent with our understanding of solute generation in boreal streams, as well as with other studies on this topic. Firstly, the results match the water table model predictions, in which the vertical patterns of water flow (very sensitive to drought) drive the lateral DOC inputs from the riparian zone towards the stream (Bishop 2004, Ledesma et al. 2017). Specifically, high DOC concentrations in streams emerge when elevated water tables intersect with organic- rich, surficial soils, for example during floods (Laudon et al. 2011). My results illustrate the flip-side of this coin, where extremely low water table levels induced by drought shifts the sources of the solutes to deeper soil layers that are more depleted in organic matter. While this mechanism is logical for the forest-dominated streams, it is interesting that drought effects were similarly strong for the mire outlet site (C4), where the vertical distribution of solutes in the mire-peats is not expected to mirror forest soils. The results may reflect the importance of preferential flowpaths through the mire (Sponseller et al. 2018), but this would require additional study. Regardless, my overall results are also consistent with others who have reported reduced DOC concentrations during drought (Schindler et al. 1997, Ylla et al.
2010). Furthermore, while these shifts in flowpaths during drought may be associated with other water quality problems (e.g., stream anoxia, Gomez-Gener et al. 2020), the reduction in DOC inputs could also have important consequences for aquatic bacterial communities’
dependant on carbon availability (Berggren et al. 2007) and so, the ecological dynamics of boreal streams.
Like DOC, the C/N ratio was also highly sensitive to drought at most sites. For example, at C4, the average C/N ratio was 83,3 during one of the wettest years (2017), but fell to 51,6 in the driest year registered for this parameter (2018). Because DON represents the dominant fraction of N pool in these streams (Blackburn et al. 2017), the C/N ratio largely reflects the stoichiometry of dissolved organic matter. Further, C/N dynamics are commonly used to assess organic matter quality related to microbial decomposition processes through the soil layers, assessing as well the soil degree of humification and relative N and C deposition (Callensen et al. 2007). In this sense, ratios of C/N are expected to decrease with soil (and peat) depth due to OM processing, reducing the carbon fraction concentrations and so, minimizing the ratio (Rumpel et al. 2011). Mineral nitrogen found on deep soil layers also contribute to the reduction of this ratio aside from the lower OM content that also affects the
11
parameter. This nitrogen fraction appears to be retained in microbial biomass during decomposition processes (Broder et al. 2012). Overall, my results are most consistent with the hypothesis that, during drought, the decreasing water tables intersect with ever deeper soil layers that are poor in carbon relative to nitrogen. In this sense, the stream C/N parameter also represents an example of how the vertical stratification of soils may affect surface water chemistry. Alternatively, it is possible that the relationship between C/N and summer drought reflects biogeochemical processes that leave a chemical signal in streams, which would otherwise be diluted under wetter conditions. For example, when anoxic groundwater hits the surface stream, interactions between DOC and iron can form particles that leave behind a dissolved organic pool that is more nitrogen-rich (Einarsdottir et al.
2020). Regardless of the mechanism, it important to consider the effects of the ratios changes under dry conditions on streams communities that could be vulnerable to the decreased carbon input. Adding to this, this organic carbon found on deep soils also would have lower quality as it’s already been processed by microorganism action, very likely resulting in low bioavailability to aquatic microbes
Unlike DOC and C/N, stream DON concentrations showed very little sensitivity to drought.
Indeed, for only one site (C4), was DON responsive to drought, and this case concentrations increased (r = +0.65), rather than decreased. Given that DOC and DON are measures of the same dissolved organic matter pool, and are often strongly correlated (Vazquez et al. 2015), it is interesting that their respective responses to flow variability were not more similar. One thing this difference highlights is that the sensitivity of C/N to drought described above is driven almost entirely by how DOC responds, rather than DON. This response is consistent with changes in soil C/N ratio with depth described above, which occur mainly as a result of carbon depletion in deeper soil layers. Accordingly, the nitrogen fraction of this ratio, represented mostly by DON, remains relatively unchanged by drought with effects only seen at sites influenced by mire drainage (e.g., C4). Headwater mires are known to accumulate reduced organic and inorganic N compounds in the deepest layers (the catotelm), where anoxic conditions and cold temperatures constrain biological processes and limit access to N by overlying vegetation (Sponseller et al. 2018). Therefore, increases in DON concentration in the mire outlet during low flow summers may be the consequence of these deeper peat layer supporting the bulk of baseflow during these periods.
Like DON, both forms of inorganic nitrogen were relative insensitive to drought, with only a few sites showing a significant response. For NH4, the expectation was that drought periods would promote the redox conditions that would favour accumulation (Helton et al. 2015;
Gomez-Gener et al. 2020), but we only observed this for two of the headwater sites. Further, this pattern was specifically found at C4 and C5; both streams fed by mire drainage from the surrounding area or from a lake source, respectively. These findings are consistent with the previously-discussed mire effects, including the storage of reduced chemical forms in their deep peat layers and the subsequent release to nearby streams. For NO3, the expectation was that drought periods would promote the low redox conditions that support denitrification, and therefore concentrations in the stream would go down during these periods. Yet, this analysis only finds support for this pattern at a single site, the main KCS outlet (C16). By correcting for changes in the overall pool size, the NH4/NO3 is perhaps a more sensitive indicator of biogeochemical changes that may favour reduced over oxidized inorganic nitrogen forms. And, in fact, the correlation analysis showed that this ratio increases during drought for a larger number of sites, although all were located in the headwaters (C2, C4, C5, and C6). From the inorganic nutrient effects found at these sites, it is also worth mentioning that stream responses appeared to be rather local, without strong effects over downstream.
For example, while low flow conditions at C4 were linked to high concentrations of NH4, this was not true for the monitoring site just downstream (C7). Such patterns indicate rapid instream transformation between monitoring stations, particularly for NH4, which can be taken up by stream microbes, but also nitrified (converted to NO3), dissipating the mire effects downstream (Lupon et al. 2020).
12
4.2 Effects of drought on landscape synchrony
Overall, there was variation in the degree of interannual synchrony in chemical responses across the network for the different measured parameters. In this case, DOC concentrations and the C/N ratio were characterized by high synchrony, based on pairwise correlations among all streams, with correlation coefficients (r) as high as 0,95 and 0,98, respectively.
This synchronous response for DOC and C/N across almost all streams reflects the dominant and broad-scale influence of summer hydrology, including drought, on these variables.
Surprisingly, the DOC response to low and high flow summers found for the different headwater sites was also shared by larger streams (e.g., C16), for which DOC concentrations are thought to be less related to connections with shallow, organic soils and more dependent on deeper groundwater sources (Tiwari et al. 2017). The chemistry of these deeper flowpaths is driven by alluvial sediments with distinctive composition and biogeochemical processes that commonly result in lower DOC and increased base cation concentrations (Peralta‐Tapia et al. 2015). In this sense, the influence of summer flow conditions on average DOC
concentrations at C16 may be driven by different mechanisms when compared to headwaters.
Specifically, this response may reflect how years with high or low discharge act to either strengthen or weaken, respectively the longitudinal connections to DOC sources in the upstream headwaters, rather than the connections to lateral soils (Fork et al. 2020).
Accordingly, during periods of elevated flow, the relative contribution of DOC from of headwaters (and thus shallow flowpaths) increases (Buffam et al., 2007). During low flow summers, these longitudinal connections are weakened, and stream chemistry takes on the character of the deeper groundwater.
Compared to DOC and C/N, interannual synchrony was weaker for DON, NO3, and NH4/NO3
concentrations, with only subsets of the sites having elevated pairwise correlations. This suggests that the dynamics from year-to-year for these solutes and ratios are more site specific, and perhaps this is not surprising given that all of these parameters were relatively insensitive to variation in summer flow conditions, including drought. Yet, one surprising result from this analysis is that interannual synchrony emerged for solutes that did not show strong sensitivity to drought. Specifically, NH4 concentrations were often strongly
synchronous across sites, with a median pair-wise correlation of 0,897. While the mire- dominated catchments were more unique, the year-to-year changes in NH4 concentrations among forest-dominated catchment in the KCS were incredibly similar. This synchrony suggests that some another driver, operating at the scale of the entire KCS, is acting to determine whether NH4 concentrations are high, low, or moderate in any given summer.
Given that nitrogen supply limits the productivity of boreal forests, and that plants and soils compete strongly to take up inorganic forms (Högberg et al. 2017), the emergent synchrony in stream chemistry may be linked to broad-scale climate conditions that regulate how well terrestrial ecosystems retain NH4 in soils, versus exporting it streams. Given that NH4
tended to be unresponsive to year-to-year summer flow conditions, the factors driving changes in concentration from may be antecedent, and connected either to winter conditions or even the plant growth conditions in the previous year. Such hypotheses obviously require more testing, but NH4 behaviour across the watershed opens up new questions about biogeochemical control of boreal streams, and highlights the idea that different chemical constituents of streams (e.g., DOC vs. NH4) can be regulated by very different sets of catchment processes.
5. Conclusion
Northern boreal landscapes are thought to be vulnerable to climate change impacts (Gauthier et al. 2015), but the consequence of increasing occurrence and severity of drought are only now getting attention (Gomez-Gener et al. 2020). Overall, my results show how one of the direct effects of climate change, hydrological drought, impact the biogeochemistry of boreal streams during summer. These effects were observed across scales and in some cases
13
influenced by the types of land cover draining into the streams. According to the analysed parameters, carbon-based forms (DOC and C/N) appear to be the most sensitive to drought, which causes a drop on their concentrations (or values), a result that was highly synchronous across a major part of the catchment network. Organic and inorganic nitrogen-based
parameters on the other hand manifest relatively weak and local responses to drought, which generated both positive and negative changes in concentrations or ratios, depending on the parameter and site. For several parameters, these results also highlight the important effects of mires during drought periods. While mires are important sources of water that may buffer against summer drought (Karlsen et al. 2016), they are also responsible generating high concentrations of reduced organic and inorganic nitrogen, as well greenhouse gases like methane (Gomez-Gener et al. 2020), during low flow periods.
Specific predictions of drought occurrence in boreal landscapes remain uncertain, but do indicate that parts of the northern Scandinavia may be particularly vulnerable (Spinoni et al.
2018). In the event that drought becomes more frequent, my results suggest that this will have clear consequence for stream chemistry during summer, particularly so for DOC and C/N. As one of the defining chemical factors of boreal streams (Laudon et al. 2011), drought- related changes in DOC could ultimately lead to a reduction of the microbial communities that are important to biogeochemical processes and food webs in these ecosystems (Berggren et al. 2007). Additionally, the rewetting of these catchments after dry periods could lead to an abrupt mobilization of solutes to these streams. In particular, my results suggest that pulses of DOC that has been accumulated in soils during drought are likely to be increasingly
common (e.g., Vázquez et al. 2015), whereas these rewetting events may be less important for other solutes. This organic matter fluxes could be of lower quality due to nutrient processing by microorganisms found on soils, resulting on a quick increase of DOC concentrations with low bioavailability for in-stream communities.
6. Acknowledgements
Thanks to Ryan Sponseller as my supervisor for his support on developing my thesis; his attentive advises and amendments on the report have been very helpful for the planification and writing of the thesis. The raw data on which this work is based were provided by the Krycklan Catchment Study (https://data.krycklan.se/). Finally, I wanted to also thank Tejshree Tiwari for providing the estimates of drought severity for each summer.
7. References
Berggren, Martin; Laudon, Hjalmar and Jansson, Mats. 2007. Landscape regulation of bacterial growth efficiency in boreal freshwaters. Global biogeochemical cycles 21.
DOI:10.1029/2006GB002844
Berggren, M.; Ström, L.; Laudon, H.; Karlsson, J.; Jonsson, A.; Giesler, R.; Bergström, A. -K.
and Jansson, M. 2010. Lake secondary production fueled by rapid transfer of low molecular weight organic carbon from terrestrial sources to aquatic consumers. Ecology Letters 13: 870–880. DOI: 10.1111/j.1461-0248.2010.01483.x
Bishop, K.; Buffam, I.; Erlandsson, M.; Folster, J.; Laudon, H.; Seibert, J. and Temnerud, J.
2008. Aqua Incognita: the unknown headwaters. Hydrological Processes 22: 1239–1242.
DOI: 10.1002/hyp.7049
Bishop, Kevin; Seibert, Jan; Kohler, Stephan and Laudon, Hjalmar. 2004. Resolving the Double Paradox of rapidly mobilized old water with highly variable responses in runoff chemistry. Hydrological Processes 18: 185–189. DOI: 10.1002/hyp.5209
Blackburn, M.; Ledesma, José L. J.; Näsholm, Torgny; Laudon, Hjalmar and Sponseller, Ryan A. 2017. Evaluating hillslope and riparian contributions to dissolved nitrogen (N)
14
export from a boreal forest catchment. Journal of Geophysical Research: Biogeosciences 122: 324–339. DOI: 10.1002/ 2016JG003535
Broder,T.; Blodau, C.; Biester, H. and Knorr, K.H. 2012. Peat decomposition records in three pristine ombrotrophic bogs in southern Patagonia. Biogeosciences 9: 1479–1491. DOI:
10.5194/bg-9-1479-2012
Buffam et al., 2007) Ishi Buffam, Ishi; Laudon, Hjalmar; Temnerud, Johan; Mörth, Carl- Magnus and Bishop, Kevin. 2007. Landscape-scale variability of acidity and dissolved organic carbon during spring flood in a boreal stream network. Journal of Geophysical Research: Biogeosciences 112 (1). DOI: https://doi.org/10.1029/2006JG000218
Burrows, Ryan M.; Jonsson, Micael; Fältström, Emma; Andersson, Jannika and Sponseller, Ryan A. 2021. Interactive effects of light and nutrients on stream algal growth modified by forest management in boreal landscapes. Forest Ecology and Management 492. DOI:
https://doi.org/10.1016/j.foreco.2021.119212
Callensen, Ingeborg; Raulund-Rasmussen, Karsten; Westman, Carl Johan and Tau-Strand, Line. 2007. Nitrogen pools and C:N ratios in well-drained Nordic Forest soils related to climate and soil texture. Boreal Environment Research 12: 681-692.
Einarsdóttir, Karólína; Attermeyer, Katrin; Hawkes, Jeffrey A.; Kothawala, Dolly; Sponseller, Ryan A. and Tranvik, Lars J. 2020. Particles and Aeration at Mire-Stream Interfaces Cause Selective Removal and Modification of Dissolved Organic Matter. Journal of Geophysical Research: Biogeosciences 125 (12). DOI:
https://doi.org/10.1029/2020JG005654
Fork, Megan L.; Sponseller, Ryan A. and Laudon, Hjalmar. 2020. Changing Source‐
Transport Dynamics Drive Differential Browning Trends in a Boreal Stream Network.
Water Resources Research 56 (2). DOI: https://doi.org/10.1029/2019WR026336 Gauthier, S.; Bernier, P.; Kuuluvainen, T.; Shvidenko, A. Z. and Schepaschenko, D.G. 2015.
Boreal forest health and global change. Science 349 (6250): 819-822. DOI:
10.1126/science.aaa9092
Gómez-Gener, Lluís; Lupon, Anna; Laudon, Hjalmar and Sponseller, Ryan A. 2020. Drought alters the biogeochemistry of boreal stream networks. Nat Commun 11, 1795 (2020). DOI:
https://doi.org/10.1038/s41467-020-15496-2
Helton, Ashley M.; Ardón, Marcelo and Bernhardt, Emily S. 2015. Thermodynamic
constraints on the utility of ecological stoichiometry for explaining global biogeochemical patterns. Ecology Letters. DOI: 10.1111/ele.12487
Högberg, Peter; Näsholm, Torgny; Franklin, Oskar and Högberg, Mona N. 2017. Tamm Review: On the nature of the nitrogen limitation to plant growth in Fennoscandian boreal forests. Forest Ecology and Managent 403: 161-185. DOI:
https://doi.org/10.1016/j.foreco.2017.04.045
Hosen, J. D.; Aho, K. S.; Appling, A. P.; Creech, E. C.; Fair, J. H.; Hall Jr., R. O.; Kyzivat, E.
D.; Lowenthal, R. S.; Matt, S.; Morrison, J.; Saiers, J. E.; Shanley, J. B.; Weber, L. C.;
Yoon, B. and Raymond, P. A. 2019. Enhancement of primary production during drought in a temperate watershed is greater in larger rivers than headwater streams. Limnology and Oceanography: 1-15. DOI: 10.1002/lno.11127
Karlsen, R.H.; Grabs, T.; Bishop, K.; Buffam, I.; Laudon, H. and Seibert, J. 2016. Landscape controls on spatiotemporal discharge variability in a boreal catchment. Water Resources Research. DOI: 10.1002/2016WR019186
Lake, P.S. 2003. Ecological effects of perturbation by drought in flowing waters. Freshwater Biology 48: 1161–1172.
Laudon, Hjalmar; Berggren, Martin; Ågren, Anneli; Buffam, Ishi; Bishop, Kevin; Grabs, Thomas; Jansson, Mats and Köhler, Stephan. 2011. Patterns and Dynamics of Dissolved Organic Carbon (DOC) in Boreal Streams: The Role of Processes, Connectivity, and Scaling. Ecosystems 14: 880–893. DOI: 10.1007/s10021-011-9452-8
Laudon, Hjalmar; Hasselquist, Eliza Maher; Peich, Mathias; Lindgren, Kim; Sponseller, Ryan; Lidman, Fredrik; Kuglerová, Lenka; Hasselquist, Niles J.; Bishop, Kevin; Nilsson, Mats B. and Ågren, Anneli M. 2021. Northern landscapes in transition: Evidence,
approach and ways forward using the Krycklan Catchment Study. Hydrological Processes 35 (4). DOI: 10.1002/hyp.14170
15
Laudon, Hjalmar and Sponseller, Ryan A. 2017. How landscape organization and scale shape catchment hydrology and biogeochemistry: insights from a long-term catchment study.
WIREs Water 1265. DOI: 10.1002/wat2.1265
Laudon, Hjalmar; Taberman, Ida; Ågren, Anneli; Futter, Martyn; Ottosson-Löfvenius, Mikaell and Bishop, Kevin. 2013. The Krycklan Catchment Study—A flagship
infrastructure for hydrology, biogeochemistry, and climate research in the boreal landscape. Water Resources Research 49: 7154–7158. DOI: 10.1002/wrcr.20520, 2013 Ledesma, José L. J.; Futter, Martyn N.; Blackburn, M.; Lidman, Fredrik; Grabs, Thomas;
Sponseller, Ryan A.; Laudon, Hjalmar; Bishop, Kevin H. and Köhler, Stephan J. 2017.
Towards an Improved Conceptualization of Riparian Zones in Boreal Forest Headwaters.
Ecosystems. DOI: 10.1007/s10021-017-0149-5
Lupon, Anna; Denfeld, Blaize A.; Laudon, Hjalmar; Leach, Jason and Sponseller, Ryan A.
2020. Discrete groundwater inflows influence patterns of nitrogen uptake in a boreal headwater stream. Freshwater Science 39 (2). DOI: 10.1086/708521
Mosley, Luke M. 2015. Drought impacts on the water quality of freshwater systems; review and integration. Earth-Science Reviews 140: 203-214. DOI:
https://doi.org/10.1016/j.earscirev.2014.11.010
Nyberg, Lars. 1995. Water flow path interactions with soil hydraulic properties in till soil at Gk-dsjijn, Sweden. Journal of Hydrology 170 (1-4): 255-275. DOI:
https://doi.org/10.1016/0022-1694(94)02667-Z
Peralta-Tapia, Andrés; Sponseller, Ryan. A; Ågren, Anneli; Tetzlaff, Doerthe; Soulsby, Chris and Laudon, Hjalmar. 2015. Scale-dependent groundwater contributions influence patterns of winter baseflow stream chemistry in boreal catchments. Journal of Geophysical Research: Biogeosciences 120 (5): 847-858. DOI:
https://doi.org/10.1002/2014JG002878
Rumpel, Cornelia and Kögel-Knabner, Ingrid. 2011. Deep soil organic matter—a key but poorly understood component of terrestrial C cycle. Plant Soil 338: 143–158. DOI:
10.1007/s11104-010-0391-5
Schindler, David W.; Curtis, P. Jefferson; Bayley, Suzanne E.; Parker, Brian R.; Beaty, Ken G.
and Stainton, Michael P. 1997. Climate-induced changes in the dissolved organic carbon budgets of boreal lakes. Biogeochemistry 36: 9–28. DOI:
https://doi.org/10.1023/A:1005792014547
Spinoni, J.; Vogt, J. V.; Naumann, G.; Barbosa, P. and Dosio, A. 2018. Will drought events become more frequent and severe in Europe? International Journal of Climatology 38 (4): 1718-1736. DOI: https://doi.org/10.1002/joc.5291Citations: 135
Sponseller, Ryan A.; Blackburn, M.; Nilsson, M.B. and Laudon, H. 2018. Headwater Mires Constitute a Major Source of Nitrogen (N) to Surface Waters in the Boreal Landscape.
Ecosystems 21: 31–44. DOI: 10.1007/s10021-017-0133-0
Tiwari, Tejshree; Buffam, Ishi; Sponseller, Ryan A. and Laudon, Hjalmar. 2017. Inferring scale-dependent processes influencing stream water biogeochemistry from headwater to sea. Limnology and Oceanography. DOI: 10.1002/lno.10738
Van Loon, Anne F.; Gleeson, Tom; Clark, Julian; Van Dijk, Albert I. J. M.; Stahl, Kerstin;
Hannaford, Jamie; Di Baldassarre, Giuliano ; Teuling, Adriaan J.; Tallaksen, Lena M.;
Uijlenhoet, Remko; Hannah, David M.; Sheffield, Justin; Svoboda, Mark; Verbeiren, Boud; Wagener, Thorsten; Rangecroft, Sally; Wanders, Niko and Van Lanen, Henny A. J.
2016. Drought in the Anthropocene. Nature Geoscience 9: 89–91. DOI:
https://doi.org/10.1038/ngeo2646
Vázquez, E.; Ejarque, E.; Ylla, I.; Romaní, A.M. and Butturini, A. 2015. Impact of drying/rewetting cycles on the bioavailability of dissolved organic matter molecular- weight fractions in a Mediterranean stream. Freshwater Science 34 (1): 263-275. DOI:
10.1086/679616
Worral, F. and Burt, T.P. 2008. The effect of severe drought on the dissolved organic carbon (DOC) concentration and flux from British rivers. Journal of Hydrology 361: 262– 274.
DOI: 10.1016/j.jhydrol.2008.07.051
Ylla, Irene; Sanpera-Calbet, Isis; Vázquez, Eusebi; Romaní, Anna M.; Muñoz, Isabel;
Butturini, Andrea and Sabater, Sergi. 2010. Organic matter availability during pre- and
16
post-drought periods in a Mediterranean stream. Hydrobiologia 657: 217–232. DOI:
10.1007/s10750-010-0193-z
