Acoustic scattering properties of a seagrass, Cymodocea nodosa: in-situ measurements

1 Introduction

Along with Posidonia oceanica (Linnaeus) Delile 1813, Cymodocea nodosa (Ucria) Ascherson 1870 is one of the most common seagrasses in the Mediterranean and other seas of the world, and is sensitive to pollution (McGonigle et al. 2011). In the Mediterranean Sea, C. nodosa lives in colonies, mainly in the infralittoral zone (Mutlu et al. 2022a,b). Besides their ecological importance, they have been used to determine the lower limit of the infralittoral of the benthic zone (Colantoni et al. 1982; Pal and Hogland 2022). Cymodocea nodosa creates an important ecological niche and plays central roles in coastal management and engineering (Brun et al. 2009; Chefaoui et al. 2016; Koch 1994; Pergent et al. 2014), being the most widely distributed and highest-abundance seagrass in the Mediterranean Sea, after P. oceanica. Therefore, it is recommended to monitor the coverage, canopy height and biometrics (leaf length, width, vertical and horizontal rhizome and sheath length, weight) of C. nodosa, in order to assess the status of the marine coastal environment.

Seagrasses are under protection in most countries, and destructive sampling methods are avoided (Gobert et al. 2020; Mutlu et al. 2023; Pergent et al. 1995). Therefore, the use of remote sensing methods for monitoring seagrass and bottom types, as well as classifying habitat types, is promoted (Gobert et al. 2020; Montefalcone et al. 2021; Zhu et al. 2021). There are a variety of remote sensing techniques, including satellites (especially Sentinel 2), video cameras, and acoustics (Jaubert et al. 2003; Mielck et al. 2014; Noiraksar et al. 2014; Randall et al. 2014, 2020; Robinson et al. 2011; Ware and Downie 2020). Like every technique, these remote sensing techniques require suitable atmospheric and sea conditions to measure data from the seagrass (Hossain and Hasmin 2019; McCarthy and Sabol 2000; Vis et al. 2003). Recent remote sensing studies have focused on the collection of quantitative measurements such as bottom coverage, classification and assessment of the biomass of seaweeds (e.g. Dimas et al. 2022; Fakiris et al. 2018; Mutlu and Olguner 2023a).

The biometric parameters of seagrasses play a crucial role in the determination of their population dynamics, seasonality, ecology, management, sustainability, protection and blue carbon content (Brown et al. 2011; Colantoni et al. 1982; Pal and Hogland 2022). Acoustic methods of quantifying biometric parameters have been advanced for marine vegetation, as they are faster, more precise, easier to use, and less dependent on sea conditions than other remote sensing methods (Brown et al. 2011; van Rein et al. 2011). Nevertheless, calibration, sea-truthing and numerical processes are needed for successful in or ex situ studies involving the acoustic quantification of seaweeds (Lurton 2002; Mutlu and Balaban 2018; Mutlu and Olguner 2023a).

In the present study area, spatiotemporal dynamics of the biometrics of C. nodosa were described using a destructive method (SCUBA) by Mutlu et al. (2022b). Excluding the number of leaves per shoot and internodal distance, the grass densities and plant traits showed regional- (across four bays) and depth-related (across a range of 5–20 m) differences. Grass densities were higher in Kekova and Kaş Bays, Turkey (Figure 1), both locations on the pathway of the rim current, compared to Finike and Antalya Bays (Figure 1b), which were almost devoid of grass and current (Figure 1). The seagrass at 5 m was exposed more to waves than at greater depths. Density (shoot density, leaf biomass and leaf area index) and plant traits (vertical rhizome, sheath, leaf lengths and widths) peaked in late autumn and early spring depending on water temperature, salinity and nutrient limitations, and were minimal during summer (Mutlu et al. 2022b).

Figure 1:

Study area (red rectangle) (a) and experimental location (red square) of seasonal “Leaf” and “Cut” experiments for Cymodocea nodosa (b), and location (blue square) for Posidonia oceanica (b, Mutlu and Olguner 2023a).

Due to the lack of biometric quantification of C. nodosa, the present study was aimed at characterizing in detail the acoustic backscattering properties of C. nodosa over time, and is the first such attempt.

2 Materials and methods

Material and methods for the present experiment were adapted from the work of Mutlu and Olguner (2023a) on Posidonia oceanica. Two different in situ experiments were performed off the coast of the Turkish Levant over four to five different months in 2011 and 2012 (Figure 1). The experiments were called “Leaf” and “Cut” experiments (Mutlu and Olguner 2023a) and are briefly described below. The acoustic data were collected with Visual Acquisition (vers 6.2) BioSonics software, using a DT-X scientific echosounder (BioSonics, Inc.) using the manufacturer’s calibration and configuration parameters during data collection (Table 1).

The data processing was described in a study published by Mutlu and Olguner (2023a). During the experiments, Elementary Distance Sampling Units (EDSU) (Simmonds and Maclennan 2005) were measured, consisting of Sv (volume backscattering strength expressed in of dB per m³), Sa (area backscattering strength, dB per m²), and TS (individual target strength, dB per m³) (Mutlu and Olguner 2023a). The leaf biometrics measured were weight, area (leaf area; LA in cm²), volume (leaf volume; V in cm³), and density (leaf density; d in g cm⁻³) estimated from measurements of the plant traits described by Mutlu et al. (2022b) for Cymodocea nodosa (Table 2).

2.2 “Cut” experiment

The “Cut” experiment applied to C. nodosa was based on the same procedure designed for P. oceanica in Mutlu and Olguner (2023a), but was conducted in a different location (Figure 1). A non-metal inclusion plastic frame (1 × 1 × 0.04 m) was placed on a flat bottom at a depth of 12–14 m (approx. 0.90–1.60 m² sampling surface area without the draft) during a period of calm seas with the ship (R/V Akdeniz) at anchor. Manual dissection of the seagrass shoots inside the frame was performed, with the process repeated five to six times by SCUBA until there was no grass left inside the frame. Similar analyses applied to P. oceanica (Mutlu and Olguner 2023a) were followed for C. nodosa, but the Visual Bottom Typer (VBT) analysis was not applied to C. nodosa in the “Cut” experiment (Figure 2) (Mutlu and Olguner 2023a).

Figure 2:

Unprocessed echogram (a), removal of weak scatterers (e.g., background noise, and zooplankton layer) through the water column, bottom echo and deadzone* (b), and removal of rhizome and sheath of Cymodocea nodosa and strong scatterers (e.g., fish) among the seagrass (c) during the “cut” calibration. Depth in range from the transducer. * “The deadzone is caused by the acoustic beam reverberating off the side of a sloping section of the bottom and occurs most prominently in areas with steep bottom topography” (Echoview 2023). deadzone = (sound speed × pulse width/2) + shortest depth/sin(90-slope of bottom)-shortest depth, where the shortest depth is the shortest distance from the transducer to the bottom between the previous and the next ping (Mutlu 2023).

Using a micrometer, the biometrics of the disc-shaped epiphyte was measured each month to estimate their contribution to EDSUs of C. nodosa by solving an elastic-shell model of the forward solution (Stanton et al. 1996). In case of the need for estimation of leaf weight related to leaf length or area, their relationships were described by Mutlu and Olguner (2023b).

3 Results

The present study suggested relationships between the acoustic and biometric variables, and the comparable possibility for detection of C. nodosa by the same acoustic frequency used for that of P. oceanica (Mutlu and Olguner 2023a). Biometrically, C. nodosa is shorter in leaf length and narrower in leaf, sheath and rhizome width than P. oceanica (Mutlu et al. 2022a,b), as reflected by their differing echograms (Figure 2) (Mutlu and Olguner 2023a).

3.1 Leaf experiment

During the leaf experiments for C. nodosa, similar analyses as those applied to P. oceanica were used to characterize its acoustic responses.

3.1.1 Leaf + sheath

The leaf + sheath alone was measured during the experiment, but vertical rhizome + sheath + leaf was not attempted for the measurements, since the vertical rhizome was too short for discrimination by the acoustics and highly likely to fall in the dead zone, unlike that of P. oceanica (Mutlu and Olguner 2023a).

The EDSUs (Sa, Sv, Sv1 and TS, Table 2) were not significantly correlated with the biometrics of leaf + sheath of C. nodosa at p < 0.05 (Figure 3). The dominant Sv varied between −50 and −40 dB, Sa between −40 and −30 dB, Sv1 between −55 and −40, and TS between −70 and −55 dB (Figure 3). The regression relationships between the acoustic and biometric variables seemed to have a trend with a logarithmical model since there was overall no significant monotony, with the exception of results for the experiment in the Finike region at p < 0.05 (Supplementary Table S1). However, the biometrics was not significantly correlated with the acoustics, and there was no seasonal difference in the relationships at p < 0.05 (Figure 3).

Figure 3:

Seasonal relationship between acoustic data of Sv (volume backscattering strength vs LAI (Leaf Area Index) (a), Sa (volume backscattering strength) versus weight (b), Sv1 (volume backscattering strength estimated with visual bottom typer, VBT) versus volume (c), and target strength (TS) versus density (d) (see Table 2 for abbreviations) for leaf + sheath of Cymodocea nodosa.

3.1.2 Leaf

A regression model of Y = b × ln(X) − a (logarithmic model) provided a significant fit to acoustic-biometric relations for the leaves of C. nodosa, with some exceptions occurring in April (Figure 4). The acoustic measurements logarithmically increased with increasing biometric values up to 20 cm² of LAI (approx. 4 g of weight, W), and then increased slightly and linearly with increasing biometric values (Figure 4). The relationship in January was found to be different than that of the other months in the plots of Figures 4 –6, but showed the same general trend.

Figure 4:

Seasonal relationship between Sv (a), Sa (b), Sv1 (see Figure 2 for description of Sv1) (c), and TS (d) and LA and weight of leaves (see Table 2 for abbreviations) of Cymodocea nodosa.

Figure 5:

Seasonal relationship between acoustic data of reduced Sv (a), Sa (b), Sv1 (c), and TS (d) versus the density of the leaves (g cm⁻³) of Cymodocea nodosa and regression equations for the seasonally pooled data (see Table 2 for abbreviations).

Figure 6:

Post-hoc test (least significant difference, LSD; horizontal bar: mean ± SE) of seasonal relationship based on regression coefficients between acoustic data (Sv, Sa, Sv1, TS) and biometric variables (LA and W) for leaf + base (leaf part within sheath) of Cymodocea nodosa (see Table 2 for abbreviations) (blue mark; to be tested among the months; 1: January, 4: April and 8: August, red: significantly different, grey: not significantly different between vertical discrete grey lines).

Reduced EDSUs versus density of leaf showed meaningful seasonal relationships (Figure 5) as compared to the trend in Figure 4. Overall, Kendall’s Tau analysis as a monotony test did not discern any colinearity between the relationship of EDSUs and biometrics at p < 0.05, with the exception of results for the experiment in the Finike region (Supplementary Table S1). The relationship seemed to be logarithmic when the data were grouped by season (Figure 4). However, leaf density was found to be greater in January than in the other months. Therefore, the relationships with the density were established to be significant using a polynomial fitting model with the following trend: reduced backscattering strengths increased with leaf density up to a density value of around 1.5–1.7 g cm⁻³ (Figure 5). Thereafter, the trend did not change with the density (Figure 5). This trend was more pronounced in the relationship between the TS and density than in the other EDSUs.

Inherently, leaf density is one important descriptor in the biometrics-acoustics relationship. The EDSUs were highly significantly correlated with the density compared to the biometrics (Figures 7 –9). In the TS-density relation, no increase occurred in the TS beyond a leaf density of 1.5 g cm⁻³ (Figure 5d).

Figure 7:

Relationship between acoustic parameters of Sv (a), Sa (b), Sv1 (c), and TS (d) and leaf area (LA) of the leaf (see Table 2 for abbreviations) of Cymodocea nodosa.

Figure 8:

Relationship between acoustic parameters of Sv (a), Sa (b), Sv1 (c), and TS (d) and leaf weight (W) (see Table 2 for abbreviations) of the leaf of Cymodocea nodosa.

Figure 9:

Principal component analysis (PCA) solution of the biometric and acoustic variables for the sheath + leaf (a), and the leaf of Cymodocea nodosa (b). Symbols indicate month number in year, and 81 denotes the second experiment conducted in the Finike region in August (see Table 2 for abbreviations).

Seasonal differences occurred between Sa and the biometrics except for the density at p < 0.05 (Table 3). However, only the density differed significantly with Sv1. The post hoc test showed that this difference was derived by the intercept of the regression equations. However, the intercept was significantly lower in August than in the other months, and none of the slopes was significantly different among the seasons at p < 0.05 (Figure 6).

Table 3:

P values from analysis of covariance (ANCOVA) for a logarithmic relationship between the biometric and acoustic parameters of Cymodocea nodosa among the seasons (see Table 2 for abbreviations).

Species, X/Y	ln(LA)	ln(W)	ln(V)	ln(d)
Sv	0.0866	0.0845	0.0866	0.5725
Sa	0.0407	0.0427	0.0407	0.6007
Sv1	0.2558	0.2770	0.2558	0.0348
TS	0.0771	0.0795	0.0771	0.6408
	log10(LA)	log10(W)	log10(V)	log10(d)
log10(sv)	0.0866	0.0845	0.0866	0.5725
log10(sa)	0.0407	0.0427	0.0407	0.6007
log10(sv1)	0.2558	0.2770	0.2558	0.0348
log10(σ bs)	0.0771	0.0795	0.0771	0.6408

1. Bold values are significantly different among seasons at p < 0.05.

Overall, Kendall’s Tau analysis did not discern any colinearity between the relationship of EDSUs and biometrics of LA and W at p < 0.05 (Figures 7 and 8, Supplementary Table S1). Seasonally pooled data showed that there were significant correlations and polynomial relationships between the EDSUs and biometrics (LAI, and W) at p < 0.05, with the exception of TS (Figures 7 and 8).

The PCA showed that biometrics of sheath + leaf did not discern colinearities with the acoustics, with the exception of the density-acoustics relationship on PCA1 (Supplementary Table S2, Figure 9a). The acoustics had linearity on PCA1 whereas the biometrics had linearity on PCA2, and two components were explained with a variance of 72.9 % (Supplementary Table S2, Figure 9a).

Similar to PCA results of the sheath + leaf, the PCA showed that there was no colinearity between the acoustics and biometrics of the leaf. Therefore, PCA1 was positively correlated with the acoustics, and the biometrics was correlated with the PCA2 (Supplementary Table S2, Figure 9b). Both components were explained cumulatively with a variance of 89.2 % (Supplementary Table S2).

Unlike the Cut experiment, the contribution of the epiphyte to the EDSUs of the leaves during the individual (Leaf) experiment was insignificant.

4 Discussion

Complementing a similar study for P. oceanica (Mutlu and Olguner 2023a), this study presents the first measurements to acoustically detect and sea-truth EDSUs with respect to the biometrics of C. nodosa. Previously published studies focused mostly on mapping, coverage, canopy height and distribution of seaweeds (e.g. Lee and Yik 2018; Olguner and Mutlu 2020; Shao et al. 2021; van Rein et al. 2011). Recently, the two seagrasses were statistically distinguished based on their acoustic parameters (Mutlu and Olguner 2023b). Leaf biomass (LAI or BLAI) is a prominent variable in monitoring the conditions of seaweeds, especially seagrasses, for determination of benthic zone, ecological status and blue carbon content (Brown et al. 2011; Colantoni et al. 1982; Pal and Hogland 2022). The acoustic approach is considered the best method if the parameters are calibrated between EDSUs and biometrics.

In C. nodosa, unlike in P. oceanica, the vertical rhizome is below ground and shorter than the sheath (Mutlu et al. 2022a,b). Therefore, the vertical rhizome of C. nodosa was not tested to determine its acoustic properties; instead, the sheath (leaf + sheath) was examined for the ground-truthing of C. nodosa during the present study. The sheath produced partially linear relationships between acoustic and biometric parameters. However, the TS was not correlated with the biometrics. An algorithm, namely “SheathFinder”, which was originally developed for P. oceanica (Mutlu and Balaban 2018), was not effective in discriminating the sheath from the leaf of C. nodosa during the “Cut” experiment. Unlike in the case of P. oceanica, relationships between the EDSUs and biometrics of C. nodosa during the “Cut” experiment were therefore similar to those found in the “Leaf” experiment. This suggested there could be an effect of the sheath on the relationships established during the “Cut” experiment.

The dominant Sv and Sa were estimated to vary between −45 and −50 dB, and −35 and −45 dB, respectively, regardless of the seasons. The dominant Sv was lower in P. oceanica (Mutlu and Olguner 2023a) than in C. nodosa. Monpert et al. (2012) estimated the same difference between P. oceanica and Zostera marina, which is structurally similar to C. nodosa, using two frequencies (38 and 200 kHz). The Sv values were similar and low, and were inversely related to the shoot density and percent coverage when using a 200 kHz-side-scan sonar (Llorens-Escrich et al. 2021). Taking other seaweeds into account, Shao et al. (2019) measured the TS of a kelp, Saccharina japonica, which was found to vary between −69.2 and −34.4 dB, corresponding to biometrics of W between 77 and 994 g and L between 0.6 and 3.2 m. Another seaweed, Sargassum horneri, had similar acoustic measurements of TS (Minami et al. 2021); TS varied between −47.7 and −22.9 dB at 120 kHz for dry weights of 0.133 g (3.13 g wet w) to 15.8 g (114.5 wet w). The TS of C. nodosa ranged from −70 to −55 dB at 200 kHz, corresponding to a wet weight of 0.1–1.25 g. Cymodocea nodosa was the strongest scatterer based on reduced TS (RTS = −57.9 to −41.2 dB) as compared to S. japonica at 200 kHz (RTS = −125.8 to −124.3 dB; Shao et al. 2019) and S. horneri at 120 kHz (RTS = −84.7 to −62.6 dB), even though S. horneri has air-sacs suspending fronds in water (Minami et al. 2021). The reduced EDSU of the seagrass, P. oceanica had low values that differed seasonally (Mutlu and Olguner 2023a). The previous studies mentioned were not conducted to measure the acoustics seasonally. Mutlu and Olguner (2023a) studied the biometrics-acoustics relation for P. oceanica. Unlike P. oceanica, there was no seasonal difference in the relationships between some acoustic parameters (TS and Sv) and the biometrics of C. nodosa, but seasonal differences occurred between Sa and the biometrics. This could be due to the echo energy averaged per unit volume for Sv and TS being lower than the echo energy summed per unit area for Sa, and the effect of the detection limit of 206 kHz (Mutlu 2003) on C. nodosa leaves, which are thinner than P. oceanica leaves (Mutlu et al. 2022a,b). One of the elements changing the acoustic impedance and reflection was the density contrast (g) of the seagrass in relation to the sound speed contrast (h) in reference to the sea water (Stanton et al. 1996). The leaf density of C. nodosa was higher than that of P. oceanica and sea water (Mutlu and Olguner 2023b) and similar to matte density of P. oceanica (Leduc et al. 2023). Furthermore, Mutlu and Olguner (2023b) estimated acoustic hardness to be higher in C. nodosa than in P. oceanica, and higher in the Finike region (around 10 m bottom depth) than in Antalya Gulf for C. nodosa. The hardness increases bottom variations (Korneliussen 2000). The photosynthetic activity of P. oceanica was maximal between May and August (Enriquez and Schubert 2004) when the P. oceanica released gas to the water and acoustic measurements were maximal. This resulted in a seasonal difference in the biometrics-acoustics relations for P. oceanica (Mutlu and Olguner 2023a), and C. nodosa showed visible bubbles on tips of the leaf blades released to the water every time of the day during a year (Randall et al. 2020). This could be a reason for why there was no seasonal difference in biometric-acoustic relations for C. nodosa. It is noteworthy here that the blades were free of the bubbles since they were first taken onboard during the “Leaf” experiment.

During the “Leaf” experiment, a calcareous red alga, Pneophyllum fragile, was found on the leaves of C. nodosa. The measurements were repeated with and without the epiphytic alga on the leaves of C. nodosa, and there were no significant differences between the two measurements. Furthermore, volume backscattering strength (Sv and then Sa) was estimated using an elastic shell model (Stanton et al. 1996) taking biometric measurements (diameter, thickness of the algal crust, abundance per square meter) into the solution for the “Cut” experiment. The solution showed that Sa was estimated to be rather low as compared to a threshold of C. nodosa (Mutlu et al. 2014). Such tiny specimens contributed to the EDSUs depending on their abundance. For instance, a sand particle had a TS value of −110 dB at 420 kHz (Wiebe et al. 1997). The TS was estimated to be higher than background noise and minimum layered zooplankton threshold as the sand abundance reached over millions per cubic meter of water (Pershing et al. 2005; Wiebe et al. 1997). Another potentially acoustically similar calcareous red alga, Hydrolithon cruciatum was the most abundant species (27.47–82.54 %) based on the percent coverage on the leaves of C. nodosa on the coast of Kavala Gulf, Greece (Tsioli et al. 2021).

During the present study, acoustic data were converted to Sv and Sa of the echo energy, since both were considered useful to convert the acoustic data to the absolute data of the seagrasses using different experiments and processing. The Sa is more useful than the Sv since seagrasses are sessile. The Sa helps more for sea-truthing the acoustic data relative to the absolute data than Sv. Sv could be considered useful for the volumetrically-studied biotic organisms such as pelagic fish and zooplankton.

In conclusion, the present study has described the acoustic backscattering properties of C. nodosa’s structures for the first time in relation to seasonal biometrics. The seagrass is only partially detected by a frequency of 206 kHz. Therefore, the higher frequency (420 kHz) is recommended to study this seagrass since a Rayleigh scattering zone occurred at low biometric values. The frequency response as echo amplitude depended on the acoustic frequency (k: wave number) × size (a: spherical radius) of the target to produce different acoustic zones (Lavery et al. 2007; Mutlu and Olguner 2023a). The geometric region occurred at the greater density (1.5 g cm⁻³), and corresponding biometrics. This occurred at a greater density than 1 g cm⁻³ for P. oceanica (Mutlu and Olguner 2023a). Different structural parts (sheaths and leaves) of the seagrass backscattered different echo energy. There were different backscattering properties between the sheaths and leaf of the seagrass. Relationships between the biometrics and acoustics during the “Leaf” and “Cut” experiments were fitted by the different regression models. Therefore, biometric estimations were also different, which could be attributed to the following explanations: i) the backscattering strengths depended on strength of biomass as a function of LAI, shoot density and volume between the two experiments, and ii) The “Leaf” experiment measured backscattering strength from individuals of specimens, and “Cut” experiment from patches or the mass of the seagrass per unit area. As suggested for P. oceanica (Mutlu and Olguner 2023a), the different equations could be used differently being derived from a generalized equation, Sv = TS + log₁₀(N per m³) for the estimates as follows:

Sv (patch) = Sv per weight + log₁₀(B per cubic meter of volume) for the “Cut” experiment,

Sa (patch) = Sa per weight + log₁₀(B per m² of area) for the “Leaf” experiment,

where Sa or Sv (c.a. weight) belongs to one bundle of species.

The equations of the “Cut” experiment were directly used to estimate the biometrics. The acoustic reflection was highly dependent on the density of the leaves in time. As a function of the leaf aging and density, the frond length of C. nodosa was greater in cold seasons than in warm seasons in the present study area (Mutlu et al. 2022b). However, Mutlu et al. (2022b) estimated all biometric measurements to be minimal at the experimental bottom depth (15 m) of the present study. Therefore, in the next study we will focus on relationships of the EDSUs and biometrics at different depths involving higher frequencies, since there was spatiotemporal variation in the biometrics (Mutlu et al. 2022b).