3.1. Hydrogeomorphological behavior
The Fertém station data show Marimbus wetland hydrogeomorphological regime and maximum potential flooding (Fig. 3a), namely:
-
the fluvial discharge;
-
the variability of water levels (seasonally and yearly extent);
-
how often it floods and dries (frequency);
-
how far the water spreads (extent);
The catchment area of the Marimbus wetland covers 10.111 km2, strongly influenced by the mountainous relief, which reduces the permanence of water in this sector of the basin and intensifies the flow. The average river discharge in the Marimbus wetland is relatively low, at 29.5 m3 s-1 and sedimentation is proportional, <5 ton.day-1. However, discharge peaks of 878 m3 s-1 are observed (Fig. 2a), which move a higher sediments load than during long periods under normal conditions, reaching 962 ton.day-1. These peaks are much lower than those found for average Brazilian and worldwide rivers, if proportionally considered the ratio between catchment areas versus solid and liquid river discharge24.
The Marimbus wetland has maintained an average water level of 1.20 m (Fig. 2b). The floodplain is quite uneven with many shoals and low margins. This means that a rise of only 2 m in water level is enough to ensure the maximum lateral flooding of the wetland, which occurs 25% of the time, with several flooding cycles exceeding 5 m in height. This frequent waterlogging led to the death of the trees and new composition of vegetation formations. Most wetlands are shallow depressions and small catchments areas8–9.
The highest elevation period of the Marimbus wetland water level extends from November to April, while the other months are considered as dry period, whose flood frequency rarely evades this pattern of seasonality (Fig. 2c). Several droughts reduced their levels drastically (December 2011-to-December 2013), but the freshwater remains. The opposite moisture event occurred between October 1977 and May 1979, when this wetland was flooded during all months, above the long-term monthly average. Both events are considered not seasonal. During the dry season, a mosaic of sandy islands emerges (Fig. 3a). During drought events, the alluvial fans are exposed and only isolated sedimentation rings remain flooded (with rounded dolines resemblance). The latter contain trapped silt and clay facies, while along the active channels there is fine to coarse sand. The overbank deposition enabled areas of shallow water to form in the adjacent backswamps. The abandoned meanders indicate these hydrodynamic energy oscillations in the flooded area25–50, creating new channels on the ଂoodplain and causing the abandonment of the old channel (avulsion).
The delimitation of the Marimbus wetland area was complex. We used a colored composition containing the band 4 (0.77-0.89 µm) commonly used for delineation of water bodies, and compared with SRTM contour lines extracted with 1 m equidistance from the 320 m above sea level (base station). Both boundaries extracted from the images show similarity in the 6 m maximum hydrological flood level (326 m asl), which occurred only in 1960 (see Fig. 2a) and its corresponding flood area of ≈58 km2.
The Marimbus surface water dynamics has been recorded from coarse-spatial-resolution satellite observations, and higher-resolution seasonality maps using all Landsat images over multiple decades have been used to map seasonality and surface water changes. Hydrogeomorphological behavior analysis showed that Marimbus is a freshwater wetland, flooding periodically, currently located in a lithological depression. The three sectors (northern, central and southern) show changes over time and space, with the southern portion and the Encantada pond being the areas with the most permanent water levels. All areas were subject to ephemeral and seasonal floods, between 1984 and 2021 (Fig. 3b). This land supports predominantly hydrophytes, the substrate is mostly undrained hydric soil, and is saturated with water or covered by shallow water at some time during the growing season of each year.
3.2 The water’s source
For the determination of the origin of the Marimbus wetland waters, isotopic analyzes were performed at several points. The samples presented values of δ18O (VSMOW) and δ2H (VSMOW) ranging from -21.1‰ to 9.0‰ and from -3.13‰ to 3.31‰, respectively. Most of the samples presented data coinciding with the global meteorological curve10, due to the marked influence of precipitation in the runoff.
Figure 4 shows the excess deuterium parameter of the waters collected in the Marimbus wetland. The values of this parameter ranged from -21‰ to + 17‰. In this figure there is a group of samples with negative values and another group with positive values of this parameter. Samples MB1 through MB6 were collected in the permanently flooded area. MB1 to MB3 in the Encantada pond, MB4 and MB5 in a large lake and MB6 in a shallow lake that, due to the smaller volume of water and water depth, undergoes more evaporation, presenting the most negative value of excess of deuterium (-21‰). Samples MB1, MB2, MB3, MB4, MB5 and MB6 showed different behaviors when compared to the other samples. They have negative excess deuterium parameter, which consists of a higher enrichment of water in δ18O with respect to δ2H. Therefore, the points located in the floodplain (annually flooded) of the north and central portions of the Marimbus contain water that presents an underground recharge component that undergoes considerable evaporation. To confirm the different isotopic compositions, a local evaporation line was developed only for samples MB1 to MB6. This curve was defined by the following linear regression: δ2H = 5.097δ18O - 8.76‰. Note that the slope coefficient of line 5,097 is less than that of the global meteoric line (GML) which is 8, which characterizes water presenting more evaporation39.
The negative excess deuterium in these samples indicate waters that have undergone considerable evaporation or waters that have underground recharge more evaporated than precipited. When evaporation occurs in water bodies, deuterium-containing water molecules evaporate more easily than 18O-containing molecules. This effect causes an 18O enrichment with respect to deuterium in the remaining water and, consequently, the calculated values of deuterium excess will become more negative as its evaporative loss increases. Samples collected between sites MB7 and MB20 show significant influence of rain water. The sample points in the chart below are very close to the GML.
Another evidence of rainfall in the isotopic composition is recorded in the values of excess deuterium that varied between +6.1 and +16.5‰, with an average value of +10.9±3.0‰, very close to the linear coefficient of the GML δ2H = 8. δ18O+10.8‰. It does not corroborate the values obtained by Sales (2017), when the Santo Antônio river acrossing carbonate rocks (see Fig. 1), where the values of excess deuterium, calculated with their isotopic data, ranged from +9.3 to +16.2‰, with an average value of +12.9±2.2‰ (Fig. 4). MB19 and MB20, collected in the Garapa and Paraguaçu rivers before reaching the Pantanal, presented the highest values of excess deuterium: + 14.8‰; + 15.3‰ and + 16.5‰, respectively.
Generally, this enrichment can be the result of the recharging of cloud rains that are not generated in the region, with several consecutive episodes. This causes a rapid depletion of the 18O in the cloud generating positive excess deuterium. The data collected in the southern portion, points MB8 to MB18, showed slightly positive deuterium values, closer to the value of the GML coefficient, indicating that the meteoric waters were their main source of recharge. The MB16 and MB17 points, collected in the southern Marimbus wetland, show the lowest values of deuterium for this data group: +7.9 and +7.6‰, respectively. The MB7 point, collected at the confluence of the São José and Santo Antônio rivers, showed the lowest value of positive deuterium due to its proximity to the sedimentation rings of the central portion of the wetland. It suggests the interaction between groundwater and river, which maintains the minimum levels in the dry period and amplifies the quotas in the rainy periods.
The similarity in isotopic information at these points is due to the marked influence of precipitation on runoff. These waters are typical of the ଂoodplain which remains saturated for extended lengths of time (backswamps) and is often isolated from the river channel as a result of aggradation occurring elsewhere on the floodplain25.
The waters in the karstic aquifer that feed into the springs of the Santo Antônio river in the Irecê carbonate basin range from calcic sodium chlorinated composition to calcium bicarbonate40. The isotopic composition of aquifer waters refers to the natural waters of an isolated karstic system, its isotopic signature distinct from the waters of the Marimbus wetland (Fig. 4). They are rounded shaped ponds typically found in wetlands and floodplains too, but theses body waters are alkaline salines and present pH values of up to 10, with the presence of bicarbonate, chlorinated and sodic waters5. Some of these ponds are isolated from pluvial surface flow and are characterized by white-sand beaches and brackish to saline water during Holocene-Pleistocene.
3.3 Radiocarbon dating geomorphic change
The radiocarbon results for each tree ring sample are presented in table 1. The piece of trunk used for radiocarbon (14C-AMS) analyses was sampled along the growth rings sequence, so that an age model could be built. This was done because the independent dating of the bark would result in a wide probability range, covering the industrial period when the input of fossil carbon has diluted the atmospheric radiocarbon concentration, preventing precise dates to be estimated47.
By dating four different samples from the bark and core, it was possible to obtain radiocarbon dates that correspond to the 18th century. For no robust dendrochronology could be performed, a simple sequence model was built, revealing a large probability that the tree lived until approximately 1700 AD (Fig. 5). Based on the dating results there would still be a slight probability of the bark reaching 1800 AD. However, the size of the sample indicates that, even if some rings were missing, it could not represent more than 100 years of growth for the Hymenolobium sp. Previous work on the growth rates of some Amazon trees revealed that such species usually grows less than 40 cm y-1 32. Therefore, the death of the tree most probably took place around 1700 AD.
Recents radiocarbon analysis showed that landslides were responsible for creating two lakes in the western United States after earthquake events, including the A.D. 1700 Cascadia earthquake. Generally, mountainous settings commonly trigger thousands of landslides, and slope failures are typically significant for landslide-dammed aquatic enviroments46. As there was never any record of earthquakes in Chapada Diamantina, we strong suggest that Marimbus wetland were formed approximately 1700 AD, by mining activities that silted the main river, leading to the impoundment of the tributary river.
It has been frequently stated that the world has lost 50% of its wetlands (or 50% since 1900 AD). The reported long-term loss of natural wetlands averages between 54–57% but loss may have been as high as 87%, since 1700 AD11. Thus, the environmental impact caused by siltation in Chapada Diamantina had the benefit of expanding the local wetlands.
The Marimbus wetland represents a scenario described in the literature, recording a few centuries of landscape changes in response to human activities, especially when the dead trees could still be found preserved in their life position. Other debates about the age of wetlands are important to know the implications of secular-and-millennial geomorphic changes, including their ecological significance. The Marimbus water level change caused the death of large trees, replacing them with aquatic vascular plants. These new aquatic plants were able to adapt to water level fluctuations in the Marimbus wetland and are concentrated in areas where water flows and levels cannot vary both seasonally and from year to year. However, the vascular plant composition in the Marimbus wetland showed low similarity to that of other wetlands present in the Brazilian territory12. This characteristic corroborates the relatively juvenile evolutionary state of this wetland. It is currently known that the floodplains of tropical South America may be considered as areas of speciation, contributing to the great species diversity in the area20. However, the low similarity index of macrophyte vegetation, recorded in the Marimbus wetland12, may be an indicator of the frequent changes in the young braided stream (avulsion), which capture the sediments inside the wetland, where plants respond to this pulsing with a large set of physiological adaptations.
There are tree possibilities to evaluate the 300-year-old Marimbus wetland: (i) insufficient time to deposit the alluvial fans, given the observed river discharges, demonstrating the anthropogenic origin of sediments, (ii) the exploration and/or use of mineral resources (not necessarily diamonds) existed since the 18th century in the Sincorá mountain range and has not been historically documented, and (iii) the vascular plant composition showed low similarity to that of other wetlands present in the Brazilian territory, corroborating the relatively juvenile evolutionary state of this wetland.
Table 1. AMS 14C data of dead trees in the Marimbus wetland
Satellite images show that alluvial fans on the Sincorá range footslope are concentrated between parallels 12º20'00”S and 13º00'00”S (Fig. 6). Historical data showed that diamond mining activities at Chapada Diamantina also concentrated approximately between these same parallels41. Outside this area there was not diamond exploration at Chapada Diamantina, and the rivers flow from the mountain without accumulating sediment.
The alluvial fans divide the Marimbus wetland into three sectors, all located at the confluences of the rivers. A narrow and elongated alluvial fan separates the northern portion from the central portion at the confluence of the São José and Santo Antônio rivers. Another alluvial fan with elongated radial axis extending for approximately 1.5 km separates the central portion from the southern portion, at the confluence of the Garapa and Santo Antônio rivers (Fig. 7). This fan system is broadly divisible into three parts: (i) an upper entry corridor, approximately 6 km long and 40 m wide; (ii) a central zone of seasonal swamps transected by several distributary channels confined by densely vegetated banks, and (iii) a lower zone of perennial swamps where flow is mostly unconfined. In the final stretch, the Santo Antônio river (tributary) was dammed by the Paraguaçu alluvial fan (main river) because, due to its intense siltation, it became topographically higher than the tributary river (Fig. 8). We suggest that the transfer of sediments from the mountains to the confluence of rivers was responsible for allowing the river system to become a wetland.
If we use radiocarbon dating of the geomorphic change of the Marimbus wetland as a marker, it is observed that the alluvial fans present exceptionally high sedimentation rates and could not have occurred in natural conditions. In this case, 300 years would be a short time to deposit this volume of sediment, where the sedimentation rate should be 6 to 8.6 cm y−1. Sedimentation rates > 1 cm y−1 are associated with great rivers (Walling and Fang 2003). Soundings indicated alluvial fans maximum thickness, between 18 to 26 m41.
The combination of intense river discharge, high slope and the mining activity may explain the alluvial fans (Fig. 8), forming placers deposits that are still active today, and where coridon, rutile, cianite, limonite and turmaline can be found in the diamonds levels. The occurrence of gold is not very significant. The volume of these alluvial fans containing only diamond pebble-supported has been estimated at 20 x 106 m3, where: Paraguaçu alluvial fan has been 6.8 x 106 m3 (Fig. 8), Santo Antônio alluvial fan has been 10 x 106 m3 and São José alluvial fan has been 3.2 x 106 m3 41.
The geomorphic change dating (this work) suggests that the exploration of mineral resources has existed since the 18th century in the Sincorá mountain range. In this scenario, it is not clear whether the existence of diamonds in the Chapada Diamantina has been known since the 18th century. It is possible that alluvial fans originated from the gold search phase and consequently formed the Marimbus wetland. Only after 1844 did Chapada Diamantina intensify exploration and attract thousands of miners to the region, where Chapada had its name linked to Diamonds (Toponymy: Chapada roughly means sedimentary plateau on Precambrian rocks, and Diamantina concerning Diamonds). This date only symbolizes one fact, due to the fact that the discovery of the mineral was kept secret and occasionally a diamond buyer was forced to reveal its existence49.
Historical documents show that between the 17th and 18th centuries, explorers circulated intensively in the coastal and semi-arid region of the State of Bahia (site of the discovery of Brazil, in 1500 AD) in search of metals and precious stones. In reality, this movement dates back to the 16th century. The first expedition entered the Paraguaçu river, in 155935 and reached Chapada Diamantina region in the 17th century. The search for mineral wealth was hard and slow and lasted a long time without major discoveries.
Early indications of Brazil’s mineral potential were sporadic, but there is evidence that crystals were found in Bahia within a century of Columbus’s discovery of the New World. In one of the earliest descriptions, historian Pero de Magalhães Gândavo (1576) mentioned the existence of “certain mines of white stones such as diamonds”. In another account, Gabriel Soares de Sousa (1587) noted that fine, eight-sided crystals (possibly diamond) had been found during the dry winter months along certain rivers48.
The search for diamonds, gold and other valuable natural products was a major driving force for the exploration and colonization of the interior of the country in the 17th and 18th centuries49. The discovery of gold in the state of Bahia was the result of much investment by the Portuguese government which, throughout the 17th century, stimulated, subsidized and gave rewards to anyone who ventured in search of metals and precious stones14.
The largest results of the expeditions were achieved first with gold, found in the 18th century, then diamond, whose first deposits were discovered in the early 19th century. The Sabugosa Viscount, Brazilian viceroy, decreed in 1731, the elaboration of precise maps on the terrestrial ways conditions, to enable the Portuguese monarchy to know the productive capacity of gold deposits, in addition to transport conditions to the coast and new mining territories of the Chapada Diamantina35–49.
Although the exact year is uncertain, the accepted discovery of diamonds in Brazil is thought to have occurred sometime between 1710 and 1730. In Minas Gerais, diamonds were officially found around 1714, and soon attracted the attention of the Portuguese monarchy, which imposed the first legislation to regulate their exploitation dated 173233. Eventually, reports of diamonds in Brazil began to reach Europe. Accounts from the colonial governor came to the attention in Portugal, and the discovery was officially announced in 17294. Portugal moved aggressively to control the area, restricting gold and diamond mining and imposing high taxes. Despite efforts by the crown, clandestine mining and diamond smuggling increased49.