Extent and Causes of Siltation in a Headwater Stream Bed: Catchment Soil Erosion is Less Important than Internal Stream Processes

Introduction

Hydrologic connectivity is the water-mediated transport of matter, energy and organisms within or between elements of the hydrologic cycle (Freeman et al., 2007). Rivers are considered highly connected, four-dimensional systems (Ward, 1989). In particular, headwater streams directly connect the upland and riparian landscape to the rest of the stream ecosystem and contribute the largest part to total stream length. Altering catchment land use and resulting changes within headwater streams thus have the potential to modify fluxes between terrestrial uplands and downstream river segments, including effects on habitat quality and biological communities (Vannote et al., 1980; Allan, 2004).

The stream bed provides an important key habitat for many aquatic species that either permanently or temporarily depend on its physicochemical properties (Findlay, 1995; Kemp et al., 2011; Boeker et al., 2016). For instance, the reproduction success of many riverine fishes such as the economically important salmonids depends on the availability of coarse substrates and high oxygen supply during egg incubation (Acornley & Sear, 1999; Sternecker & Geist, 2010; Sternecker et al., 2013). Similarly, freshwater mussels of the Unionoida group, which are considered the most imperiled taxa in streams and rivers and core targets of aquatic conservation, strongly depend on uncolmated stream beds, particularly during their juvenile stages (Brim Box & Mossa, 1999; Geist, 2010, 2011; Denic & Geist, 2015). In addition, properties of the stream bed have also been linked with nutrient turnover and to other ecosystem services (Ensign & Doyle, 2006; Studer et al., 2017), due to the effects that texture and exchange rates between open water and interstitial water exert on microbial community structure (Gleason et al., 2003; Mueller et al., 2013). Consequently, restoration of stream beds is currently high on the agenda of conservation biology and applied river restoration (Pander & Geist, 2013; Palmer et al. 2014; Geist & Hawkins, 2016; Walling & Collins, 2016), with restoration projects mostly addressing the rehabilitation of fish-spawning areas (e.g. Zeh & Doenni, 1994; Sternecker et al., 2013; Pander et al., 2015). To date, many stream bed restoration efforts still lack scientific evidence (Geist, 2015), while an understanding of the external versus internal factors and processes that govern stream siltation would be crucial.

Soil erosion by water delivers fines to the rivers, and this load increases with increasing soil loss. Soil erosion has increased around the world in recent decades due to manifold land use changes (Van Oost et al., 2000; Van Rompaey et al., 2002). Arable land use has been extended at the expense of grassland which is less prone to erosion. This happened especially on valley floors and caused shortcuts within catchments. Percentage of row crops has increased to feed animals and as energy crops; field sizes have increased (Landis et al. 2008; Schilling et al. 2010), and linear structures like hedgerows or lynchets have been removed to facilitate mechanized farming (Ankenbrand & Schwertmann, 1989; Boardman & Vandaele 2016).

It is also likely that sediment production by carbonates in bicarbonate-dominated waters has increased during the past century. The widespread use of fertilizers and lime has increased salt concentrations in groundwater (Kaushal et al., 2005; Scanlon et al., 2005; Panno et al., 2006) that feeds the rivers. Global warming, heat dumping by electrical power stations, and the removal of shading trees along the rivers have caused a pronounced increase in water temperatures (Kaushal et al., 2010; Lepori et al., 2015) which should, in turn, result in increased degassing of CO₂ and precipitation of carbonates (Neal et al., 2002). Also, eutrophication of rivers due to the input of nutrients can increase biological CO₂ consumption within the water column, which also decreases bicarbonate concentration and favors carbonate precipitation during the day (Neal et al., 2002).

These changes occurred simultaneously with the increasing colmation of stream beds and their loss in functionality; thus, in particular, soil erosion has often been proposed as the major cause of the biological deterioration of stream beds (Bogan, 1993; Da Silva et al. 2007; Sutherland et al., 2002). However, major changes also occurred in the rivers themselves since Tulla started “river corrections” along the river Rhine in the 1820s (Lammersen et al., 2002). Following these corrections, a large number of dams have been implemented along rivers of all sizes (Mauch & Zeller, 2008; Nilsson et al., 2005). Structures along the rivers to stabilize the stream bed and reduce sediment movement became necessary due to the incision of major rivers (Ward 1998; Mauch & Zeller, 2008). These changes are likely responsible for the significant decrease in land–ocean sediment fluxes in many rivers despite the trend of increasing erosion on land (Walling & Fang, 2003). It is thus questionable whether stream bed colmation is primarily controlled by the input of sediment from the catchment or by changes within the rivers such as altered flow regimes. Surprisingly, the common assumption of a causal relation between increasing colmation of stream beds and historic changes in land use and their effects on soil erosion has hardly been validated. This assumption may thus be misleading due to the multitude of other possible influences that also experienced significant changes during the past decades.

The sources of sediment are diverse, and the reasons for sedimentation are numerous. Recent advances in sediment source fingerprinting techniques have allowed identifying and apportioning the relative importance of potential sources (for review, refer to Walling, 2005), contributing to a near-exponential increase in the application of source tracking (Walling, 2013). Most studies focused on developing source fingerprinting techniques for tracing the terrestrial source (e.g. different types of land use) of fine sediment transported by streams, including mineral analysis (Klages & Hsieh, 1975; Walling et al., 1979), stable and radioactive isotopes (Walling & Woodward, 1992; Wallbrink et al., 1999; Douglas et al., 2003), rare elements (Douglas et al., 2003), and other sediment properties in combination.

The source fingerprinting approach of suspended sediment relies on the assumption of a conservative behavior in fingerprint properties during mobilization and transport (Walling, 2005). The source identification for deposited sediment with similar approaches has received comparatively little attention (e.g. Walling et al., 2003; Collins & Walling, 2007; Collins et al., 2013) despite of considerable potential (Walling, 2005). This is partly because the assumption of a conservative behavior of fingerprinting properties is less likely to be true for the deposited sediment. The deposits constitute only a small fraction of the total sediment, which increases likelihood of fractionation. Further, the deposits may be exposed to the aquatic environment for many years, which differs in ambient conditions from their terrestrial source area. They can thus undergo transformations that corrupt their fingerprint. Also, the conditions favoring transport of suspended material from terrestrial sources may differ considerably from the conditions favoring deposition. The input of erosion-derived particulate matter is highest during flooding events, that is, at times when also transport capacity of a river peaks. In contrast, carbonate precipitation will be highest during conditions favoring biological CO₂ consumption or degassing (i.e. high water temperature, high insolation), which are usually associated with low water levels and low transport capacity. The ratio for erosion-born sediment and carbonate precipitates in suspended sediment is thus unlikely identical to the respective ratio in river deposits.

These differences between transported and deposited sediments impede to reason by analogy. The same is true in biological respect. While the deposited sediment governs biological communities and species survival by affecting the exchange between the interstitial pore space with open water and oxygen supply (e.g. Geist & Auerswald, 2007; Mueller et al., 2013), suspended sediment has surprisingly little effects on the same species (e.g. Lummer et al., 2016). In this study of a pre-alpine river, the Moosach, we quantify sedimentation on the river bed and compare the contribution of external (i.e. catchment erosion) versus internal (i.e. in-stream carbonate precipitation and internal biomass production) sources mainly by their mineralogical identities. We examine the hypothesis put forward by Pulg et al. (2013), Baars et al. (2000), Pander et al. (2015), and others especially for this river that soil erosion is a major source of sediment deposited on the stream bed. By testing this hypothesis, we discuss whether erosion-reducing measures can combat the undesired siltation.

Material and Methods

Study Area and Stream System

The River Moosach, a tributary of the river Isar, is situated at the border of two contrasting landscapes (Figure 1). The Tertiary Hills constitute the left, northern part of the catchment. The Tertiary Hills are a landscape with one of the highest erosion rates in Europe (Auerswald et al., 2009; Cerdan et al., 2010), and soil erosion has thus been proposed a major reason for the siltation of this river (Pulg et al., 2013). The right, southern part of the catchment is the Munich Gravel Plain, which has roughly a square shape with a width of about 50 km (Figure 1). The Munich Gravel Plain is flat with a slight gradient to the north (about 0·1%). Coarse gravel was deposited by Alpine rivers at the end of the last glaciation on top of the underlying Tertiary materials. The gravel, several 10 m in the south and decreasing in thickness to several meters along the northern end, contains 50 to 60% carbonates (Fetzer et al., 1986). The gravel constitutes the aquifer in which groundwater originating from the Alps and the gravel plain itself flows to the north and finally drains into the rivers Moosach and Isar. Thus, the flow of groundwater increases to the north, while the surface of the gravel layer decreases in altitude. Both changes cause the groundwater table to approach the surface. As a consequence, large fen areas (several 100 km²) developed on the northern fringe of the Munich Gravel Plain, which have been drained for about two centuries and degrade since then (Fetzer et al., 1986).

The river Moosach flows mainly within this fen area. Its origin was close to Munich, but its course has undergone many changes since the 17th century (Pulg et al., 2013). Since the early 20th century, it flows within an entirely artificial bed of almost rectangular cross section that was dug into the fen and loamy riparian deposits, reaching down to the Pleistocene gravel in most places (Kohler et al., 1971). On top of this gravel bed, thick deposits of fine sediment accumulated that have to be removed in intervals of 5 to 15 year to prevent overtopping and to allow drainage of the peat land on the Munich Gravel Plain.

The hydrological characteristics of the River Moosach are given in Table 1. The Moosach has a mean energy slope of 1.3‰ with higher values in the upper reach and in the lower reach, while values even drop to 0.1‰ in the middle reach (Figure 2). From a water balance, it follows that its hydrological catchment area is about 350 km², but the exact delineation of the hydrological catchment to the south is unknown due to the groundwater flow in the gravel of the Munich Gravel Plain. The topographic catchment to the north is 75 km² in size. It is situated within the Tertiary Hills from where small, partly ephemeral streams deliver eroded material during flood events. Several publications quantify soil erosion to be practically nil on the Munich Gravel Plain. In contrast, long-term average soil loss within the Tertiary Hill catchment is about 10 t ha⁻¹ year⁻¹ on arable land and half of this for the entire area comprising all land uses (Auerswald et al., 2009; Cerdan et al., 2010).

Table 1. Hydrological properties of the river Moosach (HND, 2013; NID, 2013; Pulg, 2009)

Property		Unit
Topographic catchment areaa	175	km2
Hydrological catchment areab	351	km2
River lengthc	34.2	km
River surface area	0.27	km2
Mean low water discharge	1.94	m3 s−1
Mean high water discharge	5.68	m3 s−1
Mean discharge	2.64	m3 s−1
Mean minimum monthly temperature	5.8	°C
Mean temperature	10.3	°C
Mean maximum monthly temperature	14.5	°C

• ^a From HND (2013); 185 km² according to www.wwa-m.bayern.de/fluesse_seen/gewaesserportraits/moosach (last access October 2015).
• ^b Calculated from a water balance.
• ^c www.wwa-m.bayern.de/fluesse_seen/gewaesserportraits/moosach.

Data Analysis

Energy slope and bed area were calculated by dividing the total river length into eight rather homogeneous sections. Bed area of a section was calculated as product of section length times mean bed width determined at the upper and lower end of each section. Energy slope during average flow conditions is given by the difference in surface height divided by section length. Difference in surface height was corrected for the loss of height at weirs. Mean flow rate in the individual sections was estimated from the flow cross-sectional area and flow velocity. These flow rates were scaled to match the mean annual flow rate at the measuring station by HND (2013). The same scaling factor was used for all sections.

The mass of fine sediment deposited on the gravel bed W_fs was calculated from the volume of fine sediment V_fs and their dry bulk density ρ_fs:

The total area of the river bed was the sum of all section areas A_i. The fraction of the river bed covered with fine sediment f_fs resulted from mapping. The mean depth of fine sediment d_fs was obtained from depth soundings. Soundings yielding zero depth were omitted from averaging because they were already considered in f_fs. The dry bulk density ρ_fs resulted from the weight of dry matter W_dry after drying the sediment samples and the volume of water plus sediment, which, in turn, is given by dividing the weight of water W_water by its density ρ_water (1 kg L⁻¹) and the weight of dry matter W_dry by the density of solids ρ_solids. Solid matter densities of 2.70 kg L⁻¹ for calcite, 2.80 kg L⁻¹ for dolomite, 2.68 kg L⁻¹ for quartz, 2.83 kg L⁻¹ for illite, 2.70 kg L⁻¹ for vermiculite, and 1.30 kg L⁻¹ for organic matter were taken from Rühlmann et al. (2006) and combined according to their mean fractions in the sediment.

The mean deposition rate is then given from the mass of fine sediment by the time period elapsed since the last sediment excavation (15 years). This approach only yields the average net accumulation rate during the entire period. It neglects that accumulation rate will not be constant over time, and even sediment resuspension will occur during periods of high flow velocity, especially during snow melt events when no macrophytes are present that retard water flow during summer months (Walling and Amos, 1999).

The CO₂ partial pressure and the calcite saturation index (logarithm of the ion activity product of Ca²⁺ and CO₃²⁻ in water divided by the solubility product of calcite) were calculated with the program PHREEQC (Parkhurst et al., 1990). Concentrations of cations (Ca²⁺, Mg²⁺, Na⁺, K⁺) and (anions Cl⁻, SO₄³⁻, NO₃²⁻, F⁻, PO₄³⁻, Br⁻), water temperature, pH, O₂ saturation, and electrical conductivity were used.

This study does not aim at establishing sediment budgets of the catchment or the stream, and thus, it lacks quantification of the amount of material mobilized by dissolution or by erosion and quantification of the deposition of mobilized masses along the terrestrial and the aquatic flow paths. However, in order to be able to judge by which factor erosion would have to be lowered to limit deposition on the stream bed, we related the amount of this deposition on the stream bed to the amount introduced into the stream. For dissolved calcium, this faction f_Ca++ is given by

where f_carbonates is the mass fraction of carbonates in the sediment, Q is the annual water discharge, [Ca⁺⁺] is the calcium concentration in the river water, 2.5 is the conversion factor of calcium to calcium carbonate, and 15 is the number of years during which the sediment accumulated. This equation assumes that no carbonates arrive at the stream as erosion load. This is justified because only 7.6% of the soils in the Tertiary Hills, which are the almost exclusive source of eroded sediment, contain carbonate. Carbonate content in the A horizon of these soils is on average 9% (Fetzer et al. 1986). Taken together, the carbonate content in eroded masses should only be 0.7%.

The delivery of eroded silicates originating from the subcatchment in the Tertiary Hills is given by the sum of soil losses (ΣA) taken from Auerswald et al. (2009) and the sediment delivery ratio f_SDR taken from Auerswald (1989), which decreases with increasing size of the subcatchment. This sediment delivery ratio has been proven to be valid for the prediction of sediment transport in streams also under the local conditions of the study area (Auerswald 1992). The fraction of eroded material arriving at the river Moosach that is deposited on the river bed (f_erosion) is then given by

where f_silicates is the mass fraction of silicates in the sediment and 15 is again the number of years of sediment accumulation. Following the hypothesis that erosion is mainly responsible for the fine sediment, we calculated the upper limit of f_erosion. Thus, we neglected erosion of river banks, which was close to zero due to the stability of the river bed, and the amount of silicates arriving at the river from the subcatchment of the Munich Gravel Plain, which also is very low due to the mean gradient of 0·1% of this subcatchment. Furthermore, we calculated f_SDR from the entire size of the subcatchment situated in the Tertiary Hills. In reality, six subcatchments situated in the Tertiary Hills deliver eroded material to the River Moosach. The mean size of the subcatchments is thus only one sixth of the total size. Using the smaller size for the calculation would increase silicate delivery to the River Moosach by about 50% and, in turn, decrease the fraction deposited on the river bed f_erosion by about 40%.

Results

Substrate Composition and Fine Sediment Coverage

The streambed consisted of three different substrates: gravel, fine sediment, and oncoids. There was a clear separation along the river course regarding the prevalence of these substrates (Figure 2). Gravel dominated the upper reach where also the energy slope was highest and usually larger than 1.3‰. In the middle reach, where the energy slope was lower and even dropped to 0.1‰ in some places, fine sediment prevailed despite the already high discharge rates. In the lower third, where the energy slope was about 1‰ and where the river flows through a dense riparian forest, oncoids were most common.

A large proportion of the stream bed (51%) was covered by fine sediment (Figure 2) that often covered the entire stream bed as thick deposit where it occurred (Figure 3a). The deposits were mainly between 30 and 100 cm (mean = 58 cm) thick but could even exceed 150 cm (Figure 3b). Excluding those places that were not covered with fine sediment yielded an average 67-cm thickness. For the stream bed area (27 ha, Table 1), a total fine sediment volume of 92,000 m³ resulted.

The dry matter content of the fine sediment was low (on average 0.229 kg kg⁻¹ for n = 28) and quite variable (SD 0.060 kg kg⁻¹). The upper few centimeters, which were aerobic, had a slightly lower dry matter content (0.213 kg kg⁻¹; SD 0.059 kg kg⁻¹) than the lower anaerobic part of the sediment (0.245 kg kg⁻¹; SD 0.058 kg kg⁻¹) due to their younger age and the lower compression (absence of an overlying horizon). The difference in dry matter content over depth, however, was small compared with the variability within both horizons, and thus, consideration of both horizons separately reduced the SD only marginally. The 95% level of confidence was 0.026 kg kg⁻¹ when all horizons were combined and 0.038 kg kg⁻¹ for the upper and lower horizon when separated.

In total, the fine sediment amounted to 5530 t. The average net deposition rate since the last dredging of fine sediment was on average 361 t year⁻¹, equivalent to 1.32 kg m⁻² year⁻¹. From an erosion rate of 5 t ha⁻¹ year⁻¹, averaged over all land uses, and a sediment delivery ratio of 0.16 follows that 8400 t year⁻¹ eroded material must have reached the Moosach river from the Tertiary Hills. Only 72 t year⁻¹ or 1.3% of the sediment delivered to the Moosach were deposited within the stream. Most erosion material must thus have been transported downstream into the river Isar and have left the Moosach.

Fine Sediment Composition

The underlying gravel bed had a geometric mean diameter of 10 mm with a narrow range (Figure 4a). Less than 10% of the gravel bed mass was smaller than 1 mm. The fine sediment had a geometric diameter of 0.06 mm, again with a rather narrow range. In contrast, the colmated transition zones that were not fully covered by fine sediment exhibited a wide range of grain sizes in various quantities depending on the degree of mixing between gravel bed and fine deposits.

Carbonates contributed the largest share to the fine sediment (on average 46%). Silicates and organic matter contributed 38 and 16%, respectively, but the range was wide (Figure 4b) with no simple pattern (e.g. a continuous change along the river). Within the carbonates, calcite dominated (on average 72%; Figure 4c), but dolomite could be identified as well. The clay fraction of the fine sediment, after removal of organic matter and carbonates, was dominated by illite (on average 51%; SD 11%; n = 16) and smectite (31%; SD 13%), while vermiculite (11%; SD 6%) and kaolinite (8%; SD 2%) contributed only minor shares.

The ion concentrations (Table 2) in the free-flowing (open) water indicated that calcium bicarbonate dominated. Calculating the CO₂ partial pressure and the calcite saturation index from these data showed that the CO₂ partial pressure decreased to one tenth from the upper reach (river km 24 in Figure 5) to the lower reach close to the confluence of the Moosach with the River Isar (Moosach river km 0.5). In the same direction, the calcite saturation index increased. Except for one value, it was always above 0.5 and rose up to almost 1.5 close to the mouth. The dolomite saturation index was even higher than the calcite saturation index (on average: 1.5). From the annual water flow (Table 1) and the amount of calcite deposited annually as fine sediment, it followed that calcite sedimentation only removed about 0.8 mg L⁻¹ Ca²⁺, which was less than 1% the average Ca²⁺ concentration (105 mg L⁻¹; Table 2) despite the high calcite saturation index.

Table 2. Average water properties as used for the calculation of ion activities, CO₂ partial pressure, and the calcite saturation index with the PHREEQC model (n = 21)

Parameter (unit)	Mean	SD
pH	7.97	0.24
Bromide (mg L−1)	4.78	1.39
Chloride (mg L−1)	32.86	2.64
Sulfate (mg L−1)	44.32	8.29
Nitrite (mg L−1)	0.04	0.01
Nitrate (mg L−1)	31.56	4.80
Fluoride (mg L−1)	0.06	0.02
Phosphate (mg L−1)	0.16	0.10
Carbonate (mmol L−1)	3.00	0.26
Potassium (mg L−1)	3.35	1.06
Sodium (mg L−1)	20.51	1.08
Calcium (mg L−1)	104.88	6.05
Magnesium (mg L−1)	22.95	2.31

Discussion

As evident from the thickness of the fine sediment cover, from its spatial extent and from the textural composition, the stream under study was clearly subject to extensive colmation and siltation. Geist and Auerswald (2007) have shown that the exchange between interstitial water and free-flowing water is strongly restricted, and biological functionality is lost when the fraction <1 mm contributes 75% to the stream bed. This percentage was exceeded in most of the fine sediment samples. It is thus not surprising that several publications reported on low or degrading biological functionality of this river (e.g. Sternecker et al., 2013; Pulg et al., 2013; Sternecker et al., 2014).

Several publications (e.g. Baars et al., 2000; Pander et al. 2015; Pulg et al., 2013; Stein, 1988) and theses (Baumann, 1977; Pulg, 2009) regarded an increase of soil erosion as major cause for this siltation problem in the Moosach but also elsewhere (Soulsby et al., 2000; Sutherland et al., 2002). The conclusion that erosion is the main reason for siltation is obvious at first glance, given that part of the catchment area of the Moosach is located in one of the landscapes in Europe that are most prone to erosion. Our data however provide little evidence for this assumption, and such simplifications can thus pose serious limitations to the implementation of stream bed restoration efforts which need to become more evidence-based to be successful (refer to review by Geist & Hawkins, 2016 and references therein). Only 38% of the sediment was silicates and thus potentially originated from soil erosion of arable fields. Sediment clay mineralogy differed considerably from what is found in the top soils of arable land in the Tertiary Hills where smectite dominates in top soils (45%) while illite contributes only 37% (Niederbudde & Vogl, 1987). In the fine sediment, the ratio of smectite and illite was opposite (30 vs. 50%). This may be due to a preferential transport of the expansive and thus less dense smectite in water and would indicate that a large part of the eroded material is not deposited, but is transported further downstream. A second explanation could be that the silicates deposited in the river do not originate from eroding top soils but from input of less weathered subsoils that are higher in their illite content (Niederbudde & Vogl, 1987). The origin then could be construction sites or unpaved roads that deliver runoff even during small rain events, which do not cause flooding and high transport capacities within the stream. In-stream sedimentation is more likely during such small events. The illite content of the sediment was even higher (70%) and the smectite content smaller (30%) in the 1970s, when a gravel washing unit discharged its washing water into the river (Baumann, 1977). This gives further indication that the input of subsoil material during average or low flow conditions may be responsible for at least part of the deposited silicates.

The strongest evidence that erosion cannot explain the severe siltation of the river Moosach results from a mass balance. The amount of silicates was only 1.3% of the erosion material delivered by erosion from the Tertiary Hills to the river Moosach. Our study period (15 years since the last dredging of sediment) was long enough that many erosion events of various extents must have occurred during this time. This should also have leveled out inter-annual and intra-annual fluctuations that can be expected to occur due to the fluctuations in loading, in water flow, and due to the seasonality in growth and senescence of macrophytes (Walling & Amos, 1999). At first glance, a fraction of 1.3% appears to be much lower than fractions of 21 to 38% that were previously reported for three streams in Great Britain by Walling et al. (2006) and Collins & Walling (2007). This is due to the fact that in these publications, the total amount of fine sediment deposited on the river bed was related to the erosion amount of 1 year while the net sedimentation rates were around zero. Net sedimentations rates around zero may be more likely in many rivers because the relation of accumulation and erosion is artificially controlled in most European rivers in order to stabilize the vertical and lateral position of the river course. Otherwise, many rivers would incise as a consequence of the narrowing and straightening of the river course. In case of the river Moosach, the opposite problem occurs and dredging has to be applied to prevent overtopping and relocation of the river course, which happened in the past as historical maps suggest. This dredging was scientifically advantageous because it allowed calculating multiannual sedimentation rates. It has no consequence on our main finding that siltation can hardly be controlled by lowering terrestrial erosion rates. In other rivers that do not require dredging but still exhibit colmation, considerable less terrestrial material is required to cause the same problem.

We selected a 15-year period to determine deposition rates that do not reflect single events of accumulation or removal but long-term trends that could be related to long-term averages of soil loss. However, we do not know how deposition rates vary over time and how they would develop in the future. From the regular excavations taking place despite the high cost, it is expected that deposition would continue also in the future. Our results also provide no reason why these rates should increase or decrease considerably in the future if a sufficiently long period of time is considered to level out fluctuations. The conditions of calcite precipitation would not change due to an increasing accumulation of fine sediment. Also, the terrestrial erosion rate and delivery cannot be influenced by deposition within the stream. Finally, transport capacity of the stream will not change as long as the river does not overtop and thus change its cross-sectional area. Prevention of overtopping is the main reason for sediment dredging. Also, colmation and accumulation of fine sediment would not restrict water inflow in the stream because the water surplus of the catchment cannot disappear. A decrease of the stream bed permeability must always be compensated by higher inflows from the many ditches and tributaries draining the land.

Implementing erosion control measures on arable land, although desirable from a soil conservation perspective, can hardly have a large effect on the siltation of the river, given that only 1.3% or less of the erosion material delivered to the stream was deposited on the river bed. The dampening of runoff that is associated with on-site (e.g. Cogo et al., 1983; Gilley et al., 1987) and off-site erosion control measures (e.g. Fiener & Auerswald, 2005; Fiener et al., 2005) may even reduce transport capacity. This could counteract the reduced input of eroded material by fostering sedimentation of fines. Changes in transport conditions will likely have a larger effect than changes in erosion conditions. This is supported by the finding that the most severe siltation occurred in the middle reach where energy slope was lowest due to several dams and several diversions of the River Moosach to feed fish farms or for historic reasons. In this area, the average flow velocity near the stream bed is only 0.12 and 0.20 m s⁻¹ close to the water surface (Braun et al., 2012) and may not even exceed 0.35 m s⁻¹ during flood events.

From the amount of erosion material entering and leaving the stream, it may be classified as wash load, a concept that was described distinctly by Einstein (1950). It describes material that essentially passes straight through a stream with virtually no exchange between these particles and those remaining in the bed. Thus, these particles extract only negligible momentum from the flow. Raudkivi (1998) states that some of the wash load may at times settle on the stream bed, but this fraction remains small. The concept of wash load may thus be valid in terms of total quantities, but it disregards that retention of 5% or even less may be sufficient to deteriorate the biological functioning of the river bed. Even the assumption that wash load is hydrodynamically ineffective (Woo et al., 1986) may not hold true because Geist and Auerswald (2007) have shown that colmated areas are characterized by a high mechanical stability. Little material is necessary for filling the interstitial pore space and hardening the river bed, particularly if formation of flocs from originally smaller particles occurs after their deposition (Hill et al., 2017).

The highest proportion of fine sediment solids was represented by carbonates. Water analyses and modeling with PHREEQC clearly showed that the carbonates must have resulted from precipitation due to the decrease in CO₂ partial pressure in downstream direction. A contribution of abraded material from the carbonate gravel of the flat Munich Gravel Plain is unlikely. The calcite saturation index indicated a high probability of calcite formation, even if we consider that the calcite saturation index has to exceed 0.8 where dissolved organic matter delays calcite precipitation (Kempe & Kazmierczak, 1990). However, balancing Ca²⁺ that is transported by the river water with Ca²⁺ that is precipitated and deposited in fine sediment showed that less than 1% was retained. It is thus reasonable to assume that less carbonate is found in fine sediment than what is actually precipitated because, similar to erosion material, part of the precipitate is exported. This view is corroborated by an almost complete lack of fine sediment in the lowest reach where energy slope is steeper than in the middle reach. In the lowest reach, the calcite saturation index was by far highest and also the presence of oncoids demonstrated favorable conditions for calcite formation. This again indicated that transport processes are much more important for the deposition of fines within the River Moosach than the amount of sediment released into water.

Organic matter represented 16% of fine sediment mass. The origin of organic matter may either be internal production within the stream or input from tributaries draining fen areas. Macrophytes are common especially in the middle reach of the Moosach (Braun et al., 2012). From the deposition rate of organic matter in the fine sediment (2·4 t ha⁻¹ year⁻¹) follows that the internal biomass production must be 12 t ha⁻¹ year⁻¹ in the entire Moosach if we assume a degradability of 80% for the macrophytes. This amount seems unlikely given that the lowest third of the stream is almost completely within a dense riparian forest. Material from fen areas within the catchment thus likely also contributed to the organic matter in sediment.

The most promising measure to reduce siltation resulting from calcium carbonate precipitation in this stream likely is to increase shading by establishment of a dense tree cover along the river. This would reduce calcite precipitation due to slower warming of the water and due to lower consumption of CO₂ by macrophytes and algae. The reduced internal production of macrophyte biomass would, in turn, increase flow velocity and thus transport capacity. This prediction is corroborated by the experience that dredging had to be repeated every 5 years in the 1970s, when macrophyte growth was much more vigorous than nowadays because untreated sewage still delivered high amounts of phosphate to the river. Nowadays, dredging intervals have been extended to 20 years.

In addition, restoration of flow regimes (Poff et al., 1997) is a key component in restoring functional stream bed substrates. Peak flows have been strongly reduced in the Moosach and elsewhere as a result of damming and milder winters with absence of larger snow melt events. These changes have likely affected the sensitive dynamics of erosion and fine sediment deposition.

Conclusions and Implications for Restoration

This study of a river suffering from severe colmation and siltation illustrates that soil erosion and calcite precipitation deliver several orders of magnitude more material than what accumulates on the stream bed. Given that less than 1% of the material introduced into the stream is deposited, the severe siltation problem cannot be controlled by exclusively reducing soil erosion in the catchment. A reduction of soil erosion by several orders of magnitude would be necessary until less material would be delivered to the river than what is presently deposited. Such a reduction—at best—would only be possible by forestation of the entire catchment. The results of this study likely apply for other situations as well because their fundamental cause is that the stream bed area is small compared with the catchment area (0.16% in our case). While the findings of this study challenge the frequent perception that stream siltation and colmation problems could be solved by addressing land use management, erosion control, and connectivity within the terrestrial catchment, they should not be interpreted in the way that measures to prevent erosion and illuviation of fines and nutrients into streams from arable land would be unimportant. Instead, our findings illustrate that an understanding of siltation problems needs a critical and evidence-based evaluation to derive appropriate mitigation measures that better consider internal stream processes. Reduction of river fragmentation with its effects on erosion deposition patterns, restoration of natural flow regimes to ensure recreation of functional stream beds, as well as measures to prevent temperature increase and resulting increased calcite precipitation all need to complement appropriate catchment land use practices to tackle one of the most prominent challenges in freshwater biodiversity conservation, the restoration of functional stream beds.