**Documenting a Century of Coastline Change along Central California and Associated Challenges: From the Qualitative to the Quantitative**

#### **Gary Griggs 1,\*, Lida Davar 1,2 and Borja G. Reguero <sup>3</sup>**


Received: 21 November 2019; Accepted: 13 December 2019; Published: 15 December 2019

**Abstract:** Wave erosion has moved coastal cliffs and bluffs landward over the centuries. Now climate change-induced sea-level rise (SLR) and the changes in wave action are accelerating coastline retreat around the world. Documenting the erosion of cliffed coasts and projecting the rate of coastline retreat under future SLR scenarios are more challenging than historical and future shoreline change studies along low-lying sandy beaches. The objective of this research was to study coastal erosion of the West Cliff Drive area in Santa Cruz along the Central California Coast and identify the challenges in coastline change analysis. We investigated the geological history, geomorphic differences, and documented cliff retreat to assess coastal erosion qualitatively. We also conducted a quantitative assessment of cliff retreat through extracting and analyzing the coastline position at three different times (1953, 1975, and 2018). The results showed that the total retreat of the West Cliff Drive coastline over 65 years ranges from 0.3 to 32 m, and the maximum cliff retreat rate was 0.5 m/year. Geometric errors, the complex profiles of coastal cliffs, and irregularities in the processes of coastal erosion, including the undercutting of the base of the cliff and formation of caves, were some of the identified challenges in documenting historical coastline retreat. These can each increase the uncertainty of calculated retreat rates. Reducing the uncertainties in retreat rates is an essential initial step in projecting cliff and bluff retreat under future SLR more accurately and in developing a practical adaptive management plan to cope with the impacts of coastline change along this highly populated edge.

**Keywords:** coastline; cliff and bluff retreat; erosion rate; uncertainty; sea-level rise; adaptive management

#### **1. Introduction**

Many diverse natural forces and processes interact along the shoreline, making the coastline one of the world's most dynamic environments [1–3]. Waves, tides, wind, storms, rain, and runoff combine to build up, wear down, and continually reshape the interface of land and sea [3–5]. Through the 20th century, however, global sea-level rise, due in a large part to human-induced climate change [6–8], contributed to increase both cliff and beach erosion [9,10]. Coastline (cliff and bluff) erosion (covered in this study) is different from shoreline (beach) erosion and is defined as the actual landward retreat of a cliff or bluff. While a number of older references indicate that cliffs occur along about 80% of the world's coasts [10–12], more recent work using a GIS-based global mapping analysis and a detailed literature review suggest that coastal cliffs likely exist on about 52% of the global shoreline [13]. Cliff retreat is distinct from beach erosion in that it is not recoverable, at least within our lifetime, by any natural processes [10]. The terms cliff and bluff are often used interchangeably [10,14], but in this

*Water* **2019**, *11*, 2648

study cliff refers to coastal landforms that consist of harder and more resistant rocks that stand higher and steeper than bluffs, which are generally composed of weaker materials and stand at gentler slopes (Figure 1) [10].

**Figure 1.** (**a**,**b**): Coastal cliffs, (**c**,**d**): Coastal bluffs. Photos: © 2002-2015, California Coastal Records Project [15].

The world's coastlines will respond to global climate changes and the associated adjustment to oceanographic forcing [16]. For cliffed coasts with limited beach development, there appears to be a relationship between long-term cliff retreat and the rate of sea-level rise [17]. Satellite altimetry has shown an average rise in global mean sea level (GMSL) of ∼3.4 mm/year since 1993 [10,18,19] and this rate is increasing by about 0.08 mm annually, which implies that global mean sea level could rise at least 65 ± 12 cm by 2100 compared with 2005 [20], enough to cause significant problems for coastal cities around the planet [21]. However, more recent studies along California's coastline indicate the possibility of significantly higher sea levels by 2100, with levels at specific future dates highly dependent on future global greenhouse gas emissions [22]. Future sea-level rise will increase the frequency at which waves will attack the base of coastal cliffs and bluffs [23–28], and as a result, coastal erosion will almost certainly be accelerated during the 21st century [23,27,29]. In addition, changes in regional meteorological and climate patterns, including the frequency and intensity of El Niño events, coupled with rising sea level, are predicted to result in increasing extremes in sea level [30] and wave power [31]. Waves riding on these higher water levels will cause increased coastal erosion and shoreline damage, more than that expected from sea-level rise alone [30]. Many major coastal cities were developed in areas vulnerable to shoreline and/or coastal erosion [32]. With coastal populations and associated economic assets continuing to increase [33], cliff, bluff top, and shoreline development will be increasingly threatened by erosion and retreat [34]. This has led to an increased need for accurate information on rates and trends of coastal recession [35] in order to respond and adapt to expected future shoreline changes.

In this study, we focused on California's coast, which is experiencing well-documented sea-level rise [22,36] and related coastal impacts including coastal erosion [10,30,37]. California's coast reflects a complex geological history and the interplay of tectonic or mountain building processes, geology, climate, and the sea, and has always been identified with change [1,38]. At the close of the last ice age 18,000 years ago, the coastline stood several to as far as 50 km offshore to the west [4]. As the climate warmed, seawater expanded and ice melted. In response, sea level rose about 130 m and advanced inland, moving the cliffs, bluffs, and beaches eastward. About 8000 years ago, the rate of sea-level rise slowed from an average of about 11mm/year over the previous 10,000 years to less than a millimeter per year. Over the past century or so, however, due primarily to anthropogenic global warming, the global rate of sea-level rise has accelerated to about 3.4 mm/year (13.4 inches/century) leading to an increase in rates of shoreline and coastline retreat.

The great majority (72% or about 1272 km) of California's 1760 km coastline consists of actively eroding sea cliffs and bluffs [4], and of the 1272 km cliffed coast—including the 5.8 km (3.6 mile) long section of Santa Cruz coast covered in this paper—about 1040 km consists of low to moderate relief cliffs and bluffs ranging in height from about 10 to 100 m, which are typically eroded into uplifted marine terraces (Figure 2). Some cliff rocks are so hard and resistant, however, that photographs taken of the coast 75 years ago look identical to those of today. Elsewhere, however, coastal bluff materials are so soft and weak that the coast is being eroded at average rates of 2 m or more each year. These changes are easily recognized when comparing historic ground photographs.

**Figure 2.** Typical morphology of much of California coast with cliffs eroded into an uplifted marine terrace (photo by Gary Griggs, 2006).

The coast of California is dominated by uplifted marine terraces fronted by low cliffs, but also includes steep coastal mountains and areas of coastal lowlands, estuaries, and dunes [30]. The two sea-level rise related hazards of greatest concern to any oceanfront development along the California coast, whether public or private, are (1) coastal cliff and bluff erosion (Figure 3a) and (2) more frequent flooding of low-lying areas by storm waves and high tides (Figure 3b), followed in time by permanent inundation [39]. California's coastline is approaching a crisis point, which has resulted from a combination of natural processes and cycles, combined with human intervention and population growth [10]. California's population and economic centers are concentrated along its coast [40]. Although California's 19 coastal counties (including San Francisco Bay) make up only 22% of the state's land mass, they account for 68% of its population [41], 80% of its wages, and 80% of its GDP [42]. In addition, California's coastal population is expected to continue to grow significantly over the coming decades [43], which will only compound the erosion and flooding problems at the

edge. A recent study [27], showed that for California, the world's 5th largest economy, over \$150 billion in property, equating to more than 6% of the state's GDP and 600,000 people, could be impacted by coastal flooding by 2100.

**Figure 3.** (**a**) The erosion of a coastal cliff has removed an ocean front street about 10 km from the project site. (**b**) Flooding of a coastal parking lot during a period of elevated sea level and large waves approximately 20 km from the study area.

#### **2. Study Area**

The West Cliff Drive coastline, which has been selected as a case study in this paper, extends 5.8 km along the western edge of the city of Santa Cruz between Point Santa Cruz and Natural Bridges State Beach on the central California coast (Figure 4). This section of the coastline is somewhat unique in California in having a public street (West Cliff Drive) extending the entire 5.8 km (3.6 mile) length, along with a pedestrian/bicycle path. This allows unobstructed views of this dramatic coast without the presence of homes or other development on the top of bluff, which is more typical of many of the state's coastal communities. As a result, this road has been a popular area for residents and visitors alike for well over a century.

**Figure 4.** West Cliff Drive study area.

#### *2.1. Geologic Setting*

The striking features of California's diverse landscape, the San Andreas Fault (which lies just 25 km east of the project area) and its associated earthquakes, the rugged coastal mountains, and the uplifted marine terraces [36] and coastal cliffs that characterize much of the coastline, all have their origins in millions of years of large-scale tectonic processes that continue today [4]. The rocks exposed along the coastline and in the sea cliffs provide evidence of this complex geological history and the changes the landscape has undergone. Coastal cliffs along the state's coast may consist of granitic, volcanic, metamorphic, or sedimentary rocks.

The West Cliff Drive coastline consists of near vertical cliffs varying in height from about 6 to 12 m (20 to 40 feet), which form the outer edge of the lowest uplifted marine terrace along this coast (Figure 5a). The lower bedrock portion of the seacliff consists of two different geologic units: The older, harder, and more resistant Santa Cruz Mudstone of Miocene age (~5–7 million years old; Figure 5a) and the younger and weaker Purisima Formation of Pliocene age (~3–5 million years old) (Figure 6). The mudstone extends approximately 2100 m along the western section of the coastal area studied, while the overlying Purisima Formation makes up the approximately 3700 m long eastern section. The Purisima consists of interbedded mudstones, siltstones, and sandstones that are pervasively jointed. It is the orientation of the joints sets that exerts a major control on the erosion of the bedrock and the shape or morphology of the coastline (Figure 5b). The uppermost 2 to 4 m of the cliffs consist of much younger (~100,000 year old) poorly consolidated, marine and non-marine terrace deposits, primarily sand, gravel, and cobbles (Figure 6). The same processes that formed the coastal landscape continue to act on it today, although at a nearly imperceptible rate. More noticeable, however, is the rate of coastline erosion, as waves attack and undercut the emergent land and force the nearly vertical bluffs to recede inland.

**Figure 5.** (**a**) Unprotected section of West Cliff Drive study area showing low eroding cliffs eroded into the Santa Cruz Mudstone [15], (**b**) oriented embayments eroded along parallel joints in the Santa Cruz Mudstone (Google Earth, 2018).

**Figure 6.** This section of coastal bluff consists of three different geologic units: The Santa Cruz Mudstone, the Purisima Formation, and the overlying unconsolidated terrace deposits, which erode differentially and make selecting a bluff edge subjective.

#### *2.2. Oceanographic Conditions*

The central California coast experiences a mixed semi-diurnal tide with a maximum range of about 2.5 m (8.2 feet), ranging from +2.0 to −0.50 m (+6.6 to −1.6 feet). During years with strong El Niño events, however, water levels may be elevated as much as 30 cm above predicted water levels for days [30,44], which brings large waves closer to the cliffs, and when coincident with high tides, often produce failure of the bedrock, as well as erosion of the overlying and much more erodible terrace deposits. Current wave conditions along West Cliff Drive were defined using historical data from the Global Ocean Waves database [45] that covers the time period between 1948 and 2008. The offshore wave data was propagated to the shore using the SWAN wave propagation model based on the models developed for California with nearshore bathymetry information [46]. The wave propagation results were used to reconstruct hourly time series of wave parameters (significant wave heights, Hs; and mean periods), as described in Camus et al. [47], and calculate hourly wave energy at the 10 m depth contour along West Cliff Drive. Prevailing winter storm waves approach this stretch of coastline dominantly from the northwest and west and undergo little loss of energy through refraction as they approach the coastline along West Cliff Drive. Significant wave heights of 1 to 2 m occur frequently in the winter months (December through March). During major storms, however, wave heights may reach 4 m or more. This is a high-energy coastline with occasional severe wave attack (Figure 7).

**Figure 7.** Storm waves at high tide overtopping West Cliff Drive bluff (photos by Gary Griggs).

During the periods of largest waves and highest tides, waves are attacking essentially all of the cliffs along this entire section of coast. About 600 m of the 5800 m long stretch of sea cliffs is buffered by small pocket beaches, which come and go seasonally. These vary in length from about 30 to 300 m with maximum widths of 25–50 m in the summer months. The beaches undergo strong seasonal fluctuations in size in response to changing wave conditions, with sand levels dropping 2 m or more from summer to winter months (Figure 8a,b).

**Figure 8.** (**a**) Summer and (**b**) winter beach sand levels along West Cliff Drive (photos by Gary Griggs).

The mean and 25% and 75% percentiles of annual wave energy, and corresponding directions were then calculated using the hourly time series. There is significant variation in a high percentile of wave heights (95% percentile of significant wave height, Hs95) and wave energy intensity and direction (Figure 9). Annual mean wave energy decreases from west to east and rotates slightly anticlockwise (Figure 9a). Changes in wave energy (Figure 9b) are more marked than the differences in high-values of wave heights (Figure 9c). This is because annual wave energy not only represents conditions of a single storm, but, in addition, high wave conditions accrue proportionally more energy than calmer sea states [31]. Therefore, wave power not only shows a significant spatial variation across West Cliff as a result of wave propagation, but may also serve as an indicator of erosion potential of wave action on the cliffs over cumulative periods of time.

**Figure 9.** Spatial variation of wave climate along West Cliff. (**a**) Locations of data points with time series of wave parameters and mean direction and intensity of annual Wave Power. (**b**) Eastward variation of cumulative annual wave power. (**c**) Eastward variation of the 95% percentile of significant wave heights (Hs). The boxes represent the range between the 20% to 75% values, where the mean is indicated in red. The crosses represent values exceeding that range.

#### **3. Materials and Methods**

In this research we used both qualitative and quantitative assessments to analyze coastal cliff and bluff retreat along the West Cliff Drive coast as a case study to identify the challenges of determining historical erosion rates.

#### *3.1. Qualitative Coastline Change Assessment*

We conducted a qualitative assessment by reviewing literature and relevant documents on California coastline erosion, in addition to investigating the geological history, cliff geomorphic differences, sea-level rise, and related impacts on coastal erosion and history of coastline changes throughout the study area.

#### *3.2. Quantitative Coastline Change Analysis*

#### • Materials

We selected two historical sets of aerial photograph from 1953 and 1975 (scale = 1:10,000), and satellite imagery (Google Earth) from 2018 (Figure 10). This gave us a significant span of time to achieve useful results for extracting coastlines and comparing them to detect coastline change, identifying cliff and bluff retreat, as well as measuring the erosion rates along this coast. We also used a hill-shaded digital surface model (USGS, 2018), as well as historical ground and modern photos of the area to digitize and adjust the selected reference line in each segment as accurately as possible. GIS tools were used to perform coastline change detection.

**Figure 10.** Aerial photographs (**a**) 1953, (**b**) 1975, and (**c**) satellite imagery (2018).

• Methods

Cliff and bluff retreat were analyzed through three main stages:


**Figure 11.** The various profiles and related reference lines (coastline) along cliffed coasts.

**Figure 12.** The defined reference lines along a cliffed coast. (**a**,**b**) profile view (**c**) vertical view. (In this segment of coast, the detectable reference line on both aerial photographs and satellite imagery was "the base of bluff").

**Figure 13.** Map of the 31 identified coastal segments, armored coastline, and unprotected areas along the West Cliff Drive coast—(**a**) West side (**b**) East side.

#### **4. Results**

#### *4.1. Qualitative Coastline Change Assessment*

• Geology

Erosion along the West Cliff Drive coastline typically occurs through a combination of wave impact, weathering and abrasion of the bedrock, rainfall and terrestrial runoff to a lesser degree, and also relatively infrequent seismic shaking during large earthquakes. Bedrock erosion along weaker stratigraphic layers or joint sets leads to focused erosion and the frequent formation of undercuts, arches, caves, and embayments that made this area an early attraction for residents and visitors. The Santa Cruz Mudstone is more resistant to wave attack than the Purisima Formation and retreats at slower rates overall. As described above, it is primarily the well-developed joint in the latter formation that focuses erosion and typically leads to the failure of large joint-bounded blocks or the collapse of arches and caves. The unconsolidated sandy terrace deposits are much less resistant to wave attack, and it is during periods of large storm waves coincident with very high tides that waves overtop the bedrock and attack and erode the weaker terrace deposits. It is this erosion that has historically threatened and undermined sidewalks, West Cliff Drive, and also a historic lighthouse. Coastal erosion

or cliff and bluff retreat in the study area, as along most of the California coast, is an episodic process with most of the major cliff failures occurring during the simultaneous arrival of large storm waves and elevated sea levels.

• Coastal Protection and Erosion

Efforts to stabilize or protect this stretch of shoreline from wave erosion began in 1926 and have continued intermittently to the present. Concrete retaining walls along the upper bluff and rip-rap revetments at the base of the cliff have been the dominant type of armor emplaced, although broken slabs of old streets and sidewalks and stacked bags of concrete were also used in the early years in attempts to halt or slow cliff retreat. Today of the 5800 m of coastline studied, about 2600 m or 44.8% has been armored (Figure 13), with 91% of this armor consisting of rock revetments (Figure 14). While this has served to reduce and halt erosion, it has completely changed the natural condition and appearance of this stretch of coast. These large revetments have also covered large areas of sandy beach that are now removed from public use. The local government agency (City of Santa Cruz Department of Public Works) received emergency permits to install much of the rip-rap, but some of these permits were never finalized with the permitting agency (California Coastal Commission). The Coastal Commission has recently required that the city evaluate the coastal erosion and protection issues, public use, and economics of this 5800 m stretch of coastline and develop a long-term management plan for the future.

**Figure 14.** The placement of rip-rap over a 60 year period has reduced or eliminated bluff retreat along about 50% of West Cliff Drive, thus complicating any measurements of natural rates of bluff retreat.

• Coastal Change from Historical Ground Photographs

As soon as cameras became widely available, residents, visitors, and commercial photographers began to take pictures of the Santa Cruz coastline. The earliest dated photographs we have discovered of this coast were taken 143 years ago (1876). Certain areas, the picturesque arches, sea stacks, and distinct rock formations along West Cliff Drive, for example, were photographed frequently and memorialized in hand-colored postcards and family albums. Over the subsequent years, as winter storms have periodically battered the bluffs and beaches, and sea level has gradually risen, the coastline has slowly retreated. Some areas have changed dramatically (Figures 15a–c and 16a,b), and others have changed surprisingly little. The natural bridges, arches, and sea stacks that owe their origins to wave attack of the weaker sandstones and mudstones have been destroyed by the same forces that created them, with many fascinating and revealing photographs taken of these natural and unnatural features along the way. While it is very difficult to get any quantitative measurements of cliff or bluff retreat from old ground photographs, they do provide a clear record of the extent of change or retreat that has taken place since the time the original photograph was taken. In many cases, and for most people, a then and now set of photographs provides a more understandable record of coastal change than a rate of retreat in cm/year [48].

**Figure 15.** (**a**–**c**) Progressive erosion of an arch in the study area over a period of 50 years to ultimately form a sea-stack. Dates of photos: (**a**) Before 1888, (**b**) ~1890, (**c**) 2005.

**Figure 16.** (**a**–**c**) Promontory in foreground and Bird Rock arch (in background) in 1909, 2005, and 2019.

#### *4.2. Quantitative Coastline Change Assessment*

• Documenting Coastal Erosion from Vertical Aerial Photographs and Maps

The first aerial photographs were taken of the Santa Cruz coast in 1928 and were in stereo. This is quite amazing as Charles Lindberg had just made the first solo flight across the Atlantic Ocean the year before. These photographs were taken in order to study the route of a potential highway between San Francisco and Santa Cruz and aerial photographs were the easiest way to accomplish that. These images provide us with a photographic record extending back 90 years and can be used to determine qualitative changes; because of the only moderate resolution and lack of features from which to take measurements from, they are of limited value in quantitative assessment of historic cliff erosion. Vertical stereo photos were then taken in subsequent years along the Santa Cruz coast, which became quite regular beginning in the 1940s and extending to the present. Aerial photographs were taken more often in later years as various state and federal agencies became interested in documenting the landscape including forest cover, agriculture land use, coastal conditions, and development, highway and railway routes, among other purposes. Until relatively recently, historical aerial photographs were the most common sources for documenting or measuring rates of coastal change such as coastal bluff and cliff retreat. This required first determining the scale of the photograph, and then finding locations where the position of the cliff edge could be measured from some fixed feature (a road, building, etc.) over time. There are multiple challenges involved in this approach, including: a. The scale and resolution of the aerial photographs; b. the sharpness or ease of recognition of the cliff edge; c. the presence of vegetation obscuring the cliff edge making measurements difficult or unreliable; d. the lack of a reference point on older photographs to measure from; and e. the time period covered by the photographs.

The older aerial photographs are usually not of as high resolution as recent photos and useful reference features present in more recent images may well not have existed at the time the older photographs were taken. Even though we have aerial photographs that extend back 90 years for the Santa Cruz coast, using conventional methods (such as an optical comparator or loupe, for example) for measurement is hindered by photograph resolution and appropriate features from which to take repeated measurements.

• Documenting Coastal Erosion from Satellite Imagery and DEMs

In recent decades, the availability of satellite imagery (Google Earth, for example) and Lidar (Light Detection and Ranging) derived DEMs also provided high-resolution data sets for documenting coastal cliff or bluff erosion. With the inclusion of a series of historical satellite photographs in Google Earth, adjusted to precisely the same scale on the website, typically extending back into the early 1990s, or in some cases back to the 1980s, and a built-in measuring tool, a user can determine the distance from some landmark or feature to the cliff edge relatively easy on multiple images all of the same scale. The same issues that can affect the reliability of cliff erosion measurements from historic hard copy aerial photographs still exist, however, a landmark or feature that can be recognized on all images, the resolution and scale of the images, and the clarity or ease of recognizing the cliff edge. In addition, and this is an issue along the section of Santa Cruz coast covered in this study, there are many coastal bluffs and cliffs where the feature designated as the bluff or cliff edge is somewhat subjective because of the varying geomorphology, which is related to the differences in geologic materials (Figures 6 and 17). This makes determining erosion rates difficult.

**Figure 17.** A section of coastal bluff consisting of the overlying weaker terrace deposits that are mostly vegetated, and the underlying Santa Cruz Mudstone that is being undercut. As in Figure 6, selecting the edge of the bluff for comparative measurements is difficult and somewhat subjective.

In addition, applying the DEMs, as well as topographic maps to extract the indicator lines such as the cliff edge are other approaches to conduct coastal change analysis. The lack of high-resolution elevation data, and the differences in resolution, and scale of available elevation data, are the main constraints to conducting a time-series analysis over a relatively long-period study and would increase the uncertainties of cliff retreat rate retreat estimations.

Further complicating the measurement of changes in cliff or bluff edges is the armoring of this coast, which has gone on for about 60 years (Figures 13 and 14), and depending upon when the rip-rap was placed, this essentially brings erosion to a near halt for a number of years. As of 2019, approximately 45% of the entire West Cliff Drive area had been armored.

#### • Documenting Coastal Erosion along West Cliff Drive as a Case Study

In addition to investigating the evolution of coastline change over time, we used both aerial photographs and satellite imagery, as well as the hill-shaded DSM (USGS, 2018) to assess coastline retreat along West Cliff Drive coast. The coastline was divided into 31 individual segments that were evaluated independently (Figure 18). The results (Table 1 and Figures 19 and 20) showed the maximum coastline retreat over the studied time span occurred in coastal segment number 16 (Figure 21), where the retreat rate over 65 years ranged from 0.3 to 32 m (Figure 19), and the maximum retreat rate was 0.5 m/year. (Figure 20).

**Figure 18.** Two examples of digitized coastlines in segments 16 and 19.



**Figure 19.** Maximum coastline retreat along the 31 coastal segments (zero value indicates that the change of the segment was undetectable).

*Water* **2019**, *11*, 2648

**Figure 20.** Maximum coastline retreat rate along the 31 coastal segments (zero value shows the coastline of the segment was undetectable).

**Figure 21.** Oblique aerial photographs from coastal segment number 16 (**a**) 1972 [15], (**b**) 2018 [USGS,2018].

#### • A Review of Cliffed Coast Retreat Studies

A number of coastal researchers have endeavored to document historical coastal cliff retreat and project future retreat along the coast of California including the Santa Cruz coastline over the years from aerial photographs, satellite imagery, and DEM, with all of the inherent challenges involved. Hapke and Reid [37] completed the most comprehensive assessment. They evaluated cliff retreat using map and photographic data for more than 350 km of the California coast over a period of approximately 70 years, as part of the US Geological Survey's Assessment of Coastal Change Program. They compared one historical cliff edge digitized from old maps dating from 1920–1930, with a recent cliff edge interpreted from LIDAR topographic surveys from either 1998 or 2002. Long-term (~70 year) rates of the retreat were calculated using differences in the locations of the two different cliff edges. The average rate of coastal cliff retreat over this time period for the sections of California coast studied was 0.3 +/− 0.2 m/year. The average amount of total cliff retreat over the 70 year period was 17.7 m. Due to the regional scale of the area studied, however, the shoreline projections were not always accurate, which affected the erosion rate determinations in specific areas. The book "Living with the Changing California Coast" [1], includes cliff and bluff erosion rates where they were published or available for a number of locations along the state's coast. Moore et al. [49] utilized aerial photos corrected through softcopy photogrammetry for a detailed study of cliff erosion rates for both Santa Cruz and San Diego counties as part of a national FEMA (Federal Emergency Management Agency) assessment of coastal erosion hazards. Unfortunately, that study did not include the West Cliff Drive area. Griggs and Johnson [50] reported on cliff erosion rates along the Santa Cruz County coastline, including a few measurements along West Cliff. Their rates were based on comparative measurements from aerial photographs taken in 1940 and 1960, but were limited by the photograph issues discussed above. Young et al. [51] detected 30 individual cliff edge failures and maximum landward retreats from 0.8 to 10 m along the 7.1 km of unprotected coastal cliffs near Point Loma in San Diego over a 5.5-year period (2003–2009). In a recently published study, decadal-scale coastal cliff retreat in southern and central California [52], cliff erosion was detected along 44% of the 595 km of shoreline evaluated, while the remaining cliffs were relatively stable.

Revell et al. [53] evaluated potential future erosion hazards along the coast of California by 2100 under a 1.4 m sea-level rise scenario. For cliff-backed shorelines future potential erosion is projected to average 33 m, with a maximum potential erosion distance of up to 400 m. Young et al. [54] studied cliff and shoreline retreat considering sea-level rise in southern California, and based on their model's results, mean and maximum scenario cliff retreat over 100 years ranged from 4–87 and 21–179 m, respectively. Barnard et al. in a study on coastal vulnerability [16], demonstrated that El Niño events result in wave directional shifts, elevated wave energy, and severe coastal erosion for the Central Pacific and California. Limber et al. [29] applied a multimodel ensemble to project time-averaged sea cliff position of the 475 km long coastline of Southern California over multidecadal time scales and large (> 50 km) spatial scales. Results showed that future retreat rates could increase relative to mean historical rates by more than twofold for the higher SLR scenarios, causing an average total land loss of 19–41 m by 2100.

#### **5. Discussion**

One of the most important types of information needed prior to initiating or approving coastal human and natural communities' protection plans, as well as any development along the coast, whether private or public, are the long-term rate at which the cliffs or bluffs are eroding. The longer the period of record covered by the aerial photographs or other data sources, the more representative will be the erosion rate calculated. Additionally, in recent decades, the challenges of a continuing rise in sea level at an accelerated rate, and changes in wave climate, which will affect the long-term cliff erosion or retreat rates, have increased the demand for historical erosion rate data that can be used to project future cliff and bluff positions in highly developed areas. Knowing where the edge of the coastal bluff or cliff is likely to be over time is important for future planning. However, most coastal change studies conclude that there are always different levels of uncertainty in erosion rate measurements and coastline retreat projection results. In this study, some of the identified challenges that could increase the result uncertainties were:


**Figure 22.** A hill-shaded Digital Surface Model—DSM (USGS-2018) shows the complex profile of cliffed coasts, as well as various lines which could be used in cliff and bluff retreat analysis.

Appropriately addressing the identified challenges in coastal cliff erosion studies (both qualitative and quantitative assessments) could reduce the uncertainty of the historical erosion rate measurements (cliff or bluff retreat) and, as a result, would improve the future bluff or cliff retreat projections. Erosion models include substantial uncertainty, not only derived from the definition of future sea levels and waves but also from the estimates of historical coastline retreat rates. Cliff erosion brings even more complexity and uncertainty given the interaction of the coastal geology with sea levels and waves, which produce different coastal sections with collapsing and eroding modes. Therefore, coastline change models should be contrasted with historical rates from remote sensing and historical imagery as a ground truth and expected erosion potential. While this approach may not provide quantitative definition of the future coastline, it is adequate to identify the historical impacts, delineate erosion

hotspots, and establish priorities for management, today and in the foreseeable future, both from regular oceanographic conditions and episodic cycles such as El Niño.

**Figure 23.** Seacave (**a**) and undercuts (**b**) along West Cliff Drive.

#### **6. Conclusions**

As sea level continues to rise at an accelerated rate, the intensive development and infrastructure along California's coastline is under an increasing threat. Whether construction on coastal bluffs or cliffs, or along low-lying shoreline areas, higher sea levels combined with storm waves and high tides will lead to increased rates of cliff and bluff retreat and more frequent coastal flooding. Planning and adapting to a new but uncertain coastline position is and will continue to be a major challenge for many coastal communities. However, there are challenges and uncertainties in both accurately documenting historic cliff and bluff retreat and in projecting these values into the future. There are significant obstacles for developing and implementing future sea-level rise adaptation strategies, managed retreat for example, along the coast, in particular cliffed coastal regions. Reducing or avoiding the problems and concerns identified in this study in determining erosion rates as much as possible through using the most reliable data sets and applying appropriate approaches is an essential process in developing a roadmap for the future management of the area.

Coastal cliff retreat is the product of a complex interaction between the (1) intrinsic properties of the cliff or bluff materials (lithology or rock type, internal rock weaknesses such as joint patterns, and stratigraphic variations, for example) that combine to resist erosion, and (2) the extrinsic processes (rock weathering, rainfall, wave energy, tidal range, storm frequency and intensity, and sea-level rise, for example) that work to weaken the cliff materials and produce failure or erosion. While the historic coastline data from ground and aerial photographs, maps, and satellite imagery can be used with caution and experience to provide the most accurate measurements possible of past changes, with the increase in sea-level rise rates and other aspects of climate change, conditions are shifting. Implementing a detailed coastal monitoring program to document and track the present location and condition of cliff and bluff edges, delineating armored and unarmored sections of coastline, documenting rates of sea level change, and identifying erosion hazard zones will, over time, provide a more robust foundation for future decision making.

**Author Contributions:** G.G., designed research; G.G., L.D., and B.G.R., performed research; G.G. and L.D., field survey and data collection; G.G., L.D., B.G.R., writing, review, and editing.

**Funding:** This research received no external funding.

**Acknowledgments:** We would like to express our appreciation to Patrick Barnard for providing the scientific advice, as well as logistic support, and Alexander Snyder for his help with our field survey to collect the GCPs coordinate. Daniel Hoover for facilitating the field survey, Mark Fox and Jeff Nolan for their valuable contribution to a brainstorming exercise on indicator coastlines.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Influence of a Reef Flat on Beach Profiles Along the Atlantic Coast of Morocco**

#### **Mohammed Taaouati 1, Pietro Parisi 2, Giuseppe Passoni 3, Patricia Lopez-Garcia 2, Jeanette Romero-Cozar 2, Giorgio Anfuso 2, Juan Vidal <sup>2</sup> and Juan J. Muñoz-Perez 2,\***


Received: 7 February 2020; Accepted: 11 March 2020; Published: 12 March 2020

**Abstract:** The North Atlantic coast of Morocco is characterised by a flat rocky outcrop in the south (Asilah Beach) and a sandy beach free of rocky outcrops in the north (Charf el-Akab). These natural beaches were monitored for a period of two years (April 2005–January 2007) and two different profiles (one for each beach) were analysed based on differences in the substrate. Topographic data were analysed using statistics and empirical orthogonal functions (EOFs) to determine beach slope and volumetric changes over time. Several morphologic phenomena were identified (accretion/erosion and seasonal tilting of beach profiles around different hinge points), attesting to their importance in explaining variability in the data. Periods of accretion were similar in both profiles, but the volumetric rate of change was faster in the sand-rich (SR) profile than in the reef flat (RF) profile. Moreover, the erosion rate for the SR profile was greater than the RF profile (135.18 m3/year vs. 55.39 m3/year). Therefore, the RF acted as a geological control on the evolution of its profile because of wave energy attenuation. Thus, special attention should be given to the RF profile, which has larger slopes, less amounts of mobilised sand, and slower erosion/accretion rates than the SR profile.

**Keywords:** EOF; beach profiles; reef flat; coastal dynamics; sand rich; accretion; erosion rate

#### **1. Introduction**

The presence of rocky platforms on beaches is found worldwide. An RF beach is the name for beaches that are perched on hard landforms. The US Army Corps of Engineers [1] and Larson and Kraus [2] define this as a hard-bottom beach. Morphological changes on beaches due to the existence of an RF are not well studied. A few investigations have focused on shape changes, such as Black and Andrews [3] in New Zealand and New South Wales, and Sanderson and Eliot [4] in Australia. Other authors have studied temporal changes, reporting winter erosion and summer accretion rates over a limestone platform near Perth in south-west Australia [5]. Rock and coral landforms on beaches can dissipate wave energy, as confirmed by researchers in Galicia (north-west Spain [6]), St. Martin's Island (Bangladesh [7]), and the fringing reef along Kaanapali Beach in Maui [8].

Sea level rise and other anthropic phenomena induce coastal recession worldwide [9], and coastal researchers and engineers are interested in studying coastal evolution to properly design mitigation and/or remediation measures [10]. Short-term and long-term morphological variability must be considered in the design and evaluation of beach nourishments [11]. Evaluation over seasonal time scales (months or years) is important to determine the rate of erosion and therefore determine

future land use in areas adjacent to beaches. Medium-term responses, such as seasonal oscillations in winter–summer profiles, provide information about the across-shore dimension of the berm and may play an important role in the location of beach services such as showers, litter bins, toilets, as well as ramps and bridges for wheelchair accessibility [12].

Levelling of beach profiles is a widely-used tool to monitor the evolution of the coast, and various formulae have been proposed to calculate a general expression (e.g., Dean's formula [13]), although some authors have questioned their validity when an underlying shoreface geology exists [14]. Thus, several researchers have presented results of the influence of coastal reefs on the spatial and temporal variability of beach morphology [15–19]. Other characteristics, such as wave attenuation over reef platforms [20], wave-setup and water-level fluctuations [21], interannual changes in beach morphology [22], modification of the A parameter of Dean's formula [23,24] or sediment flux [25] along reef-protected profiles, have also been discussed.

Nevertheless, few comparisons can be found between the behaviour of profiles on adjacent beaches subject to the same wave conditions but with different geological substrates or boundary conditions. It is worth noting that Muñoz-Perez and Medina [12] compared the behaviour of two beach profiles from Victoria Beach (Cadiz, Spain) over a five-year period where one profile was perched on a rock platform. The northernmost zone presented a rocky platform that emerged during low tide and acted as a geological boundary for profile development, whereas the southern zone had no such platform. Some differences in erosion and subsequent accretion rates were observed.

Thus, the aim of this paper is to compare how beach profiles change (volume and slope) over time on two adjacent beaches (under the same climatic conditions), one of which is a sand-rich beach (Charf el-Akab, SR) and the other beach is supported by a reef flat (Asilah, RF). Monitoring by beach profiling was performed to analyse their morphological differences over a period of two years (April 2005–January 2007) to observe seasonal changes between summer and winter in order to draw useful conclusions regarding the behaviour of beach morphology as it relates to differences in the seabed.

#### **2. Study Area**

#### *Beach Location*

The sites investigated in this paper are located along the North Atlantic coast of Morocco (Figure 1). Two adjacent beaches were chosen. The northern beach is Charf el-Akab (35◦46' N, 5◦48' W), close to Tanger. Immediately to the south is Asilah Beach, which takes its name from the homonymous city (35◦28' N, 6◦2' W). Both beaches have an NNE–SSW orientation and are composed of the same quartz sand. The main difference between the two sites is that Asilah presents an almost horizontal rocky platform situated around the low tide level, which influences the dynamics of the coast, while Charf el-Akab is a completely sandy beach.

(c)

**Figure 1.** (**a**) Location of Charf el-Akab and Asilah beaches on the north-west coast of Morocco facing the Atlantic Ocean. Wave and climate data were collected from a virtual SIMAR buoy in front of Asilah (www.puertos.es); (**b**) view of Charf el-Akab sandy beach; (**c**) view of the Asilah beach supported by a reef flat. Photographs taken by the authors.

#### **3. Methods**

#### *3.1. Meteorological Data*

Meteorological data were collected from a wave prediction point (SIMAR point 5,041,003 from www.puertos.es) located in front of the monitored stretch of coast (Figure 1a). The area is mesotidal, with a tidal range of 2.7 m and a semidiurnal periodicity [26]. The hydrodynamic conditions are principally controlled by storms approaching from occidental quadrants [27]. The predominant winds, named "Chergui", blow from the east (i.e., from land) 27% of the time and are especially abundant in spring and summer, reaching maximum velocities of 130 km/h. Secondary winds ("Rharbi") blow from the west (i.e., from the Atlantic Ocean) 16% of the time. Rharbi winds are wet and prevail in winter and autumn [26].

The SIMAR database is obtained through numerical wave modelling from wind time series by solving the equation of energy balance. This virtual database does not come from direct measurements. However, it has been validated by numerous studies and used in practical applications along the Spanish coast [28].

Both beaches are subject to the same wave and wind climatic regimes because of their proximity to each other. The nearshore areas are uniform, and both beaches face the same direction. Figure 2 reports temporal series of wave height and period, wind speed, and wind direction from April 2005 to January 2007.

**Figure 2.** Temporal series of (**a**) wave height (m); (**b**) wave period (s); (**c**) wind speed (m/s); (**d**) most frequent wind direction: N = 0◦, E = 90◦, S = 180◦, W = 270◦ (adapted from www.puertos.es). The dotted blue line refers to the average value, while the red one is the maximum value registered.

#### *3.2. Field Data Surveying*

The morphological changes of Charf el-Akab and Asilah beaches were studied through a topographic monitoring program carried out every two months during a two-year period, from April 2005 to January 2007. Data were collected on emerged beaches at low tide with a total station. The vertical datum (or zero elevation surface) matches the lowest low-water level (LLWL). Some fixed positions were selected and monitored at both beaches. Five profiles were taken in Charf el-Akab (Figure 1b) and seven in Asilah (Figure 1c). The beach profile spacing was 50 m, following

recommendations found in the literature [29–32], whereas the distance between adjacent points along one profile was 5 m. However, because there were no appreciable differences between the profiles of the same beach (the variance ranges from 0.015 for Asilah to 0.061 for Charf el-Akab), only two mean profiles (one from each beach) were studied. Afterwards, analyses of the data collected were performed by statistical way (Statgraphics Centurion software) and EOF (multivariate statistical analysis package (MVSP)).

#### *3.3. Statistics*

Following Jimenez and Sanchez-Arcilla [33], a previous study by Anfuso et al. [34] chose to use least-squares linear regression to analyse the evolution of the profile. Similar methodologies to the one used by Anfuso et al. [34] were carried out to obtain the accretion/erosion volumes in this study. The analysis of the mean profiles was carried out using the statistical software to find differences between the behaviour of the RF and the SR profiles. The beach face slope and accretion/erosion volumes of sand per unit of beach length were calculated. The area between two profiles of adjacent dates is the accretion/erosion rate volume (m3/m). The slope was obtained as the mean of slopes calculated at 5 m intervals along each profile. The use of EOFs enabled the identification of morphological changes [35] and allowed us to obtain additional results from the data that better explained the spatial and temporal variability of the beach profiles.

#### *3.4. Empirical Orthogonal Functions (EOFs)*

EOFs are a mathematical method and have been widely used in coastal geomorphology since Winant et al. [36] studied variability in beach profiles. Other researchers have applied this technique to different aspects of coastal morphology; for example, longitudinal variations in contour lines [37–39], sand transport in a transverse direction [40], or the distribution of sediment grain size along a transverse profile [41]. In addition, other phenomena have also been investigated using EOFs: responses to beach nourishment at different times and spatial scales [11], behavioural changes in profiles over a fortnightly tidal cycle [42], the capacity of this technique to identify modes of shoreline variability [38,39], and changes in coastal dune profiles [43].

EOFs, also known as principal components analysis (PCA), provide a technique for separating the spatial and temporal variability of beach-profile data; a detailed description of the method can be found in statistics textbooks [44]. In brief, if a function h = (x,t) represents the profile elevation at a particular position and time, such a function may then be defined as a linear combination of a few spatial, Xn (x) and temporal, Tn (t), eigenfunctions (and their associated eigenvalues) as follows:

$$\mathbf{h}\_{\overline{\mathbf{i}}\} = \mathbf{h}\begin{pmatrix} \mathbf{x}\_{i} \ \mathbf{t}\_{\rangle} \end{pmatrix} = \sum\_{\mathbf{l}=1}^{N} \mathbf{X}\_{\mathbf{l}}(\mathbf{x}\_{\mathbf{i}}) \ast \mathbf{T}\_{\mathbf{l}}\begin{pmatrix} \mathbf{t}\_{\mathbf{i}} \end{pmatrix} \ast \mathbf{a}\_{\mathbf{l}} = \mathbf{a}\_{1} \ast \mathbf{X}\_{1}(\mathbf{x}\_{\mathbf{i}}) \ast \mathbf{T}\_{1}\begin{pmatrix} \mathbf{t}\_{\mathbf{i}} \end{pmatrix} + \mathbf{a}\_{2} \ast \mathbf{X}\_{2}(\mathbf{x}\_{\mathbf{i}}) \ast \mathbf{T}\_{2}\begin{pmatrix} \mathbf{t}\_{\mathbf{i}} \end{pmatrix} + \dots \tag{1}$$

Eigenfunctions are ranked according to the percentage of variability they explain, defined as the mean squared value (MSV) of the data. In some cases [36], the mean value is of such importance in explaining the variability that it must be removed from the original data to allow for the better and clearer identification of other smaller, but important, changes. Then, the MSV becomes part of the variance. The first eigenfunction explains most of the MSV in the data, the second eigenfunction explains the greater part of the remaining MSV, and so on. The MVSP software was used to calculate the EOFs. Furthermore, according to Aubrey [45], assuming that a physical process provides most of the variability, the corresponding eigenfunction would be related to that physical process.

#### **4. Results and Discussion**

#### *4.1. Topographic Profile Analysis*

Topographic mean profiles carried out from April 2005 to January 2007 at the RF beach at Asilah and the SR beach at Charf el-Akab were investigated in order to assess how the profiles change over time and to compare the two types of beaches. A representation of the mean profiles over time is shown in Figure 3. Since it is not easy to see a rational behaviour or trend, as previously mentioned in Section 2, a statistical analysis was carried out with Statgraphics software to obtain the results presented in Table 1 (i.e., net sand volume variations and rates of accretion represented by positive values and erosion by negative values).

**Figure 3.** Topographic profiles. These are the mean profile of each beach over time from data collected in the field. The profiles are used to perform the statistical calculations and the complementary empirical orthogonal function (EOF) analyses. Profile dates are reported in the legend. High, mean, and low water levels are represented as HWL, MWL, and LWL, respectively. Levellings of the mean profile (h) are presented on the vertical axis.


**Table 1.** Volume rates of the Asilah (reef flat (RF)) and Charf el-Akab (sand-rich (SR)) beaches' mean profiles. The accretion volume between dates is representedpositive values (m3/m) and the volume eroded as negative values (m3/m). Cross-shore transport speed (m3/m per day) has been calculated as the net value

 as of

The results presented in Table 1 show the typical erosion/accretion cycle for both the SR and RF beaches. The accretion phase took place during the "summer" season (usually from April to September) due to the prevalence of relatively calm conditions. The erosion phase occurred during "winter" months (from October to March), when high-energy events caused erosion along the foreshore.

The erosion rate (m3/m per year) was calculated as, first of all, the sum of the net volume eroded (negative values) in the 21 months of study. Then, the erosion rate was multiplied by the correction factor of 12/21 to compute the annual rate. Similarly, the accretion rate (m3/m per year) was also estimated as the sum of the net accretion volume (positive values) across the entire time interval. The erosion rate for Asilah (RF) resulted in 38.90 m3/m per year; this value was 77.25 m3/m per year for Charf el-Akab (SR). The accretion rate was 36.46 m3/m per year for Asilah (RF), and a similar value (40.30 m3/m per year) was recorded for Charf el-Akab (SR). Therefore, cross-shore transport for Asilah (RF) ranged from <sup>−</sup>0.37 to 0.67 m3/m per day and from <sup>−</sup>0.98 to 0.35 m3/m per day for the SR beach. The slopes ranged from 3.1 to 6.2% for Asilah (RF) and from 1.5 to 2.2% for Charf el-Akab (SR).

Charf el-Akab (SR) lost twice the volume of sand per year than that for Asilah (RF). The RF dissipates the wave energy due to friction, causing less beach erosion. Nevertheless, Charf el-Akab recorded higher accretion rates than Asilah, but the erosion rate at Asilah was faster than its accretion speed; this favoured a negative sediment budget trend. Moreover, the slope of the RF beach was double (and sometimes triple) that of the SR beach (Figure 4). Once again, wave energy reduction (due to the friction on the RF) is the cause of the higher slope of the beach.

**Figure 4.** Slope (%) of the two kinds of beaches (reef flat, RF, is blue and sand-rich, SR, is red) during the April 2005–January 2007 period.

On the other hand, the slope of the regression line ("m") estimates the volumetric rate of change during the surveyed period. Therefore, the "m" values in Figure 5 express the erosion/accretion per month (m3/month). The higher the slope of the fitted line, the clearer the profile trend. In this way, the SR beach presented high values of "m"; that is, clear tendencies. Both the SR and RF beaches have low R-squared values due to seasonal variability and episodes of erosion and accretion (Figure 5). The seasonal variability is clearly distinctive for the SR beach when summer periods have positive volumetric changes (accretion) and winter periods have negative trends (erosion). Even though the seasonal behaviour is similar for the RF beach, there were smaller changes in volume. The volume changes over time for the SR beach presented a more marked tendency. Anfuso et al. [34] stated that low correlation coefficients between volume changes and time indicate a high degree of beach variability.

**Figure 5.** Evolution of sand volume of two types of beaches (reef flat, RF, and sand-rich, SR) during the April 2005–January 2007 period. Units of the linear regression slope (m) are in m3/m per month.

Comparisons between SR and RF beaches from another country (Spain) were also carried out by Muñoz-Perez and Medina [12] using EOF methodology. The values are similar between RF and SR beaches from two different countries (Victoria Beach in southern Spain vs. Asilah and Charf el-Akab beaches in Morocco). The results are shown in Table 2 in order to verify that both countries present a similar behaviour depending on the kind of beach. The sand-rich profile of Victoria beach shows losses of less volume per year than Charf el-Akab. Nevertheless, the reef flat part of Victoria beach loses less volume per year than Asilah. Therefore, the rocky platform from Asilah experiences less change in volume because it is wider than the one at Victoria. Wide rocky platforms offer more friction, resulting in more wave energy dissipation. In the case of a sandy beach, the absence of a reef flat causes even more erosion because it is not protected from wave energy on the bottom. Significant differences between SR beaches and RF beaches indicate the importance of the presence and typology of the rock platform


**Table 2.** Comparison of erosion and accretion volume rates between reef flat and sand-rich profiles. The beaches are Victoria (Cadiz, Spain), Asilah and Charf el-Akab (Morocco).

#### *4.2. EOF Analysis*

To date, the analyses of the mean profiles have been statistically performed simply to identify differences between the RF and SR profiles on slope and accretion/erosion rates. As mentioned in Section 2, the EOF analysis allowed us to obtain more information from the data; in this way, it helped explain the variability in temporal and spatial profiles. The following results were obtained using EOFs applied to the profiles by subtracting the mean profile.

The first and second spatial components from the EOFs are plotted in Figure 6a (Asilah) and Figure 6c (Charf el-Akab), and the mean profile is presented with levelling on the right axis. Temporal components from EOFs are shown in Figure 6b,d. The variance described by the first component was bigger in the sand-rich profile than in the reef-protected profile, at 77.3% vs. 57.9%. Therefore, the second spatial component explained the greater weight in the variance for Asilah (33.6%) as compared to Charf el-Akab (9.6%). Each spatial component described how the data collected changed along the profile. Thus, the maximum and minimum points mark where either the accumulation or erosion of the beach was observed. Taking this into account, zero means there was no transport of

material at that point, which is called the "rotation point." The temporal components describe the beach erosion/accumulation cycle.

**Figure 6.** (**a**) Spatial EOF in Asilah (RF); (**b**) temporal EOF in Asilah (RF); (**c**) spatial EOF in Charf el-Akab (SR); (**d**) temporal EOF in Charf el-Akab (SR). The height of the mean profile is reported on the vertical axes. The percentages of variance described by the components are reported in the legends.

The first spatial component in Asilah beach (RF) crossed the rotation point at x = 75 m, which corresponded to h = 0.67 m in the mean profile. The second component was always negative except for a small part between x = 25 m and x = 40 m, which corresponded to h = 2.68 m and h = 2.18 m in the mean profile, respectively.

The first spatial component in the Charf el-Akab beach (SR) presented a small part where it changes sign between x = 20 m and x = 50 m and which corresponded to h = 3.64 m and h = 2.88 m in the mean profile, respectively. The second component presented two points that passed through the zero point: the first one at x = 105 m corresponded to h = 2.05 m in the mean profile, while the second one at x = 155 m corresponded to h = 0.93 m.

Oscillations around null axes can be observed in the temporal graphs and cannot be associated with seasonal variations. Indeed, two relative maximums and minimums were observed over the period of two years, indicating that the beach profiles changed only once every year. The peaks of the first component were observed in June 2005 and March 2006 for Asilah Beach, while the most important peak in the second component occurred in April 2006. The peaks of the first component in Charf el-Akab were recorded in September 2005 and March 2006, while the second component presented an oscillation without relevant peaks.

#### *4.3. Physical Interpretation of the Changes*

The first spatial component was associated with the general seasonal change in beach profiles as observed at Asilah; that is, typical of "storm" and "calm" conditions [13]. Similar characteristics were observed at beaches in Cadiz [12,46]. The accretion/erosion periods and the portion of the profile that reflected such changes were identified by the combined analysis of the spatial and temporal components. Therefore, if the rotation point for the first component in Asilah Beach (RF) was x = 75 m, which corresponded to h = 0.67 m in the mean profile, the accretion periods were above h = 0.67 m in summer and below this level in winter (Figure 7a). Moreover, Charf el-Akab Beach (SR) presented a small part in which the trend changed, between x = 20 m and x = 50 m, which corresponded to h = 3.64 m and h = 2.88 m, respectively. Therefore, great volumetric changes observed at Charf el-Akab took place from h = 2.88 m to the submerged zone, which corresponded with a period of accretion during 2005 and erosion during the first half of 2006. After that, the beach seemed to acquire equilibrium (Figure 7b).

**Figure 7.** Sketch of seasonal changes in beach profiles at (**a**) Asilah and (**b**) Charf el-Akab beaches.

The behaviour of the second component was completely different. The increments of this component were associated, as for the first component, with a change in the significant wave height, but also with the prevalence of winds from the east. This condition produced aeolian transport on the beach, but not a large wave regime because the fetch was small and did not allow wave formation [47]. Thus, it is clear that the second component was affected by different variables and not solely controlled

by sea and wave conditions. Hence, the second component was responsible for the shape of the profile and was related to different interactions among several variables, essentially sea regime and wind conditions. Thus, this component influenced the morphology of the profiles under the combined effects of wave and aeolian transport. Taking this into account, it is possible to state that the wind in Charf el-Akab did not produce notable changes in the beach profile because of the shield of topography and presence of some edification. Meanwhile, that was not the case for Asilah beach, which was affected by sand transport when east winds were strong. The presence of tall buildings in Victoria Beach is probably the reason why this behaviour was not observed in the Spanish coast. Beach morphological changes have been described in other papers [48], but more detailed knowledge about the interactions between wave transport and aeolian transport is less well described. Nickling and Davidson-Arnott [40] associated the shape of beach profiles with usable sand volume.

#### **5. Conclusions**

The aim of this paper is to assess differences in profile morphology (volume changes and slope variations of two beaches), taking into account the influence of the nature of the seabed (i.e., the presence or absence of a reef flat). Two kinds of beaches along the north-west coast of Morocco were studied; one is a sandy beach, and the other one is supported by a hard-bottom reef flat. The comparison of the profiles over time helps us to understand the importance and influence on the beach behaviour of the presence of a hard, rocky substrate. EOFs were used to determine the components that describe the behaviour of the beach profiles. The first component was associated with seasonal changes: erosion in winter months, when high-energy events cause significant erosion across the foreshore, and accretion in summer months because of calm conditions. This was particularly evident in the RF beach. On the other hand, the second component was associated with combined wave and wind sediment transport, as observed in many cases in which increments and movement of sand are associated with high winds and low wave heights. These results are similar to other studies based on the use of EOFs. In addition, the behaviour of the beaches is comparable with a previous study of Victoria Beach in the Gulf of Cadiz.

A typical yearly cycle of erosion and accretion can be observed at both beaches. Charf el-Akab (SR) beach lost twice as much sand per year as Asilah (RF) beach. When a beach has a reef flat, the beach profile suffers less erosion than a sand-rich beach where wave energy is not reduced by friction when in contact with a rocky bottom. For the same reason, the slope of the RF beach is twice that of the slope of the SR beach. Despite this, Charf el-Akab has higher yearly accretion rates than Asilah, but the erosion rate for Asilah was faster than the accretion rate, producing an erosional trend, whilst the Charf el-Akab beach seems to have reached a state of equilibrium.

**Author Contributions:** Conceptualization, G.P. and J.J.M.-P.; Data curation, M.T., P.P., P.L.-G. and J.R.-C.; Formal analysis, M.T., P.P., P.L.-G., J.R.-C. and J.V.; Funding acquisition, G.A.; Investigation, G.A.; Methodology, M.T., P.P., G.A. and J.J.M.-P.; Project administration, M.T. and G.A.; Resources, G.P.; Software, P.L.-G. and J.V.; Supervision, G.P. and G.A.; Writing—original draft, P.L.-G. and J.J.M.-P.; Writing—review & editing, G.P., J.R.-C., G.A, J.V. and J.J.M.-P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Mangrove Forests Evolution and Threats in the Caribbean Sea of Colombia**

#### **Diego Andrés Villate Daza 1, Hernando Sánchez Moreno 2, Luana Portz 3,\*, Rogério Portantiolo Manzolli 3, Hernando José Bolívar-Anillo 2,\* and Giorgio Anfuso <sup>4</sup>**


Received: 11 March 2020; Accepted: 11 April 2020; Published: 15 April 2020

**Abstract:** Colombia has approximately 379,954 hectares of mangrove forests distributed along the Pacific Ocean and the Caribbean Sea coasts. Such forests are experiencing the highest annual rate of loss recorded in South America and, in the last three decades, approximately 40,000 hectares have been greatly affected by natural and, especially, human impacts. This study determined, by the use of Landsat multispectral satellite images, the evolution of three mangrove forests located in the Colombian Caribbean Sea: Malloquín, Totumo, and La Virgen swamps. Mangrove forest at Mallorquín Swamp recorded a loss of 15 ha in the period of 1985–2018, associated with alterations in forest hydrology, illegal logging, urban growth, and coastal erosion. Totumo Swamp lost 301 ha in the period 1985–2018 associated with changes in hydrological conditions, illegal logging, and increased agricultural and livestock uses. La Virgen Swamp presented a loss of 31 ha in the period of 2013–2018 that was linked to the construction of a roadway, alterations of hydrological conditions, illegal logging, and soil urbanization, mainly for tourist purposes. Although Colombian legislation has made efforts to protect mangrove ecosystems, human activities are the main cause of mangrove degradation, and thus it is mandatory for the local population to understand the value of the ecosystem services provided by mangroves.

**Keywords:** mangrove; coastal dynamic; salinization; *Rhizophora mangle*; *Avicennia germinans*; *Laguncularia racemosa*

#### **1. Introduction**

Mangrove forests are composed of unique plant species, that is, halophilic trees and shrubs that have specific morphological, physiological, and reproductive characteristics that enable them to survive in a critical interface among terrestrial, estuarine, and near-shore marine ecosystems in tropical and subtropical regions around the world. They are considered one of the most productive natural ecosystems on earth because of their relevant ecosystem services and ecological functions, such as being a nesting habitat for fishes, birds, marine mammals, crustaceans, amphibians, and reptiles. They also act as effective nutrient filters, support numerous rural economies, and protect coastal communities from storms and floods by acting as windbreaks and wave barriers, reducing coastal erosion [1–7]. Last but not least, mangrove forests, due to their great biomass (above- and below-ground) and capacity of accumulation of sediments, are able to store more carbon (on average 22 <sup>±</sup> 6 Tg year<sup>−</sup>1) than terrestrial forests, making them one of the most carbon-rich ecosystems in the tropics, with an estimated value of USD 194,000 per hectare per year [8].

Mangrove forests currently occupy less than 14 million hectares, representing <1% of the world's coastal areas, of which more than two-thirds are located in 18 countries: Indonesia, Brazil, Australia, Mexico, Nigeria, Malaysia, Myanmar, Bangladesh, Cuba, India, Papua New Guinea, Colombia, Guinea Bissau, Mozambique, Madagascar, the Philippines, Thailand, and Vietnam [1,3]. It is estimated that approximately 35% of mangrove forests disappeared during the last two decades of the 20th century, mainly due to their direct conversion to different land uses [1,6,9,10] such as aquaculture, agriculture, urbanization, and impacts due to alterations in the hydrology of river basins and changes in fluvial sediment inflow, among others [3,5,7,11,12]. Although the rate of mangrove forest loss has decreased significantly in the last two decades, it is still worrying, with rates of up to 3.1% per year in some countries—this could lead to a loss of their functionality in less than 100 years. In addition, only 6.9% of the world mangroves are protected, and hence it is mandatory to establish new areas of protection in an effort to reduce the rate of loss [1,4,10,13]. It is estimated that between 0.02 and 0.12 Pg per year of carbon have been released into the atmosphere as a consequence of mangrove degradation, which represents 10% of the total emissions resulting from deforestation [7]. Therefore, global net loss of mangroves would require the successful rehabilitation of about 100,000 ha per year unless the necessary measures were taken to halt current mangrove losses [14]. Further, mangroves work as a transitional intertidal ecosystem that is particularly vulnerable to the effects of climate change, mainly those linked to rising sea level, surface water warming, warming and changes in the composition of the atmosphere, and changes in rainfall, among others [5,12]. In different regions of Latin America and the Caribbean, climate change is considered to be the main driver of environmental impacts on mangroves [12]. Natural and anthropogenic stressors may interact in an additive or synergistic manner, which could lead to accelerated and massive alterations of these ecosystems [11]. However, mangrove forests are considered as a highly resilient ecosystem that has the capacity to adapt and adjust to changing conditions [6], and hence it can play a fundamental role in the design of climate change adaptation strategies.

Mangrove forests observed in Latin America and the Caribbean represent around 26% of the total amount recorded at world scale. They cover an area between 3.58 and 4.54 <sup>×</sup> 106 ha, of which 80% is found in six countries: Brazil, Mexico, Cuba, Colombia, Venezuela, and Honduras [12]. On the western side of South America, the largest mangrove forest cover is found in the tropical zone of the Colombian Pacific coast and in northern Ecuador [15]. Colombia is characterized as being the only country of South America with coasts in the Pacific Ocean and the Caribbean Sea with an extension of 1200 and 1800 km, respectively. Differences in precipitation and tidal range between both coasts favor the existence of almost continuous strips of mangroves along the Pacific coast, whereas in the Caribbean this ecosystem is closely linked to freshwater sources [16].

Overall, Colombia has a mangrove forest cover of around 379,954 ha, with 292,724 and 87,230 ha respectively located on the Pacific and Caribbean coasts [15,17,18]. They show a total amount of eight species: *Rhizophora mangle*, *Rhizophora harrisonii*, *Rhizophora racemosa*, *Laguncularia racemosa, Conocarpus erectus*, *Avicennia germinans*, *Avicennia harrisoni*, *Pelliciera rhizophorae*, and *Mora oleífera*.

On the Caribbean coast of Colombia, in the Department of Atlántico (Figure 1), there are currently around 613.3 hectares of mangrove forests located in different municipalities [19] and, along the coast of the Department of Bolívar, they currently occupy a surface of around 7000 hectares [20]. This paper determines the evolution, during the last decades, of the most extended and representative mangrove forests on the Colombian Caribbean coast between Barranquilla and Cartagena de Indias (departments of Atlántico and Bolívar)—Mallorquín, La Virgen, and Totumo mangrove swamps (Figure 1). This study took into consideration both natural changes (due to coastal erosion/accretion, etc.) and those produced by anthropic activities [21], which have influenced the evolution of the aforementioned mangrove forests, in order to design adequate plans for their environmental improvement and sound conservation strategies. The present paper investigates their evolution and their human (deforestation, road construction, etc.) and natural impacts (salinity variations, coastal erosion, etc.) on three mangrove forests located at the departments of Atlántico and Bolívar, located in the Caribbean coast of Colombia (Figure 1). Such areas are characterized by five mangrove species that are common along the Colombian Caribbean Sea [21]: *Rhizophora mangle*, *Avicennia germinans*, *Laguncularia racemosa*, *Conocarpus erectus*, and *Pelliciera rhizophorae.*

**Figure 1.** Location map with studied mangrove forests.

#### **2. Study Area**

The Caribbean coast of Colombia is a tropical environment with seasonal variations in rainfall (Figure 2) from the dry season (December–March) and the transitional seasonal (April–July) to the rainy season (August–November) [22,23].

**Figure 2.** Rainfall and temperature variations at Colombian Caribbean coast (data recorded in the 1980–2010 period- IDEAM [24]): (**a**) Department of Atlántico; (**b**) Department of Bolivar.

The main regulator of rain cycles throughout the Colombian territory is the Intertropical Convergence Zone [25,26]. This low pressure equatorial system follows the synchronization of the sun [27] and a southern movement (between 27.5◦ S and 27.5◦ N), having the largest amount of solar energy received by the planet [28]. Other oscillations, such as Maiden and Julian, as well as the variability associated with the El Niño and La Niña phenomena (Southern Oscillation—ENSO) in the Pacific, contribute especially to the intensity of the precipitation anomalies on the Colombian Caribbean coast [26]. Tide has mixed semi-diurnal periodicity and microtidal variability, with maximum values of 60 cm [22]. Wind average velocity is <12 m/s, with the strongest winds blowing from the northeast in December–March and weakest values associated with easterly approaching winds, usually blowing in September–November. The dominant sea surface current is the *Caribbean current* that flows during almost all the year from east to west; an opposite current, the *Darien* or *Colombia* current, flows from Panama northeastward [29]. Significant wave height is between 1 and 2 m and wave climate is dominated by swell waves approaching from the northeast from November to May, and sea (smaller) waves, approaching from the northwest, west-southwest, and southwest, during the remainder of the year. Predominant longshore sediment transport is southwestwardly directed and a minor reversal takes place during rainy periods when southerly winds achieve more importance, giving rise to short erosive waves [30]. Coastal erosion is essentially linked to the impact of hurricanes and cold fronts—the former events impacting the coast from June to November, and latter events from January to March [31].

**Mallorquín Swamp** is located in the northwest part of the Department of Atlántico (Figure 1), on the western bank of the Magdalena River, close to Barranquilla. It is a shallow estuarine coastal lagoon with an area of around 650 ha, surrounded by floodplains and sand dunes. Three species of mangroves have been reported in Mallorquín Swamp: *A. germinans, R. mangle,* and *L. racemosa* with a maximum average height of 15.7 m. The most abundant species is *A. germinans* (71%), followed by *L. racemosa* (21%), and finally *R. mangle* (8%) [32]. Salinity range was 9%–22% in surface water and 14%–50% in interstitial water, that is, at a depth between 50 and 100 cm [32]. The mangrove leaves had cuts and perforations associated with herbivory. The presence of climbing plants was also observed—they generate overweight and strangulation of stems and branches [32]. The system has been affected by coastal erosion, increased sedimentation processes, and anthropogenic activities linked to the urbanization of nearby areas and contamination processes due to solid waste and sewage discharges [33–37].

**Totumo Swamp** (Figure 1), located between the departments of Bolívar and Atlántico, is composed by a main body of water with an approximate extension of 1361.06 ha and presents several mangrove patches on the borders of the swamp. Totumo Swamp presents average salinity values from 0.1% to 2%, and thus it is considered a fresh water body [38]. The vegetation adjacent to Totumo Swamp belongs to the dry tropical forest and very dry thorny scrub. Lastly, along the swamp flood plains, an abundant vegetation of *Typha dominguensis* was observed [38]. The swamp is partially protected by a 5 km long spit, namely, Galerazamba spit, made up of a wide beach with a gentle slope from 2◦ to 5◦, which has undergone major morphological changes since its initial description by Francisco J. Fidalgo in 1805 [39–41]. The spit encloses inland lagoons (e.g., La Redonda) that are fed by rainfalls, as well as directly by the sea during the stormy season. La Redonda lagoon, inside the Galerazamba spit, has an average salinity of 23%, suitable for mangrove development [38]. On and nearby the spit, there are several mangrove areas composed of *L. racemosa* (77%), which is located on the edge of the lagoon, and *C. erectus* (23%), which is found at the mouth of the swamp. Mangroves have a shrubby growth, with maximum average heights of 6.1 m for *L. racemosa* and 5.7 m for *C. erectus*. The vegetation is currently under anthropogenic pressure, mainly due to the expansion of the agricultural frontier that has been increasing in recent years, negatively impacting the mangrove forest by illegal logging and successive land reclamation. Presently, the original brackish vegetation is changing towards a typical freshwater vegetation ecosystem because of the reduced inputs of marine waters due to the construction of gates (that are usually closed) at the lagoon inlet entrance [19,38].

North of Cartagena de Indias (Department of Bolívar, Figure 1), there is one of the most important coastal wetland of Colombia, namely, **La Virgen Swamp,** which has an approximate surface of 20 km2 and a mangrove forest cover of around 824 ha, with a main drainage network consisting of 8 streams in the rural area and 20 channels in the urban perimeter of Cartagena de Indias. Four species of mangrove have been reported in La Virgen Swamp: *A. germinans* (67%), *R. mangle* (30%), *L. racemose*, and *C. erectus* that, together, make up 3% of the swamp. *R. mangle* is located on the internal border of lagoons and channels; *A. germinans* is found in the less intervened sites, reaching the limit with the tropical dry forest; *L. racemosa* is mainly found on the edges of abandoned ponds; and *C. erectus* is only found at the border between the mangrove forest and the mainland vegetation [42]. Salinity varies considerably, and its most frequent peaks are in the range of 0% to 35% [42]. The average height of the mangrove plants varies from 1 to 10 m, according to the forest sector [21,43].

The area is subject to various threats such as illegal logging; artificial filling; and terracing for the implementation of fish farming, waste disposal, urbanization, and pollution; for example, La Virgen Swamp receives about 60% of the Cartagena de Indias wastewaters, around 114,000 m3/day. Further, because of the reduced capacity of water exchange between the lagoon and the Caribbean Sea, several problems have arisen in the last decades, such as eutrophication, increased salinity, and fish mortality [20,44,45].

#### **3. Methodology**

The methodology applied to assess the mangrove area changes was supported by multi-date remote-sensed data. Using satellite images, it is possible to identify, calculate, and monitor mangrove areas, as well as the surfaces affected by erosion processes [46–49].

#### *3.1. Data Used*

To determine the extent of mangrove ecosystems, we used Landsat 5 and 8 multispectral satellite images (TM—thematic mapper, and OLI—operational land) with a spatial resolution of 30 m in their optical channels available at the United States Geological Survey [50]. The dates of the images varied between 1985 and 2018 (Table 1). The dataset was composed of the OLI sensor bands, already subjected to a complex algorithm of adjustment of the atmospheric effects that was based on parameters estimated from the same sensor bands and the application of the model of the second simulation of the satellite signal with the code known as solar vector spectrum (6SV) [51].

The images presented a good resolution due to the lack of cloudiness and atmospheric disturbances. Similarly, Google Earth Pro images and thematic cartography available in DIMAR-CIOH [52], INVEMAR [53], and a layer from the Humboldt Foundation cartographic base were used.

#### *3.2. Digital Images Analysis*

The manipulation of Landsat images was carried out using ARGIS 10.4. Initially, the coordinate system was defined by processing the georeferencing according to the parameters defined for land mapping in Colombia with the Magna Sirgas Datum Bogotá coordinate system.

The analysis, interpretation, and quantification procedures of the images used were performed using band composition, as well as the normalized difference vegetation index (NDVI), which was calculated using the following equation:

$$\text{NDVI} = (\text{IVP} - \text{V}) / (\text{IVP} + \text{V}) \tag{1}$$

where IVP and V represent the reflectance values in the near-infrared and red infrastructure bands, respectively. The NDVI varies from −1 to +1, with negative values and zero representing areas without vegetation [54]. The procedure for delimiting the areas identified as mangrove forests was carried out by manual vectoring on the classified images. This method requires a great deal of interpretation and is time consuming, as each polygon must be determined individually.

Ortho-rectified satellite images were also used to map shoreline position. Shoreline migration was analyzed in ARGIS 10.4 by the Digital Shoreline Analysis System—DSAS 5.0 [55].


**Table 1.** Details of multispectral remote sensing data and other documents used in this study.

<sup>1</sup> Atlas Geomorfológico del Caribe Colombiano [52]. <sup>2</sup> Sistema de información para la gestión de los manglares de Colombia SIGMA [56]. <sup>3</sup> Ordenamiento Ambiental de la Zona Costera del Departamento del Atlántico [53].

#### *3.3. Field Visits*

Detailed field visits were made to the three mangrove forests investigated in this study in order to determine actual predominant plant species, their distribution, their conditions (leaf characteristics and plant average height), and the typology of anthropic activities (road emplacement, expansion of urbanized areas, disposal of waste materials, illegal logging, etc.) and their effects on the environment. Characteristics of mangroves in past investigated time spans were essentially reconstructed from unpublished reports carried out by environmental departmental authorities and research institutes.

#### **4. Results and Discussion**

#### *4.1. Mangrove Swamps Evolution*

The evolution of three most relevant mangrove forests along the Department of Atlántico and Bolívar, that is, the Mallorquín, Totumo, and La Virgen swamps, is presented in the following sub-sections.

#### 4.1.1. Mallorquín Mangrove Swamp

The mangrove forest of Mallorquín Swamp presented small changes in the period of 1985–1998, a reduction of around 51 ha from 1998 to 2013, and an increase of around 34 ha during the 2013–2018 period (Figure 3, Table 2).

**Figure 3.** Evolution of the mangrove forest (green) in the Mallorquín swamp (**a**) 1985, (**b**) 1998, (**c**) 2013, and (**d**) 2018.


**Table 2.** Variation of the mangrove forest cover (ha) in the Mallorquín, Totumo, and La Virgen swamps.

Until the beginning of the 1940s, the swamp had an estuarine regime constituted of different coexisting and connected systems, that is, the Cantagallo, Mallorquín, La Playa, and Manatíes swamps, which presented a wide variety of ecosystems and fishing areas [57,58]. After the construction of the western jetty at the Magdalena River mouth, during the 1925–1935 time span, the sand bar that encloses Mallorquín Swamp suffered—between 1939 and 1987—an erosion rate of 65 m/year, finally acquiring its current configuration. The swamp recorded a strong hydrologic imbalance due to a deficit in fresh water supplies that were essentially limited to the León stream and direct rainfall [53,59]. The deficiency in water supplies significantly affected the different ecosystems, impacting the fishing activities and the mangrove forest health [57,58].

Concerning the period from **1985 to 1998**, fairly unrepresentative changes were reported, that is, a very small increase (Table 2) as a result of the mangrove loss observed at the sand bar that encloses the swamp and the mangrove cover increase recorded in the southern part of the swamp (Figure 3a,b). Despite such small changes, the swamp reflected a stressful environmental condition, mainly due to the diminution of hydrodynamic processes. The effects of such an unfavorable situation continued in following years, and were more evident at the beginning of the 2000s.

In this same period, the area of Mallorquin Swamp was reduced considerably. The bar migrated landward on average 29.5 m/year, resulting in a period of relevant morphological changes and loss in mangrove cover.

Changes in the hydrodynamic conditions, as observed since the end of the 1980s by Galvis et al. [19,60], were linked to the few continental and marine water supplies, because of the reduced fresh water supplies from the León stream and the lack of permeability of the western jetty at the Magdalena River mouth. This brought to the modification of the mangrove ecosystem, and *A. germinans* and eurihaline vegetation became the predominant species compared with *R. mangle*, reflecting a process of salinization that continued in following years, reaching interstitial salinity values from 20% to 30%, as reported by Galvis et al. and Ulloa-Delgado et al. [60,61]. However, the loss of *R. mangle* was not only associated with the salinity increase but also with erosion processes that led to the loss of trees and beach surfaces, which made the implantation of new *R. mangle* propagules impossible.

In the period from **1998** to **2013**, around 51 ha of forest were lost (Table 2) as a result of urban area expansion in the southern part of the swamp [19,61], where illegal forest cutting by local inhabitants, accumulation of solid waste materials, and artificial infilling works were observed (Figure 3b,c). On the other hand, mangrove growth was recorded in the sand bar because of the reduction of coastal erosion (Figure 4) that allowed bar stabilization and propagules implantation and growth.

The most exploited species was *L. racemosa*, used for the construction of huts to provide shade for tourists visiting local beaches. As for solid waste materials, which prevent the proper development of seedlings and pneumatophores and hence favor a decrease in the self-healing capacity of the forest [62], they essentially came from the Magdalena River that drains a basin of 257,430 km2, wherein 724 municipalities are located, representing 80% of the Colombian population [63]. Further mangrove losses were related to the diminution of fresh water supplies, especially during the 1997–1998 period in which Colombia was greatly impacted by the El Niño phenomenon that led to a generalized decrease of rainfall and river discharges. During the most critical dry period (October 1997 to January 1998), the flow of the Magdalena River fluctuated between 45% and 70% in its lower basin [64], considerably decreasing its water contribution to Mallorquín Swamp and the consequent increase of salinity. The

very narrow strip of *R. mangle* observed in 1998 in the southern, northeastern, and northwestern parts of the swamp was later replaced by a monospecific forest of *A. germinans*, with the presence of few individuals of *L. racemosa* and *C. erectus* and xerophytic plants (*Prosopis juliflora*, *Capparis odorantissima*, *Sarcostema* sp., and *Stigmaphyllon* sp.) [61]. In general, the forest presented mature individuals, but of little height (average: 3.86 m) and with leaves characteristic of xerophytic plants [61]. Such a trend continued in 2005, as reported by INVEMAR [53] and GTA [57], but slightly decreased in 2007 when salinity achieved values between 20% and 35% that allowed *A. germinans* to survive [65] and led to a slow recovery in the general mangrove forest cover [53]. In addition, in 2007, reforestation programs were initiated by the fishing communities with the support of local environmental authorities [53]. The stabilization of the hydrological conditions and the reforestation programs had positive effects that were observed in the following period investigated.

**Figure 4.** Evolution of the mangrove forest (green) in the Totumo swamp: (**a**) 1985, (**b**) 1998, (**c**) 2013, and (**d**) 2018.

In the period from **2013** to **2018**, an increase in mangrove coverage of around 34 ha was located on the west part of the sand bar and in the southern part of the swamp (Figure 3c,d). In this period, the stability of the sandy bar was observed, as a consequence of the growth of a spit, as being connected to the Magdalena western jetty, located to the north of the swamp. This new formed feature protected the bar from incoming waves (Figure 3c,d).

Additionally, the recovery of the mangrove could be related to the reforestation campaigns carried out by different entities [53] and the maintenance of the different channels that communicate the swamp with the Magdalena River. In 2016, the Corporación Autónoma Regional del Atlántico (Local Environmental Authority) carried out activities aimed at the environmental recovery of the swamp by directly cleaning the channels and by installing meshes in the boxcoulverts to reduce waste materials entering in the channels communicating the swamp with the Magdalena River [58]. Presently, an ongoing project is being developed to declare Mallorquín Swamp a protected area [66].

#### 4.1.2. Totumo Mangrove Swamp

The mangrove forest of Totumo Swamp presented a reduction of ca. 280 ha in the period from **1985 to 1998**, an increase of ca. 13 ha in the period from **1998 to 2013** and a loss of ca. 33 ha during the **period from 2013 to 2018** (Figure 4 and Table 2).

The large loss of mangrove cover recorded in the first period (from **1985 to 1998**) was related to the transformation of the forest into a freshwater vegetation ecosystem [38]. Most relevant changes were recorded in the northern part of the swamp (Figure 4a,b). Totumo Swamp communicated with the Caribbean Sea through a natural inlet that allowed the entry of salt water, generating ideal conditions for the growth of mangrove [38,53]. However, in the 1970s, gates emplaced at the inlet entrance, 20 m in width, greatly limited sea water entrance, turning much of Totumo Swamp into a freshwater system (Figure 4) [38,67], which receives supplies from numerous streams, such as Punta Antigua, Lata, Calabria, and Bombo [61]. As a result, during the rainy season, salinity was close to zero and, during dry seasons (i.e., January–May and July–September), salinity increased up to 15% [68]. Mangrove plants presented shrubby and stunted forms and did not exceed 2 m in height. There were observed specimens of *R. mangle*, *L. racemosa*, *C. erectus,* and *A. germinans,* the latter being the most abundant. A considerable decrease of *R. mangle* was observed at the mouth of the swamp, and *Typha dominguensis* was recorded on the west bank, alternating with *C. erectus* with rhizomes on the roots, stems, and lower branches [61].

An increase of around 13 ha in mangrove surface was recorded in the period from **1998 to 2013** as a result of mangrove growth on the western side of the swamp, on the Galerazamba spit, as well as a decrease on the eastern side (Figure 4b,c).

Galerazamba spit has undergone relevant morphological changes since its original description by Francisco J. Fidalgo in 1805 [39–41]. Anfuso et al. [69] reported spit evolution from 1947 to 2013, recording an accretion of around 0.7 km<sup>2</sup> and a down drift (i.e., southward) migration of about 80 m. Hence, the increase in the mangrove area was directly related to the growth of the spit that enclosed numerous small lagoons, among which the larger one was La Redonda Swamp, fed by rainfalls and by the sea during stormy season [39,40].

On the eastern side of the swamp, INVEMAR [19] observed in 2004 that the mangrove cover was reduced to a strip of approximately 5 to 10 m in width, mainly composed of *C. erectus* and *L. racemosa.* At this area, the natural mangrove cover was progressively displaced by pastures and macrophytes and human activities as livestock and agriculture [53]. On the southwestern side of the swamp, only a very narrow line of *C. erectus*, with plants about 7.5 m high, was recorded; no natural regeneration or flowering of mangroves was observed [19].

The mangrove forest of Totumo Swamp degraded in the period from **2013** to **2018**, with a loss of around 33 ha (Figure 4c,d). The decrease was mainly observed on the eastern side, whereas the western side showed only a small increase associated with mangrove growth in the lagoons at Galerazamba split. The overall reduction in mangrove cover was undoubtedly related to the loss of water exchange with the Caribbean Sea. This turned Totumo Swamp into a freshwater ecosystem, wherein water is nowadays mainly used for agricultural and livestock activities. The economic interests linked to the maintenance of such activities retain the gates at the inlet entrance that was closed to prevent seawater entrance, thus prohibiting the restoration of the original swamp ecosystem.

#### 4.1.3. La Virgen Mangrove Swamp

La Virgen mangrove forest presented a reduction of around 40 ha in the period from **1985 to 1998**, an increase of around 255 ha in the period from **1998 to 2013**, and again a reduction of around **34 ha** in the period from **2013 to 2018** (Figure 5, Table 2).

**Figure 5.** Evolution of the mangrove forest (green) in La Virgen swamp: (**a**) 1985, (**b**) 1998, (**c**) 2013, and (**d**): 2018.

The mangrove forest of La Virgen presented a reduction of approximately 40 ha in the period from **1985 to 1998** (Table 2). This reduction was mainly observed in the northern part of the swamp (Figure 5a,b) and was associated with different factors that were reported since 1984 by CARDIQUE [70], among them are (i) the accumulation of waste materials, mainly plastics, observed especially at the

roots of *R. mangle* [21]; (ii) the illegal logging—for the construction of houses by the communities living on the banks of the swamp—of *L. racemose*, which consequently suffered a relevant pressure, with mortality values of 92.9%; and (iii) the land filling for urban infrastructure emplacement, such as the construction of a connection runway to the local airport and the roadway Cartagena-Barranquilla, among others [45,70]. In the following years, the mentioned roadway greatly affected water exchanges between the swamp and the Caribbean Sea, reducing the existing inlets to a single point located at La Boquilla [43,71]. Therefore, water exchange capacity was significantly reduced, limiting the oxygenation process and altering the constant flow of sediments and organic material, mainly due to the periodic infilling of the inlet, which generated problems such as eutrophication, increased salinity, and fish mortality [44]. In addition, the road construction at the border of the mangrove forest will probably lead to the replacement of native vegetation by other vegetation types, as observed in Punta Mala Bay, Panama [72]. During this investigated period, along and nearby the mentioned roadway, the construction of many resorts and hotels also took place, which was done to satisfy the tourism market demand, as well as the implementation of aquaculture activities (e.g., cultivation of tarpon fish) [71].

In the southern sector, the loss of mangrove areas occurred mainly through landfill processes and, consequently, substitution by urbanization (Figure 5).

Concerning the period from **1998 to 2013**, an increase of 255 ha was observed essentially along the eastern border of the swamp (Figure 5b,c). This mangrove surface increase was linked to the execution of a project, at the end of 2000, which stabilized the La Bocana area to guarantee the permanent water entrance from the Caribbean Sea. The project consisted in the emplacement of different structures at the inlet entrance and along the feeding channels (sand traps, gates, etc.) that favored the recirculation of water within the swamp, improving the level of oxygenation and salinity; in this way, the lagoon recovered the natural ecosystem services [73]. INVEMAR [19] recorded an increase of mangrove vegetation in 2004, which was dominated by *R. mangle* and *A. germinans* (that together accounted for 97% of the total amount), followed by *L. racemosa* and *C. erectus*. *L. racemosa* was rare and was usually found on the edges of abandoned ponds, whereas *C. erectus* was the least frequent and was only located at lagoon borders, along with the mainland vegetation. In addition, in 2004, a reforestation program was carried out, wherein 401,564 plants (of four species of mangroves) were planted in an area of 44.3 ha [19]. Summing up, the reestablishment of the hydrodynamic of the swamp and the implementation of reforestation programs seemed to be the main reasons for the increase in mangrove coverage between 1998 and 2013.

Concerning the last investigated period, from 2013 to 2018 (Figure 5c,d), the swamp presented a reduction of mangrove forests of approximately 30 ha, essentially due to the conclusion of the construction, in 2015, of the Cartagena-Barranquilla roadway, as well as a new roadway that included the emplacement of a viaduct 5.39 km in length, 4.73 km of which passes over La Virgen swamp [74]. The environmental assessment prior to the construction of the viaduct stated that the structure would impact around 52 ha of swamp surface and 41 ha of mangroves [75]. The National Environmental Licensing Authority (ANLA) authorized in 2015 the logging of 1673 trees, among which 1158 were mangrove trees [75]. The construction company has to compensate such actuations by creating a new green area of 177 ha, which could in the next few years increase the total forest coverage [75]. In 2015, the Institute Humboldt and the Foundation Omacha observed, as is the case in the northern zone of the swamp, there is only *R. mangle* (64%) on the banks and *A. germinans* (33%) on the mainland, whereas four species of mangrove were recorded in the eastern sector: *A. germinans* (65%), *R. mangle* (33%), *C. erectus* (0,6%), and *L. racemosa* (0.3%). Inside the forest, there were no permanent flooding events, and hence *A. germinans* dominated (with 97%), and both *C. erectus* and *L. racemosa* were rare and interspersed with *A. germinans* [76]. Further losses of mangrove coverage in the period between 2013 and 2018 were observed in the southern and southeastern areas of the swamp that have been progressively filled in and occupied by human illegal settlements and aquaculture ponds that are generally linked to the illegal occupancy by people displaced by violent events from different territories in Colombia and in Venezuela [43,45]. These different factors could together be the cause of the

mangrove loss recorded in the analyzed time frame. Nowadays, the swamp shows around 775 ha of mangroves, including *R. mangle, A. germinans,* and *L. racemosa* [45].

#### *4.2. Environmental and Natural Concerns*

Oceanographic, climatic, and geomorphological aspects; soil conditions; level and duration of flooding; salinity; and sediment load determine the structure and floristic composition of mangrove forests [77].

**Salinity** is one of the most important abiotic factors influencing the structure, seedling establishment, and function of mangrove ecosystems, and small changes in salinity can lead to abrupt ecological changes [6,78]. Most mangrove species can grow in fresh water, but growth is stimulated under saline conditions, with optimal seawater concentrations ranging from 5% to 50%, depending on the mangrove species [78,79]. High salinity soils limit seed germination and reduce plant growth [80]. The low rates of water loss from the leaves, as a strategy for surviving under hypersaline growth conditions, limit rates of carbon gain affecting the plant growth; as a result of this, individuals of low stature and stunted growth (scrub mangroves) are observed [79]. Therefore, the degree of tolerance to salinity determines the dominance of species within the mangrove forest. Mangrove forests dominated by *R. mangle* can cope with wave strength and salinities close to those of sea water (about 35%), those dominated by *A. germinans* can tolerate and develop under salinity conditions of 60–65% and sandier substrates. Although *L. racemosa* tolerates salinities similar to *R. mangle*, it requires a more open canopy, which allows a greater availability of light for regeneration, and therefore is generally associated with disturbed areas [77]. The effect of the increased salinity on the mangrove species was clearly observed in Mallorquín Swamp, which suffered a loss of diversity—*A. germinans* and *R. mangrove* were replaced by *A. germinans* of low stature and a few individuals of *L. racemosa* [19]. The same condition was also described by Sánchez et al. [65] in a mangrove forest located in the municipality of Puerto Colombia (Colombian Caribbean Sea). The authors detected that in soils with high salinity (76% at 0.3 m), the dominant species was *A. germinans*; meanwhile, *R. mangle* was absent or very scarce. A similar trend, that is, the predominance of *A. germinans* associated with a few individuals of *L. racemosa,* was also observed in the forest along the Indian River Lagoon (Florida, USA), which is developed on high salinity soils due to low sea water inputs and high evaporation rates [79]. Mangrove mortality due to hypersalinity conditions was reported in the Ciénaga Grande de Santa Marta (Colombia), where sites with dead or dwarfed vegetation had an average soil salinity of 74%, with values between 52% and 100%. Basal area and forest biomass volume were inversely correlated with soil salinity [81]. Currently, the Mallorquin swamp presents salinity values between 14%–50% in interstitial water, which allow the growth of *A. germinans*, *R. mangle,* and *L. racemosa* [32]. In addition, new sandy bodies are being formed in front of Mallorquín Swamp, where mangrove seedlings are observed (Figures 3d and 6a). This could generate ideal conditions prone to the increase of this mangrove forest. However, the presence of solid waste is one of the main problems that this forest presents; therefore, cleaning and educational campaigns should be carried out in the surrounding communities and also should be devoted to the tourists that visit the forest (Figure 6b).

**Figure 6.** (**a**) Mangrove seedlings growing in the new sand bar, and (**b**) plastic waste observed in the mangrove forest of Mallorquín Swamp.

The mangrove, as a facultative halophyte plant, can tolerate fresh water conditions for a limited time but not during its entire life cycle [82,83]. The flowing of seawater into the mangrove forest system reduces the possibility of survival for plant species that are not salt tolerant [82]. The mangrove forest of Totumo Swamp was transformed on its eastern side into an ecosystem dominated by freshwater plants (*Typha dominguensis* and *Eichhornia crassipes*) due to the loss of seawater input that brought salinity to zero (Figure 7a); meanwhile, mangrove remnants were recorded on its western side that is characterized by brackish water conditions (Figure 7b). La Virgen Swamp was affected mainly by the construction of a road that impedes the normal exchange of water with the Caribbean Sea. The swamp receives about 60% of the city wastewater, which has an approximate volume of 114,000 m3/day [44,45]. The lack of communication with the Caribbean Sea generated an increase in wastewater concentration, generating eutrophication processes that affected the mangrove forest.

**Figure 7.** (**a**) Eastern side of Totumo Swamp characterized by fresh water conditions; (**b**) mangrove remnants (*Rhizophora mangle*) in the western side of Totumo Swamp characterized by brackish water conditions.

**Deforestation** is considered one of the main anthropogenic causes of mangrove degradation [84]. Colombia has the highest annual rate of deforestation in South America, showing values between 1.1% and 0.6% for 1980–1990 and 2000–2005 periods versus average values of 0.69% and 0.18% observed in South America [85]. In Colombia, mangrove areas have been converted mainly into agricultural land, ponds for aquaculture, and urban development (mainly for tourism). Illegal logging has been reported for the three mangrove forests investigated in this paper (Mallorquín, La Virgen, and Totumo), but each one was associated with different activities.

In Mallorquín Swamp, the logging was particularly linked to an increase in urban growth, mainly in the southern part of the swamp (Figure 3d). In La Virgen Swamp (Figure 8a) urban growth for

tourism, road construction (Figure 8b), aquaculture, and the use of wood for house building were the main factors associated with the felling of the mangrove forest. In Totumo Swamp, the expansion of the agricultural frontier and the increase in livestock activity on the margins of the swamp were the main factors associated with the felling of the mangrove forest. With respect to land uses in Totumo Swamp, they include extensive livestock farming (5209.43 ha), annual or seasonal crops (53.75 ha), pasture cultivation (3049.85 ha), and aquaculture (155.82 ha) [38].

**Figure 8.** (**a**) Mangroves (*R. mangle*) in good condition in the west sector of La Virgen Swamp; (**b**) mangrove cutting due to the construction of the viaduct over La Virgen Swamp.

Deforestation modifies the mangrove phytogeographic landscape and reduces its biological diversity [84]. Deforestation not only decreases the number of mangrove specimens, but also produces ecological effects on the forest, such as stunted and shrubby growth, canopy opening, stem mortality, decreased regeneration of harvested species, and changes in species composition, among others [86]. The extraction of woody and non-woody products has degraded many areas of mangroves, resulting in the development of plants of low height and thin diameter [85], as reported mainly in the Mallorquín and Totumo swamps. Mangrove deforestation is linked to the lack of adequate methods to assess their economic value; this has led people to consider them as worthless wastelands and as unhealthy risk areas that should be adapted to alternative and more lucrative uses [84].

Mangrove forests are also subject to natural disturbance related to changes in the amount and seasonality of rainfall. This is one of the most relevant factors in the Colombian Caribbean coast, an example being the intensification of the El Niño phenomenon, which generates decreases in the productivity of mangrove forests dominated by *R. mangle* and, in the long term, their replacement by *A. germinans* [77]. This was observed in the mangrove forest of Mallorquín Swamp that, due to alterations in its hydrology caused by anthropic activities (e.g., construction of the western jetty at the Magdalena River mouth) and natural phenomena such as a relevant reduce of precipitations due to El Niño (1997–1998), recorded not only alterations/degradation in its cover, but also in its composition becoming a monospecific forest consisting only of *A. germinans.*

Mortality events associated with climate extremes (e.g., tropical cyclones and El Niño and La Niña phenomena) could increase in the coming years because mangroves are important sentinels of global climate change processes [87].

#### **5. Conclusions**

The results obtained in the framework of this paper show how, during different periods of time, mangrove forests at localities investigated in the Colombian Caribbean Sea have been impacted by diverse and complex anthropogenic activities and natural disturbances. Although natural disturbances such as the El Niño phenomenon have greatly affected, at times, the cover and the structure of mangrove forests, human activities were the main cause of degradation and loss. Alterations in the hydrology of swamps, which lead to increases or decreases in salinity, urban growth, illegal logging, expansion of agricultural frontiers, and road construction, were the main causes associated with the loss of mangrove cover.

Although losses of mangrove covers were observed in all investigated sites, it is worth noting that different activities carried out by local environmental authorities, together with local inhabitants, favored mangrove forest cover stabilization and, at places and times, led to an increase, such as in the Mallorquín and La Virgen swamps. In Totumo Swamp, different economic interests, essentially linked to agricultural activities, prevented mangrove forest recover. At many places, illegal activities such as logging continue to affect the mangrove forests investigated in this paper, and thus it is mandatory for the Colombian environmental authorities to develop strategies aimed not only at protecting and recovering these ecosystems, but also at raising awareness among the local inhabitants concerning the ecological value of these ecosystems, as well as their importance in coastal adaptation and mitigation function of climate change-related processes.

**Author Contributions:** Conceptualization, H.S.M., D.A.V.D. and H.J.B.-A.; Data curation, G.A., R.P.M.; Formal analysis, L.P., D.A.V.D., H.S.M. and H.J.B.-A.; Methodology, D.A.V.D., L.P., R.P.M. and G.A.; Writing–original draft, L.P., R.P.M., H.J.B.-A. and G.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** This research is a contribution to the Andalusia PAI Research Group RNM-328, the RED PROPLAYAS network, Universidad Simón Bolívar (Barranquilla, Colombia), Escuela Naval de Suboficiales ARC Barranquilla, Universidad de la Costa (CUC), and the Center for Marine and Limnological Research of the Caribbean CICMAR (Barranquilla, Colombia).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Dune Systems' Characterization and Evolution in the Andalusia Mediterranean Coast (Spain)**

**Rosa Molina 1, Giorgio Manno 2,\*, Carlo Lo Re <sup>2</sup> and Giorgio Anfuso <sup>1</sup>**


Received: 26 May 2020; Accepted: 21 July 2020; Published: 23 July 2020

**Abstract:** This paper deals with the characterization and evolution of dune systems along the Mediterranean coast of Andalusia, in the South of Spain, a first step to assess their relevant value in coastal flood protection and in the determination of sound management strategies to protect such valuable ecological systems. Different dune types were mapped as well as dune toe position and fragmentation, which favors dune sensitivity to storms' impacts, and human occupation and evolution from 1977 to 2001 and from 2001 to 2016. Within a GIS (Geographic Information System) project, 53 dune systems were mapped that summed a total length of ca. 106 km in 1977, differentiating three dune environments: (i) Embryo and mobile dunes (Type I), (ii) grass-fixed dunes (Type II) and (iii) stabilized dunes (Type III). A general decrease in dunes' surfaces was recorded in the 1977–2001 period (−7.5 <sup>×</sup> <sup>10</sup><sup>6</sup> m2), especially in Málaga and Almería provinces, and linked to dunes' fragmentation and the increase of anthropic occupation (+2.3 <sup>×</sup> <sup>10</sup><sup>6</sup> m2). During the 2001–2016 period, smaller changes in the level of fragmentation and in dunes' surfaces were observed. An increase of dunes' surfaces was only observed on stable or accreting beaches, both in natural and anthropic areas (usually updrift of ports).

**Keywords:** dune characterization; anthropic occupation; fragmentation index; dune surface

#### **1. Introduction**

Human interest in coastal processes and evolution has greatly increased in recent decades due to the increment of human developments recorded in coastal areas [1] and the impacts of extreme events, such as hurricanes and storms [2,3], the effects of which are enhanced by sea level rise and other climatic change-related processes, such as the increasing height of extreme waves, or changes in the tracks, frequency and intensity of storms [4–7]. Coastal development, which is essentially linked to tourism—one of the world's largest industries [8]—continues to increase, and some 50% of the world's coastline is currently under pressure from excessive development [1]. In Europe, the rapid expansion of urban artificial surfaces in coastal zones during the 1990–2000 period [9] has occurred in the Mediterranean and South Atlantic areas, namely Portugal (34% increase) and Spain (18%), followed by France, Italy and Greece.

Activities and infrastructures related to tourism and other human developments too (e.g., fishing and industrial activities) are significantly affected by the impacts of storms and hurricanes that, over the past century, have caused huge economic losses along with high mortality rates along the world's coastlines [10–13]. Coastal erosion and flooding processes have reduced beach and dune ridges' width and produced the loss of associated touristic, aesthetic and natural values [14–17]. Beach erosion/accretion cycles are often recorded at an inter-annual time scale and are related to

seasonal wave climate variations due to temporal and spatial distributions of high-latitude storms and hurricanes/typhoons [3,18–22]. Erosion is observed after storm events, at high latitudes recorded during winter months, but beach recovery takes place during fair weather conditions, which is known as "seasonal" beach behavior [10,23], and in general happens at longer time intervals, from weeks to months [24,25]. Hence, natural beach recovery guarantees the reformation of a wide beach and its associated protection function and touristic use, but dune response to erosive events is very different. Meanwhile, dune erosion is always very rapid and time located accretion is a process that usually takes place with low rates over a long time, from several months to years in the Andalusia Atlantic littoral [26]. Hence, the determination of coastal dune characteristics, behavior and evolution need special attention in the attempt to reduce erosive/flooding processes' impacts on natural and urbanized coasts. Several recent investigations [27,28] have identified dunes as one of the most relevant coastal ecosystems as a natural defense able to reduce flood sensitivity/vulnerability and hence, dunes' maintenance/emplacement has been considered as an effective coastal protection measure included among possible "Disaster Risk Reduction" (DRR) strategies in several European directives [28–32]. In fact, dune ridges protect large sections of low-lying coasts against flooding during extreme storms [9,33,34], and hence, lateral dune continuity and level of fragmentation are extremely relevant [35–38].

This paper is the first one that deals with the characterization and evolution of all dune systems along the Mediterranean coast of Andalusia (South of Spain), and this is a first step to assess their great ecological value, sensitivity and relevance in coastal flood protection. Different dune types have been mapped as well as their level of fragmentation (by means of a new index proposed in this paper) and present human occupation and evolution from 1977 to 2016. Results obtained are of relevance to enhance the general database on dune characteristics along the Mediterranean coast of Andalusia and the possibility of use ecosystem-based solutions in coastal protection along with, or instead of, traditional engineering approaches [27,28].

#### **2. Study Area**

Located in South Spain, the Mediterranean coast of Andalusia administratively belongs to the provinces of Cádiz, Málaga, Granada and Almería (Figure 1). It has a prevailing rectilinear E-W outline, with two NE-SW easterly facing sectors, i.e., the Gibraltar Strait and the Almería easternmost coastal sector (Figure 1). It is a micro-tidal environment (tidal range < 20 cm) with a total length of ca. 546 km, including rocky sectors (ca. 195 km) and intermediate to reflective beaches (ca. 350 km) [39], usually composed of medium to coarse dark sands and/or pebbles. Dune systems, which have a total length of ca. 76 km, are especially observed along the provinces of Cádiz and Almería ( Figures 1 and 2) [25,40,41].

The Betic Range, a tectonically active mountain chain that, at places, reaches relevant elevations higher than 2200 m above sea level (m a.s.l.) close to the coast, determines coastal orography and morphology, forming cliffs, embayments and promontories. Several small coastal plains are especially extended at the mouth of short rivers and *ramblas* (seasonal streams) draining the chain, the most important being the Guadiaro, Guadalhorce, Guadalfeo, Adra and Andarax rivers (Figures 1 and 2). In the past decades, river basin regulation plans devoted to water management for tourist and agricultural purposes have enhanced the construction of dams and reservoirs that have reduced sediment supplies to the coast and have promoted coastal retreat in most deltas [39,42–44].

Large coastal towns are Málaga (>500,000 inhabitants), Almería (ca. 200,000 inhabitants) and the tourist towns along the western part of Costa del Sol area, namely Marbella (150,000 inhabitants), Fuengirola (80,000) and Torremolinos (70,000). Málaga is the province that has experienced the most important coastal occupation, in particular due to the construction of structures related to national and international tourism [45]. Main commercial ports are located at Almería, Algeciras, Cádiz and Málaga, and several marinas at Costa del Sol [46–48].

**Figure 1.** Location of the study area and average wind speed roses at 5 points of SIMAR (SImulación MARina), from the wave reanalysis model by Puertos del Estado (PdE).

Concerning weather characteristics, the provinces of Cádiz, Málaga and Granada have a Mediterranean climate with sub-tropical characteristics, with coastal orientation and the Betic Chain favoring average annual temperature of ca. 13 ◦C and, in July and August, the average is 19 ◦C. Annual rainfall ranges from 400 to 900 mm, with the most abundant values observed at Gibraltar Strait. The province of Almería presents a Mediterranean climate with sub-desert characteristics, i.e., rainfall is extremely limited (ca. 200 mm/year), average annual temperature is 21 ◦C and in July–August, temperature is 26 ◦C [49].

The coast is generally exposed to winds blowing from E to W and from NNE to SW in the easternmost part of Andalusia (i.e., at Carboneras, Figure 1), with minimum and maximum velocities ranging from 0.4 to 9.0 m/s [50]. The wave climate and storm energy are very variable [13,50]: the coast of Cádiz province is mainly affected by eastern storms, Málaga, Granada and (partially) Almería provinces are exposed both to western and eastern storms, whereas the easternmost portion of the coast of Almería province is primarily exposed to eastern storms [13,50].

The mean duration of storm events ranges from 0.9 to 7.0 days, despite their intensity. Waves show a clear seasonal behavior with storm conditions being recorded during November–March, i.e., the winter season [42,50,51], with mean values of significant wave height that reaches 5.18 m in extreme storm conditions [50]. A storm characterization for the studied area was developed by Molina et al. [50], using the Energy Flux parameter to classify storm events into five classes, from weak (Class I) to extreme (Class V). They observed that the most energetic area was the central part of the Mediterranean Andalusian coast, i.e., the coast between Málaga and Almería provinces, highly exposed to storms belonging to all classes, and specially to most energetic ones that can have a great impact on both natural and urbanized sectors [50].

Due to shoreline orientation, predominant easterly winds (Figure 1) and associated storm waves give rise to sea wave conditions generating a prevailing westward littoral drift [51]; meanwhile, an opposing drift is particularly important in certain coastal sectors and/or periods [42,50].

**Figure 2.** Location of the studied dune systems. Natural protected areas are in green. The capital letters A–G in the main subplot refer to the zoomed areas in the other subplots.

#### **3. Materials and Methods**

Aerial orthophotographs dated 1977, 2001 and 2016 (Table 1) were used to map different dune systems, and to quantify their surface evolution, level of fragmentation and the progression of human occupation. Aerial orthophotographs were obtained from the Web Map Services (WMS) of the Open Geospatial Consortium (OGC) of the Andalusia Regional Administration [52].


**Table 1.** Aerial orthophotos used [52], 2001 and 2016 orthophotos are from Plan Nacional de Ortofotografía Aérea (PNOA).

Orthophotographs were elaborated within a GIS project (reference system WGS84–UTM 30N) by means of the ArcMap application from ArcGIS Desktop, Release 10 Redlands, CA: Environmental Systems Research Institute. All dune systems with a minimum longshore seaward front of 100 m in length were mapped, summing a total of ca. 106 km in 1977, including 53 systems (Figure 2). Within each system, three dune environments were mapped according to Sanjaume Saumel and Gracia Prieto [25], who defined, on the base of the most important coastal dune habitats in Spain [53], i.e., (i) embryo and mobile dunes (Type I), (ii) grass-fixed dunes (Type II) and (iii) stabilized dunes (Type III). Coastal dune habitats described at the study area corresponded to the Sites of Community Importance (SCI's) of the European Commission Habitat Directive listed in Table 2.

**Table 2.** Sites of Community Importance (SCI's) described in the study area and their correspondence with the dune typologies mapped in this work [54].


The first group (Type I) comprises embryo and mobile dunes, which are the first band of colonizing vegetation and the first important continuous sandy relief. The second group (Type II) comprises grass-fixed dunes, which develop in a more stable soil and form a more continuous plant cover based on lawns or even some woody plants and bushes. The third group (Type III) comprises the stabilized dunes, is the innermost band of the dune system, and is made up of fully fixed vegetated dunes, with structured and stabilized soils. Its vegetation evolves into forests and a dense and diverse vegetation cover is developed.

The main characteristics used to distinguish between each dune type was the color and vegetation density, so that as systems evolve, color darkens and plant density increases, i.e., embryo and mobile dunes are often called "white" or "yellow" dunes and fixed dunes are called "grey" dunes because of their characteristic color. Díez-Garretas et al. [55], in their study on spatio-temporal changes of coastal ecosystems in Southern Iberian Peninsula (Spain), used a similar classification, taking into account the phytosociological plant communities present at the location studied and the habitat code. They related the habitat code with the ecological units present in their study, including mobile dunes, semi-fixed dunes and stable dunes. Pintó et al. [56] recognized the distinct habitats present in their study area and related them to the sea-to-land ecological gradient and the Habitats of Community Interest. Their classification is more detailed, attending more to morphological than ecological criteria.

The position, evolution and fragmentation of the dune toe position was also reconstructed, and the latter aspect favors dune sensitivity to storms' impacts [40,57–59]. Further, the total surface of each one of the 53 system and dunes' surfaces occupied by human structures/interventions was calculated. The proxy used to map the dune toe was the seaward dune vegetation line, manually detected by a GIS operator [13,39,48,60–62]. To calculate dune fragmentation, a database was obtained for each dune system containing three shape files: the first file included a polyline of the total length of the dune toe line, the second file included the length of all breaks observed along the dune toe line of each system, and a third file, which was the result of the differences between the two previous shape files. The level of fragmentation was calculated by determining the ratio between the length of all breaks and the whole dune toe length at each dune system and year. These values were normalized according to a constant length of 100 m by dividing the total length of all breaks in the shorefront dune toe ("l") by the entire length of the dune toe ("L"):

$$\mathbf{F} = \frac{1}{\mathbf{L}}\tag{1}$$

The F Index is a new index proposed for the first time in this paper vaguely based on the coefficient of infrastructural impact "K" [63]. It was applied along unitary coastal sectors of 100 m in length in order to reduce the importance of dune seaward length. The F Index was calculated for the systems present in all investigated periods (37 out of 53 systems), that is, the dune systems that disappeared in the second period were not taken into account to avoid interpreting a decrease in fragmentation when, in the reality, the entire system was lost. Values of the F Index used to express the fragmentation level were classified into three classes using the Natural Breaks Function [64], from Class 1 ("Null or very low fragmentation", 0.00 < F < 0.06), Class 2 ("Medium fragmentation", 0.06 < F < 0.16) to Class 3 ("High fragmentation", 0.16 < F < 0.41).

#### **4. Results**

#### *4.1. Dune Systems' Distribution and Evolution*

Of the 53 dune systems investigated in the Mediterranean coast of Andalusia, 10 belonged to natural protected areas and, from an administrative point of view, 10 were located in Cádiz province, 18 in Málaga, 2 in Granada and 23 in Almería province (Figure 2, Appendix A Table A1). In Cádiz province, dune systems were equally divided between the Algeciras Bay and an exposed, rectilinear shoreline, including both natural and urbanized areas (Figure 2A). In Málaga province, they were mainly located at the westernmost part of the littoral (Figure 2B), and south of the Guadalhorce river mouth and west of the Vélez river delta (Figure 2C). In Granada province, only 2 dune systems were observed, located in an area updrift of the port of Motril and at Carchuna (Figure 2C). In Almería province, 5 systems were located close to delta areas, namely at Adra and, especially, at Andarax river delta (Figure 2E,F), and 8 were located at rectilinear coastal sectors limited by ports, promontories or river deltas (Figure 2E–G). Very developed dune systems were located in the relevant protected area of Punta Entinas-El Sabinar (Figure 2E); meanwhile, several systems were located at the easternmost area of Almería province and the most relevant system was observed in a large pocket beach (Los Genoveses) (Figure 2F,G).

A total of 15 dune systems disappeared from 1977 to 2016, 7 of them located in Málaga, 7 in Almería and 1 in Cádiz provinces. Dune systems' extension was changing during the periods studied without a clear trend; meanwhile, a clear decrease in size was evident for the three largest dune systems (Appendix A Table A2). Surfaces of "Embryo and mobile dunes" (Type I), "Grass-fixed dunes" (Type II) and "Stabilized dunes" (Type III) were calculated within each one of the 53 dune systems and per each time span considered. The progressive decrease of typologies I and II was observed, meanwhile, "Stabilized dunes" (Type III) recorded a decrease from 1977 to 2001 and a slight increase from 2001 to 2016 (Figure 3).

**Figure 3.** Yearly surface values of each dune typology, i.e., "Embryo and mobile dunes" (Type I), "Grass-fixed dunes" (Type II) and "Stabilized dunes" (Type III). Value on top represents the surface of all dune typologies and in brackets the surface that was lost with respect to the previous year is reported.

The distribution of the different dune typologies within each dune system varied during the studied period (Figure 4). At all provinces (but Cádiz), a reduction of all dune typologies was recorded during the 1977–2001 period; meanwhile, a decrease in all provinces (but Almería) of types I and III and an increase of Type II was recorded in the 2001–2016 period (Figure 4).

**Figure 4.** Distribution of dune surface typologies, i.e., "Embryo and mobile dunes" (Type I), "Grass-fixed dunes" (Type II) and "Stabilized dunes" (Type III), by province during the studied periods. The value on top represents the sum of dune surfaces (in m2) and in brackets the value of lost dune surface with respect to the previous year is presented.

The comparison of the total amount of eroded/accreted surfaces recorded in each system for the periods 1977–2001, 2001–2016 and 1977–2016 showed a clear negative balance for 49 systems and a positive one for 4 systems (Figure 3). The dune systems that showed a positive balance were located at Playa del Rinconcillo (System no. 2, 38,884.4 m2), at the Guadarranque (no. 5, 19,262.8 m2) and at the Guadalquitón (no. 9, 260,531.7 m2) rivers' mouths in Cádiz province (Figure 2A), and in the Albufera de Adra (no. 31, 6708.5 m2), a natural protected area at the Adra river delta (Almería province, Figures 1 and 2E). The most eroded dune systems were located at Ensenada de San Miguel (System no. 34, <sup>−</sup>557,765.0 m2), Punta Entinas–El Sabinar (no. 35, <sup>−</sup>4,166,157.9 m2) and Vera (no. 53, <sup>−</sup>567,841.2 m2) areas, in Almería province.

Comparing the evolution of each system in the 1977–2001 and 2001–2016 periods, it was observed that 3 systems recorded accretion and 23 erosion in both periods, and the others showed different behaviors. Regarding the distribution of these records, the 3 accreting dune systems were located in Cádiz province and most of the dune systems that recorded erosion or disappeared were located in Málaga province, and dunes in Granada province presented erosion for both periods. In Almería province, the systems located in Almería Bay and at the easternmost part of the province presented a negative trend for both periods, and the group located from the Adra river delta to Ensenada de San Miguel presented erosion and then accretion. The two dune systems located at Roquetas de Mar disappeared in 2001–2016 (Figures 2E and 5).

**Figure 5.** Dune systems urbanized in Málaga and Almería provinces. System no. 14 Playa Nueva Andalucía in 1977 (**A**) and 2016 (**B**), and System no. 36 Playa de Roquetas in 1977 (**C**) and 2016 (**D**).

#### *4.2. Anthropic Occupation Evolution*

Surfaces occupied by human structures/interventions were calculated within each one and per each year of the 53 dune systems (Figure 6). The greatest increase (ca. 2.3 <sup>×</sup> 106 m2) was observed in the 1977–2001 period.

**Figure 6.** Evolution of the surfaces occupied by anthropic interventions during the studied periods. The number on top of histograms represents the total value of surface occupied, and in brackets shows the increment recorded among successive periods.

Comparing the evolution of human occupation of each system in the 1977–2001 and 2001–2016 periods, it was observed that 21 out of 53 systems presented an increase of human occupation in both periods and 2 systems a decrease due to the removal of small installations (Figure 7). Further, 4 systems recorded an increase in the first period and a decrease in the second due to the urbanization of a part of the dune system, and the removal of small installations, and the opposite was true for 1 system due to coastal erosion problems since the shoreline retreatment forced the removal of human structures (Figure 7).

**Figure 7.** Removal of small constructions in Cádiz and croplands in Almería provinces. System no. 9 Guadalquitón 1977 (**A**), 2016 (**B**), with a decrease of the occupation area of 112.90 m<sup>2</sup> in 1977–2001 and 263.31 m2 in 2001–2016. System no. 43 Las Algaidas-Las Marinas in 1977 (**C**) and 2016 (**D**), with a decrease of 117.14 m2 in 1977–2001 and 184.07 m<sup>2</sup> in 2001–2016.

At places in System no. 2 Playa del Rinconcillo (in Cádiz province), the increase of dune surface and anthropic occupation was linked to the formation of a new beach at the northern side of the port of Algeciras (Figure 8). Summing up, the decrease of occupation was essentially due to the removal of buildings and was always very small.

**Figure 8.** System no. 2 Playa del Rinconcillo registered an increase of dune surface of 732.26 m2 in 2001–2016. (**A**) 1977, (**B**) 2001 and (**C**) 2016.

#### *4.3. Dune Fragmentation*

Analysis of the dune toe fragmentation was carried out for such systems (37 out of 53) that were observed in all investigated periods and a general increase of fragmentation was evident (Figure 9), confirming the trend observed for the evolution of human occupation.

**Figure 9.** Fragmentation Index. The number on top of histograms represents the total value of fragmentation per year and in brackets is the increment recorded among successive periods.

Considering the 1977–2016 timespan, 23 dune systems presented an increase of fragmentation, 3 systems recorded a decrease (they were located in erosive coastal sectors within natural protected areas) and 11 presented no variations. The two systems that recorded a major increment of fragmentation were Punta del Río (no. 41), in Almería province, and Playa de las Chapas (no. 20), in Málaga province, with an increase of +0.25 and +0.22, respectively. Conversely, the two systems that recorded the major decrease of fragmentation were Playa de Río Real (no. 16) and Playa de la Misericordia (no. 26) in Málaga province, with a decrease of −0.20 and −0.09, respectively. Comparing the evolution of fragmentation at each dune system in the 1977–2001 and 2001–2016 periods, it was observed that only 7 out of 23 presented an increase of the fragmentation at both periods, in general due to coastal zone urbanization (Figure 10A–C). Only 1 dune system showed a decrease of fragmentation (no. 16) (Figure 10D–F), and 8 presented no variation in both periods.

**Figure 10.** System no. 20 Playa de las Chapas, in Málaga province, recorded an increase of fragmentation due to the increment of anthropic pressure, changing from (**A**) Class 1 (F = 0.003) in 1977 to (**B**) Class 2 (F = 0.12) in 2001 and to (**C**) Class 3 (F = 0.22) in 2016. System no. 16 Playa de Río Real, in Málaga province, where the modification of the coastal zone and the emplacement of a touristic urbanization produced the destruction of the already fragmented dune system. Remnant dune systems presented a lower Fragmentation Index, changing from (**D**) Class 3 (F = 0.20) in 1977 to (**E**) Class 2 (F = 0.15) in 2001 and to (**F**) Class 1 (F = 0.0) in 2016. Red line represents dune toe position.

Other dune systems presented a different behavior at both periods: 4 systems recorded an increase in the first period and a decrease in the second and the opposite was true for 1 system (Figure 11). In general, the increase in fragmentation occurred along with the increase of urbanization and anthropic

pressure, while the opposite was observed in natural protected areas. At places where systems were already fragmented, their erosion implied a reduction in their fragmentation since: (i) very fragmented sectors often disappeared and the remaining ones presented low fragmentation (Figure 10D–F and Figure 11A,B) and (ii) coastal erosion produced the loss of the most fragmented part of dune toe (Figure 11B,C). An increase of fragmentation in 7 dune systems was due to the increment of erosion processes and/or the formation of pedestrian pathways.

**Figure 11.** System no. 10 Torreguadiaro (in Cádiz province) recorded a slight decrease of fragmentation in the 1977–2001 period and an increase in the 2001–2016 period due to the reduction of a high fragmented dune sector that was replaced by the port structure. The dune system located in front of the natural protected Torreguadiaro Lagoon was not modified. In the 2001–2016 period, the growth of new dunes at the northern part of the system and the increase of pathways at both sides of it was observed. This resulted in a decrease of fragmentation that changed from (**A**) Class 1 (F = 0.08) in 1977 to (**B**) Class 2 (F = 0.03) in 2001 and an increase of fragmentation to (**C**) Class 3 (F = 0.20) in 2016. Red line represents dune toe position.

#### **5. Discussion**

#### *5.1. Erosional Dune Systems*

Erosion or complete disappearance of dune systems can be produced by human activities or natural processes [65–67]. Anthropic impacts were related to: (i) urban developments, mainly due to the coastal tourist demand, and the associated opening of pathways on dune ridges, which was especially evident in Málaga province (Figure 5) [41,46,48,55,68], (ii) dunes' occupation due to the demand for agricultural uses, as observed at different locations in Andalusia, and reported by References [67,69–73] in other Mediterranean Spanish areas (in Catalonia, [56]) or on the Mediterranean coast of Morocco [74] and (iii) the decrease of sediments' inputs to coastal environments due to the construction of ports and harbors, as observed along the study area by Malvárez et al. [46] and Manno et al. [48], and the reduction of the sedimentary load of rivers due to the construction of dams in river basins, especially in Málaga and Almería provinces [43,46,75], also observed in other Mediterranean rivers, e.g., for the Ebro [76] and the Arno [77] rivers.

Among natural processes, there are the impacts of chronic erosion processes and of extreme storms, the impacts of which are often enhanced by climatic change-related processes, e.g., an increase of storm intensity and frequency and Sea Level Rise [11,35,37,67,78–83]. Specifically, for the studied area, storm characterization was described by Guisado et al. [42] and Molina et al. [50]; meanwhile, it seems that Sea Level Rise is not relevant at the studied area [84–86].

Of the 53 dune systems studied, all but 4 recorded a reduction of their surface, or even disappeared, and this was especially evident where the systems were affected by hard human interventions [41,55,68–73] and, secondarily, by shoreline erosion [71,87,88]. The greatest loss of dune surface was recorded in the 1977–2001 period due to the massive urban occupation of coastal areas, although in the 2001–2016 period, a decrease in the loss of dunes' surfaces was observed because the main causes of their destruction recorded in the previous period partially ceased. Cases of disappearance due to urban occupation were still observed, especially in Málaga province [41], but the anthropic pressure derived from the tourist use of beaches and the decrease in river contributions were not so evident as in the 1977–2001 period [46,68–73].

The loss of dune surface was at places and times linked to the progressive fragmentation of the dune toe (i.e., the increase of dune discontinuity), which is a factor that has to be considered in order to estimate coastal and dune vulnerability [40,57–59] since a fragmented dune system is more vulnerable to temporary flooding and hence, it is less effective against storm surges [35,40,58,59,89–91]. In this study, the most fragmented (and hence susceptible sectors) were observed at the west side of the Andarax river delta in Almería province (no. 41, Figures 1 and 2), which was the most fragmented dune system located in a natural area (Appendix A Table A2) and the system at Las Chapas beach in Málaga province (no. 20, Figures 1, 2 and 10), located in a strongly developed urban area.

At almost all sectors, dunes' fragmentation was mainly due to the opening of pathways and to their progressive expansion due to marine- and wind-induced erosion processes, as also observed by Gracia et al. [40], Pintó et al. [56], Rangel-Buitrago and Anfuso [58] and Rizzo et al. [59]. Due to the accuracy of the orthophotos used in this study, dune discontinuities caused by overwash processes were only detected at few places (Figure 12). Such processes were distinguished from other types of fragmentation due to the absence of vegetation at the areas presenting the characteristic shape of a washover fan; meanwhile, pathways showed narrow rectilinear shapes.

Summing up, the majority of the dune systems that showed an aerial decrease were affected by anthropic factors, highlighting the importance of urban and agricultural occupations that were very relevant in Málaga and Almería administrative districts.

**Figure 12.** Details of the central area of System no. 9 Guadalquitón (Cádiz province), where washover fans and pathways were observed. Red line represents dune toe position. (**A**) 1977, (**B**) 2001 and (**C**) 2016.

#### *5.2. Accretional Dune Systems*

Conditions for dune formation and development were discussed by a large number of authors who agree that the temporal variation of the sedimentary contribution and the wind regimes are the most important factors controlling the beach–dune system relationship [65,92–95].

The increase of systems located in the Bay of Algeciras (Figure 2A) was associated with the sedimentation processes recorded in such beaches [96] that receive the sediment supplies of the Palmones and Guadarranque rivers [73]. Such beaches are located next to two large coastal protection structures that promote sedimentation processes. In the case of the system observed at El Rinconcillo in 2001, it began to form after the expansion of the port of Algeciras (Figure 8). Instead, systems at Guadalquitón and Albufera de Adra (Figure 2) were located in areas that registered an important erosion [39,96] and a significant human occupation linked to urban development in the case of Guadalquitón and intense agricultural occupation in Albufera de Adra. The Guadalquitón dune system recorded the highest increase during the 1977–2001 period, and it was due to the degradation of the vegetation that facilitated the inland dune migration. The formation of large mobile dunes in this area was also due to strong east winds (Figure 1), especially on the east-facing beaches [73]. In the case of the Albufera de Adra system, an important loss of dune surface in the 1977–2001 period was caused by shoreline erosion and the significant anthropic pressure (intense agricultural activities) [39,43,96]; however, the sedimentation produced at the north side of the system [39,96] supported the development of mobile dunes.

#### *5.3. Dune Types' Evolution*

Several dune systems studied in this paper were described by different authors [41,55,68–73,87,88], but none of them provided a description of all dune systems along the whole Mediterranean coast of Andalusia.

Unlike tidal-influenced coasts, in which the sedimentary contribution can be obtained through periodic exposure of the intertidal plain, on micro-tidal coasts such as the Mediterranean one, the beach itself is the main source of sedimentary contribution to the dune systems. In addition, when the beaches are composed of gravels, as is the case of many beaches of Málaga and Almería, it is more difficult to ascertain the source of sandy sediment necessary for the dune systems, so the rivers become the main sediment suppliers of the system [72]. As stated before, river systems at the Mediterranean coast of Andalusia are mostly short or seasonal streams and, in general, provide a coarse grain size on the beaches. Further, the accentuated relief observed nearby the coast and the presence of reflective beaches represent great limitations for the development of coastal dunes [72].

Further points to be taken into consideration are the intensity and direction of predominant winds that, to be effective in dune formation, should be shore normal. Due to their coastal orientation, which is normal to predominant winds, the provinces of Cadiz (especially) and Almeria constitute areas favorable for dune formation. According to Bardají et al. [72] and Gracia et al. [73], the central part of the Andalusia coastline is parallel to predominant winds that give rise to a relevant longshore transport that supplies different dune systems, e.g., at Artola-Cabopino [72,88].

Analyzing the evolution of each dune typology is of relevance since each typology represents a clear evolution state from Embryo and mobile dunes (Type I) to Stabilized dunes (Type III) [65,92]. The increment over the 1977–2016 period of the Stabilized dunes (Type III) (Figure 3) was due to the progressive evolution of Grass-fixed dunes (Type II), a natural process described by Hesp [65,92].

Surface variations of the different types of dunes' systems were relatively homogeneous (Figure 4). With the exception of the province of Cádiz, the rest of the provinces showed a decrease of the three types of dunes in the first period and, in the second period, a decrease of types I and III and an increase of Type II in all provinces except Almería. The general decrease recorded in the period 1977–2001 was mainly due to urban occupation, intensive agricultural exploitation and the extraction of sand—such activities were not regulated until the approval of the Coastal Law in 1988 [41,46,47,55,68–73]. Dune destruction was especially evident in Málaga and Almería provinces, where entire dune systems disappeared: in Málaga province, a total surface of 1,766,711 m<sup>2</sup> was lost, of which ca. 1 <sup>×</sup> 10<sup>6</sup> m2 were Type II dunes and ca. 600,000 m<sup>2</sup> were Type III dunes, and in Almería, ca. 56,300,000 m2 of dune surface was lost, of which ca. 4,360,000 m<sup>2</sup> were Type II dunes. Some examples of papers that quantified the loss of dune surfaces in specific areas were by Viciana Martínez-Lage [71], who

quantified a loss of 262 ha of dunes in Punta Entinas–El Sabinar, in Almería province, due to sand extractions, or Gómez Zotano [41], who quantified a reduction of 44.5% of the dune surface in Saladillo area, in Málaga province, during the 1956–2007 period.

The increase, in the 2001–2016 period, of the Type II in Málaga province was linked to the degradation of Type III dunes, especially evident in an area west of Marbella (Figure 2B) that was greatly impacted by urban developments, a quite common trend in Málaga province [41,46,47]. The increase of Type III in Almería was due to the stabilization of Type II dunes, especially in the area from Albufera de Adra to Almerimar and at Cabo de Gata (Figure 2E,F), which are areas where the shoreline is stable [39]. Overall, in Cádiz province, a slight increase of Type III was observed, and the other dunes' types recorded small variations (Figure 4). Such behavior was due to the low human pressure, the stable or even accreting conditions of the area [39,96] and the action of strong east winds (Figure 1) that favored dunes' growth and mobility [73].

#### **6. Conclusions**

This study analyzed the evolution of the dune systems along the Mediterranean coast of Andalusia, focusing on their characterization, level of fragmentation and anthropic occupation, for the 1977–2001 and 2001–2016 periods. Within a GIS project, there were 53 dune systems mapped that summed a total length of ca. 106 km in 1977 and ca. 76 km in 2001 and 2016.

Of the 53 dune systems, all but 4 recorded a reduction of their surface, or even disappeared, and this was especially evident in 1977–2001 when dune systems were affected by hard human interventions, such as the emplacement of buildings and touristic constructions, especially at Málaga province, and agricultural expansion at Almería province, and secondarily, at places by shoreline erosion processes.

Dunes' loss was at places and times linked to the progressive fragmentation of the dune toe, mainly due to the opening of pathways and to their progressive expansion due to marine- and wind-induced erosion processes. An increase of dunes' surface was observed in both natural and anthropic areas in Cádiz and Almería provinces, in accreting and stable beaches, usually on the updrift side of ports or due to strong east winds on the east-facing beaches.

Concerning the evolution of the Embryo and mobile dunes (Type I), Grass-fixed dunes (Type II) and Stabilized dunes (Type III), most of the provinces showed a decrease of the three types of dunes in the 1977–2001 period and, in the 2001–2016 period, a decrease of types I and III and an increase of Type II in all provinces. The increase of Type II dunes was linked to the degradation of Type III, observed in the 2001–2016 period at very anthropized areas; meanwhile, an increase of Type III was observed in stable and accreting areas.

Results obtained could be used to enhance the general database on dune characteristics along the Mediterranean coast of Andalusia and the possibility of utilizing ecosystem-based solutions in coastal protection, along with, or instead of, measures based on traditional engineering approaches. The methodology used in this study could be applied in other locations with a similar database.

**Author Contributions:** Data curation, R.M., G.M. and C.L.R.; Formal analysis, R.M., G.M., C.L.R. and G.A.; Methodology, R.M., G.M., C.L.R. and G.A.; Software, R.M., C.L.R. and G.M.; Geological and geomorphological supervision, G.M.; Hydraulic analysis, C.L.R.; Supervision, G.A.; Writing—original draft preparation, R.M., G.M., C.L.R. and G.A.; Writing—review and editing, G.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** This work is a contribution to the Andalusia Research Group PAI RNM-328 and to the PROPLAYAS network and has been partially developed at the Centro Andaluz de Ciencia y Tecnología Marinas (CACYTMAR), Puerto Real (Cádiz, Spain).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Appendix A**


**Table A1.** Name and protection typology of each dune system and balance for the entire studied period.


**Table A1.** *Cont.*

Typologies of protection: (1) Natural Park, (2) Natural Site, (3) Special Plan for the Protection of the Physical Environment, (4) Natural Monument, (5) Natural Reserve. Fragmentation index was not calculated for periods where the dune system disappeared.


#### **Table A2.** Results obtained at each dune system.


**Table A2.** *Cont.*

Fragmentation index was not calculated for periods where the dune system disappeared.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
