Insights on the evolution of the living Great Amazon Reef System, equatorial West Atlantic

The Great Amazon Reef (GARS) is an extensive mesophotic reef ecosystem between Brazil and the Caribbean. Despite being considered as one of the most important mesophotic reef ecosystems of the South Atlantic, recent criticism on the existence of a living reef in the Amazon River mouth was raised by some scientists and politicians. The region is coveted for large-scale projects for oil and gas exploration. Here, we add to the increasing knowledge about the GARS by exploring evolutionary aspects of the reef using primary and secondary information on radiocarbon dating from carbonate samples. The results obtained demonstrate that the reef is alive and growing, with living organisms inhabiting the GARS in its totality. Additional studies on net reef growth, habitat diversity, and associated biodiversity are urgently needed to help reconcile economic activities and biodiversity conservation.

. Location of the study area, sectors as defined by Moura et al. (2016), and samples analysed in this study. Source of information for samples: Red cross 8 , Brown triangle 3 , Black cross 37 , Yellow dot (this paper). turbid-water reefs include the most extensive coral reefs in the SW Atlantic (Abrolhos Bank, central Brazilian coast), which are surrounded mostly by siliciclastic sediments 14,15 . Brazilian biogenic reef ecosystems are recognised for being well adapted to high turbidity levels due to terrestrial input (via large rivers discharge) and sediment resuspension 14,16 .
The GARS is located mostly at the middle, and outer shelf of the Foz do Amazonas and Pará-Maranhão marginal sedimentary basins within a depth range of 70 to 220 m 4 . The area is substantially influenced by the Amazon River, with a mean annual water discharge on the order of 6.0 10 12 m 3 and annual suspended-sediment load on the order of 1.2 10 9 tons 17 , affecting all physical and chemical aspects of the water column (e.g., salinity, light availability, pH, dissolved nutrients) over the shelf including the GARS. Most of the sediment travels north in a plume driven by the northwest-flowing North Brazil and Guiana Currents, which are forced by local wind patterns 18 .
The Amazon River plume has a very dynamic character and even though it is generally driven to the northwest, seasonal variations of winds and currents on the shelf lead to some southeastward expansion of the plume, as a result of displacements caused by the Inter-Tropical Convergence Zone (ITCZ), with substantial rainfall and NE Trade Winds between January and June. In contrast, between July and December rainfall is reduced, and SE Trade Winds prevail 19 . There are four distinct zones based on light regimes reaching the bottom, three of them with constant light regimes (dark coastal zone under the permanent influence of the plume; dim-light zone in the deeper northern shelf and high-light zone in the shallower southern shelf) and one zone with seasonal changes in benthic light regimes (northern mid-to-outer shelf) 10 .
Waves reach the region mainly from the east and northeast, because of the NE-SE trade winds, with dominant offshore wave heights between 1 and 3 m. The most energetic waves approach the region between December and March, with strong NE Trade Winds offshore 20 . Besides the oceanic currents and waves, tidal water level variation and tidal currents are also substantial in the area, with tidal amplification on the shelf resulting in macrotidal conditions at the coast 21 .
In this paper, we aim to summarise published and unpublished radiocarbon dating from carbonate samples from the GARS in order to: 1) evaluate its modern or relict nature and 2) contribute to the understanding of its evolution by proposing a theoretical model from the end of Marine Isotope Stage 2 until modern ages. Table 1 summarizes the radiocarbon ages presented in this work.

Results
Radiocarbon ages can be divided into three main groups ( Fig. 3): i. Contains samples exclusively from the northern sector, at water depths between 104 and 150 meters, encompassing samples with ages from the MIS2, as defined by 22 , with three exceptions, the first corresponding to an age of ca. 41,800 cal BP, and two others, with ages corresponding to the Younger Dryas, as delimited by 23 . The oldest ages correspond to the dating of oolites, published by 8 ; ii. Encompasses samples from the North and Central sectors, and presents a period from ca. 7,100 cal BP (Mid-Holocene), to historical ages (non-modern radiocarbon ages). Water depth from these samples varies from 50 to 95 meters, and; iii. Corresponds to samples with modern radiocarbon ages, as defined by 24 and contains samples from the three sectors. The water depth range of these samples varies from 23 to 100 meters; the shallowest samples are located in the Southern sector.

Discussion
Cross-shelf profiles of sediment samples obtained between 2017 and 2019 show that the morphology and the sedimentary characteristics from the central and southern sectors of the GARS support the ages compiled here (Fig. 4). Considering the mesophotic depth range of 30 to 150 m, local depths and the sea level variation 25 , the area of the potential occurrence of mesophotic reefs during the Last Glacial Maximum (LGM) was rather narrow and restricted to the continental slope. As sea level rose quickly from LGM towards Mid-Holocene, there was a period of restriction of areas within the mesophotic depth range, before the drowning of the shelf. The bathymetric profile from the central sector shows a substantially wider shelf, within a deeper outer shelf (Fig. 4), favouring reef development during Mid-Holocene. On the other hand, the distance from the Amazon River mouth and the shallowness of the shelf at the south favoured relatively shallow reef development there, with shallow reefs expanding their occurrence towards the central sector. A recent study focusing on fisheries in the Amazon River mouth suggests that reef structures may occur in areas that are much shallower than previously anticipated 26 .
A recent review of the formation of oolites 27 confirms the intertidal and shallow subtidal formation of most of the marine oozes but emphasizes the complexity of the biogeochemical processes involved in their formation. When comparing the position of the oolites dated by 8 with the sea-level change curve (Fig. 3), we agree with the observations by 11 about the lack of reliability of oolite materials as indicators of ancient shores on the Amazon margin.
In this sense, we state that the MIS2 and MIS 3 ages, presented by 8 and used by 7 as for the onset of the GARS cannot be used for an evolutionary model for the reef system. Assuming that the oldest ages must be analysed with care, the first reliable dating to be considered for the GARS correspond to the samples, located on the Northern sector, that lie between 14,680 and 12,100 cal BP, at a water depth of 120 meters, in synchronicity with the Heinrich H1 28 .
Considering these ages and the sea-level curve shown in Fig. 2 25 , we propose a model of the evolution of GARS throughout the Late Quaternary, comprising three major phases:   www.nature.com/scientificreports www.nature.com/scientificreports/ ii. After the first phase, there is a gap in radiocarbon ages, corresponding to the interval between 12,100 and 7,100 cal BP, followed by the beginning of the occurrence of reef material in the Central sector. This gap roughly corresponds in time to the acceleration of the sea-level rise after the Younger Dryas (Melt Water Pulse 1B) 30,31 . Further, the time of the gap also follows the period when the Amazon River sediment load was developing a prominent plume, whereas during the previous period the sea-level was close to the shelf break and most of the Amazon sediment was transported into the deep sea by way of the Amazon Submarine Canyon and smaller channels 32 . Within the strong plume development, light penetration in the water column would be strongly attenuated, impairing the reef development. This process attenuated as the sea-level rise went on and the Amazon River plume moved landward and northwestwards. iii. Finally, modern ages are found in samples from the three sectors, indicating the spread of the reef complex, from Northwest to Southeast. Also, reef constituents are found in areas as shallow as 20 meters (in the southern sector), and as deep as 220 meters 4 .
Our radiocarbon dating study is coherent with previous studies performed on rhodoliths from the Pacific and the Atlantic oceans where metagenomics demonstrated that the majority of living material corresponds to microbes (bacteria) capable of growing in a variety of extreme environmental conditions for carbonate precipitation 33,34 . Also, it was clear from these previous studies, that rhodoliths have a fraction of living carbonate cryptofauna and/or epifauna, including Foraminifera (Homotrema rubrum), Polychaeta (encrusting calcareous tubes), Bryozoa, and Mollusca (Vermetidae) phyla 33 .
The GARS is considered an ecotone of biodiversity between Brazil and the southern Caribbean, possibly acting as an ecological corridor between the South and North Atlantic. It comprises a high diversity of habitats and large areas dominated by healthy reef-building organisms (mostly crustose calcareous algae) 4 . Reef growth and high structural complexity are presently concentrated in the central and southern sectors, which are shallower and with higher light incidence over the bottom wealth. Net reef growth estimates (i.e. considering both, reef accretion and erosion) are scarce even for shallow reefs. Only recently, this question started to be addressed for mesophotic reefs 35,36 . Similar studies are urgently needed for the GARS, particularly considering its ecological importance and the threats posed by large scale oil and gas exploration in the region 4 .

conclusions
The analysis of previously published and new radiocarbon data from the GARS allowed us to recognize a Northwest-Southeast growing trend of the reef complex. Older ages correspond to samples from the northern sector, in areas located below the present plume of the Amazon River. These ages refer to the carbonates that sustained the reef during MIS2 and to the beginning of the last deglacial. After a gap, occurred between ca. 12,100 and 7,100 cal BP, the reef extended to the Central and Southern sector of the area. This gap corresponds to the time interval of acceleration of sea-level rise and plume development, after the Younger Dryas (Melt Water Pulse 1B).
Modern radiocarbon ages, obtained in rhodoliths and sponges, are present in the three sectors, indicating that differently than suggested, living organisms inhabit the GARS in its totality.

Methods
As a convention, in this work, we keep the same proposal of three sectors (North, Central, and South) as stated by 3 . Besides our new radiocarbon data, we gathered dating information from other sources 3,8,37 in order to evaluate the modern or relict character of the reef complex. When available, all of the conventional radiocarbon datings were recalibrated using software Calib 7.1 38 , using the Marine13 calibration curve 39 and a global reservoir effect. Ages published in 37 are presented "as is" (only calibrated ages, without the 2σ interval), since we did not have access to the original data. We also dated ten other samples (Table 1) of carbonates from different marine organisms from the three sectors at Beta Analytic (Miami, USA).
In order to compare our data with sea-level changes, and due to a lack of a reliable Late Quaternary sea-level curve for northern Brazil, we used the global curve proposed by 22 .
The location of the samples is shown in Table 1 and Fig. 1.