Size/Age Models for Monitoring of the Pink Sea Fan Eunicella Verrucosa (Cnidaria: Alcyonacea) and a Case Study Application

Home , Eunicella, Eunicella verrucosa, Gorgoniidae, Holaxonia, Isidella, Savalia savaglia

Journal of Marine Science and Engineering

Article Size/Age Models for Monitoring of the Pink Sea Fan Eunicella verrucosa (Cnidaria: Alcyonacea) and a Case Study Application

Giovanni Chimienti 1,2,* , Attilio Di Nisio 3 and Anna M.L. Lanzolla 3

1 Department of Biology, University of Bari Aldo Moro, Via Orabona 4, 70125 Bari, Italy 2 CoNISMa, Piazzale Flaminio 9, 00196 Roma, Italy 3 Department of Electrical & Information Engineering (DEI), Polytechnic University of Bari, 70125 Bari, Italy; [email protected] (A.D.N.); [email protected] (A.M.L.L.) * Correspondence: [email protected]; Tel.: +39-080-544-3344

Received: 26 October 2020; Accepted: 19 November 2020; Published: 22 November 2020

Abstract: The pink sea fan Eunicella verrucosa is a habitat-forming octocoral living in the East Atlantic and in the Mediterranean Sea where, under proper circumstances, it can form large populations known as coral forests. Although these coral forests represent vulnerable marine ecosystems of great importance, these habitats are still poorly known, and their monitoring is almost non-existent to date. For this reason, we compared two diﬀerent models to infer the age of E. verrucosa based on nondestructive measurements of the colonies’ size, in order to highlight strengths and weaknesses of the existing tools for a potential application in long-term monitoring. We also applied the two models on a case-study population recently found in the northwest Mediterranean Sea. Our results showed which model was more reliable from a biological point of view, considering both its structure and the results obtained on the case study. However, this model uses solely the height of the colonies as proxy to infer the age, while the total branch fan surface area could represent a more appropriate biometric parameter to monitor the size and the growth of E. verrucosa.

Keywords: Anthozoa; gorgonian; corals; coral habitat; coral forest; VME; mesophotic; modeling; biometry; Mediterranean

1. Introduction The pink sea fan Eunicella verrucosa (Pallas, 1766) is an octocoral species (order Alcyonacea, family Gorgoniidae) living in the East Atlantic, from Ireland to Angola, and in the Mediterranean Sea [1–5]. This species can be found at a depth of 2 to 60 m in the Atlantic Ocean, while it can reach a depth of 200 m in the Mediterranean Sea [1,4–7]. Under proper oceanographic conditions including strong currents and temperate/cold waters, E. verrucosa can form large colony aggregations known as coral forests, settling on hard bottoms sometimes covered by a thin sediment veneer [5]. Similar to other animal forests (sensu [8]), octocoral assemblages provide biomass, structural complexity and aesthetic value to coastal communities, sustaining a rich biodiversity [5,9–17]. Coral forests, comprised of E. verrucosa represent a Vulnerable Marine Ecosystem (VME) because of their vulnerability to human pressures, according to the Food and Agricultural Organization (FAO) [18,19]. This vulnerability is based on the rarity, the functional significance, the fragility (both physical and functional), and the structural complexity of the coral forest, as well as the species’ life-history traits (e.g., slow growth rate, late age of maturity, low or unpredictable recruitment, and extended life expectancy) that makes recovery difficult after a fishing impact. Indeed, bottom-contact fishing gears such as trawl nets, dredges, longlines, and artisanal fishing nets (e.g., gillnets and trammel nets) can have both direct and

J. Mar. Sci. Eng. 2020, 8, 951; doi:10.3390/jmse8110951 www.mdpi.com/journal/jmse J. Mar. Sci. Eng. 2020, 8, 951 2 of 12 indirect impacts on these coral forests, representing one of the main anthropogenic threats together with pollution and climate change [5,20–26]. Furthermore, the stranding of E. verrucosa has been recently documented on the southwest coast of England, where the coral appeared to be entangled in lost fishing gears, as well as in domestic marine litter [23]. From the 1960s to the 1970s, colonies of E. verrucosa were collected as souvenirs in the northeast Atlantic, where the species has been currently protected from intentional damage [27]. It represents a major heritage species [28,29] and it is listed as “vulnerable” on the Red List of Threatened Species by the International Union for Conservation of Nature and Natural Resources (IUCN) [30]. More locally, the status of E. verrucosa in the Mediterranean Sea seems to be slightly better than elsewhere, as it is listed as “near threatened” according to IUCN [31]. In fact, records of E. verrucosa are increasing in this basin thanks to the implementation of deep diving techniques, the carrying of Remotely Operated Vehicle (ROV) explorations, as well as the application of a citizen science approach involving scuba divers and fishermen [5–7,25]. The emerging picture is that of a species widespread in the Mediterranean Sea, although large populations constituting coral forests are still poorly known [5]. Due to its vulnerability and ecological importance, E. verrucosa is receiving increasing attention and, similar to other forest-forming corals, further studies are likely to unveil new populations of this species in the mesophotic zone. Together with the finding of new coral forests aiming to assess the overall distribution of these VMEs, their monitoring is essential to understand ongoing and future changes (both natural and anthropogenic) [32]. Underwater observations and size measurements carried out on living colonies have allowed the development of two almost contemporary and independent models of growth and size/age relationship for E. verrucosa, one in the Atlantic Ocean [33] and one in the Mediterranean Sea [6]. In this study we compared the two existing models to infer the age of E. verrucosa based on the size of the colonies, in order to analyze pros and cons for long-term monitoring. Moreover, we applied the two models on a recently described forest of E. verrucosa in the northeast Mediterranean Sea [5] as a case study. New indications to improve age modeling and to obtain more reliable data on E. verrucosa populations are also provided.

2. Materials and Methods

2.1. Models Considered We considered two different models for the correlation of size and age in E. verrucosa. Model A [33] was developed based on 70 colonies tagged and surveyed between 2006 and 2008 in the Gulf of Morbihan (northwest Atlantic coast of France), at a depth of 8–20 m. The authors measured the colonies with in situ photographic recordings, and then used imaging analysis to estimate the total Branch Fan Surface Area (BFSA, cm2), namely, the measure of the surface that is occupied by the organism on photographs in two-dimensional (2D) projection (Figure1). The BFSA was calculated considering both height and width, and linearly correlated to the BFSA with different coefficients (Figure2a,b). A correction factor for branch overlapping was introduced considering that E. verrucosa colonies could exhibit a three-dimensional (3D) morphology. Individual growth was assessed by measuring the increase in the BFSA of tagged colonies during the three years of study, and the following model that correlated the age of the colony with the BFSA was proposed (Figure2a) [33]:

BFSA = 1.25 age1.73. (1) × J. Mar. Sci. Eng. 2020, 8, x FOR PEER REVIEW 3 of 13 J. Mar. Sci. Eng. 2020, 8, 951 3 of 12 J. Mar. Sci. Eng. 2020, 8, x FOR PEER REVIEW 3 of 13

Figure 1. Colony of Eunicella verrucosa with indication of height, width, and branch fan surface area (BFSA). Figure 1.1. Colony of Eunicella verrucosa with indication of height, width, and branch fan surface area (BFSA).(BFSA).

(a) (b)

Figure 2. Model A [33(].a) (a) Estimated behavior of the BFSA (branch fan( surfaceb) area) versus age Figurein Eunicella 2. Model verrucosa A [33]. ;((a)b Estimated) Linear correlation behavior of between the BFSA the (bra BFSAnch and fan heightsurface (blue) area) orversus width age (red) in of EunicellatheFigure colonies. verrucosa 2. Model; ( bA) [33].Linear (a )correlation Estimated betweenbehavior theof theBFSA BFSA and (bra heightnch (blue)fan surface or width area) (red) versus of theage in colonies.Eunicella verrucosa; (b) Linear correlation between the BFSA and height (blue) or width (red) of the The model was based on colonies having an initial BFSA between approximately 3 and 50 cm2. colonies. TheThe latter model size was corresponded based on colonies to an estimated having an age initial of about BFSA 30 between years and approximately an estimated height3 and 50 of cm 282 cm.. The same study reported the following linear relations between BFSA and height (h) and width The latterThe size model corresponded was based to on an colonies estimated having age of an about initial 30 BFSA years between and an estimated approximately height 3 ofand 28 50cm. cm 2. (w) (Figure2b), that have been considered in this study: TheThe latter same size study corresponded reported the to followingan estimated linear age relations of about between 30 years BFSAand an and estimated height ( hheight) and widthof 28 cm. (w) (FigureThe 2b), same that study have reported been considered the following in this linear study: relations between BFSA and height (h) and width h = 0.02 BFSA + 18.08. (2) (w) (Figure 2b), that have been consideredℎ = 0.02 in ×this× 𝐵𝐹𝑆𝐴 study: + 18.08, (2) wℎ= =0.03 0.02 ×BFSA 𝐵𝐹𝑆𝐴+ 14.29. + 18.08, (3) × (2) 𝑤 = 0.03 × 𝐵𝐹𝑆𝐴 + 14.29. (3) Model B [6] was developed based on 15 colonies surveyed every two years between 1997 and 2006 𝑤 = 0.03 × 𝐵𝐹𝑆𝐴 + 14.29. inModel the Riou B [6] archipelago was developed (Mediterranean based on coast15 colonies of France), surveyed at a depth every between two years 20 and between 40 m. 1997 The authorsand (3) measured in situ the maximum height of the colonies (between 5 and 40 cm) using a decimetre with 2006 in theModel Riou B archipelago[6] was developed (Mediterranean based on coast15 colonies of France), surveyed at a depth every between two years 20 between and 40 m. 1997 The and a millimetre scale. Individual growth rate was assessed by measuring the increase in the height of authors2006 inmeasured the Riou inarchipelago situ the maximum (Mediterranean height coast of the of France),colonies at(between a depth between5 and 40 20 cm) and using 40 m. aThe tagged colonies during the nine years of study, and the following Von Bertalanﬀy growth function that decimetreauthors with measured a millimetre in situ scale. the Individualmaximum growthheight rateof the was colonies assessed (between by measuring 5 and the 40 increase cm) using in a correlated height (h) and age of the colonies was proposed (Figure3)[6]: thedecimetre height of withtagged a millimetre colonies duringscale. Individual the nine yegrowthars of ratestudy, was and assessed the following by measuring Von theBertalanffy increase in growththe height function of thattagged correlated colonies height during (h) theand nine age ofye thears coloniesof study, was and proposed the following (Figure Von 3) [6]: Bertalanffy h = 17.94 ln(age) 18.39. (4) growth function that correlated height (h) and ×age of the− colonies was proposed (Figure 3) [6]:

J. Mar. Sci. Eng. 2020, 8, x FOR PEER REVIEW 4 of 13

J. Mar. Sci. Eng. 2020, 8, 951 ℎ = 17.94 × 𝑙𝑛 (𝑎𝑔𝑒) .− 18.39 4 of(4) 12

Figure 3. HeightHeight estimation estimation of of the the coloni colonieses as as a function of age in Eunicella verrucosa according to model B [[6].6].

For both models, the behavior of height as a function of age was analyzed to compare the performance of the two models.models. In In particular, particular, the following relationship was obtained from the modeling proposed in [[33]:33]: h = 18.08 + 2.5 age1.7. (5) ℎ = 18.08 + 2.5× × 𝑎𝑔𝑒.. (5) 2.2. Study Site

2.2. StudyThe extensiveSite population of E. verrucosa recently found off the city of Sanremo (Ligurian Sea, northwest Mediterranean) [5] was considered as a case study. High-resolution photos of E. verrucosa The extensive population of E. verrucosa recently found off the city of Sanremo (Ligurian Sea, colonies with a size reference (resolution of 1 cm) were collected by technical scuba dives in order northwest Mediterranean) [5] was considered as a case study. High-resolution photos of E. verrucosa to estimate height and width of the colonies. A total of 82 colonies, found at a depth of 65–70 m, colonies with a size reference (resolution of 1 cm) were collected by technical scuba dives in order to were photographed. Height and width values were estimated for each colony by means of image estimate height and width of the colonies. A total of 82 colonies, found at a depth of 65–70 m, were analysis with a raster graphics editor. photographed. Height and width values were estimated for each colony by means of image analysis The empirical cumulative distribution functions of height and width in the observed population with a raster graphics editor. were calculated and compared. Skewness and kurtosis, calculated for height and width, as well as for The empirical cumulative distribution functions of height and width in the observed population age, were not corrected for bias since no assumptions on distributions were made. were calculated and compared. Skewness and kurtosis, calculated for height and width, as well as The relation between width w and height h for the population of Sanremo was analyzed by for age, were not corrected for bias since no assumptions on distributions were made. calculating the linear regression of width on height for two models, i.e., w = a h + b and w = a h. The relation between width 𝑤and height ℎ for the population of Sanremo× was analyzed ×by The standard error of regression coefficients, their 95% confidence intervals, and p-values were calculating the linear regression of width on height for two models, i.e., 𝑤=𝑎×ℎ+𝑏 and 𝑤= evaluated by means of a t-statistic. The linear regression was compared with a scatter plot for the 𝑎×ℎ. The standard error of regression coefficients, their 95% confidence intervals, and p-values were population off Sanremo and with a regression illustrated in [33] relevant to model A. evaluated by means of a t-statistic. The linear regression was compared with a scatter plot for the The prediction of age from height for the population off Sanremo was calculated according to population off Sanremo and with a regression illustrated in [33] relevant to model A. models A and B. For h = 18.08 cm, age was zero according to model A. Hence, the result of that model The prediction of age from height for the population off Sanremo was calculated according to for colonies below 18.08 cm was considered to be zero years (indicating colonies younger than one models A and B. For ℎ = 18.08 cm, age was zero according to model A. Hence, the result of that model year). Age distribution according to the two models was analyzed by means of histograms with five for colonies below 18.08 cm was considered to be zero years (indicating colonies younger than one year bin widths, whose values were normalized to obtain a probability density function (i.e., relative year). Age distribution according to the two models was analyzed by means of histograms with five frequencies were divided by bin width). Histograms were smoothed by fitting a Gaussian kernel with year bin widths, whose values were normalized to obtain a probability density function (i.e., relative a bandwidth of 5 years, and the enforcement of positive age values were obtained by means of the frequencies were divided by bin width). Histograms were smoothed by fitting a Gaussian kernel with reflection method implemented by the ksdensity MATLAB function. a bandwidth of 5 years, and the enforcement of positive age values were obtained by means of the reflection3. Results method implemented by the ksdensity MATLAB function.

3.3.1. Results Comparing Two Diﬀerent Models Models A and B have a very diﬀerent structure. The comparison of the height as a function 3.1. Comparing Two Different Models of colonies’ age shows that model A is characterized by an exponential increase of height with age (Figure 4), a feature that is poorly representative from a biological point of view. On the contrary, J. Mar. Sci. Eng. 2020, 8, x FOR PEER REVIEW 5 of 13

J. Mar. Sci.Models Eng. 2020 A, 8 ,and x FOR B PEERhave REVIEW a very different structure. The comparison of the height as a function 5 of 13 of J.colonies’ Mar. Sci. Eng. age2020 shows, 8, 951 that model A is characterized by an exponential increase of height with5 ofage 12 model(Figure B seems 4), a to feature better thatrepresent is poorly the corals’representative trend of growth,from a biological with small point colonies of view. having On a thefirst contrary, stage of fastmodel growth B seems and to a flatteningbetter represent curve thewith corals’ increasing trend age of growth, (Figure with4). small colonies having a first stage model B seems to better represent the corals’ trend of growth, with small colonies having a ﬁrst stage of fast growth and a flattening curve with increasing age (Figure 4). of fast growth and a ﬂattening curve with increasing age (Figure4).

Figure 4. Comparison of height estimation as a function of age in Eunicella verrucosa by using model A andFigureFigure B. 4.4.Comparison Comparison of of height height estimation estimation as as a a function function of of age age in inEunicella Eunicella verrucosa verrucosaby by using using model model A andA and B. B. 3.2. Case Study 3.2. Case Study 3.2. Case Study The forest of E. verrucosa recently found off the city of Sanremo, in the Ligurian Sea, represented The forest of E. verrucosa recently found oﬀ the city of Sanremo, in the Ligurian Sea, represented a a good caseThe studyforest ofto E.test verrucosa the two recentlymodels duefound to offthe the presence city of ofSanremo, both small in the and Ligurian large colonies Sea, represented [5]. good case study to test the two models due to the presence of both small and large colonies [5]. a goodThe empirical case study distribution to test the functiontwo models of the due colonies to the ’presence size showed of both that small most and of the large population colonies had[5]. The empirical distribution function of the colonies’ size showed that most of the population had similar valuesThe empirical of height distribution and width, function while large-sizeof the colonies colonies’ size tended showed to that be widermost ofthan the highpopulation (Figure had similar values of height and width, while large-size colonies tended to be wider than high (Figure5a). 5a).similar On the values contrary, of heightcolonies and of smwidth,aller whilesize (likely large-size to be coloniesjuveniles) tended were generallyto be wider higher than than high wide, (Figure On the contrary, colonies of smaller size (likely to be juveniles) were generally higher than wide, considering5a). On the that contrary, 12% of thecolonies colonies of sm wasaller less size than (likely 10 cm to wide, be juveniles) while only were 6% generally of the colonies higher was than less wide, considering that 12% of the colonies was less than 10 cm wide, while only 6% of the colonies was thanconsidering 10 cm high that (Figure 12% of 5b). the coloniesIn fact, youngwas less colonies than 10 arecm wide,usually while unbranched only 6% ofor the with colonies very shortwas less less than 10 cm high (Figure5b). In fact, young colonies are usually unbranched or with very short branches,than 10 while cm highthey generally(Figure 5b). develop In fact, a fan young shape colonies in later years.are usually This is unbranchedin accordance or with with the very model short branches, while they generally develop a fan shape in later years. This is in accordance with the model proposedbranches, by whileCoz et they al. [33] generally in the Northdevelop Atlantic. a fan shape in later years. This is in accordance with the model proposed by Coz et al. [33] in the North Atlantic. proposed by Coz et al. [33] in the North Atlantic.

(a) (b)

Figure 5. (a) Empirical distribution functions of height and width of the Eunicella verrucosa colonies oﬀ (a) (b) FigureSanremo; 5. (a) (Empiricalb) Histogram distribution of height function and widths of for height classes and 10 width cm wide. of the Eunicella verrucosa colonies off Sanremo; (b) Histogram of height and width for classes 10 cm wide. MaximumFigure 5. (a height) Empirical and widthdistribution were function 58 and 61s of cm, height respectively. and widthThe of the size-frequency Eunicella verrucosa distribution colonies was mesokurtic,Maximumoff Sanremo; with height a(b slight) Histogramand positivewidth of were height skewness 58 and and width (Table 61 cm, for1). classesrespectively. 10 cm wide. The size-frequency distribution was mesokurtic, with a slight positive skewness (Table 1).

J. Mar. Sci. Eng. 2020, 8, x FOR PEER REVIEW 6 of 13

Maximum height and width were 58 and 61 cm, respectively. The size-frequency distribution was mesokurtic, with a slight positive skewness (Table 1). J. Mar. Sci. Eng. 2020, 8, 951 6 of 12 Table 1. Size statistics for the Eunicella verrucosa population monitored off Sanremo.

Standard TableMean 1. Size statisticsMedian for the Eunicella verrucosa populationSkewness monitored off Sanremo.Kurtosis Deviation (cm) (cm) (cm3/cm3) (cm4/cm4) Standard Deviation2 Mean (cm) Median (cm) (cm ) Skewness (cm3/cm3) Kurtosis (cm4/cm4) (cm2) Height 27.6 28.0 12.8 0.4 2.61 Height 27.6 28.0 12.8 0.4 2.6 1 WidthWidth 27.9 27.9 26.5 14.6 14.6 0.4 0.4 2.4 2.4 1 1ThisThis value value is wronglywrongly reported reported as as0.4 −0.4 in [ 5in]. [5]. − The test of the linear regression of width 𝑤 on height ℎ (𝑤=𝑎×ℎ+𝑏) on the population off The test of the linear regression of width w on height h (w = a h + b) on the population off Sanremo showed a slope value of 𝑎=1.05, with a standard error of 5%× and a 95% confidence interval Sanremo showed a slope value of a = 1.05, with a standard error of 5% and a 95% confidence interval equal to 0.95–1.14. The obtained intercept was 𝑏=− 1.0 cm. The 95% confidence interval was large, equal to 0.95–1.14. The obtained intercept was b = 1.0 cm. The 95% confidence interval was large, (i.e., −4.0, 1.9 cm), with p-value = 0.5 for the t-statistic− of the hypothesis test that the coefficient is null. (i.e., 4.0, 1.9 cm), with p-value = 0.5 for the t-statistic of the hypothesis test that the coefficient is Considering− the obtained results, the data were fitted by means of a linear regression without null. Considering the obtained results, the data were fitted by means of a linear regression without intercept, as illustrated below. The latter model is also more significant from a biological point of intercept, as illustrated below. The latter model is also more significant from a biological point of view, view, considering that at time zero, when the coral larva settles down, the size of the colony is null. considering that at time zero, when the coral larva settles down, the size of the colony is null. The regression of width on height (𝑤=𝑎×ℎ) for the measured colonies was characterized by a The regression of width on height (w = a h) for the measured colonies was characterized by a slope 𝑎=1.02 with a standard error of 2% and× a 95% confidence interval equal to 0.98–1.06 (Figure slope a = 1.02 with a standard error of 2% and a 95% confidence interval equal to 0.98–1.06 (Figure6). 6). The linear relation between width and height relevant to model A, obtained from [33], is The linear relation between width and height relevant to model A, obtained from [33], is characterized characterized by a slope of 1.5 (Figure 6). Considering the scatter plot of width and height (Figure 6), by a slope of 1.5 (Figure6). Considering the scatter plot of width and height (Figure6), it is evident that it is evident that each colony tended to be more developed in width at large sizes (records mostly each colony tended to be more developed in width at large sizes (records mostly above the quadrant above the quadrant bisector), while small colonies were higher than wide. This confirmed, at the level bisector), while small colonies were higher than wide. This confirmed, at the level of each colony of each colony (Figure 6), what was observed with the empirical distribution functions (Figure 5a) (Figure6), what was observed with the empirical distribution functions (Figure5a) where, instead, where, instead, width and height were treated independently. width and height were treated independently.

. Figure 6. Height and width of the Eunicella verrucosa colonies off Sanremo, with linear regression (red line).Figure In 6. yellow Height the and regression width of from the Eunicella model A verrucosa [33]. colonies off Sanremo, with linear regression (red line). In yellow the regression from model A [33]. Application of the Two Models 3.2.1. Application of the Two Models The age of the E. verrucosa colonies monitored off Sanremo, calculated on the basis of the height values,The showed age of dithefferent E. verrucosa results colonies between monitored model A and off Sanremo, model B (Figure calculated7). In on particular, the basis of model the height A is notvalues, valid showed when height different is less results than 18between cm, since model in that A and case model (5) cannot B (Figure be solved. 7). In This particular, happened model for 29%A is ofnot the valid colonies, when which height were, is less therefore, than 18 estimatedcm, since in to that have case an age(5) cannot of zero be years solved. (i.e., youngerThis happened than one for year,29% seeof the left colonies, side of Figure which7). were, Although therefore, all the estimated size classes to have from an 6 to age 58 cmof zero were years present (i.e., in younger the studied than populationone year, see (Figure left side5b), theof Figure prediction 7). Although from model all th Ae jumped size classes from from several 6 to colonies 58 cm were younger present than in one the yearstudied to a colonypopulation 16 years (Figure old(Figure 5b), the8a), prediction proving tofrom be scarcelymodel A e ffjumpedective for from small several sizes. colonies On the contrary, younger model B showed a gradient of predicted age from four to nine years old for the same small colonies, although the smallest colony (6 cm in height) could be likely to be younger than four years old. J. Mar. Sci. Eng. 2020, 8, x FOR PEER REVIEW 7 of 13 J. Mar. Sci. Eng. 2020, 8, x FOR PEER REVIEW 7 of 13 than one year to a colony 16 years old (Figure 8a), proving to be scarcely effective for small sizes. On thethan contrary, one year model to a colony B showed 16 years a gradient old (Figure of predicted 8a), proving age fromto be fourscarcely to nine effective years for old small for the sizes. same On smallthe contrary, colonies, model although B showed the smallest a gradient colony of (6 predicted cm in height) age from could four be likelyto nine to yearsbe younger old for than the samefour yearssmall old. colonies, although the smallest colony (6 cm in height) could be likely to be younger than four J.years Mar. Sci. old. Eng. 2020, 8, 951 7 of 12

Figure 7. Age of the Eunicella verrucosa colonies off Sanremo, obtained from the height measure, using FiguretheFigure two 7. considered7.Age Age ofof thethe models.Eunicella Eunicella verrucosaverrucosacolonies colonies o offﬀ Sanremo, Sanremo, obtained obtained from from the the height height measure, measure, using using thethe two two considered considered models. models. age (y) age age (y) age

(a) (b) (a) (b) Figure 8. Age of the Eunicella verrucosa colonies off Sanremo estimated using model A and B. (Figurea) Empirical 8. Age distribution of the Eunicella function; verrucosa (b) Box colonies plot showing off Sanremo median estimated (red), first using and thirdmodel quartile A and (blue),B. (a) rangeEmpiricalFigure (whiskers), 8. distributionAge of andthe outliersEunicellafunction; (red verrucosa(b) crosses).Box plotcolonies Here,showing off outliers Sanremomedian are (red), valuesestimated first at a andusing distance third model abovequartile A and the (blue), thirdB. (a ) quartilerangeEmpirical (whiskers), or belowdistribution theand first outliers function; quartile (red ( greaterb )crosses). Box thanplot Here, 1.5showing times outliers themedian interquartileare values (red), firstat range. a distanceand third above quartile the (blue),third quartilerange (whiskers), or below the and first outliers quartile (red greater crosses). than Here, 1.5 times outliers the interquartile are values at range. a distance above the third Withquartile an or increasing below the first size, quartile the two greater models than 1.5 diverged times the considerably, interquartile range. with model B being more conservativeWith an thanincreasing model size, A (Figure the two8b). models On the highestdiverged size considerably, found, i.e., 58with cm, model the models B being showed more almostconservativeWith the samean than increasing estimated model Asize, age (Figure (approximately the two8b). Onmodels the 70 highest yearsdiverged old). size considerably, This found, is likely i.e., 58 towith becm, closemodel the tomodels theB being maximum showed more heightalmostconservative thatthe sameE. verrucosathan estimated modelcan ageA reach, (Figure (approximately according 8b). On to the present70 yearshighest knowledge, old). size This found, is and likely i.e., is the to58 be pointcm, close the of to contactmodels the maximum between showed theheightalmost two that the curves E. same verrucosa (Figure estimated 7can). Larger reach, age (approximately sizesaccording could to eventually present 70 years knowledg showold). This discordancee, isand likely is the to between point be close of thecontact to the two maximum between models, consideringtheheight two thatcurves E. that verrucosa (Figure model 7). can A Larger reach, predicts sizes according a youngercould to even present agetually as knowledgcompared show discordancee, withand is model the between point B for of the contact heights two models,between above 58consideringthe cm two (Figure curves that7). (Figuremodel A7). predicts Larger sizesa younger could age even astually compared show with discordance model B between for heights the above two models, 58 cm (FigureconsideringThe 7). largest that colonies model A were predicts of a fewa younger large, oldage onesas compared that were with included model in B thefor boxplotheights above of model 58 cm A (Figure(Figure8 b),7). while they were present as outliers for model B. In the latter, the interquartile range was smaller as compared with model A, highlighting the higher conservativism of model B. The predicted age-frequency distributions are shown in Figure9a,b as histograms and as fittings with a Gaussian kernel. It further highlighted the presence of a gap for small sizes in model A (Figure9a), and a distribution of the data that seemed to suggest that model B was more reliable from a biological point of view. In fact, the latter showed the presence of a highly skewed and leptokurtic J. Mar. Sci. Eng. 2020, 8, x FOR PEER REVIEW 8 of 13

The largest colonies were of a few large, old ones that were included in the boxplot of model A (Figure 8b), while they were present as outliers for model B. In the latter, the interquartile range was smaller as compared with model A, highlighting the higher conservativism of model B. The predicted age-frequency distributions are shown in Figure 9a, b as histograms and as fittings with a Gaussian kernel. It further highlighted the presence of a gap for small sizes in model A (Figure 9a), and a distribution of the data that seemed to suggest that model B was more reliable from a J. Mar. Sci. Eng. 2020, 8, 951 8 of 12 biological point of view. In fact, the latter showed the presence of a highly skewed and leptokurtic distribution (Figure 9b and Table 2), with many colonies younger than 20 years old and a long right taildistribution of older colonies (Figure9 representedb and Table2 ),by with those many broadly colonies higher younger than 35 than cm. 20 years old and a long right tail of older colonies represented by those broadly higher than 35 cm.

(a) (b)

Figure 9. Age-frequency distribution of the Eunicella verrucosa population monitored oﬀ Sanremo, by Figuremodel. 9. ( aAge) Model-frequency A; (b) distribution Model B. The of histograms the Eunicella have verrucosa beennormalized population tomonitored become comparableoff Sanremo, with by model.the estimated (a) Model probability A; (b) M densityodel B. The function. histograms The bin have width been of normalized the histogram to become and the comparable bandwidth ofwith the theGaussian estimated kernel probability are both density set to 5 years.function. The bin width of the histogram and the bandwidth of the Gaussian kernel are both set to 5 years. Table 2. Age statistics for the Eunicella verrucosa population monitored oﬀ Sanremo, y = years.

Table 2. Age statistics for the EunicellaStandard verrucosa Deviation population monitored off Sanremo, y = years. Mean (y) Median (y) Skewness (y3/y3) Kurtosis (y4/y4) (y2) Standard Deviation Model AMean 27.9 (y) Median 31.8 (y) 21.9Skewness 0.1 (y3/y3) Kurtosis 1.9 (y4/y4) Model B 17.0 13.3 14.3(y2) 2.0 6.8 1 Model A 27.9 31.81 This value is wrongly21.9 reported as 4.2 in [5]. 0.1 1.9 Model B 17.0 13.3 14.3 2.0 6.81 1 This value is wrongly reported as 4.2 in [5]. 4. Discussion 4. DiscussionThe two size/age models available, thus far, for monitoring of E. verrucosa [6,33] were developed in twoThe different two size/age areas, i.e., models the northwest available Mediterranean, thus far, for monitoring Sea and the of northeast E. verrucosa Atlantic [6,33] Ocean. were Thesedeveloped areas inare two characterized different areas by di, i.e.,fferent the environmental northwest Mediterranean conditions (e.g., Sea temperature,and the northeast water Atlantic currents, Ocean. and depth These of areasoccurrence) are characterized which affect by trophism different and environmental coral physiology, conditions influencing (e.g., the temperature, growth rate water of E. verrucosa currents,[ 6and,33]. depthThus, of the occurrence) difference betweenwhich affect the trophism results of and the twocoral models physiology, observed influencing in this studythe growth could rate be partlyof E. verrucosadue to the [6,33]. different Thus, areas the ofdifference application. between However, the results a standardized of the two method models to observed estimate thein this age study of the couldcolonies be partly based due on theirto the size different should areas be adjusted of application. also considering However, a the standardized study area, method at least atto theestimate basin thelevel age (e.g., of the introducing colonies based a di ffonerent their correction size should factor be adjusted for the Atlanticalso consideri Oceanng and the study the Mediterranean area, at least atSea the populations). basin level (e.g. Model, introducing B seemed toa be different stronger correction than model factor A, considering for the Atlantic both Ocean the structure and the of Mediterraneanthe formula (Figure Sea populations4) and the) results. Model of B theseemed case to study be stronger (Figure than7). This model is relatedA, considering to the fact both that, the structureindependently of the fromformula the study(Figure area, 4) and the relationshipthe results of between the case size study and (Figure age in animals,7). This is including related to corals, the factfollows that, a independently von Bertalanff y from curve the where study young area, stages the relationship grow very betweenfast with size a decrease and age in growthin animals, rate according to age [34–36]. Notwithstanding, the height of the colonies used in model B can be limiting, considering the importance of width measures in E. verrucosa, at least for the larger colonies (Figure6). The importance of width measures has also been highlighted in other congeneric species, such as E. singularis [33,37]. In fact, modular colonial organisms such as E. verrucosa and other sea fans form colonies by polyp iteration, and therefore the total branch fan surface area (BFSA), which is a function of both height and width, is probably the most appropriate estimate of its growth and the more suitable measure to be linked to the age [33,38]. J. Mar. Sci. Eng. 2020, 8, 951 9 of 12

The greatest limitation of both model A and B is that the predicted age of the colonies is based on indirect information rather than direct measures. For this reason, it is essential to improve these models or to identify a new one based on the empirical measure of the colonies’ age using geochemical dating methods (e.g., radiocarbon), already adopted for the ageing of other coral species [39–41], as well as growth ring analysis, as carried out on other Eunicella species, although with some limitations due to poorly differentiated rings at the periphery of the skeleton of large colonies [42,43]. These methods are based on the analysis of the basal cross-section of the colony, being unsustainable for long-term monitoring surveys, as well as incompatible for ethical and conservation reasons, considering that E. verrucosa represents a vulnerable species [30]. This important VME indicator taxa (sensu [19]) is affected by bottom-contact fishing gears and it might occur as bycatch during fishing operations. In fact, the exclusion of demersal towed equipment has benefited E. verrucosa populations and other benthic taxa in the Lyme Bay Marine Protected Area (MPA; southwest England) [44], emphasizing the role of MPAs for the conservation of vulnerable and threatened species. However, bottom longlines and artisanal fishing nets (e.g., trammel nets and gillnets), often considered of scarce impact on benthic communities, also have a relevant bycatch rate on mesophotic and deep-sea coral species [24,45,46]. These gears can have a significant impact on E. verrucosa, as they can bycatch the colonies during fishing operations, but also lost fishing gears can be entangled in the colonies and damage them (ghost fishing) [5,23]. The exploitation of fishing data can allow for the identification and recording of vulnerable coral communities [47–51]. The involvement of fishermen could help to keep these accidentally collected colonies, in order to maximize the bycatch and provide specimens for invasive analysis aiming to understand the longevity and the growth rate of this species. These data could also be used to evaluate the impact of fishing gear on this vulnerable species, while non-destructive survey methods (i.e., ROV and technical diving) should be incentivized to identify and to monitor coral populations as a valid alternative to destructive sampling. The possibility to directly measure the age of E. verrucosa colonies based on samples represents an opportunity to correlate size and age in this species, in order to build a stronger model. Moreover, the finding of a rich E. verrucosa forest would represent a population in the Atlantic Ocean that could provide a better comparison with the Mediterranean ones studied thus far. Such information is essential to the development of a standardized and non-invasive method for long-term monitoring of E. verrucosa and its population structure. This could lead to visual methods for identification, quantification, and eventually, measurement of the colonies. For instance, visible, multispectral, or hyperspectral imaging could be used, as already has been carried out for other coral species [52–55], as well as in other fields of research, from the study of animals and plants [56–58] to marine biogeochemistry [59].

Author Contributions: Conceptualization, G.C., A.D.N., and A.M.L.L.; methodology, G.C., A.D.N., and A.M.L.L.; software, A.D.N. and A.M.L.L.; validation, G.C.; formal analysis, A.D.N. and A.M.L.L.; investigation, G.C.; resources, G.C., A.D.N., and A.M.L.L.; data curation, G.C.; writing—original draft preparation, G.C.; writing—review and editing, G.C., A.D.N., and A.M.L.L.; visualization, G.C., A.D.N., and A.M.L.L.; supervision, G.C., A.D.N., and A.M.L.L.; project administration, G.C.; funding acquisition, G.C. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by the Italian Ministry of Education, University and Research (Ministero dell’Istruzione, dell’Università e della Ricerca; Programma Operativo Nazionale, PON 2014–2020), grant AIM 1807508-1, Linea 1, and by the Italian Ministry for Environment, Land and Sea Protection (Ministero dell’Ambiente e della Tutela del Territorio e del Mare), as part of the Italian monitoring program for the implementation of the Marine Strategy Framework Directive (European Union, 2008/56/EC). Acknowledgments: The authors wish to thank Maurizio Delfini for the support with underwater videos and photos. 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. J. Mar. Sci. Eng. 2020, 8, 951 10 of 12

References

1. Carpine, C.; Grasshoff, M. Les Gorgonaires de la Mediterranean. Bull. Inst. Océanogr. Monaco 1975, 71, 1–140. 2. Carpine, C. Contribution à la connaissance des Gorgones Holaxonia de la Mediterranean occidentale. Bull. Inst. Océanogr. Monaco 1963, 60, 1–52. 3. Weinberg, S. Revision of the common Octocorallia of the Mediterranean circalittoral. I. Gorgonacea. Beaufortia 1976, 24, 63–104. 4. Grasshoff, M. Die Flachwasser-Gorgonarien von Europa und Westafrika (Cnidaria, Anthozoa); Courier Forschungsinstitut Senckenberg: Berlin, Germany, 1992; pp. 1–135. 5. Chimienti, G. Vulnerable forests of the pink sea fan Eunicella verrucosa in the Mediterranean Sea. Diversity 2020, 12, 176. [CrossRef] 6. Sartoretto, S.; Francour, P. Bathymetric distribution and growth rates of Eunicella verrucosa (Cnidaria: Gorgoniidae) populations along the Marseilles coast (France). Sci. Mar. 2012, 76, 349–355. [CrossRef] 7. Di Camillo, C.G.; Ponti, M.; Bavestrello, G.; Krzelj, M.; Cerrano, C. Building a baseline for habitat-forming corals by a multi-source approach, including Web Ecological Knowledge. Biodivers. Conserv. 2018, 27, 1257–1276. [CrossRef] 8. Rossi, S.; Bramanti, L.; Gori, A.; Orejas, C. An overview of the animal forests of the world. In Marine Animal Forests; Rossi, S., Ed.; Springer International Publishing: Berlin, Germany, 2017; pp. 1–26. 9. Gili, J.M.; Coma, R. Benthic suspension feeders: Their paramount role in littoral marine food webs. Trends Ecol. Evol. 1998, 13, 316–321. [CrossRef] 10. Cerrano, C.; Danovaro, R.; Gambi, C.; Pusceddu, A.; Riva, A.; Schiaparelli, S. Gold coral (Savalia savaglia) and gorgonian forests enhance benthic biodiversity and ecosystem functioning in the mesophotic zone. Biodivers. Conserv. 2010, 19, 153–167. [CrossRef] 11. Angeletti, L.; Taviani,M.; Canese, S.; Foglini, F.; Mastrototaro, F.; Argnani, A.; Trincardi, F.; Bakran–Petricioli, T.; Ceregato, A.; Chimienti, G.; et al. New deep-water cnidarian sites in the southern Adriatic Sea. Mediterr. Mar. Sci. 2014, 15, 225–238. 12. Ponti, M.; Perlini, R.A.; Ventra, V.; Grech, D.; Abbiati, M.; Cerrano, C. Ecological shifts in Mediterranean coralligenous assemblages related to gorgonian forest loss. PLoS ONE 2014, 9, e102782. [CrossRef][PubMed] 13. Ponti, M.; Turicchia, E.; Ferro, F.; Cerrano, C.; Abbiati, M. The understorey of gorgonian forests in mesophotic temperate reefs. Aquat. Conserv. Mar. Freshw. Ecosyst. 2018, 28, 1153–1166. [CrossRef] 14. Grinyó, J.; Gori, A.; Ambroso, S.; Purroy, A.; Calatayud, C.; Dominguez-Carrió, C.; Coppari, M.; Iacono, C.L.; López-González, P.J.; Gili, J.M. Diversity, distribution and population size structure of deep Mediterranean gorgonian assemblages (Menorca Channel, Western Mediterranean Sea). Prog. Oceanogr. 2016, 145, 42–56. [CrossRef] 15. Mastrototaro, F.; Chimienti, G.; Acosta, J.; Blanco, J.; Garcia, S.; Rivera, J.; Aguilar, R. Isidella elongata (Cnidaria: Alcyonacea) facies in the western Mediterranean Sea: Visual surveys and descriptions of its ecological role. Eur. Zool. J. 2017, 84, 209–225. [CrossRef] 16. Chimienti, G.; De Padova, D.; Mossa, M.; Mastrototaro, F. A mesophotic black coral forest in the Adriatic Sea. Sci. Rep. 2020, 10, 1–15. [CrossRef] 17. Chimienti, G.; Angeletti, L.; Furfaro, C.; Taviani, M. Habitat, morphology and trophism of Tritonia callogorgiae sp. nov., a large nudibranch inhabiting Callogorgia verticillata forests in the Mediterranean Sea. Deep Sea Res. I Oceanogr. Res. Pap. 2020, 165, 103364. [CrossRef] 18. Food and Agriculture Organization (FAO). International Guidelines for the Management of Deep-Sea Fisheries in the High Seas; FAO: Rome, Italy, 2009; pp. 1–21. 19. FAO (Food and Agriculture Organization). Abandoned, lost or otherwise discarded gillnets and trammel nets: Methods to estimate ghost fishing mortality, and the status of regional monitoring and management. In FAO Fisheries and Aquaculture Technical Paper No. 600; Gilman, E., Chopin, F., Suuronen, P., Kuemlangan, B., Eds.; FAO: Rome, Italy, 2016; pp. 1–79. 20. Watling, L.; Norse, E.A. Disturbance of the seabed by mobile fishing gear: A comparison to forest clearcutting. Conserv. Biol. 1998, 12, 1180–1197. [CrossRef] 21. Stone, R.P. Coral habitat in the Aleutian Islands of Alaska: Depth distribution, fine-scale species associations, and fisheries interactions. Coral Reefs 2006, 25, 229–238. [CrossRef] J. Mar. Sci. Eng. 2020, 8, 951 11 of 12

22. Bo, M.; Bava, S.; Canese, S.; Angiolillo, M.; Cattaneo-Vietti, R.; Bavestrello, G. Fishing impact on deep Mediterranean rocky habitats as revealed by ROV investigation. Biol. Conserv. 2014, 171, 167–176. [CrossRef] 23. Sheehan, E.V.; Rees, A.; Bridger, D.; Williams, T.; Hall-Spences, J.M. Strandings of NE Atlantic gorgonians. Biol. Conserv. 2017, 209, 482–487. [CrossRef] 24. Chimienti, G.; Mastrototaro, F.; D’Onghia, G. Mesophotic and Deep-Sea Vulnerable Coral Habitats of the Mediterranean Sea: Overview and Conservation Perspectives. In Advances in the Studies of the Benthic Zone; Soto, L.A., Ed.; IntechOpen: London, UK, 2019; pp. 1–20. 25. Enrichetti, F.; Dominguez-Carrió, C.; Toma, M.; Bavestrello, G.; Betti, F.; Canese, S.; Bo, M. Megabenthic communities of the Ligurian deep continental shelf and shelf break (NW Mediterranean Sea). PLoS ONE 2019, 14, e0223949. [CrossRef][PubMed] 26. Morato, T.; González-Irusta, J.M.; Dominguez-Carrió, C.; Wei, C.L.; Davies, A.; Sweetman, A.K.; Taranto, G.H.; Beazley, L.; García-Alegre, A.; Grehan, A.; et al. Climate-induced changes in the suitable habitat of cold-water corals and commercially important deep-sea fishes in the North Atlantic. Glob. Chang. Biol. 2020, 26, 2181–2202. [CrossRef][PubMed] 27. HMSO Wildlife and Countryside Act. 1981 (London). Available online: www.legislation.gov.uk/ukpga/1981/ 69/section/28P (accessed on 29 September 2020). 28. Hiscock, K. Where have all the corals gone? Mar. Conserv. 2003, 6, 8–9. 29. Pikesley, S.K.; Godley, B.J.; Latham, H.; Richardson, P.B.; Robson, L.M.; Solandt, J.L.; Trundle, C.; Wood, C.; Witt, M.J. Pink sea fans (Eunicella verrucosa) as indicators of the spatial efficacy of Marine Protected Areas in southwest UK coastal waters. Mar. Policy 2016, 64, 38–45. [CrossRef] 30. The IUCN Red List of Threatened Species. Version 2020-2. Available online: www.iucnredlist.org (accessed on 29 September 2020). 31. Otero, M.M.; Numa, C.; Bo, M.; Orejas, C.; Garrabou, J.; Cerrano, C.; Kružić,P.; Antoniadou, C.; Aguilar, R.; Kipson, S.; et al. Overview of the Conservation Status of Mediterranean Anthozoans; IUCN: Málaga, Spain, 2017; pp. 1–73. 32. Chimienti, G.; Bo, M.; Mastrototaro, F. Know the distribution to assess the changes: Mediterranean cold-water coral bioconstructions. Rend. Lincei Sci. Fis. Nat. 2018, 29, 583–588. [CrossRef] 33. Coz, R.; Ouisse, V.; Artero, C.; Carpentier, A.; Crave, A.; Feunteun, E.; Olivier, J.M.; Perrin, B.; Ysnel, F. Development of a new standardised method for sustainable monitoring of the vulnerable pink sea fan Eunicella verrucosa. Mar. Biol. 2012, 159, 1375–1388. [CrossRef] 34. Mistri, M.; Ceccherelli, V.U. Growth of the Mediterranean gorgonian Lophogorgia ceratophyta (L., 1758). Mar. Ecol. 1993, 14, 329–340. [CrossRef] 35. Coma, R.; Ribes, M.; Zabala, M.; Gili, J.M. Growth in a modular colonial marine invertebrates. Estuar. Coast. Shelf Sci. 1998, 47, 459–470. [CrossRef] 36. Andrews, A.H.; Cordes, E.E.; Mahoney, M.M.; Munk, K.; Coale, K.H.; Caillet, G.M.; Heifetz, J. Age, growth and radiometric age validation of a deep-sea, habitat-forming gorgonian (Primnoa resedaeformis) from the Gulf of Alaska. Hydrobiologia 2002, 471, 101–110. [CrossRef] 37. Skoufas, G.; Poulicek, M.; Chintiroglou, C.C. Growth variation of Eunicella singularis (Esper, 1794) (Gorgonacea, Anthozoa). Belg. J. Zool. 2000, 130, 125–128. 38. Weinbauer, M.; Velimirov, B. Population dynamics and overgrowth of the sea fan Eunicella cavolinii (Coelenterata: Octocorallia). Estuar. Coast. Shelf. Sci. 1996, 42, 583–595. [CrossRef] 39. Roark, E.; Guilderson, T.P.; Dunbar, R.B.; Ingram, B. Radiocarbon-based ages and growth rates of Hawaiian deep-sea corals. Mar. Ecol. Prog. Ser. 2006, 327, 1–14. [CrossRef] 40. Bo, M.; Bavestrello, G.; Angiolillo, M.; Calcagnile, L.; Canese, S.; Cannas, R.; Cau, A.; D’Elia, M.; D’Oriano, F.; Follesa, M.C.; et al. Persistence of Pristine Deep-Sea Coral Gardens in the Mediterranean Sea (SW Sardinia). PLoS ONE 2015, 10, e0119393. [CrossRef][PubMed] 41. Lartaud, F.; Mouchi, V.; Chapron, L.; Meistertzheim, A.L.; Le Bris, N. Growth patterns of Mediterranean calcifying cold-water corals. In Mediterranean Cold-Water Corals: Past, Present and Future; Orejas, C., Jimennez, C., Eds.; Springer International Publishing: Berlin, Germany, 2019; pp. 405–422. 42. Grigg, R.W. Growth rings: Annual periodicity in two gorgonians corals. Ecology 1974, 55, 876–881. [CrossRef] 43. Goffredo, S.; Lasker, H.R. Modular growth of a gorgonian coral can generate predictable patterns of colony growth. J. Exp. Mar. Biol. Ecol. 2006, 336, 221–229. [CrossRef] J. Mar. Sci. Eng. 2020, 8, 951 12 of 12

44. Sheehan, E.V.; Stevens, T.F.; Gall, S.C.; Cousens, S.L.; Attrill, M.J. Recovery of a temperate reef assemblage in a Marine Protected Area following the exclusion of towed demersal fishing. PLoS ONE 2013, 8, e83883. [CrossRef] 45. Sampaio, I.; Braga-Henriques, A.; Pham, C.; Ocaña, O.; De Matos, V.; Morato, T.; Porteiro, F.M. Cold-water corals landed by bottom longline fisheries in the Azores (north-eastern Atlantic). J. Mar. Biol. Assoc. UK 2012, 92, 1547–1555. [CrossRef] 46. Mytilineou, C.; Smith, C.J.; Anastasopoulou, A.; Papadopoulou, K.N.; Christidis, G.; Bekas, P.; Kavadas, S.; Dokos, J. New cold-water coral occurrences in the eastern Ionian Sea: Results from experimental long line fishing. Deep Sea Res. Part. II Top. Stud. Oceanogr. 2014, 99, 146–157. [CrossRef] 47. Kenchington, E.; Murillo, F.J.; Lirette, C.; Sacau, M.; Koen-Alonso, M.; Kenny, A.; Ollerhead, N.; Wareham, V.; Beazley, L. Kernel density surface modelling as a means to identify significant concentrations of vulnerable marine ecosystem indicators. PLoS ONE 2014, 9, e109365. [CrossRef] 48. Terribile, K.; Evans, J.; Knittweis, L.; Schembri, P.J. Maximising MEDITS: Using data collected from trawl surveys to characterise the benthic and demersal assemblages of the circalittoral and deeper waters around the Maltese Islands (Central Mediterranean). Reg. Stud. Mar. Sci. 2016, 3, 163–175. [CrossRef] 49. Chimienti, G.; Angeletti, L.; Rizzo, L.; Tursi, A.; Mastrototaro, F. ROV vs. trawling approaches in the study of benthic communities: The case of Pennatula rubra (Cnidaria: Cennatulacea). J. Mar. Biol. Assoc. UK 2018, 98, 1859–1869. [CrossRef] 50. Bastari, A.; Pica, D.; Ferretti, F.; Micheli, F.; Cerrano, C. Sea pens in the Mediterranean Sea: Habitat suitability and opportunities for ecosystem recovery. ICES J. Mar. Sci. 2018, 75, 1722–1732. [CrossRef] 51. Chimienti, G.; Di Nisio, A.; Lanzolla, A.M.L.; Andria, G.; Tursi, A.; Mastrototaro, F. Towards non-invasive methods to assess population structure and biomass in vulnerable sea pen fields. Sensors 2019, 19, 2255. [CrossRef][PubMed] 52. Gleason, A.C.R.; Reid, R.P.; Voss, K.J. Automated classification of underwater multispectral imagery for coral reef monitoring. In Proceedings of the OCEANS 2007, Vancouver, BC, Canada, 29 September–4 October 2007; pp. 1–8. 53. Phinn, S.R.; Roelfsema, C.M.; Mumby, P.J. Multi-scale, object-based image analysis for mapping geomorphic and ecological zones on coral reefs. Int. J. Remote Sens. 2012, 33, 3768–3797. [CrossRef] 54. Foglini, F.; Grande, V.; Marchese, F.; Bracchi, V.A.; Prampolini, M.; Angeletti, L.; Castellan, G.; Chimienti, G.; Hansen, I.M.; Gudmundsen, M.; et al. Underwater Hyperspectral Imaging for seafloor and benthic habitat mapping in the southern Adriatic Sea (Italy). In Proceedings of the 2018 IEEE International Workshop on Metrology for the Sea, Learning to Measure Sea Health Parameters (MetroSea), Bari, Italy, 8–10 October 2018; pp. 201–205. 55. Foglini, F.; Grande, V.; Marchese, F.; Bracchi, V.A.; Prampolini, M.; Angeletti, L.; Castellan, G.; Chimienti, G.; Hansen, I.M.; Gudmundsen, M.; et al. Application of Hyperspectral Imaging to Underwater Habitat Mapping, Southern Adriatic Sea. Sensors 2019, 19, 2261. [CrossRef] 56. Dumke, I.; Purser, A.; Marcon, Y.; Nornes, S.M.; Johnsen, G.; Ludvigsen, M.; Søreide, F. Underwater hyperspectral imaging as an in situ taxonomic tool for deep-sea megafauna. Sci. Rep. 2018, 8, 12860. [CrossRef] 57. Di Nisio, A.; Adamo, F.; Acciani, G.; Attivissimo, F. Fast Detection of Olive Trees Affected by Xylella fastidiosa from UAVs Using Multispectral Imaging. Sensors 2020, 20, 4915. [CrossRef] 58. Volent, Z.; Johnsen, G.; Sigernes, F. Kelp forest mapping by use of airborne hyperspectral imager. J. Appl. Remote Sens. 2007, 1, 011505. 59. Johnsen, G.; Volent, Z.; Dierssen, H.; Pettersen, R.; Ardelan, M.V.; Søreide, F.; Fearns, P.; Ludvigsen, M.; Moline, M. Underwater hyperspectral imagery to create biogeochemical maps of seafloor properties. In Subsea Optics and Imaging, 1st ed.; Watson, J., Zielinski, O., Eds.; Elsevier: Amsterdam, The Netherlands, 2013; pp. 508–540.

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional aﬃliations.

© 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/).