Inverse Estuaries in West Africa: Evidence of the Rainfall Recovery?

: In West Africa, as in many other estuaries, enormous volumes of marine water are entering the continent. Fresh water discharge is very and 26,500 km² respectively) is to be compared with the basins of their two main neighbor basins, the Gambia River and the Senegal River, which provide significant fresh water discharge to their estuary.


Problematics, State of the Art
West African Sahelian and Sudanian areas commonly have very flat coast plains. Therefore, rivers have long estuaries in front of their low hydraulicity. The Gambia River estuary, 450 km long, is considered the second longest in the world after that of the Amazon River, although its basin has an area 100 times and a discharge 1000 times smaller. In West Africa, only great basins provide enough fresh water to reach pushing saline waters seasonally out of the river mouths. In this constraining context, it is very easy for marine salt water to enter profoundly in the continent, and it is also easy for the tide to influence deeply the river water level within the continent. An inverse estuary is an estuary in which freshwater input is less than the losses due to evaporation; such estuaries contain hyperhaline water (e.g., Laguna Madra in Texas; [1]). Inverse estuaries were also defined by Pritchard [2] as the ones where salinity increased with distance to the mouth. In an inverse estuary, sea water enters the estuary from downstream to upstream to compensate for the losses due to evaporation, carrying salt, which raises concentration [3].
Some of these estuaries have very low fresh water income and are located in areas with very high evaporation. This leads to an increase in water density and a hypersaline density downwelling. Therefore, a surficial water flow is noticed upwards, and another flow is noticed near the bottom downwards [4]. In the case of the Saloum River, the increase is observed to be in a roughly linear fashion with distance to the sea [5].
These inverse estuaries are commonly found in semiarid or arid areas, such as southern Australia [6], northern Australia [7], some areas of Texas [1], Baja California (NW Mexico) [8,9], and in the Sahel [10], among others. Such hypersaline estuaries are relatively rare on the Earth. Other examples are the Red Sea and the Persian Gulf in the northern Indian Ocean [11], Shark Bay and Exmouth Gulf in Western Australia [12], and the hypersaline Coorong estuary/lagoon in South Australia [13,14]. The characteristics of Spencer Gulf, South Australia, are that evaporation exceeds precipitation all year round and that the spring-neap tidal cycle is greatly exaggerated [6].
In West Africa, as in the other estuaries, enormous volumes of marine water are entering the continent [15]. Fresh water discharge is commonly strongly linked to rainfall level [5]. During the Great Sahelian Drought , some main rivers completely dried up; this was the case of the Niger River in Niamey in May 1985 [16]. In extreme cases, "a negative water budget has even more drastic effects on smaller rivers: discharge becomes negative, (and) seawater may invade the estuary which becomes hyperhaline" [17]. As an example, "the Casamance River estuary, in a dry year, and during the dry season, can be changed into an evaporation basin, concentrating marine salts coming from [the] Atlantic Ocean and becoming a threat to fluvio-marine areas soils" [17].
Southward Dakar Senegambian estuaries are subject to [this] "unusual hydrodynamical regime caused by weak or absent run-off" [18]. Such a process has been occurring in two coastal "rivers" of Senegal, the Casamance and the Saloum Rivers, which are both actually tide-influenced "inverse estuaries" [5]. Therefore, the West African Sudano-Sahelian coast is one of the regions in the world where inverse estuaries are observed [19].
West Africa has shown a great interdecadal rainfall variability. Few data before 1950 allow highlighting two dry periods in 1910-1915 and 1940-1944. After the Second World War, the number of rain gauges increased significantly, and the following evolution can be described [20][21][22]: a very rainy period from 1950 to 1967 -a long and very dry period from 1968 to 1993 (whole West Africa) and from 1968 to 1998 in Senegambia and Mauritania -a rainfall recovered period (1994 to 2018 in WA, 1999 to 2018 in Senegambia), which was characterized by a rainfall annual amount close to the 1918-2017 average and an intensification of rainfall: an increase in rainfall intensity [23] and in the number of extreme rainfall events [24] are observed.
The evolution in coastal Senegal is similar to the regional one. Figure 1 shows the location of Senegambia (Figure 1a) and the location of rain gauge stations (Figure 1b). Figure 2 gives the rainfall evolution at different spatiotemporal scales. Figure 2a shows the evolution at three Senegalese coastal stations from 1917 to 2017, and Figure 2b shows the one in the inverse estuaries areas, southward from Dakar, from 1950 to 2017. Figure 2c,d give the SPI (Standard Precipitation Index) respectively for Senegambia and for the Casamance River basin. This confirms also the conclusions of Faye and Sané (2015; [25]), who observed the end of the long dry period in 1996 for the Casamance River Basin.
Teleconnections with the ITCZ (Inter Tropical Convergence Zone) fluctuations and their impact on flooding were analyzed by Mahé et al. (2013) [26]. When a river bed is close to the sea level, sea water may ingress during the low-water seasons. If the freshwater inflow is less than the loss through evaporation, salinity becomes higher than that in the sea. During the drought (1968-1998) Binet et al. 1995 [27] wrote, "This happens in the Senegal and the Casamance, particularly during the "current" drought".
Similar to the Casamance and Saloum estuaries, the semidiurnal microtidal Somone River estuary (80 km southeast of Dakar), where the maximum tidal range is about two meters, is characterized by an inverse salinity gradient [30,31]. Fluvial flows in the Somone estuary are null. Therefore, the salinity gradient is inverse; the only fresh water incomes observed during the 2007-2010 period are provided by rainfall and groundwater [31].
Hydrochemical analysis confirms that the Casamance River and Saloum River are "inverse estuaries", in which the water is salted during most of the year (at least 9 months) and hypersaline at the end of dry season [32].
Some estuaries within dry tropical areas (e.g., in Australia) show a combination of estuarian "normal" and "inverse" modes [4]. Inverse functioning of the Saloum River estuary, where salinity can reach 130 g/L [33], strongly affects fish and all marine species populations.
To conclude, Baran (1994) [34] summarized the context explaining that "Gambia river has a normal estuary, i.e., with decreasing salinity upwards. The Casamance estuary is inverse (rising salinity upwards) during the dry season and normal during the rainy season, while the Sine-Saloum is a "ria" where fluvial flows are null and the estuary is always inverse".
This study proposes to describe the current behavior of two West African inverse estuaries and to compare it with that observed during the Sahelian dry period of the end of the 20 th century.

Methodology
In order to document the current functioning of the Saloum and the Casamance estuaries (see location in Figure 3), two measurement devices were implemented: a twice a year direct measuring campaign through the Saloum and the Casamance estuaries since the end of 2016: This measuring campaign included a network of measurement sites in both Saloum and Casamance estuaries, salinity measures with a refractometer PCE© 2020 (67250 Soultz-sous-forêt, France) and, for values lower than 20 mS/cm, a conductimeter HANNA© HI 98130 (Woonsocket, RI, the USA), which also gives temperature and pH. This campaign was carried out at the end of the dry season (may) and at the end of the rainy season (November); a settled ensemble of five multi-sensor devices localized in the Casamance River estuary only, since Jan 2014; These devices are CTD sensors (Conductivity, Temperature, Depth) model Decagon© CTD 10 (Pullman, WA, the USA), each one was coupled with a Decagon© EM50 data logger.
According to Noblet (2012) [35], salinity is calculated as follows: where: Equation (2) was validated with the measured values with both a refractometer and field conductimeter. Since the bolons' water temperature always ranged between 21 and 28 °C, the deviation between measured and calculated values was low, rarely exceeding 2.5% (the highest observed difference was 4.8%). Table 1 summarizes the data collected thanks to the implemented instruments. CTD data were regularly calibrated with that of the conductimeters (up to 20 mS/cm) and that of the refractometer (from 5 g/L, i.e., approximately 3 mS/cm, up to 100 mS/cm).

-
Refractometers were calibrated with distillated water at the beginning and at the end of each measurement fraction of the day. -Conductimeters were calibrated with standard dilution products supplied by the provider at the beginning and at the end of each measurement fraction of day in order to ensure the quality of measured data.

Results
Figure 4a (Saloum estuary) and 4b (Casamance estuary) give the location of the observations and implementation devices.
Despite the different periods of observation in each one of the settled devices, some general observations can be made about the current behavior of the salinity in the estuaries. -

Casamance Estuary
The salinity annual variation increases from the mouth (located 2 km downstream from Karabane station) to the upstream part of the estuary.  The following characteristics are observed: -There is along all the year a fresh water income at the upstream entry of the main branch of the Casamance estuary (at Diopcounda Bridge); the same observation is made at the upstream origin of its main tributary, the Soungrougrou (Diaroumé Bridge); however, fresh water discharge is significantly lower in this river; -There is always an estuarine turbidity maximum (ETM, [36,37] Downstream of this moving peak, salinity decreases all year long; then, Casamance river has an inverse estuary, however, its upper part has a normal functioning during a few kilometers in the dry season and over some tens of kilometers in rainy season; o In the main branch (Casamance), a second salinity peak is observed during some seasons at the confluence with the Soungrougrou river, due to the upper salinity values of the latter; o As observed in Figure 4a-f, salinity values are lower in the rainy season and higher in the dry season in the tributary bolons than in the main reach of the Casamance river estuary.

Saloum Estuary
Figure 7a-f documents the salinity of the Saloum estuary at the end of the rainy season ( Figure  7a,c,e for 2016, 2017, and 2018, respectively) and at the end of the dry season (Figure 7b,d,f for 2017, 2018, and 2019 respectively). The functioning of this estuary is simpler than that of the Casamance.

-
The fresh water discharge in the rainy season is quasi null and thus completely negligible; -Values are lower during the rainy season due to lower evaporation, rain fallen within the wide estuary zone, and the sum of many small inputs by surficial runoff and small bolons; - The estuary has an inverse behavior all year long; - The salinity always increases upwards; the maximal values are always measured completely upstream, at Kaolack bridge in the Saloum and at Fatick Bridge in the Sine river; - The salinity is higher in the north branch of Saloum estuary than that in the mid one (Diomboss) and overall than that in the southern one (Bandiala) (see location Figure 4a); - The Bandiala bolon is provided in fresh water by the Nema Bah river, which is a small permanent fresh water river; water comes from the abundant water table of the southern Saloum plateau. The temporal interannual variability is discussed in the last part of the discussion.

A Comparison with Historical Data
One of the main references of past salinization (during the dry period 1968-1993) is the documentation about the Baila Bolon [3,38,39]. These authors highlighted the hypersalinization of the Baila bolon at its peak in 1984, as well as within the whole Senegambian valley [29]. In the Baila bolon, the mean salinity value was 96.8 g/L in 1983-1984; in the following years, salinity decreased until reaching 45.4 g/L in 1988-1989. Figure 8 shows the evolution of salinity at Baila (Baila bolon) during two periods. Salinization is not irreversible, and the rainfall recovery after 1988 allowed a quasi-complete desalinization at the end of the rainy season at the upstream sites of the estuary. The groundwaters have been recharged, and they flow through the salted rivers. They recharge the aquifers, rising water tables and allowing the conservation of all its fresh water [3]. The salinity at Baila station during the current period (Figure 5d), although presenting high interannual variability, is closer to the level observed during the 1978-1979 and 1988-1989 periods [3,32] than that measured during the 1982-1987 period [3]. The latter corresponds to the driest period of the Great Drought, the end of the second peak of dry years, which therefore is the period when the cumulated rainfall deficit was the highest. The 1978-1979 years are the period with a temporal rainfall recovery between the two "peaks" of droughts (1972-1974 and 1982-1984); the 1988-1989 period is also considered as the beginning of the rainfall recovery after the drought. Table 2 shows the evolution of salinity at different points of the Casamance estuary, during two periods:  This suggests that in spite of the rainfall recovery, the salinity did not decrease completely after the second "peak" of the drought, during the 1980s. Figure 9 allows the same conclusion about the Saloum estuary; values of salinity are higher in the 2016-2019 period than during the 1980-1982 one. It is likely due to the strong increase in salinity caused by the second peak of the drought (1983)(1984)(1985); the rainfall recovery since the 1990s was not sufficient to complete a true desalinization. Figure 9. Salinity (g/L) vs. distance to mouth in the Saloum estuary, after [15]. Debenay and Pagès (1987) [40] defined the Casamance estuary, and they noticed five areas to be distinguished from downstream to upstream: -From 0 to 50 km: a marine domain, with salinity, tides, and behavior close to those of the sea; -From 50 to 85 km: an intermediary area, with increasing salinity; -From 85 to 175 km: an hyperhaline area where salinity can reach 100 g/L; -From 175 to 225 km: an alternative domain where salinity can vary from 0 to 100 g/L between rainy and dry season and during a few weeks only; -Upstream from 225 km: fresh water with low discharge of the continental area.
The rainfall recovery in Casamance [25] and in the whole Senegambia [21] did not modify this general functioning; however, it caused a significant decrease in salinity in both estuaries of the Saloum and Casamance rivers.

An Integrative Indicator: The Mangrove
The mangrove forest is located at the boundary between the ocean and continent. The global mangrove area is declining mainly due to human activities [41,42]. Inversely, it is worth noticing that in West Africa, after a decline period during the Great Drought of 1968-1993, the mangrove area was significantly rising [43][44][45]; it almost reached its ancient extension level within the Saloum estuary and it even exceeds it within the Casamance estuary. It is expanding quickly [44,46] due to rainfall recovery and overall due to sea level rise. An improvement of mangrove governance and reforestation (unfortunately, mainly proceeded with Rhizophora in places where Avicennia was most indicated) locally could have contributed to this extension (only 4% of the increase in the mangrove area is due to reforestation, with the other 96% being spontaneous in the Saloum estuary [46]; respectively, these figures are 7% and 93% in the Casamance estuary [45]).
The mangrove is very sensitive to salinity levels; it can resist a few weeks or months to very high or very low salinity; however, if these peaks are repeated each year or if salinity values remain very high for more than 7 or 8 months a year, mangrove would eventually die. Gilman et al. (2008) [47] describe the role of sedimentation on the mangrove stability and the ways that sedimentation may impact mangrove resilience (such as sediment accretion and erosion, biotic contributions, belowground primary production, autocompaction, fluctuations in water table levels, and pore water storage) but conclude that there is no correlation between sedimentation rates and sea level rise (SLR). Osland et al. (2018) [48] show how the accelerated SLR could favor sedimentation and biotic contributions and thus the extension upland or upriver of the mangrove forest. Mangrove forests in arid and semiarid climates are known to be particularly vulnerable to changes in rainfall and freshwater availability [48]; the SLR probably accelerated the desalinization of the West African bolons, and this is probably the main explaining factor of the current mangrove expansion.
As an example, and contrary to the conclusion of Dièye et al. (2013) [49], the natural opening of the "fleche de Sangomar" sand spit in 1987, at the northern side of the Saloum estuary, allowed the entry of important volumes of marine water (35 g/L) in an hyperhaline estuary; then, it provoked a reduction of the salinity in the inland estuary. The mangrove regeneration was firstly observed near Djiffère in 1992 (Mamadou Sow, personal communication), where the opening of the "fleche de Sangomar" allowed a strong relative desalinization and the mangrove recovery. Mangrove decline in the lower Saloum estuary after 1987 and the end of the drought (rainfall annual total amount began increasing after 1985) observed by Dieye et al. (2013) [49] must be due to factors other than salinization.
SLR seems to be influencing the mangrove extension by leading to a relative desalinization of inverse estuaries water and by helping greater volumes of sea water entering in estuaries with much higher salt rates. The rainfall recovery since the end of the 1980s made the Casamance estuary show normal behavior during an increasing proportion of the year; this recovery did not allow at the time the Saloum River to have periods with normal functioning. However, there is no evidence that Saloum was not yet, before the drought, a completely inverse estuary at least since the African Humid Period, before 4000 BP.
The current expansion of the mangrove is the result of the significant relative desalinization of the inverse West African estuaries.

Low Discharges Explaining the Inverse Estuaries
Saloum and Casamance rivers have very low discharge values due to geology (sedimentary originated mainly sandy soils), the topography is very flat and, in addition, in Casamance, vegetation is very dense and rice cropping is organized to store most of the rainwater during the rainy season. Table 3 indicates the runoff coefficient values to be compared with that of surrounding basins. Clearly, runoff coefficients are lower in the Saloum and Casamance basins compared with other basins that have a comparable total rainfall amount. This is one of the main explaining factors of their inverse functioning.
The Casamance River basin has a runoff coefficient just slightly higher than that of the Senegal River basin, although it receives 55% more rainwater than the latter. The runoff coefficient of the Saloum River basin is half of that of the Casamance River. The Geba River basin also has a low runoff coefficient compared with that of the other basins with similar rainfall amount.
These three basins have little runoff due to geological and topographical factors. In the two first cases, runoff is significantly enhanced by the rain water fallen directly in the estuary; this constitutes 44% of the total discharge of freshwater in the Saloum basin and 58% of that for the Casamance basin.

Conclusions
These observations about salinity spatiotemporal evolution carried out within two West African inverse estuaries allow us to attempt a few statements: Firstly, about the functioning of estuaries, we can confirm that: -The Saloum River estuary has a total inverse behavior, with salinity increasing upwards; - The Casamance estuary has a spatially partial inverse functioning, with a point of maximum salinity migrating from 20-30 km of the upstream end estuary at the end of the dry season to 50-80 km downstream from this point at the end of the rainy season. -Therefore, about the spatial variability, we observed: -decreasing salinity downwards from the peak of the ETM in the Casamance estuary and from the upstream entry of the estuary at Kaolack in the Saloum river; -Increasing salinity seasonal variability in the tributary bolongs; -A similar behavior in the bolongs than in the Casamance upper estuary.
Finally, the salinity seasonal variability increases with the distance to the ocean. A relative desalinization is attested by the comparison of measurement realized since 2013. Besides, the progressive post-drought rainfall recovery led to a decreasing bolongs water salinity but answering with a 10-15-year delay. The mangrove spectacular recovery is a good indicator of the progressive reduction of the bolongs' water salinity.
However, the low discharge and runoff coefficient values in the continental part of the basins is one of the main and persistent explaining factors of this inverse functioning.