Inland waters impart considerable influence on nutrient cycling and budget
estimates across local, regional and global scales, whilst anthropogenic
pressures, such as rising populations and the appropriation of land and water
resources, are undoubtedly modulating the flux of carbon (C), nitrogen (N)
and phosphorus (P) between terrestrial biomes to inland waters, and the
subsequent flux of these nutrients to the marine and atmospheric domains.
Here, we present a 2-year biogeochemical record (October 2011–December 2013) at
biweekly sampling resolution for the lower Sabaki River, Kenya, and provide
estimates for suspended sediment and nutrient export fluxes from the lower
Sabaki River under pre-dam conditions, and in light of the approved
construction of the Thwake Multipurpose Dam on its upper reaches (Athi
River). Erratic seasonal variation was typical for most parameters, with
generally poor correlation between discharge and material concentrations, and
stable isotope values of C (
The acknowledgement of the vital role inland waters play in carbon (C)
cycling and budget estimates at local, regional and global scales has
progressed steadily over the past three decades (e.g. Meybeck, 1982; Cole et
al., 2007; Tranvik et al., 2009). For example, inland waters not only act as
a conduit for the delivery of significant quantities of terrestrial organic
C to the coastal zone and open ocean, they are typically sources of
greenhouse gases (GHGs: e.g. CO
Given that recent reports assert a similar order of magnitude to the lateral
C input to inland waters (
Over the past decade, momentum has gathered towards a broader
understanding of the nutrient cycling within sub-Saharan inland water
ecosystems (e.g. Coynel et al., 2005; Abrantes et al., 2013; Bouillon et
al., 2014). However, Africa has experienced the highest annual population growth
rate over the past 60 years (
The potential perturbation of the biogeochemistry of tropical inland waters
by climate and land-use change (Hamilton, 2010), and that of Africa
specifically (Yasin et al., 2010), has received some attention. Given a
projected warming of a
The Athi–Galana–Sabaki River basin:
British settlement brought European land-use practices to the Kenyan highlands early in the 20th century, triggering severe soil erosion in, and elevated sediment fluxes from, the Athi–Galana–Sabaki (A-G-S) River basin (Champion, 1933; Fleitmann et al., 2007). These terrigenous sediments have had a significant impact on the environment surrounding the outflow of the Sabaki River in the Indian Ocean, for example, by increasing coral stress (van Katwijk et al., 1993) and spreading seagrass beds on local reef complexes, as well as siltation and infilling of the Sabaki estuary and the rapid progradation of nearby shorelines (Giesen and van de Kerkhof, 1984). In order to alleviate regional water scarcity, construction of reservoirs on the Athi River has been under consideration for decades, the implementation of which could modify the magnitude of sediment delivery to the coastal zone (van Katwijk et al., 1993) as previously observed in the neighbouring Tana River (Finn, 1983; Tamooh et al., 2012).
The lower Sabaki (also known as Galana) River forms after the confluence of
the Athi and Tsavo rivers, and has been shown to be strongly influenced by
nitrogen inputs from the greater Nairobi area (Marwick et al., 2014a), yet
annual fluxes of particulate and dissolved elements have not been measured
in detail. In light of the planned construction of the Thwake Multipurpose
Dam (currently awaiting tender approval; see
The Athi–Galana–Sabaki River basin is the second largest drainage basin
(
Precipitation ranges between 800 and 1200 mm yr
Physico-chemical parameters of the Sabaki River were monitored biweekly
(i.e. fortnightly) approximately 2 km upstream of Sabaki Bridge
(approximately 5 km upstream of the river outlet to Malindi Bay) for the
period October 2011 to December 2013. This site was chosen since it is close
to the outflow to the ocean and thus integrates the yields over the entire
basin; however, it is not influenced by salinity intrusion or tidal influence. Water
temperature, conductivity, dissolved oxygen (O
Samples for total suspended matter (TSM) were obtained by filtering 60–250 mL
of water using pre-combusted (4 h at 500
TA was analysed by automated electro-titration on 50 mL samples with 0.1 mol L
Filters (25 mm) for POC, PN and
Our dataset for CH
Historical discharge observations and daily gauge height data for the
sampling period were provided by the Water Resource Management Authority
(WRMA), Machakos, Kenya. Due to the poor resolution of discharge and gauge
data at the Sabaki Bridge north of Malindi (gauge no. 3HA06) over the
monitoring period, the finer fidelity record from the Baricho station (gauge
no. 3HA13) was used, situated approximately 50 km upstream of our
biogeochemical monitoring station (i.e. site S20 from basin-wide sampling
campaigns; see Marwick et al., 2014a). With discharge measurements from 2006
and 2007 (
The Baricho gauge height dataset contains a 2-month period of no measurements (1 February to 31 March 2013). For this period, the daily discharge was estimated as the average discharge for that day over the previous 10 years (2003–2012). Since this period falls within the dry season when flows are relatively stable and low, we expect any bias deriving from this approximation to have no major effect on our annual flux estimates.
Annual flux estimates for suspended sediments and the various riverine fractions of particulate and dissolved C, N and P were calculated with the discharge data above. We interpolated linearly between the concentrations measured on consecutive sampling dates in order to establish concentrations for every day of the study period. The daily concentrations were then multiplied by daily discharge and summed over the study period to establish annual flux estimates.
All data (excluding results for NH
Throughout the results and discussion we use discharge values of
Discharge (solid grey line) and 2 years of monitoring the
Water temperature varied from 24.1 to 33.9
The concentrations of TSM, POC, particulate N (PN) and total particulate
phosphorus (TPP) are shown in Fig. 3, as well as the stable isotope
composition of POC and PN, with most variables showing no pronounced
relationships with discharge across the hydrological year. The Sabaki River
exported TSM varying in concentration from 50.0 to 3796.7 mg L
The relationship between the percent contribution of particulate
organic carbon to the total suspended load and
Discharge and 2 years of monitoring the dissolved
Two years of monitoring concentrations of dissolved
The dissolved organic C (DOC) concentration fluctuated from 3.3 to 9.3 mg L
The
Sampling of NH
The river was consistently oversaturated in dissolved CH
Annual material flux estimates to the coastal zone for TSM and various C, N
and P fractions are provided in Table 1. Briefly, our data suggest a mean
flux of 4.0 Tg TSM yr
Annual dissolved nutrient flux estimates (Table 1) were 2.3 Gg NH
Summary of annual fluxes, element ratios and annual yields for the Athi–Galana–Sabaki River basin from data reported here and from the NEWS2 export model (see Mayorga et al., 2010), as well as data for 2012 and 2013 from the neighbouring Tana River basin at Garsen (Geeraert et al., 2018).
Various surface area estimates are reported for the A-G-S Basin, ranging
from 40 000 km
Taking the above basin area estimate and the flux values detailed above, we
estimate mean annual yields of 84.6 Mg TSM km
Although previous studies provide estimates of annual suspended sediment
fluxes at the Sabaki outlet as well as annual yield estimates for the A-G-S
Basin (GOK-TARDA, 1981; Munyao et al., 2003; Kitheka, 2013), their primary
research focus lay elsewhere, and none provide the comprehensive
biogeochemical record at a comparable temporal scale as presented here. The
following discussion revolves around the main objectives of our study,
including (i) the quantification of annual suspended matter, C, N and P
fluxes and sediment yield, (ii) characterising the sources of particulate
and dissolved fractions of C and N and (iii) providing indications to the
water–atmosphere transfer of important greenhouse gases (CH
To the best of our knowledge, and excluding suspended matter, the estimates
provided in Table 1 are the first quantifications of material fluxes from
the A-G-S system, although we stress that our material flux estimates may
not be the most robust since (i) hydrological data are incomplete and
discharge data rely on an limited number of measurements to construct a
rating curve and (ii) our study covered a period of 2 years, while annual
discharge in this system is likely to show substantial interannual
variability. A suspended sediment flux of
If we normalise the basin area of
Some have reported that prior to 1960 the suspended sediment load of the
A-G-S Basin was
The SY reported here is low compared to the global average of 190 Mg km
The annual POC yield (1.5 Mg POC km
The annual DOC yield from the A-G-S Basin (0.5 Mg DOC km
In contrast to some other C
The findings from the basin-wide campaigns reported in Marwick et al. (2014a) led to the suggestion that the concentration of DIN in export from
the A-G-S Basin likely peaks during the wet season, due to the significant
processing and removal of DIN in the upper to mid-basin during the dry
season and which resulted in significantly lower DIN concentration at the
monitoring station (i.e. site S20 from Marwick et al., 2014a) relative to
wet season observations. Our higher-resolution dataset, however, suggests a
more complex relationship between DIN concentrations, seasonality and
discharge, given that peak DIN concentrations were also observed during low-flow conditions (Fig. 6b, c). In particular, a prominent NH
The Global Nutrient Export from Watersheds 2 (NEWS2; see Mayorga et al.,
2010) provides flux and yield estimates for TSM and particulate and
dissolved fractions of organic and inorganic forms of C, N and P for
> 6000 river basins through hybrid empirical and conceptual based
models relying on single and multiple linear regressions and
single-regression relationships. Comparatively, our flux estimates are in
general considerably lower than the NEWS2 estimates (Table 1), except for
the dissolved PO
The combination of high-frequency sampling and long-term monitoring of
dissolved CH
Nitrous oxide in rivers is sourced from either nitrification or
denitrification, and although the interest in N
The biogeochemical cycles and budgets of the Athi–Galana–Sabaki river system have been considerably perturbed by the introduction of European agricultural practices in the early 20th century and the expanding population of Nairobi living with inadequate waste water facilities (Van Katwijk et al., 1993; Fleitmann et al., 2007). These factors have had considerable impact on riverine sediment loads (Fleitmann et al., 2007), instream nutrient cycling (Marwick et al., 2014a) and near-shore marine ecosystems in the vicinity of the Sabaki outlet (Giesen and van de Kerkhof, 1984; Van Katwijk et al., 1993). Recent modelling of nutrient export to the coastal zone of Africa to the year 2050 foreshadows continued perturbation to these ecosystems, with the extent dependant on the land management pathway followed and mitigation strategies in place (Yasin et al., 2010). Although suspended sediment fluxes are estimated to decrease over Africa in the coming 40 years, the projected increase in dissolved forms of N and P and decreases in particulate forms of C, N and P as well as dissolved OC (Yasin et al., 2010) will further augment nutrient stoichiometry within the inland waters of the A-G-S system.
Although no large reservoirs have been developed within the A-G-S Basin,
approval has been given for the construction of the Thwake Multipurpose Dam
on the Athi River, though commencement has been delayed by tender approval
for the project. The total surface area is expected to be in the vicinity of
29 km
The full dataset can be found in the Supplement accompanying this article.
TRM, lead author, conceived research, performed field sampling, performed sample and data analysis and wrote the paper. FT performed field sampling and sample analysis. BO performed field sampling. AVB conceived research, performed sample analysis and wrote the paper. FD performed sample analysis. SB conceived research, performed sample and data analysis and wrote the paper.
The authors declare that they have no conflict of interest.
Funding for this work was provided by the European Research Council (ERC-StG
240002, AFRIVAL,