*4.2. Linking VLM and Coastal Erosion*

The coastline changes and VLM findings are contrasting. Whereas erosion processes underlie most of the study area, the signal of VLM is highly variable along the study area. In other words, not all areas with annual shoreline erosion rates larger than the uncertainty

threshold (i.e., ±5.0 m/yr) were associated with local subsidence. This indicates that subsidence is a factor that can exacerbate the impact of ongoing erosive processes. We explored the influence of scale by examining aggregations of the data at 1.5- and 5-km scales (see Supplementary Materials). At all scales considered, the sector with high coastline erosion rates of around −10 m/yr coincided spatially with subsidence rates of up to −1.0 cm/yr (i.e., stretch between Kangarú and Cuatro Bocas in Figure 5). Regions that exhibited coastal erosion within the range of uncertainty (± 5m/yr, e.g., seaward from the CGSM) showed a variable VLM signal at all scales. A similar interplay is described for a stretch of coast comprised of marshes and mud in the Mekong Delta in Vietnam, where secular coastal erosion values larger than −50 m/yr are paired with subsidence rates exceeding −1.5 cm/yr [17]. Similarly, high coastal erosion trends in an abandoned lobe of the Yellow River delta in China were associated with average subsidence rates of −0.7 cm/yr and up to −2.1 cm/yr for the period 1992–2000 [13]. These observations highlight the regional variability and complexity of drivers behind the coastal change that demand weighing the different drivers of change even within one deltaic system.

In addition to the natural compaction caused by sediment overloading, the site-specific subsidence rates observed along the stretch of coast between Kangarú and Cuatro Bocas might be triggered by the effect of localized dehydration and mortality of mangrove forest (Figure 1), following the highway construction in the early 1950s [23]. In this sector, as the mangrove decayed after the highway construction, soil salinity has hampered the establishment of vegetation [23], giving place to barren or sparsely vegetated swales (see Figure 9c). Yuill et al. [8] report that a large component of subsidence in organic-rich soils (e.g., peats associated with mangroves) is due to the exposure of soils to atmospheric oxygen, which reduces soil volumes by converting organic carbon into carbon dioxide gas (Figure 1). In coastal Louisiana, at least 60% of subsidence takes place within the uppermost 5–10 m, where woodpeat is widespread [11]. Subsidence in the former mangrove forest within the study area (Figures 9 and 10) may therefore be higher with this additional factor at play and is at its root caused by the anthropogenic alteration of the wetland. Other factors causing site-specific variability in subsidence rates are related to the uneven distribution of compressible sediments, hydraulic properties in aquifer systems, and groundwater extractions from aquifer systems [56,57]. In the Mississippi delta, the variability of the underlying Holocene lithology modulates compaction rates and their spatial variability [11]. Such factors may explain the variability in VLM and LOS velocity observations east of Kangarú (Figure 5). A lack of detailed knowledge of the Holocene stratigraphic sequence at this site hinders establishing any current relationship between subsidence rates and underlying substrates, but surficial geological cartography, developed by the Colombian Geological Survey [26], indicates that subsidence takes place in alluvial and fluvial-lacustrine sediments that make up a floodplain landform [24].

A rise in relative sea level as a product of either subsidence or an increase of GMSL creates more space for sediments to accumulate (i.e., accommodation space) [64], which is an underlying condition for transgression to occur. In other words, for any given interval of time, accommodation space is created by a rise in sea level and/or subsidence. Thus, where rates of accommodation space are larger than sediment deposition, there is a landward shift in sedimentary environments, an expansion of the subtidal zone [65], and a landward displacement of the barrier [34]. Although stratigraphic evidence of a transgressive succession was not gathered in this work, previous mapping of chronic erosion in conjunction with frequent overwashes between Kangarú and Cuatro Bocas [24] indicate that transgression might occur along this stretch of coast. Transgression was identified by two processes acting in tandem on the landscape: (i) storm overwash deposition resulting in temporal accretion of sediments landward of the contemporaneous shoreline, followed by (ii) a period of more gradual erosion interrupted by a new overwash cycle. Thus, we suggest that subsidence velocities of up to −1 cm/yr in the observed stretch of coast are a factor that underpins transgression of the coastline and the occurrence of overwashes, resulting in a landscape composed of scarped dunes and eroding beaches overlain by washover fans

and surrounded by overwash channels. The spatial variability of VLM and LOS velocities east of Kangarú (see Figure 5) refrain us from extending the hypothesis of a transgressive shift of the coastline to the stretch of coast seaward from the CGSM.
