**Geomorphology of Canyon Outlets in Zrmanja River Estuary and Its E**ff**ect on the Holocene Flooding of Semi-enclosed Basins (the Novigrad and Karin Seas, Eastern Adriatic)**

**Ozren Hasan 1,\*, Slobodan Miko 1, Dea Brunovi´c 1, George Papatheodorou 2, Dimitris Christodolou 2, Nikolina Ilijani´c <sup>1</sup> and Maria Geraga <sup>2</sup>**


Received: 28 August 2020; Accepted: 1 October 2020; Published: 10 October 2020

**Abstract:** Detailed multi-beam bathymetry, sub-bottom acoustic, and side-scan sonar observations of submerged canyons with tufa barriers were used to characterize the Zrmanja River karst estuary on the eastern Adriatic coast, Croatia. This unique karst environment consists of two submerged karst basins (Novigrad Sea and Karin Sea) that are connected with river canyons named Novsko Ždrilo and Karinsko Ždrilo. The combined use of high-resolution geophysical data with legacy topography and bathymetry data in a GIS environment allowed for the description and interpretation of this geomorphological setting in relation to the Holocene sea-level rise. The tufa barriers had a predominant influence on the Holocene flooding dynamics of the canyons and karst basins. Here, we describe the possible river pathways from the basins during the lowstand and the formation of a lengthening estuary during the Holocene sea-level rise. Based on the analyzed morphologies and the relative sea-level curve for the Adriatic Sea, the flooding of the Novsko Ždrilo occurred 9200 years before present (BP) and Karinsko Ždrilo was flooded after 8400 years BP. The combination of high-resolution geophysical methods gave an accurate representation of the karst estuarine seafloor and the flooding of semi-isolated basins due to sea-level rise.

**Keywords:** eastern Adriatic coast; estuary; sea-level rise; tufa; multi-beam; sub-bottom profiler; holocene

#### **1. Introduction**

The rapid development of swath acoustic techniques has enabled seabed mapping at high spatial resolutions and accuracies. The results of the high-resolution acoustic technologies, such as multibeam echosounder (MBES) bathymetry, MBES backscattering, sub-bottom profiling (SBP), side-scan sonar (SSS), and their derivatives, represent an excellent platform for geomorphological and geological classifications of the seabed [1–3]. Despite the problems regarding the acquisition of the data in shallow water or problems related to the incident angle due to steep slopes [4], MBES bathymetry data and quantitative terrain indices, such as the slope, curvature, and roughness, prove to be very useful as a tool for seabed characterization (best summarized by [5,6]). Backscatter data of the acoustic intensity scattered by the seabed collected during MBES surveys [7] gives us valuable information about bottom-type sediment characteristics [1,8]. GIS-based classification techniques and packages for MBES bathymetry and MBES backscattering data have become numerous and available, and have undergone significant development and improvements in the past decade [7,9–12]. Sub-bottom profiles, in conjunction with SSS and bathymetry data and its derivatives, can give insight into depositional dynamics, therefore enabling geomorphological reconstructions [3,13].

This study investigated a unique karst geomorphological setting that was recognized along the Croatian karst on the eastern Adriatic coast; it consists of two canyon-type outlets (Novsko Ždrilo and Karinsko Ždrilo) connecting two semi-isolated basins (Novigrad and Karin Seas) to each other and the open sea. We used the above-mentioned acoustic techniques and the ArcGIS tools available in its Spatial Analyst extension (slope, curvature, maximum likelihood classification (ML), interactive unsupervised classification (uISO), and segment mean shift classification (SMS)), as well as the ArcGIS package Benthic Terrain Modeller (BTM) [12] in order to estimate the timing of the Holocene marine flooding of the semi-isolated sedimentary basins and to make paleoenvironmental reconstructions. During the last glacial maximum (LGM) lowstand, the Novigrad and Karin Seas acted as karst poljes, i.e., interior valleys [14]. After the post-LGM transgression [15–17], the present-day landscape was formed by creating a submerged karst landscape with drowned canyons (including Zrmanja River) and karst poljes [14]. Geomorphological conditions that could prevent marine flooding of the area in the form of submerged barriers within the narrow channels (called ždrila) had to be determined before the reconstruction was possible. Prior to the present study, the existence of submerged tufa barriers was described by scientist divers [18]. The calcareous tufa barriers are unique karst geomorphological features that are common in the karstic region of Croatia [19–22]. They are formed as rheophilic algae and mosses are encrusted by carbonate in waters with a high concentration of dissolved CaCO3, forming porous sediment that grows laterally and vertically, creating a series of obstacles, lakes, and waterfalls [22,23]. Here, we estimated the timing of the flooding of the Novigrad and Karin Seas based on the available relative sea-level curves [15–17] and multibeam-derived elevations of barriers within the drowned Novsko Ždrilo and Karinsko Ždrilo. The submerged canyon, Novigrad Sea, and the Zrmanja River are the three major elements that influenced the formation of the Holocene Zrmanja estuary and its geomorphology during the sea-level rise. An extensive literature review has shown very few examples of similar semi-enclosed marine environments or lakes, with most studies focusing on biota and habitat mapping [23–26]. Furthermore, these studies were performed in non-karstic environments.

We hypothesized that the geomorphological features present in this area had a predominant influence on the Holocene sea flooding. Furthermore, MBES bathymetry, MBES backscattering, SBP, and SSS data and their derivatives proved to be complementary, enabling both surface and subsurface characterizations of the sediments. Therefore, the present study benefited from the use of various remote sensing techniques combined with GIS classification tools in order to obtain a comprehensive overview of the seafloor's physical and geomorphological properties. This type of study can be applicable to other estuarine environments, as well as other coastal environments.

#### **2. General Geological and Geomorphological Features**

The Novigrad and Karin Seas are two small semi-enclosed marine bays located in northern Dalmatia on the eastern Adriatic coast, Croatia (Figure 1a). The bays are interconnected via a submerged river canyon called Karinsko Ždrilo (KZD) (Figure 1b). The Novigrad Sea is on the northern side, connected to Velebit Channel and open sea via another canyon-like passage called Novsko Ždrilo (NZD). The bays have a flat bottom with average depths of 26 to 30 m in the Novigrad Sea and 12 to 13 m in the Karin Sea [27]. Both canyons are oriented in the NNW–SSE direction (Figure 1b). Novsko Ždrilo is 3750 m long and 150 m to 250 m wide at sea level. Its steep slopes rise to 150 m above sea level (a.s.l.). The water depth within the NZD channel is 25 to 46 m. Karinsko Ždrilo has a less dramatic morphology with slopes rising 20 to 40 m a.s.l., whereas the canyon width is 115 m to 250 m at sea level. The water depth in KZD ranges from 15 to 25 m. Due to the region's geographic importance in the connection of the interior with northern Dalmatia, two bridges cross NZD. The larger highway bridge is located at the central part of NZD, while the smaller local Maslenica Bridge is at the southern end of the channel (Figure 1b).

**Figure 1.** (**a**) General study site location in the red square. (**b**) Multibeam echosounder (MBES) bathymetry data for Novsko Ždrilo (NZD) and Karinsko Ždrilo (KZD) draped over a shaded relief and an orthophoto map [27]. The study areas are highlighted with red squares and presented in detail in Figures 2 and 3. (**c**) Geological map with the plotted sub-bottom profiling (SBP) track lines.

Freshwater enters the Karin Sea via several periodical torrential rivers, but mainly through Karišnica and Bijeli Potok. The karst river Zrmanja, with a total length of 69 km, feeds the Novigrad Sea through a river canyon. It forms an estuary, which extends from Novsko Ždrilo up to 14 km inland until it reaches the tufa waterfall Jankovi´ca Buk [28]. The estuary is highly stratified most of the year [29,30]. The river is characterized by many tufa barriers that were formed as autogenous calcite deposits on macrophytes and microphytes [31]. Tufa barriers across the Zrmanja River were previously studied from biological, geochemical, fossil evolution, and chronological points of view [21,31–33]. Tufa growth forms barriers, barrages, or waterfalls across the river valley. When carbonate-rich water falls vertically, vertical cascade tufas result. If water flows over steep slopes, the tufa occurs in the shape of fans, cones, or mounds [19,23,32]. Often, tufa barrier growth forms dams and lakes. Some barriers in the Zrmanja River are still active and some are fossilized. Available studies from the Dinaric karst and the Zrmanja River show that the majority of them are of the Holocene age, with the most intensive growth in the last 7000 years [31,33,34]. The existence of tufa barriers within Novsko Ždrilo has already been confirmed by scientist divers [18], who claimed barrier heights of 10–20 m. The river's complex karst hydrological characteristics are well described by Bonacci [35]. Despite the numerous monitoring measurements, it was not possible to determine the exact karst aquifer (catchment) extent, but it is suggested that the Zrmanja River is connected with the neighboring Krka River through complex karst underground flows [35]. There are also several periodical rivers feeding the Novigrad Sea from the west and south. The composition of sedimentary records has been analyzed from a geochemical perspective on short cores from the Novigrad Sea and Zrmanja River in order to determine the distribution of trace elements and differentiate the anthropogenic impacts from the natural background values [36–38].

**Figure 2.** (**a**) Bathymetry image of NZD draped over the DTM and the orthophoto image [27]. Prominent barriers are marked with numbers I–V. (**b**–**f**) Profiles showing the barriers and deepest parts of the NZD canyon with steep and high sides with depths of over 45 m and elevations of up to 150 m a.s.l. (**g**) The lowest profile through NZD showing a possible path of the water flow (out of and into the Novigrad Sea) during the lowstand and sea-level rise. The same barriers from (**a**) are marked with numbers I–V. The two bridges across NZD are illustrated as references.

As a result of their isolated geomorphological location, sea currents and waves have a slight influence on the bays [39]. Conversely, the Zrmanja River brings 2–3 times more freshwater annually than the total volume of the Novigrad Sea [36]. Together with the freshwater flowing into the Karin Sea by Karišnica and Bijeli Potok, this amount of water can cause strong outflow currents in narrow channels, such as NZD and KZD.

The study area is a part of the Croatian karst Dinarides and consists of a thick carbonate succession deposited from the Late Palaeozoic to the Eocene. During the period from the Mesozoic to the Cenozoic, the area was a part of the large Adriatic–Dinaric Carbonate Platform (ADCP, [40,41]). The disintegration of the ADCP started in the Late Cretaceous and is characterized by the development of flysch basins and carbonate deposition on the margins. The transition from the Cretaceous to the Paleogene was marked by the regional emergence of the entire platform, followed by dynamic tectonics in the Paleogene. The final uplift of the entire Dinaric area culminated in the Oligocene

and the Miocene as a result of the collision of Adriatic and Dinaric segments of the Platform [40,41]. The geological background of the studied area consists mainly of Mesozoic, Paleogene, and Neogene carbonates and clastic deposits [42–46] (Figure 1c), with *terra rossa* soils and cambisols on limestone as the dominant soil types. The Mesozoic comprises Jurassic limestones and dolomites at the base, with a succession of Cretaceous limestones, dolomites, and carbonate breccias. Eocene limestones; dolomites and flysch; Oligocene limestones, conglomerates, and marls transgressively overlie Mesozoic rocks [42–46] (Figure 1c). Occurrences and deposits of bauxites can be found in the study area and the wider region, as well as a disused bauxite processing factory in the city of Obrovac [42,47,48]. Prior to the rapid Pleistocene−Holocene transgression, the present-day Novigrad and Karin Seas acted as karst poljes. They were subsequently submerged, creating a typical drowned karst landscape together with the drowned canyon of the Zrmanja River [14].

**Figure 3.** (**a**) MBES bathymetry image of KZD draped over the DTM and orthophoto image [27]. Prominent barriers are marked with numbers I-IV. (**b**–**f**) Profiles showing the barriers and deepest parts of KZD canyon with depths of 25 m and elevations up to 40 m a.s.l. (**g**) The lowest profile through KZD showing a possible path of water flow (out of and into the Novigrad Sea) during the lowstand and sea-level rise. The barriers from (**a**) are marked with numbers I-IV.

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

To study the morphology of the submarine canyons, we used pinger profiles and shipborne multibeam bathymetric data collected during the two surveys conducted in 2015 and 2019. A 2015 campaign comprised SBP and SSS surveys. We used a 3.5 kHz pinger (ORE), GeoAcoustics Ltd. (Great Yarmouth, UK) GeoPulse Transmitter model 5430A, and a GeoAcoustics Ltd. (Great Yarmouth, UK) GeoPulse Reciever model 5210A. SBP data was logged using a Triton SB-Logger (v 7.3, Triton Imaging Inc., Capitola, CA, USA). The signal penetration was limited by the water depth and shallow limestone bedrock and never exceeded 23 ms. Assuming a sound velocity of 1500 m/s, the vertical signal penetration was up to 17 m. A towfish EG&G 272 TD TVG (EG&G Inc., Gaithersburg, MD, USA) was towed 3 m below the sea surface and SSS data were recorded with an EdgeTech 4100P Topside Processor unit (EdgeTech Inc., Escondido, CA, USA). The positioning was obtained usinga Hemisphere DGPS (Hemisphere GNSS, Inc., Scottsdale, AZ, USA). The equipment was mounted on a 6 m long vessel moving at an average speed of 3.5 knots. Afterward, the SBP data were exported in a SEG-Y format (Society of Exploration Geophysicists Exchange Tape Format) and further interpreted in the Geosuite Allworks software (version 2.6.7., Geo Marine Survey System, Rotterdam, Netherlands).

The second campaign dataset was taken in 2019 comprised MBES mapping of the canyons. For this purpose, we used a WASSP S3 MBES (Furuno ENL, Auckland, New Zealand), which is capable of recording multibeam and backscatter data. It was side-mounted on a vessel moving at an average speed of 3.5 knots. The used MBES system works at a typical frequency of 160 kHz with an effective beam width (angular coverage) of 120 degrees using 224 beams. The beam width is 4.4 × 3.2 (PS/FA) with a transmitting voltage response of 155 dB and a receiving voltage response of -194 dB. The vessel motion was corrected for with a WASSP Sensorbox (Furuno ENL, Auckland, New Zealand) inertial measuring unit (IMU), which makes corrections for the pitch, roll, and heave. The IMU worked in conjunction with the Hemisphere Vector V103 DGPS compass antenna (Hemisphere GNSS, Inc., Scottsdale, AZ, USA) used for positioning. The data were acquired with WASSP CDX software (version 3.9, Furuno ENL, Auckland, New Zealand). Cleaning, processing, and validation of the MBES data were performed with the hydrographic software BeamworX Autoclean (version 2020.1.1.0., BeamworX BV, Utrecht, Netherlands).

#### *Morphometric Analyses*

The final MBES bathymetry and MBES backscatter grids were exported from BeamworX as a 1 m pixel ASCII grid for further analysis in ArcGIS (version 10.2.1, ESRI inc., Redlands, CA, USA). To create a more meaningful base for the GIS analyses, we gridded together the MBES bathymetry data with the available onshore and bathymetry data digitized from topographic maps 1:25,000 [27] as line and point data into a georeferenced digital terrain model (DTM) witha1m pixel size.

Following an extensive literature review [1,3,5,6,12,49], we calculated a range of secondary features to classify and interpret the collected MBES bathymetry, MBES backscatter, and SSS data. ArcGIS with the Spatial Analyst extension to do the DTM, shaded relief, and slope analyses. We used the joined DTM to do the shaded relief and slope analysis, while other morphometric analyses were applied only for the MBES bathymetry/MBES backscatter data. A multidirectional hillshade was created to highlight the morphological features of the terrain, including channels and the bottom morphology. A slope analysis, which is relevant in a geomorphological context linked to the acceleration of currents, the stability of sediments, and erosion [6], was calculated using the standard ArcGIS algorithm proposed by [50]. To describe the heterogeneity of the studied canyons, we used a vector ruggedness measure (VRM). It was calculated using a Benthic Terrain Modeler ArcGIS tool package (version 3.0) [12]. The calculated values are dimensionless and range from 0 (no variation) to 1 (complete variation). Typical values are small (up to 0.4) in natural data [12]. Variations in the range were better observed when calculated for the MBES bathymetry data resampled to a 10 m cell size.

Curvature (the second derivative of the bathymetric surface, or the first derivative of the slope) was calculated using the standard ArcGIS tools according to the method proposed by [51]. The curvature can be calculated parallel to the slope (profile curvature), where it describes the acceleration or deceleration of the flow, or perpendicular to the slope (plan or planiform curvature), which describes the convergence or divergence of the flow. The planiform curvature can be useful when defining ridges, valleys, and slopes along the side of the features [5]. While values close to 0 indicate that the surface is flat, moderate relief values vary from −0.5 to 0.5 and extreme relief values vary between −4 and 4 or more. Physically, the calculated attributes can affect the marine flow, internal waves, and current channeling [12]. Variations in the curvature were also better observed when calculated for the MBES bathymetry data that was resampled to a 10 m cell size.

The morphometric analyses included a combination of the fill DTM, flow direction, and flow accumulation needed to determine the flowline in the channels. The flowline presents the lowest possible pathway for water flowing out of the poljes during the sea-level lowstand, and a pathway for the sea to enter the poljes during the sea-level rise. By applying this methodology, it is possible to determine a correct relative sea level, and consequently, the timing of the Novigrad and Karin Sea drownings.

We made several attempts to classify the backscatter data using ArcGIS tools Spatial analyst tools (version 10.2.1, ESRI inc., Redlands, CA, USA). The best results of the unsupervised backscatter acoustic classifications were achieved using the ML uISO, and SMS. In ML, the mean vector and the covariance matrix characterize each class. A statistical probability can be calculated for each class based on these two cell values. This leads to the determination of the membership of the cell to a specific class [49]. The procedure is based on Bayes' theory of joint probabilities, which accounts for marginal distributions of datasets and their respective internal correlations under the assumption of multivariate normality in N-dimensional Euclidean space [52]. uISO provides a quantitative unsupervised clustering using the functionalities of the Iso Cluster and Maximum Likelihood Classification tools. SMS determines clusters in the MBES backscatter raster by grouping adjacent pixels with similar spectral characteristics. The mean shift algorithm is a non-parametric clustering method for image segmentation. After applying the function, all convergence points are found and clusters are built from them. All convergence points closer than the range defined in the spatial domain are grouped. The number of significant clusters present in the feature space is automatically determined by the number of significant modes detected [53].

#### **4. Results**

The use of available high-resolution bathymetry data (bathymetry, seismic, and side-scan sonar data) incorporated with the already available topographic data enabled us to undertake spatial and morphometric analyses of the Novsko Ždrilo and Karinsko Ždrilo channels.

#### *4.1. Bathymetry and Morphometric Analyses*

Both studied channels were characterized by elongated geometries and steep slopes (Figure 1). The MBES bathymetry results showed a very distinct seabed within the channels, with multiple barriers, which is typical for a karst morphology (Figures 2 and 3). This is well depicted in the profile lines (Figures 2d and 3d), where multiple pronounced barriers are visible. The water depth at the northern entrance to Novsko Ždrilo was 39 m, whereas, on the SSE end of the channel, the water depth was 37 m. There was an S-shaped bend at the northern entrance to NZD, where the first barrier in NZD could be observed (marked as I in Figure 2). After the bend toward the south, there was a deep part of the canyon with depths of over 40 m below sea level (b.s.l.) extending to the next barrier, which rose to 24.5 m b.s.l. in the lowest part (marked as II in Figure 2a,d). The central part of the canyon was shallower, with two joining barriers at depths of 25–30 m b.s.l. in the lowest part of the crown (marked as III in Figure 2a,d). To the south, this shallow part deepened steeply to the deepest part of the canyon below 45 m b.s.l., then rose steeply again to 26 m b.s.l. (marked as IV in Figure 2). This was the most pronounced barrier as the channel deepened beyond this barrier toward

the south, below 40 m b.s.l. The bottom elevated to form two minor barriers near the exit (marked as V in Figure 2), then flattened toward the southern end.

The northern part of KZD was the deepest, with the depth at the northern end reaching 24 m b.s.l. (Figure 3a,d). The bottom gradually elevated in the middle part of the KZD canyon, where the first barrier could be observed (marked as I in Figure 3a,d). Altogether, there were four barriers at the southern part of the canyon (marked as I-IV in Figure 2a,d), all with similar heights (14–16 m b.s.l.) and declining to similar depths (21–22 m b.s.l.). The depth at the southern end of the channel was also the deepest part of the Karin Sea, with a depth of 20.6 m b.s.l.

The slope analysis of the broader area surrounding the channels revealed what was already described: steep slopes rose to 150 m a.s.l. and the continuation of these slopes underwater, which reached almost to the bottom of the channels, where they flattened. The sidewalls in NZD had a maximum inclination of up to 44 degrees on the western side of the canyon close to barrier II (Figures 2a and 4a). In the rest of the canyon, the slopes were still steep and inclined at 28–35◦, flattening further at both channel ends.

**Figure 4.** (**a**) Slope analysis of NZD and its surroundings. (**b**) Vector ruggedness measure of NZD. (**c**) Curvature analysis of the NZD area. (**d**) Planiform curvature calculated for the NZD area; the two bridges across NZD are illustrated as references. (**e**) Slope analysis of KZD and its surroundings. (**f**) Vector ruggedness measure of the KZD. (**g**) Curvature analysis of the KZD area. (**h**) Planiform curvature calculated for the KZD area.

The steepest slopes in KZD barely reached the inclinations observed with the NZD sidewalls, with a maximum inclination of 27 degrees, while the average inclinations were 15 to 20◦. The steepest parts of the KZD were canyon slopes on both sides of the central-to-southern part of this channel (Figure 4e).

We used the BTM VRM to present the surface roughness, where higher values should represent rockier surfaces (Figure 4b,f). In the analysis of the generalized MBES bathymetry data with a 10 m cell size, the higher surface roughness was clearly visible on the sides in most of the NZD channel. In some areas in the southern and northern parts of the channel (red colours in Figure 4b), high roughness areas extend throughout almost the whole width of the channel. The northern part of KZD, as well as the farthest southern flat-bottom parts of the KZD channel, exhibited low roughness values (Figure 4f). The central part of the channel had high roughness values through its whole width (Figure 4f).

Curvature analysis of the NZD 10 m MBES bathymetry data produced high values (blue and red colors, Figure 4c) near the Maslenica bridge and 500 m to NW, under the highway bridge, and N toward the canyon exit. The curvature of the rest of the channel was low. The planiform curvature (which is meant to emphasize convex or concave forms in the relief) especially emphasized the area around the Maslenica bridge. The curvature analysis of the KZD exhibited elevated positive or negative values in the central and southern parts of the channel (Figure 4g). The values in the central part were somewhat higher. The planiform curvature highlighted the central part of the channel as a part that had elevated values (Figure 4h).

#### *4.2. Acoustic Backscatter Characteristics and Its Derivatives*

The backscatter intensity ranged from −10 db to −45 dB for 99.9% of the data in NZD, and from −16 dB to −38 dB in KZD (Figure 5a,e). The backscatter physiography of the channels consisted of low acoustic backscatter surfaces at the canyon ends. Within the channels, the backscatter intensity increased, especially from the steep canyon sides.

A segmentation classification was created with 10 classes, out of which, 5 classes with values higher than 202 were relevant to NZD (Figure 5b). The classification that was derived from the backscatter intensity in NZD resulted in diversification of the central part of the channel, while the NW and SE channel ends had lower values. Areas under the bridges had the highest values. Elevated values extended toward the north and south of the Maslenica bridge and north of the highway bridge. uISO created seven classes, where based on the created dendrogram, it was possible to further reduce the number of classes (Figures 5c and 6). Classes with distances lower than 1 were merged, namely, classes 3 and 4 and classes 5 and 6, creating a classification with five classes (Figures 5c and 6). The derived layer showed three classes within the channel, with a majority of the channel covered by class 6, while class 7 covered the areas near the bridges and north of them. Classes 3 and 4 were limited to the entrance/exit areas of the channel. A raster classification using ML derived seven classes. The classification showed that three classes were dominant within the channel. A majority of the channel was covered with class 7, while class 8 covered the areas near both bridges and north of them. Classes 3, 4, and 5 were limited to the entrance/exit areas of the channel. ML composed very similar visual results to uISO.

The segmentation classification for KZD created 10 classes. Classes with lower values (176–199) were dominant at both channel ends. At the NW end of KZD, classes with values 176–199 reached well within the channel, approximately 800 m from the NW end (Figure 5). Going southward, classes with green and light-yellow colors dominated (values 199–210), while the highest values could be detected in the central part and on the eastern channel sides along most of the channel length. The uISO for KZD consisted of seven classes. A reduction to five classes based on the dendrogram (Figure 6) produced a similar result. Classes 1, 2, and 3 were dominant at the channel ends, and were more pronounced in the NW area. Classes 6 and 7 dominated the rest of the channel. Class 8 appeared on the eastern steep sides of most of KZD and on the western sides of the central part of the channel. ML derived seven classes. The classification showed that classes 1 and 3 were dominant at the channel ends. At the NW end of KZD, classes 1 and 3 reached 800 m within the channel from the NW end. Classes 6 and 7 covered the rest of the channel, and class 8 covered the channel sides.

**Figure 5.** (**a**) Backscatter data of the NZD channel. (**b**) Segment mean shift classification of the NZD area. (**c**) Interactive unsupervised classification (uISO) of the NZD area. (**d**) Maximum likelihood classification of the NZD area; the two bridges across NZD are illustrated as references. (**e**) Backscatter data of the KZD channel. (**f**) Segment mean shift classification of KZD. (**g**) uISO classification of KZD. (**h**) Maximum likelihood classification of KZD.

**Figure 6.** Correlation of determination (R2) for the linear Pearson correlation measure between seven variables created using uISO for NZD and KZD.

#### *4.3. Side-Scan Sonar*

A visual analysis of the SSS mosaic revealed several morphological characteristics that helped to interpret the MBES bathymetry, MBES backscatter, and SBP data.

The central-bottom part of the NZD canyon entrance and the deeper central parts of the channel exhibited low reflectivity of the planar surface in the SSS mosaic. The sides of the channel showed high reflectivity throughout the whole length of the channel. The eastern steep sides appeared to have more exposed boulders and rocky outcrops on the slopes (Figure 7). This was especially highlighted on the sharp bends within the canyon. There was a collapsed steel structure visible in the southern part of the canyon, spreading across the whole width of the channel (Figure 7). This underwater construction was the remains of a bridge demolished during the War of Independence in 1991 [54]. It is located below the present steel bridge that was constructed in 2005 [54]. After the bridge demolition, a floating bridge was constructed at the southern entrance to the canyon. Its concrete supports were still well visible on the SSS mosaic image (Figure 7b). The shallows at the southern end of the SSS mosaic had rocky outcrops and were partially covered with sediments with visible waveforms (Figure 7b).

**Figure 7.** Side-scan sonar mosaic data showing the morphologic characteristics and anthropogenic impact in (**a**) the NZD channel; (**b**) enlarged details of the SSS mosaic showing the collapsed bridge construction (marked with arrows), the flat bottom with the pipeline crossing the channel, and the remains of the concrete slabs used to secure a floating bridge; (**c**) the KZD channel; (**d**) northern end with a flat bottom and rocky sides, a sharp bend with rocky outcrops, and the southern end of the channel with a flat bottom and several waveforms.

Similar to NZD, KZD had a flat bottom-central part of the channel with a lower reflectivity on the SSS mosaic, while steep sides exhibited a higher reflectivity (Figure 7). A low reflectivity was the most pronounced at the northern part of the canyon entrance (for approximately 800 m southward). Rocky outcrops and solitary boulders were exposed, especially on the steep sides of the sharp bends (Figure 7d).

The effort to classify the SSS mosaic using ArcGIS tools failed to provide useful results, creating only a small number of classes.

#### *4.4. Sub-Bottom Profiler*

The penetration of the SBP seismic signal was limited due to a very thin sediment cover over the limestone bedrock and shallow water depth causing the occurrence of multiples. Three acoustic units could be determined at the SE end of NZD on the profile perpendicular to the channel (Figure 8b). The uppermost seismic unit (unit 1, Figure 8) was acoustically semi-transparent. Some internal parallel reflectors with weak amplitudes were visible at the base of this unit. The lower sediment unit (unit 2) was characterized by high acoustic amplitudes. The acoustic signal penetrated for less than 10 ms through this unit, indicating coarse sediments. Units 1 and 2 were separated by a high amplitude

unconformity with an irregular surface. The acoustic basement (unit 3) was interpreted as bedrock. Due to its relative position in the stratigraphic succession and surrounding surface geology, it is safe to presume that it was bedrock constituted of karstified limestones. The acoustic profile through NZD showed a very dynamic morphology of the upper surface with many barriers in the form of ridges, where some were covered with sediment. Side-echo refractions, caused by the steep adjacent barriers, were observed throughout the profile. A thin overlay of acoustically semi-transparent sediments was visible in the central to northern part of the profile (Figure 8a). It was separated from the bedrock by a very weak amplitude reflector. Due to the water depth, multiples could also be observed throughout the whole SBP profile.

**Figure 8.** High-resolution seismic profiles surveyed in the canyons. (**a**) Profile A–A . Only a thin sediment cover was visible over the karstified NZD seabed with multiple side echoes. A thicker sediment succession was visible at the channel ends and in the central part. (**b**) Profile B–B . A thicker sedimentary succession with two distinct units was visible at the southern NZD end. (**c**) Profile C–C . The northern end of the KZD with a thicker sediment succession. (**d**) Profile D–D . The northern part of KZD was covered with sediment (unit 1); in the southern half, rocky barriers stood out with sediment infill between them. (**e**) The SBP track lines of the interpreted profiles are marked with red lines on the maps. The barriers from Figures 2 and 3 are marked with Roman letters.

The same three acoustic units could be determined at the SE end of KZD on the profile perpendicular to the KZD channel (Figure 8c). The uppermost unit was acoustically semi-transparent (unit 1) with some weak reflectors on the bottom of the unit. The underlying unit 2 had high acoustic amplitudes and attenuated the signal penetration. Unit 3 exhibited sharp and steep ridges in the middle of the profile and along the base of the western side of the profile, which were draped by units 1 and 2.

An SBP profile along KZD indicated a very dynamic bathymetry. Several steep carbonate ridges (unit 3) penetrated through the sediment cover to the seabed surface (Figure 8d). Most of the bedrock was covered with the acoustically semi-transparent sediments of unit 1. On the NW third of the profile, the bedrock was draped with the thicker sediment succession of unit 2. Southward, several carbonate

ridges pointed out to the surface, while the space in between was partially filled with acoustically semi-transparent sediments. A thicker unit 2 succession was determined at the SE end of KZD.

#### **5. Discussion**

Integration of the acoustic and morpho-bathymetric surveys enabled us to reconstruct a detailed bottom morphology of the two narrow karst canyons that connect two semi-enclosed bays, the Novigrad and Karin Seas. Merging the available high-resolution hydroacoustic data-bathymetric, high-resolution seismic, and side-scan sonar data with the already available topographic data enabled us to make spatial and morphometric analyses and create maps to describe the unique environment that acted as a river discharge passage during the sea lowstand, as well as an inlet of the sea into the basins and estuary. Steep slopes and a pronounced bottom morphology characterized the canyons.

#### *5.1. Morphology of the Canyons*

Analysis of the MBES bathymetry measurements in NZD revealed six barriers extending along the channels to their steep sides. The steepness of the channel sides was highlighted in the slope and curvature analyses that provided typical values for extreme relief [55]. Three barriers rose to a depth of 25 m b.s.l., with a height difference of 15–20 m, extending to a maximum depth of 45 m b.s.l. The bottom morphology of NZD was very irregular, as depicted on the bathymetric profile and as highlighted by many side-echo refractions visible on the SBP profile (Figure 8a). Adjacent steep rocky outcrops or steep sidewalls cause side-echo refractions [13]. The sediment thickness was higher in the bays, as evidenced by the SBP profile perpendicular to the NZD channel. Within the channel, the sediment overlay was thin or non-existent on the most barriers, and a significant sediment build-up was only noticed on barriers 2 and 3 in the central part of the channel. Thin sediment cover was emphasized by the increased surface roughness of the steep channel sides, depressions, and some barriers (Figure 4). This lack of sediment cover was most likely caused by strong present-day currents in the narrow channels. Strong sea-bottom currents can be caused by a significant input of freshwater into two bays, as the Zrmanja River alone brings 2–3 times more water annually than the total volume of the Novigrad Sea [37]. To this volume, we must add the contribution of the rivers Karišnica and Bijeli Potok flowing into the Karin Sea, as well as karst underground flows ending as submarine springs in the Novigrad Sea [35]. Tidal currents have a negligible effect on the estuary, as tides are rather weak, with M2 amplitudes below 10 cm [28]. As the MBES backscatter signal differs due to the bottom type and its physical characteristics, namely, its hardness or softness [1,8]. Thus, it was possible to classify the MBES backscatter data. A thicker sediment succession at the ends of the NZD channel, which was visible on the SBP profiles, was well delineated in the MBES backscatter derivatives (Figure 5) due to different characteristics compared to sediments within the channel. Within the NZD channel, the MBES backscatter intensity increased, pointing to a rockier surface with a high acoustic backscatter. Sediments in the deepest areas or depressions were also well defined as a class with different sediment characteristics.

The bathymetry data for KZD showed the deeper and more even bottom of the northern part, while the southern half of the channel revealed five barriers. The barriers were equally deep and had similar heights rising to a depth of 14–16 m b.s.l. The SBP data pointed to the fact that most of the channel was covered with at least several meters of sediment, with only peaks of the barriers in the southern part of the KZD comprising a thin sediment overlay (Figure 8). The northern part of the channel bedrock was covered with a thicker sediment succession that increased toward the Novigrad Sea. Higher surface roughness in the central part of KZD, as well as toward the southern part, highlighted the more uneven morphology of the southern part of the channel. The diversification in the MBES backscatter signal derivatives due to the difference in the physical characteristics of the sediment [1,8] was most pronounced in the northern part of KZD, where a thicker sediment succession was visible on the seismic profiles. Similar characteristics could be observed on the southern end of the channel.

Despite the similarity in their appearance, it seems as if the KZD was a "reduced" version of NZD in many ways. The steep sides of NZD rose 150 m a.s.l., with depths below 40 m, while the KZD slopes rose to 40 m a.s.l. and the channel was only up to 20 m deep. The pattern was similar in the case of the bottom morphology, which was more prominent in NZD. One of the reasons for the milder morphology was the thicker sediment cover in KZD. There were several reasons for the preservation of the thicker succession of sediments within KZD: the currents were not as strong as in NZD due to the reduced freshwater influx that only came from the short periodical rivers Karišnica and Bijeli Potok, which is in contrast to NZD, where the volume of the freshwater influx was significantly higher as a result of rivers flowing into the Karin and Novigrad Seas, including the Zrmanja River [37]. Another reason can be found in the easily erodible flysch sediments abundant in the Karišnica and Bijeli Potok watersheds [45] (Figure 1c).

It is clear that the majority of submerged barriers within the canyons were karstic limestone forms. What is still unclear is whether all the submerged barriers in the canyons are made of tufa. The reason for the assumption of tufa barriers in the canyon is the existence of many relicts and recent tufa barriers in the Zrmanja River [21,31–33]. Therefore, favorable conditions for tufa growth in the studied canyons also existed during the lowstand. Ultimately, the morphology of tufa deposits is controlled by the topography and water flow regime [32]. The growth and calcification of rheophilic algae and mosses produce porous hardened substrates and results in a lateral displacement that extends across the river, forming dams with lakes behind them [22]. The tufas in Zrmanja River are described as waterfall and barrage tufas, with some waterfalls being more than 8 m high [21,32]. Tufa barriers higher than 10 m have been detected in the NZD by scientist divers, with one barrier reaching 20 m high with a crest at a depth of 26 m b.s.l. [18]. Five barriers could be detected in the presented data with crests at the highest depths of 16–30 m b.s.l. in NZD, and four barriers with crowns at the highest depths of 14–16 m b.s.l. in KZD (Figure 9; see also Figures 2, 3 and 8).

**Figure 9.** (**a**) Detailed MBES bathymetry map of barriers III and IV in NZD, with the outlined barriers and their crests (dashed line). (**b**) SBP profile of barriers III and IV that are overlain with the actual bathymetry profile over the central part of the canyon (red line).

The SSS mosaic proved to be very useful for determining rocky areas, as well as single boulders collapsed from the steep canyon sides. The anthropogenic effect on the NZD bottom was best recorded on the SSS mosaic, documenting the remains of a collapsed metal bridge construction, which comes from past war efforts during the 1990s War of Independence [54]. The anthropogenic effect was also visible in the form of concrete blocks delineating the path of a temporary floating bridge that was constructed at the southern entrance of the NZD due to the collapse of the pre-1990 bridge [54].

#### *5.2. The Role of the Channels and the Barriers in the Holocene Flooding of the Novigrad and Karin Seas*

There are many definitions of an estuary, where many include not only its present state under the influence of the river and the sea but also its morphogenetic origin [56,57]. In this way, Dalrymple et al. [58] define an estuary as "the seaward portion of a drowned river valley system which receives sediment from both fluvial and marine sources and which contains facies influenced by tide, wave and fluvial processes." Evans and Prego [56] conclude that estuaries were produced by a relative rise in sea level and drowning of a previous erosional depression produced via fluvial erosion. Due to the rapid late Pleistocene–Holocene transgression, the river canyons and the poljes in the present-day Novigrad and Karin Seas were submerged [14]. Based on the data gathered in this study and the available data on the relative sea-level rise, we can make hypotheses regarding the evolution of the lower part of the Zrmanja River estuary during the Holocene sea-level rise (Figure 10). During that period the sea level rose 65 m [15,16,59], and eventually formed today's Zrmanja River estuary along with the Novigrad and Karin Seas. The tufa barriers presented a significant factor for the flooding of poljes, as they created 10-20 m high barrages that prevented water from flowing directly (Figure 10). The similarity with the present Zrmanja River is evident, whose present estuary ends at Jankovi´ca Buk, a tufa waterfall that creates a border between an estuary and a river [30,35]. The present river flow is intermittent with plenty of active and fossil tufa barriers [1,31]. The created flowline, as the lowest possible water path in NZD and KZD, allowed us to determine the pathway through the canyon and the relative sea level during the flooding of the Novigrad and Karin Seas. The sea level reached the lowest part of the crest of the barriers in NZD at the present depth of −24.5 to −25 m, and afterward, flooded the Novigrad Sea (Figure 10a–c). Flooding of the polje in the Novigrad Sea area enabled the formation of the Zrmanja River underwater fan, as the sea-level rise caused the deposition of river sediment at the exit of the canyon (Figure 10a). When the sea level rose to the depths of −14 m to −16 m, it reached the crest of the barriers in KZD (Figure 10a,b,d). As the sea level continued to rise, it flooded the Karin Sea until it reached the present level (Figure 10a,b,d). Based on the relative sea-level curve [16] and the heights of the barriers, the estimated time of the flooding of NZD, and consequently Novigrad Sea, was after 9200 BP, while the time of the flooding of KZD, and consequently Karin Sea, can be estimated as having occurred after 8400 BP (Figure 10b).

Further investigations of the sediments deposited in these basins (including sediment cores and high-resolution acoustic methods) will provide more conclusive answers about the timing of the Holocene sea-level rise.

**Figure 10.** (**a**) Schematic of the coastline position in the study area at sea level at −35 m b.s.l. before the canyons were flooded (white polygon with a green outline), at −25 m b.s.l. (light blue polygon with a red outline), at −15 m b.s.l. (medium blue polygon with a yellow outline), and at the present sea level (dark blue polygon with a blue outline) based on the recent bathymetry data and the Holocene sea-level rise [59]. (**b**) Sea-level curve (modified from [59]) with relative depths of the lowest position of barrier crowns in NZD (red line) and KZD (yellow line) that prevented flooding into the Novigrad and Karin Seas, respectively, as well as the sea level at −35 m b.s.l. before the flooding (green line) and the present sea level (blue line). (**c**) Profile based on the available bathymetry data with sea levels before the NZD canyon was flooded at −35 m b.s.l. (white polygon with a green outline) and at sea level at −25 m b.s.l. (light blue polygon with a red outline). (**d**) Profile based on available bathymetry data with the sea level at −15 m b.s.l. after both canyons were flooded (medium blue polygon with a yellow outline). (**e**) Profile based on the available bathymetry data with the present sea level (dark blue polygon with a blue outline).

#### **6. Conclusions**

High−resolution MBES bathymetry, MBES backscatter, SBP, and SSS measurements were carried out in two canyons to provide the first insights into the contemporary physiography of this unique environment consisting of two semi-isolated basins (the Novigrad and Karin Seas) connected with narrow steep channels Novsko Ždrilo and Karinsko Ždrilo.

NZD connects the Novigrad Sea with the open sea. It is also a part of the 15-km-long estuary of the Zrmanja River flowing into the Novigrad Sea and brings large volumes of freshwater. NZD canyon comprised steep and high side slopes extending up to 150 m a.s.l. Six barriers were detected within NZD, extending from the top of the barriers at −25 m b.s.l. to the bottom at −45 m b.s.l. The barriers were most probably all made of tufa, as detected by divers and as an analog with past and present tufa growth in the Zrmanja River. Those barriers prevented marine flooding during the Holocene sea−level rise since during the lowstand, the semi−isolated Novigrad and Karin Seas acted as karst poljes. Strong outflow currents prevented significant sediment build-up. A thicker sediment succession was detected at the ends of the channel extending to the open sea and bay. This was well depicted in SBP profiles and delineated in the MBES backscatter intensity and its derivatives. KZD connects the Karin Sea with the Novigrad Sea. The KZD canyon characteristics were less prominent compared to NZD, with lower and less steep sides. The five barriers detected in KZD were mainly covered in sediment and extended from 14 to 16 m b.s.l.

The post-LGM sea-level rise drowned the coastal karstic landscapes in the Eastern Adriatic Coast, including the Zrmanja River canyon and two poljes, the present-day Novigrad and Karin Seas. Determination of the lowest possible water path in NZD and KZD allowed us to determine the elevation of the relative sea-level rise. The sea level reached the top of the barriers in NZD at a present depth of −24.5 m to −25 m, and in KZD from −14 m to −16 m. The timing of the flooding of the channels and the bays was estimated based on the relative sea−level curve for the Adriatic Sea after 9200 BP in NZD and after 8400 BP in KZD.

SSS proved useful for determining the anthropogenic effect on the NZD bottom, where the remains of metal construction of the collapsed bridge, as well as concrete supports of the floating bridge, were detected.

Knowledge of the geomorphology of the aforementioned karst features is the most important for the relative sea-level reconstruction. When merged with additional investigations of the sediments deposited in the studied basins, which include sediment cores and high-resolution acoustic methods, these results will provide new insights into the timing of the rapid Holocene relative sea-level rise in the eastern Adriatic coast, as well as in other Mediterranean coastal areas [60].

**Author Contributions:** Conceptualization, O.H. and S.M.; methodology, O.H., S.M. and D.B.; software, O.H.; validation, O.H., S.M. and D.B.; formal analysis, O.H.; investigation, O.H., S.M., G.P., D.C., M.G., D.B. and N.I.; resources, S.M.; writing—original draft preparation, O.H., S.M. and D.B.; writing—review and editing, O.H., D.B., S.M., N.I., G.P., D.C. and M.G.; visualization, O.H., S.M. and D.B.; supervision, S.M.; project administration, S.M. and N.I.; funding acquisition, S.M., O.H. and N.I. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Croatian Science Foundation (HRZZ) under the project "Lost Lake Landscapes of the Eastern Adriatic Shelf" (LoLADRIA), grant number HRZZ-IP-2013- 11-9419, and the EMODNet Geology project, grant number EASME/EMFF/2018/1.3.1.8/Lot1/SI2.811048.

**Acknowledgments:** The authors would like to thank Nikos Georgiou, Xenophon Dimas, George Ferentinos, and Margarita Iatrou from the University of Patras for their support during the seismic survey and during the interpretation of the seismic data. The authors would like to acknowledge the anonymous reviewers and editors for their valuable comments, which helped to improve the quality of this manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **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* **Bridging Terrestrial and Marine Geoheritage: Assessing Geosites in Portofino Natural Park (Italy)**

#### **Paola Coratza 1, Vittoria Vandelli 1,\*, Lara Fiorentini 2, Guido Paliaga <sup>3</sup> and Francesco Faccini 3,4**


Received: 28 August 2019; Accepted: 8 October 2019; Published: 11 October 2019

**Abstract:** Interest in geoheritage research has grown over the past 25 years and several countries have issued laws to encourage improvement and conservation. Investigations on geosites are prevalently carried out on land environments, although the study of underwater marine environments is also of paramount scientific importance. Nevertheless, due to the constraints of underwater environments, these sites have been little explored, also on account of the higher costs and difficulties of surveying. This research has identified and assessed the terrestrial and marine geosites of the Portofino Natural Park and Protected Marine Area, which are internationally famous owing to both the land scenic features and the quality of the marine ecosystem. The goal was to pinpoint the most suitable sites for tourist improvement and fruition and identify possible connections between the two environments. In all, 28 terrestrial sites and 27 marine sites have been identified and their scientific value as well as their ecological, cultural, and aesthetic importance has been assessed. In addition, accessibility, services, and economic potential of geosites has also been taken into account. Both the updated database of terrestrial and marine geosites in the Portofino protected areas and the assessment procedure adopted can become useful tools for the managers of these sites and provide decision-makers with possible strategies for tourist development.

**Keywords:** underwater geoheritage; geosites; geomorphological survey; geotourism; Portofino Park; Italy

#### **1. Introduction**

Geoheritage and geosite studies have assumed growing scientific importance in the past 25 years, and territorial legislative initiatives have emerged all around the world. Geoheritage studies have usually been carried out in terrestrial environments: Mountain areas (e.g., [1–6]), coastal areas (e.g., [7–11]), karst areas (e.g., [12–16]), fluvial areas (e.g., [17–19]), and volcanic areas ([20–23]). Recently, a great deal of interest has concerned also geoheritage in urban areas (e.g., [24–30]).

For what concerns the definition of geosites and their different types of values, they have been much debated within the scientific community (cfr., [31,32] and reference therein). Up to now, two main approaches can be distinguished for defining what geosites are: A restrictive and a broader definition.

According to the restrictive definition, geosites are in situ elements with high scientific value [33], i.e., sites "having particular importance for the comprehension of the history of the Earth and of its present and future evolution" [34,35]. According the broader definition, geosites—or geodiversity sites

(*sensu* [33])—are defined as geological elements that present a certain value due to human perception or exploitation, e.g., elements with high scientific, educational, aesthetic, and cultural value. Often, geosites are included in protected areas even if their institution is, in most countries, related to the biological aspects more than the geological ones. In fact, geology has often been inadequately accounted for in parks creation, planning and management. Nevertheless, after decades of focus on the protection of biological heritage, a great deal of progress has been made in the last 20 years (cf., [36] and reference therein). In this respect, particularly notable is the UNESCO Global Programme, which intends to "promote a global network of geoparks safeguarding and developing selected areas having significant geological features" [31,37,38]. Moreover, natural disasters and their tangible evidence in landscape may be important geosites, ideal to promote geological education [39] and geotourism [40,41].

In Italy as elsewhere, the nature conservation in coastal and marine environment is provided by marine protected areas whose nature conservation policy primarily addresses the biodiversity, often underestimating or nearly neglecting abiotic features. Among the European legislative framework worthy of note are the EU Birds Directive (1979), the Habitats Directive (1992), the OSPAR Convention (1992), and the EU Marine Strategy Framework Directive (2008), which have focused the attention towards the marine environment.

Concerning underwater geoheritage and marine geosites [42], despite their importance, only few studies have been developed; this is particularly true when compared with studies on marine biotic heritage ([43] and references therein). This is mainly due to the physical constraints of the marine environment that influence the high costs of underwater surveys and the difficulty of investigating near shore areas, where navigation is not possible. In addition, as highlighted by Burek et al. [44], there are general differences in attributes related to sites of geological and geomorphological interest in terrestrial and marine environments. In a marine environment, geological heritage is largely invisible, except in clear and shallow water, and hardly accessible. These characteristics have reduced the opportunities for promotion, education, and interpretation activities for the public, but at the same time, they have also reduced vulnerability to man-made damage. Furthermore, the different perception and enjoyment of abiotic features of the aquatic environment by tourists has led to a delay in developing common schemes and approaches to the identification, assessment, and improvement of submarine geosites [43].

While many studies have dealt with emerged shorelines [45–47], geoheritage research in underwater environments still lacks common investigation schemes and approaches, again especially in comparison to studies on marine biotic features [48,49]. Specific studies on submerged geoheritage are few and were developed mainly by Italian researchers [43,50–55]. In particular, in Orrù et al. [52], the selection of sites of geomorphological interest was carried out by considering several significant valences as: (i) Model of geomorphological evolution; (ii) exemplarity; (iii) paleo-geomorphological testimonial; and (iv) ecological valence. In the same work, the geosite assessment was carried out considering their scientific interest and other types of interest such as cultural, educational, and historical interests. Similar to this approach was the one used by Rovere et al. [43]; in fact, they evaluated underwater geomorphological heritage in two Mediterranean marine areas by considering two sets of values, that were the scientific and the additional values. The two sets were further divided in subcategories inspired by those proposed for terrestrial environment (e.g., [47,56,57]). Recently, Flores-de la Hoya et al. [58] prosed a method to rapidly assess coastal underwater spots to be used as recreational scuba diving sites. In the latter work, the assessment was based on several criteria inspired by the methodology provided by Ramos [59] for the evaluation of diving site attractiveness in the Algarve region.

As regards marine geoconservation, a growing interest has been recently observed, especially in the UK, where geoheritage has started to be integrated in the management of protected areas ([44,60,61]) and a methodology to assess geodiversity key areas on the seabed has been developed.

From a geoheritage viewpoint, submerged areas are particularly interesting for several reasons:


According to Rovere et al. ([43] and references therein), a complete approach in the studies of geoheritage in coastal zones should necessarily include the description of both the shore and inner continental shelf, according to the fact that two environments showing common processes and landforms must be considered as a single feature [71]. The need for integrating terrestrial and submerged datasets in geomorphological studies is not new. Examples of studies coupling land and sea data available in literature have considered several aspects, such as: (i) Archaeological investigations (e.g., for the northern coast of Ireland by Westley et al. [72] and Harff et al., [73]); (ii) paleo-environmental reconstruction (e.g., Quaternary geomorphological evolution of the Tremiti Islands, southern Italy [74,75])—especially in fluvial environments; (iii) marine spatial planning [76]; (iv) coastal hazards assessment and risk reduction (e.g., mitigation of the risk due to tropical cyclones, tsunamis, floods, and sea-level rise along the Mozambique coasts [77,78]); (v) integrated geomorphological mapping of emerged and submerged areas (e.g., [79] in the Netherlands; [80] in the Tremiti Islands, southern Italy; [70,81] in the Maltese Archipelago).

The goal of this study is to identify and assess terrestrial and marine geosites—intended, in a broad sense, as component of the cultural heritage of a territory [82,83]—in the Portofino Natural Park (Liguria Region, northern Italy), in order to select sites more suitable for a geotourism exploitation, pinpointing a potential morphogenetic bridge between terrestrial and marine features. These latter are poorly known by the general public especially from geological and geomorphological perspectives. The Portofino Natural Park, which comprises a terrestrial protected area, established in 1935, and a marine protected area, established in 1999, is well known at an international level thanks to its landscape, environmental, and cultural characteristics (Figure 1). Over 1 million people a year visit the sea hamlets of Portofino and Camogli, as well as the coast between Rapallo and Portofino, whereas the 80 km long footpath network is trodden throughout the year by over 100,000 hikers [11]. In recent times, scuba diving activities, managed by the protected marine area administration, have significantly increased. Scuba divers arrive at properly chosen buoys starting from the diving centers of Santa Margherita, Camogli, and San Michele di Pagana (located between Santa Margerita Ligure and Rapallo). The remarkable environmental and cultural features conserved both in terrestrial and marine areas of the Portofino Natural Park led the study area to become an ideal site for the development of geotourism, defined according to the broader approach of the National Geographic in the United States as "tourism that

sustains or enhances the geographical character of the place being visited, including its environment, culture, aesthetics, heritage and the well-being of its residents" [84].

**Figure 1.** Location map of the study area (modified from [11]).

#### **2. Geographical Setting**

The Promontory of Portofino breaks the continuity of the coastline between Genoa and La Spezia, along a perimeter of 13 km and an area of 18 km2. The orography is characterized by rather high peaks, considering the short distance from the sea [85]. There is a WNW–ESE oriented relief, the culmination of which corresponds to the Mount of Portofino (610 m). Hydrographic catchments are less than 1 km2 wide, with channels of the second order at the most [86]. Among the most important catchments of the southern slope, the following can be quoted: Cala dell'Oro catchment, located west of San Fruttuoso bay; San Fruttuoso catchment; Ruffinale and Vessinaro catchments, both located between San Fruttuoso Bay and Portofino promontory. Whereas, on the eastern side, the Rio del Fondaco at Portofino and Fosso dell'Acquaviva at Paraggi are found [87,88].

Due to the torrential regime, the flow rates of watercourses are substantially nil for most of the year; in the case of heavy rainfall of short duration (not infrequent in the area), the maximum flow rates, for return times of 200 years, range between 20 and 40 m3/sec (flow rate unit contribution of 40 m3/sec/km2 for catchment areas of less than 1 km2).

The Portofino Park protects the area of the promontory bearing the same name, which is located less than 20 km away to the east of Genoa. To date, the protected area is 1056.26 ha, out of which 58.61 ha make up the integral reserve, 597.31 ha are the general reserve area, and 362.50 ha are the protection area. The remaining 37.84 ha belong to the economic promotion area [11]. The contiguous territory adds an extra 932 ha to the Park (Figure 2).

**Figure 2.** Geological and geomorphological sketch of the study area (modified from [11,86–88]): (**1**) Conglomerate; (**2**) marly limestone flysch; (**3**) bed attitude; (**4**) fault; (**5**) tectonic lineament; (**6**) landslide and debris covers; (**7**) anthropic fill; (**8**) soil slip; (**9**) edge of sea cliff scarp; (**10**) incising channel; (**11**) pocket beach; (**12**) spring; (**13**) cave; (**14**) submerged cliff with boulders, sands in the small bays (San Fruttuoso, Portofino and Paraggi); (**15**) sea bottom with sands and muddy sands; (**16**) submerged channel; (**17**) submerged spring; (**18**) submerged cave; (**19**) submerged peak; (**20**) coralligenous reef; (**21**) seagrass meadow.

The area of the Park stretches over the municipal territories of Camogli, Portofino, and Santa Margherita Ligure, whereas the contiguous area is part of the municipal territory of Rapallo. The residing population of the Park is around 750 inhabitants. The presence of tourists is high throughout the year: At the village of Portofino, there are over 1 million tourists per year, whereas at San Fruttuoso, tourist boats carry some 400,000 tourists/year around the Gulfs of Tigullio and Paradiso [11,86]. Apart from seaside tourism, there is also a high presence of hikers along the over 80 km long footpaths: Just the stretch from Portofino Vetta to Pietre Strette is trodden by over 70,000 hikers per year.

Thanks to its landscape, natural, and cultural values [89,90], the Promontory of Portofino has been protected since 1935 by Italian Law no. 1251 (Establishment of the local authority of Mount of Portofino). Since 1995, it has been managed as 'Ente Parco', established by Ligurian Regional Law no. 12/95 (Reorganization of protected areas), which redefined the borders of the protected area with Regional Law no. 29/2001 (Identification of the perimeter of the Portofino Natural Regional Park).

The Marine Protected Area of Portofino, established by the Italian Ministry for the Environment, was added to the Park with the Decree of 26/04/1999, which implemented the Italian Law no. 979/1982 (Measures for the Sea Protection). The marine area is subdivided into three zones of safeguard (A, B, and C), in which free navigation, hunting or catching of fauna, underwater fishing, and diving are forbidden. In addition, all underwater activities that require contact with the seabed are forbidden, as well as the anchoring of any boat [91]. Zone A (Integral Reserve) comprises the sea area of Cala dell'Oro bay (west of San Fruttuoso bay). Access to this area is permitted only for emergency rescue and authorized scientific research. Zone B (General Reserve) stretches from the Portofino lighthouse

point to Punta Chiappa, excluding the access corridor to the harbor of San Fruttuoso. Less restraining issues characterize this zone: Authorized sport fishing is allowed for residents, scuba diving is allowed for diving centers and authorized private subjects, whereas bathing is free. This marine area is very popular among scuba divers, who are attracted by the considerable natural beauty of the seabed and, in particular, by the great number of violescent sea-whips (*Paramuricea clavata*) and the richness of sea fauna. Zone C (Partial Reserve) stretches between the two sides of the Promontory of Portofino and owes its fame to the vast prairies of Mediterranean tapeweed (*Posidonia oceanica*). Bathing, scuba diving, and sport fishing are allowed. On the whole, over 70,000 scuba divers per year plunge into the water of the Portofino Protected Marine Area [92].

Recently, in December 2017, the Portofino National Park was established. It comprises both the terrestrial area and the protected marine area. By the end of 2019, the Italian Ministry for the Environment will establish the new borders of this National Park with a specific law.

#### **3. Geological, Geomorphological, and Hydrogeological Setting**

The geology of the Portofino National Park is known at an international level owing to the presence of Portofino Conglomerate, whose lithological nature and geological and geomorphological significance have in fact been widely studied (e.g., [93–95] and references therein) as well as its geomechanical behavior (e.g., [88,96]). The conglomerate forms the trapezoid-shaped promontory between Punta Chiappa to the West and the Portofino lighthouse to the East. The geological root of the Portofino Mount, between Camogli and Rapallo, is characterized by a marly-calcareous flysch (Mt. Antola Flysch). The boundary between these two geological formations (pudding stone and flysch) is partially ascribable to tectonic causes and shows a WNW–ESE trend (Figure 2). The Promontory morphology is derived from a structure bounded by normal faults, typical of a continental margin subject to disjunctive tectonics [97,98].

The Portofino Conglomerate is made up of marly-calcareous clasts and, to a lesser degree, sandstones, ranging in size from centimeters to meters, arranged in several-meter thick layers with rare sandstone intervals, often accompanied by thin coal layers. Ophiolite, limestone, cherts, and gneiss clasts are also found, although less frequently. This conglomerate, which lacks a fossil record, was dated doubtfully to the Oligocene due to the scarcity of biostratigraphic records [97,98].

On the whole, the structural setting of the Conglomerate shows a SE to SW dip, with a less than 20◦ inclination. The rock mass is affected by various joint systems, easily identifiable at a meso- and macro-scale. The NW–SE and NE–SW oriented systems, which are ascribable to normal faults, are the most important. At a slope scale, the intersection between the various joint systems produces the subdivision of the conglomerate into several decameter-thick blocks [99].

Mt. Antola Flysch, dating to the Cretaceous, is made up of calcareous marls and marly limestones, marls with argillite levels, siltites, and calcarenites. The structural setting of the flysch is constrained by diverse deformation phases, both ductile and brittle, which affected this rock mass. An isoclinal-fold arrangement was identified in this formation; it shows a SSW vergence with a WNW–ESE oriented axis [95].

Landforms in the study area are controlled by geological-tectonic setting and conditioned by meteo-climate conditions [87,88]. Rocky cliffs up to 200 m high, the highest of the Mediterranean coast, characterize the southern slope of the Promontory of Portofino [93]. The average inclination of the slope is 45◦ to 65◦, although many are the coastal stretches characterized by vertical cliffs [94]. The action due to swell is important and is determined by both SE wind ('Scirocco', dominant wind), and SW wind ('Libeccio', prevailing wind). Sea storms are rather frequent, with wave heights exceeding 5 m; they can cause serious damage to buildings and infrastructures, as in the event of 27–29 October 2018, which affected the Promontory eastern coast, between Rapallo and Portofino.

The profile of the emerged cliff continues underwater up to a depth of some 70 m. Up to the margin of the shelf, some 140 m deep, the inclination of the seabed is rather homogeneous and gentle. The margin of the shelf, which is not influenced by the presence of the promontory, is found at a distance of 3.5 to 4 km from the coast [100].

The base of the narrow continental slope is found at a depth of 0.6 to 1 km, in correspondence with a furrow named 'Canyon della Riviera di Levante', which stretches in an E–W direction, with the confluence of a small canyon formed in the West front of the Promontory of Portofino [85,100].

The morphologically significant tectonic alignments, which contour Mount Portofino with landforms such as saddles, towers, and triangular facets, continue in the submerged portion. Some of the faults are considered active, since they disrupt the seabed in their underwater part.

In the high conglomerate cliffs of the southern slope, there are often rock falls, even along very steep fluvial channels, mostly of the first order, as in the case of the torrents Ruffinale and Vessinaro [96]. On the western slope, the cliff has been prevalently modelled in Mt. Antola Flysch, attaining heights exceeding 100 m. This stretch of coast is subject to a SW swell, which is one of the main causes for occurrence of rapid slope movements, such as debris/mud flows and rock avalanches, which often have a high destructive power [94]. There are also slow slope movements with surface of rupture in the marly-calcareous bedrock. In this case, numerous morphotectonic clues suggest a process of the mountain slope deformation type [101]. Along the boundary between the conglomerate and flysch, there are landslides of diverse origin and state of activity owing to the contrast of resistance and deformability between adjacent rock masses [11]. Among the landslide bodies surveyed, worthy of note is the accumulation found at Sotto Le Gave, on the eastern slope, which is partially due to mountain slope deformation and has affected buildings and infrastructures even in the recent past.

In the submerged area comprised within 200 m from the coastline, morphological rises linked to neotectonic modelling are found, as South of Punta Chiappa (*Secca dell'Isuela*), E of San Fruttuoso (*Secca Gonzatti*), and SE of Punta Portofino. This portion of the seabed reveals exceptional biodiversity [91], also resulting from geomorphological features. The widespread coralline biocoenosis and tapeweed prairies, which characterize most of the seabed near the coast, are developed on large rock blocks (>1 m).

The meteo-climatic characteristics of this area are linked to the cyclogenesis of the Gulf of Genoa, which causes events of short but intense precipitation (less than 6 h, with rain peaks exceeding 50 mm/h) between mid-summer and mid-autumn [93,102,103]. Consequently, the most common effects at ground level are flash floods, hyper-concentrated fluxes, and debris/mud flows. Among the most significant and destructive events in living memory, those of 1915, 1961, and 1995/1996 should be mentioned. Also, in the 2000–2018 period, many extreme hydro-meteorological events occurred on Portofino Promontory, causing important effects at ground level with considerable damage to buildings and infrastructures: The average, on a historical basis, is over one event per year [11].

The Ligurian Speleological Registry lists 20 caves in the Portofino Conglomerate [104]. Their origin is prevalently tectonic although, to a much lesser extent, is due to chemical–physical dissolution or processes linked to the sea wave action. In addition, several natural caves have been surveyed in the submerged portion of the cliff, up to a depth of 60 to 70 m; also, their genesis is a result of tectonic modelling.

The intense joint network of the conglomerate, the contrast of hydraulic conductivity with the marly-calcareous flysch, and the climate characteristics of the territory cause significant effective infiltration with widespread presence of groundwater and springs [99]. Effective infiltration ranges from 350 mm/y at sea level up to over 500 mm/y at higher elevations. The water springs are located either along the contact between the Conglomerate and the marly-limestone Flysch, or in the Conglomerate rock mass, along tectonic lineation, or along the interface with the sandy interlayers. Underground aquifers are extremely fragmented, with annual intermittent flow rates ranging from less than 1 L/min in dry summer to over 10 L/s in late autumn. Some of these springs have been used for a long time and today still feed local water-supply systems [93]. There are also significant springs underwater, along the submerged cliffs, and at the connection with the shelf. The latter is an important morphological element indicating the position of the sea-level at the end of the Würm regression. Furthermore, these features bear witness to the neotectonic activity taking place in the Plio-Quaternary, with uplifting and lowering phenomena affecting the seabed.

Among anthropic forms, drywall slope terracing is a very common farming technique, which dates back to ancient times. Terracing has deeply modified the geomorphological, vegetation, and dwelling landscape at a slope scale. Well-preserved examples of slope terracing are found in the Valloni di Paraggi, Portofino, and San Fruttuoso; they make up an important cultural and landscape asset.

#### **4. Materials and Methods**

The increasing interest in the promotion of geotourism requires the selection and assessment of geosites in order to determine priorities in site management and geoconservation strategies. Based on these premises, a research program for the identification and assessment of geosites at the Portofino Natural Park has been developed.

#### *4.1. Geosites Identification*

Research on geosites at Portofino Natural Park has taken advantage of the numerous thematic maps and scientific publications on the geology and geomorphology of the study area, concerning both emerged and submerged areas of the Park, as well as tourist maps and guidebooks. Some milestone publications have been particularly significant for the aims of this study, such as Ristori [105] on the Conglomerate and groundwater regime at Mount Portofino and Pellati [85] on the geomorphological characteristics of the Promontory of Portofino. As for geological and petrographic features of the conglomerate, the contributions by Giammarino et al. [97] and Giammarino and Messiga [98] should be mentioned. As concerns geomorphological features, in recent times there have been contributions on geomorphological hazard and tourist vulnerability along the Park footpaths [86], on the landslides of the western slope of Mount Portofino [94], and on geomorphological mapping of San Fruttuoso and Portofino [87,88]. In addition, other publications have been taken into account: The debris flows along the coast [93], the hydrogeology of the Caselle springs [99], and the terracing of the Park considered as a cultural asset [90]. Salmona and Varardi [91] discuss the socioeconomic aspects of the protected marine area, whereas other contributions deal with underwater tourism and related impact on the ecosystem. Cerrano et al. [92] stress the importance of volunteer scuba divers for scientific activities aiming at the conservation of Mediterranean natural resources and [106] describe the success of scuba diving in the Portofino protected marine area. Furthermore, Lucrezi et al. [107] illustrate the contribution of scuba divers in the management of protected marine areas and, again, Lucrezi et al. [107] pinpoints the correct balance between scuba diving activities and environmental sustainability. Saayman and Saayman [108] discuss the economic benefits resulting from scuba diving in protected marine areas, and Di Carro [109] describes an approach for assessing human impact on the Portofino protected marine area. Finally, Markantonatou et al. [110] develops a study on social networks and the flow of information for responsible and sustainable planning in the Portofino protected marine area.

For the selection of terrestrial geosites (Table 1 and Figure 3), this study took advantage of the inventory developed by Faccini et al. [11] where geosites have been selected and classified according to their main scientific relevance in: Geological, geomorphological, mineralogical-petrographic, hydrogeological geosites, and viewpoints (*sensu* [111]).

A geoheritage inventory for the underwater part of the area investigated was lacking. Therefore, marine geosites were selected (Table 2) by combining geological and geomorphological data in strict collaboration with park managers. The sites were classified according to their main scientific relevance as geomorphological, speleological, and hydrogeological geosites (Tables 1 and 2 and Figure 3). Since geosite assessment is important for the promotion of the area from a geotourism perspective, two marine sites of cultural interest and great tourist potential have been included (*Cristo degli Abissi*—ID 21S; and Mowak Deer shipwreck—ID 13S). These sites show a complex relationship between the natural and/or human heritage of the Portofino Park [11]. The underwater geosites have been classified into two categories according to the skills of the visitors: (i) Snorkeling sites (more or

less available to everybody) and (ii) sites equipped for qualified scuba divers. Snorkeling sites have been classified based on direct observations and comprise practically all the free bathing sites of the Portofino Protected Marine Area. They are pocket beaches, often fed by annual beach nourishments, apart from the Punta Chiappa site, which is a rocky cliff. The scuba diving sites are managed by the Portofino protected marine area and are identified by 21 signaling buoys, where one or two boats can be moored. The diving sites have been classified into three categories, according to their technical difficulties. The scientific data reported in the cards of each diving point, which have been elaborated by the Portofino protected marine area [112], have been updated with new original observations on each diving point up to a maximum depth of 45 m.



#### *4.2. Geosite Assessment*

Evaluation of geosites has been developing since the 1990s (cf., [32,113–115]). In spite of many published methods about the assessment of sites, the scientific literature reveals that there is still a great debate concerning values and criteria to be used in the geosite assessment process (see [31,116] and reference therein) and there is no general accepted method. One of the most popular approaches for geosite assessment is the comparative analysis of geosites within a given area, by applying numerical evaluation of their values, based on several criteria and respective indicators (e.g., [33,56,117–121]). The aim of a quantitative assessment is to reduce subjectivity [122] associated with any evaluation procedure, since the intrinsic value of these environmental elements cannot really be measured. Indeed, the scientific quality of a geosite is a purely indicative numerical quantity, which can be subject to variations determined by the subjectivity of the operators and the general characteristics of the area under examination.

**Figure 3.** Examples of geosites in the Portofino Natural protected areas: (**1**) Grotta dell'Eremita (44.31797 N 9.15120 E, marine geosites 3S and cave on the sea cliff); (**2**) Cristo degli Abissi (44.314038 N 9.174979 E, marine geosites 21S); (**3**) Mt. Campana (44.31973 N 9.15529 E, terrestrial geosites 26T); (**4**) High cliffs at Vitrale (44.30389 N 9.20164 E, terrestrial geosites 7T); (**5**) Punta Cervara stack (44.31355 N 9.21291 E, 'lo scoglio della Carega', terrestrial geosites 8T); (**6**) Rock fall boulders at Castello di Paraggi (44.31112 N 9.21241 E, marine geosites 26D). Image 1 and 2 from Portofino Marine Protected Area archive [100].

**Table 2.** Marine geosites (minimum and maximum depth are expressed in meters). The eighth column refers to the qualitative level of difficulty to reach a certain submerged geosite through scuba diving (source: [112]).



**Table 2.** *Cont.*

Some methods for quantitative assessment of geosites are based on combined numerical indices to obtain a final score, often named Q-value or global value (e.g., [57,123,124]). This index corresponds to the combination of three sets of criteria relevant to: (i) intrinsic characteristic of a geosite (e.g., degree of scientific knowledge), (ii) potential for use, (iii) need for protection.

Other authors preferred methodologies based on independent criteria (e.g., Brilha's [33] methodology) without the determination of a final score, but considering the results of each set of criteria relevant to a given site. This is because the criteria considered are independent of each other and because the independent numerical evaluation for each criterion enables the individual analysis of each geosite. Specific geosite assessment procedures vary in terms of both the number and type of criteria considered as well as weighing individual parameters and indicators. The criteria generally used for geosite and geomorphosite assessment can be classified into five categories as follows [120,125]:


The criteria are preferably adapted to the geological and geomorphological context of the study area.

In the present study, the recognized terrestrial and marine geosites have been quantitatively assessed by applying a methodology that has been specifically set up on the basis of previous works ([47,56,57,119,126]), which concerned the evaluation of terrestrial sites of geomorphological interest and which were applied also to underwater geosites [11]. Although the coastline marks the boundary between terrestrial and marine environments, there is a continuity of geological and geomorphological features across this boundary [61]. In order to fulfil this continuity between land and sea the same assessment methodology for terrestrial and marine geosites was applied. This methodology is based on three sets of values relevant to scientific, additional, and potential for use (Tables 3–5) and the evaluation process builds on bibliographical data and on the detailed and well consolidated knowledge of the geological and geomorphological features of the study area acquired by authors. Scientific value was divided into four sub-criteria (Table 3): Integrity (INT), representativeness (REP), rareness (RAR), and paleogeographic model (PAL). Additional value was divided into three sub-criteria (Table 4): Ecological (ECOL), aesthetical (AEST), and cultural (CULT). Potential for use value was divided into three sub-criteria (Table 5): Accessibility (ACC), services (SER), and economic potential (ECON).

A score between 1 and 5 was attributed to each sub-criterion. For each geosite, the total scientific/additional/potential for use value (*Totval*) was estimated by summing the score of each sub-criterion (*ai*) and dividing by the number of sub-criteria (*na*) for each set of values (cf. Equation (1)):

$$Total\,\,val\,\,=\,\frac{\sum\_{i}a\_{i}}{n\_{a}}.\tag{1}$$

The aesthetic value is the most subjective one and for the definition of criteria and its assessment, research on landscape perception (see e.g., [34,127] for a review) has been taken into account. Table 4 specifies which features are to be considered in order to assess the aesthetic value of a given geosite. According to Coratza et al. [119], these features are: (i) panoramic quality, (ii) colour diversity, (iii) vertical development, iv) naturalness. The cultural value is the more heterogeneous sub-criterion (Reynard et al., 2007). Therefore, in order to guide the assessment procedure, the features considered to estimate the cultural value for a given geosite are specified in Table 4. Considering the tourist vocation of the area, attention was devoted to the assessment of the potential for use value. In particular, for the estimation of accessibility (ACC), two sets of sub-criteria were taken into account for terrestrial and marine geosites, as shown in Table 5. In order to estimate the economic potential of a site, the number of visitors per year has been taken into account. In fact, it can be assumed that the greater the number of visitors, the more the economic income. In particular, for the study area, two different sets of thresholds, regarding the number of visitors per year, were considered in the assessment of the economic value of terrestrial and marine geosites, respectively. An exception was made for estimating the economic value of marine geosites accessible via snorkeling. For these, the same visitor thresholds as the terrestrial geosites have been here considered, since they are comparable to terrestrial geosites in terms of number of visitors. The data on visitor influx along the footpath network across the present geosites, which have been given by the Park authority, are an indispensable element for judging the economic potential of the area. For this purpose, eco-counters aiming to monitor hikers have been installed. As for the number of visitors to marine geosites accessible with scuba diving equipment, the management is ruled by an agreement between diving centers and the administration board of the protected marine area, which has also provided us with attendance data concerning each diving point.


#### **Table 3.** Sub-criteria used for the numerical assessment of geosite scientific value.

<sup>1</sup> Rare at the scale of Portofino Natural Protected area.




**Table 5.** Sub-criteria used for the numerical assessment of geosite potential for use.

#### **5. Results**

Twenty-eight terrestrial geosites and 27 marine geosites were identified and assessed (Figure 4 and Tables 1 and 2). The terrestrial geosites are mainly sites of geomorphological interest of tectonic origin or gravity-induced slope landforms or even coastal landforms, all strictly linked to each other in terms of origin and geomorphological evolution. The footpath network follows the distribution of the terrestrial geosites, which are widespread all over Portofino Park. Marine geosites are mainly concentrated between Punta Chiappa di Levante and Punta Portofino, which is the largest outcrop area of the Portofino Conglomerate.

**Figure 4.** Location of terrestrial and marine geosites.

The geosites selected were evaluated considering their scientific value. Moreover, the research assessed numerically geosite additional value in terms of ecological and eventually cultural importance of the sites as well as their aesthetic quality. Additionally, the potential for use was estimated taking into account the visit conditions, proximity, and availability of services and economic potential of each site. If we compare the average scientific, additional, and potential for use values in both marine and terrestrial geosites (Figure 5), it can be noticed that these values are similar and comparable to each other and no significant variation was identified. Notwithstanding the adoption of the same assessment methodology for terrestrial and marine geosites, some sub-criteria (accessibility, services, and economic potential) have been adapted considering the different characteristics between terrestrial and marine sites. This approach has allowed a balanced evaluation of geosites between the terrestrial and marine area. Moreover, the relationships between the two environments have been better defined.

**Figure 5.** Compared graphs referring to the average of the scientific, additional, and potential for use total values (cf. Section 4.2), calculated considering all the assessed terrestrial (inner circle) and marine geosites (outer circle), respectively.

At the same time, the use of independent assessment criteria permitted the individual analysis of the geosites and the identification of opportunities, weaknesses, and restrictions on tourism development. Indeed, these are fundamental steps for the enforcement of Park management strategies [129].

Multivariate representation (Figures 6 and 7) allows one to compare geosites to each other. Different colors indicate the main scientific interest (geomorphological, speleological, or hydrogeological) of each geosite and the form of the cartograms indicates groups of geosites (e.g., geosites with high scientific value but low potential for use value).

Regarding the scientific value, the assessment of the sub-criteria has revealed that the majority of geosites both terrestrial and marine are well preserved. This confirms the success of conservation strategies applied in the area by the Park authorities. Moreover, most of the geosites are fair to good examples of geological/geomorphological processes and landforms, in terms of level of representativeness (REP). Both terrestrial and marine geosites can be considered as rare at a regional scale, whereas the *Cristo degli Abissi* site (ID 21S) is exceptional at an international scale. High cliffs, more than 150 m a.s.l., set up along active normal faults, continue below sea-level and evolve due to retrogressive erosion, as witnessed by submerged rock fall deposits located on the sea bottom at different distances from the cliffs. These rock fall deposits are good examples of retrogressive processes favored by intense faulting and originated by gravity-induced processes. In addition, good examples of paleo-processes are offered by submerged caverns of structural origin, which are the continuation of land caves. Terrestrial conglomerate outcrops and submarine reliefs are good representatives of past paleo-environments. In particular, submarine reliefs (e.g., the *Secca Gonzatti* geosite—ID 8S) and saddles are ascribable to deep-seated gravitational slope deformations. Many terrestrial springs gushing in the conglomerate or at the boundary between conglomerate and marly-calcareous flysch, are found also below sea-level and witness the different uplift rate due to neotectonic activity between the emerged and submerged areas of the Park.

**Figure 6.** Multivariate representation of the results of terrestrial geosite assessment. Along the axes, the quantitative results of the assessment of the scientific, additional, and potential for use values and their sub-criteria are shown for each terrestrial geosite (for axis legend, see Figure 7). The colors represent different scientific interests, as specified in the figure legend.

**Figure 7.** Multivariate representation of the results of marine geosite assessment. Along the axes, the quantitative results of the assessment of the scientific, additional, and potential for use values and their sub-criteria are shown for each marine geosite. The colors represent different scientific interests, as specified in the figure legend.

The majority of terrestrial geosites is protected at a regional level, since it is included within the Portofino Regional Natural Park. The great majority of marine geosites is characterized by high ecological value; most of them are included in the protected marine area of type B, while the others are located within the type C protected area. Marine sites have a high ecological importance owing to the presence of coralline biocoenosis and prairies of *Posidonia oceanica* both on sandy and rocky seabeds. Furthermore, the presence of rocky blocks resulting from rock falls and topples originating in the overhanging conglomerate and, to a lesser extent, marly-calcareous flysch cliffs, favors exceptional biodiversity recognized at an international level.

Viewpoints generally have the greatest additional value due to their scenic quality and color diversity (aesthetic value, AEST). Among marine geosites, the *Baia di San Fruttuoso* (ID 23D), *Baia di Paraggi* (ID 25D), and the *Cristo degli Abissi* site (ID 21S) have the highest additional value due to their aesthetic quality and cultural relevance. For instance, the *Baia di San Fruttuoso* hosts the abbey bearing the same name, which dates back to the 10th century CE, while the internationally famous *Cristo degli Abissi* statue was placed in 1954 and has now assumed historical relevance.

The assessment of the potential for use is crucial in choosing management strategies aimed at the promotion of geotourism in the area; therefore, particular attention has been devoted to choosing the most suitable sub-criteria for this evaluation. The latter revealed that terrestrial geosites are generally easily accessible except from the cliffs, of course, which require climbing skills for their fruition, while panoramic points are generally the easiest accessible terrestrial geosites. The most accessible marine geosites are the ones reachable by snorkeling, while all the other sites are accessible only to second-level certified scuba divers.

Low values are recorded for services in the proximity of terrestrial geosites. In fact, despite the support services being generally located at a walkable distance, they are subject to seasonal availability (mainly in spring and summer). For marine geosites, the services—i.e., boarding docks—are mostly located at a distance of 4 to 7 km. Every year, tens of thousands of visitors choose to hike along the footpaths of the Portofino Natural Park (from 20,000 to more than 70,000 visitors per year), making the economic value of terrestrial geosites very high. As for marine geosites, the number of visitors, in terms of scuba dives per year, is two orders of magnitude lower than the terrestrial geosite visitors, except for snorkeling sites, which are attended as much as land ones. It should be mentioned that data on scuba diving are underestimated due to the presence of illegal non-registered scuba divers.

#### **6. Conclusions**

The Portofino Natural Park boast some of the most impressive sceneries of the Mediterranean area, displaying a large variety of geological landscapes as well as unique ecological systems, both in terrestrial and marine environment.

This research has led to the identification and assessment of 28 terrestrial and 27 marine geosites of the Portofino Natural Park and protected marine area, pinpointing the most suitable sites for geotourism promotion, for both their contribution to the understanding of the geological processes acting through time on landscapes as well as their aesthetic importance.

In fact, the area is a well-known seaside resort and the present economy is almost exclusively based on onshore and offshore tourism. Nevertheless, tourism activities focus mainly on the rich marine biota and habitats and for recreational purposes while the geological and geomorphological features are usually neglected. Instead, these features, including submarine ones, could play a relevant role in developing a sustainable and safe tourism fruition, thanks to a deeper understanding of the complex geological and geomorphological contexts.

Moreover, for the first time, geosite assessment has been performed by applying a common methodology to both terrestrial and marine geosites. Some sub-criteria (accessibility, services, and economic potential) have been adapted considering the different characteristics between terrestrial and marine sites. This approach has allowed us to emphasize the relationships between the terrestrial and marine environments. The selected geosite network is meant to show common processes and

landforms between these environments creating the ground for diversified but common and more efficient management and conservation actions and policies.

**Author Contributions:** Conceptualization, P.C. and V.V.; methodology, P.C. and V.V.; formal analysis, P.C. and V.V.; investigation, L.F., G.P., and F.F.; data curation, L.F., G.P., and F.F.; writing—original draft preparation, P.C., V.V., and F.F.; writing—review and editing, P.C., V.V., L.F., G.P., and F.F..; visualization, P.C., V.V., L.F., G.P., and F.F.; supervision, P.C. and F.F.

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

**Acknowledgments:** The authors would like to thank Giorgio Fanciulli and Alberto Girani for all the material provided and for the profitable discussions on the geosites of the Portofino Park and Portofino Marine Protected Area.

**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*
