Field Campaign of Bourgneuf Bay 2025 - REWRITE
Field trip organisation
The summer field campaign in Bourgneuf Bay took place on September 11th, 2025, at the Barbâtre meadow. Low tide occurred at 14:20 local time, with a tidal range of 4.75 m. We arrived on site around 11:00. Two distinct sampling areas were selected based on the historical consistency of meadow density over the past 40 years. One area, where the meadow has consistently shown more than 50% seagrass cover every year for the last 40 years, was selected (Zone H), as well as an area where meadow cover has been highly inconsistent over the same period (Zone L). In each zone, 30 cores were collected, covering a gradient of seagrass cover from low (SPC < 20%) to high (SPC > 80%). Each core was 20 cm deep to harvest seagrass leaves, rhizomes, and associated fauna.
What has been done in each area ?
Zone L :
Photo quadrat
Coring
Above and Bellow ground biomass (Dry weight)
Bivalves Diversity and Density
Sediment Sampling
Zone H :
Photo quadrat
Coring
Above and Bellow ground biomass (Dry weight)
Bivalves Diversity and Density
Carbon Estimation
Hyperspectral measurments
Sediment Sampling
Above Ground vs Below Ground biomass
The samples were sorted to separate seagrass above-ground biomass (AGB) from below-ground biomass (BGB), and bivalves were counted and identified. After sorting, the leaf and rhizome fractions were dried for 48 h at 60 °C in a drying oven and then weighed on a precision balance to obtain dry mass for each core. Figure 2 shows the relationship between BGB and AGB for all 60 cores collected in the field.
The relationship between the below ground and above ground biomass of Zostera noltei in Bourgneuf Bay can be describe as follow:
\[ Below_{Ground} = 0.64 * Above_{Ground} + 10.04 \]
Site L and Site H have a similar relationship between those two variables (Figure 2).
Hyperspectral signature zone H
Only in Zone H, before coring, a hyperspectral signature of the sample was taken using an ASD Fieldspec HandHeld 2. The instrument was calibrated in the field for reflectance acquisition using a Spectralon with a 99% reflective Lambertian surface (No radiance measurement has been done).
Figure 3 shows the hyperspectral signature of all the cores of the area H. Spectra are showing differente level of reflectance, with core H_04 having a relatively strong overall reflectance, with an small absorption peak around 665 nm due to the presence of chlorophyll-a. On the other hand the spectra H_30 is showing a lower overall reflectance, with really low reflectance in the visible and a strong absorption peak of chlorophyll-a around 665nm.
About the reflectance anomaly around 760 nm.
An unexpected reflectance feature can be observed around 761 nm in most of the spectra (Figure 3). This feature appears as a positive anomaly in some cases, such as H_04 or H_03, or as a negative anomaly, as in H_02 or H_19. The most likely explanation is an instrumental artifact, for example related to calibration issues of the ASD spectroradiometer. Nevertheless, it is worth noting that at 761 nm vegetation can also produce a narrow reflectance peak due to sunlight-induced chlorophyll fluorescence (SIF). This fluorescence signal arises from the broad chlorophyll a emission around 740 nm that partly fills the atmospheric O₂ absorption band, and has been documented both in terrestrial and aquatic vegetation (Lu et al. 2016, Meroni & Colombo, 2006).
To investigate the origin of this feature, the Fluorescence Line Height (FLH;Lu et al. 2016) will be calculated.
\[ \mathrm{FLH}(761) = R_{rs}(761) - \left[ \frac{769-761}{769-757} \big(R_{rs}(757)-R_{rs}(769)\big) + R_{rs}(769) \right] \]
The FLH values will be compared with NDVI, as proxy of the biomass, with the green leaf index proxy of the greeness of seagrass leaves as well as with the amount of time the ASD had been operating at the moment of spectrum acquisition.
Figure 4 is showing the ralationship between the FLH, measuring the height of the peak visible at 761 nm, and the NDVI and the GLI. No relationship between the size of the anomaly and those radiometric indices can been seen from the data acquired on site H. No matter the biomass, and no matter the greenness of leaves, the peak at 761 nm can be more or less pronounced.
To estimate, for each spectrum, how long the instrument had been running at the time of acquisition, we extracted the capture time of the corresponding photo-quadrat from the images’ EXIF metadata. Because the photo-quadrats were taken a few minutes before the hyperspectral measurements, their timestamps provide a close proxy for the acquisition time; aligning these timestamps with the instrument start time yields the elapsed operating time for each spectrum. Sample H_01 was acquired first. By convention, we set its instrument run time to 0 (min), and the run time for all subsequent samples is expressed relative to H_01—that is, as the difference between each sample’s acquisition timestamp and that of H_01.
Figure 5 shows the relationship between ASD run time and the fluorescence line height at 761 nm (FLH). Sample H_01 was acquired immediately after startup (T0), and the last sample in the sequence, H_08, was collected 7,316 s (≈ 122 min) after T0. A negative linear trend can be seen (grey 95% CI shown): FLH values are highest at the beginning of the session and decline as the instrument continues to run, with several late measurements approaching zero or slightly negative values.
Radiometric indices vs Biomass
From those spectral signatures, a range of spectral indices were tested to identify the one with the strongest relationship to above-ground biomass. All the indices listed below were evaluated, but only the results for the best-performing index and the NDVI are shown.
\[ \begin{aligned} \text{NDVI} &= \frac{R_{NIR} - R_{Red}}{R_{NIR} + R_{Red}} &\quad \text{GNDVI} &= \frac{R_{NIR} - R_{Green}}{R_{NIR} + R_{Green}} \\[6pt] \text{SR} &= \frac{R_{NIR}}{R_{Red}} &\quad \text{VARI} &= \frac{R_{Green} - R_{Red}}{R_{Green} + R_{Red} - R_{Blue}} \\[6pt] \text{MSR} &= \frac{\tfrac{R_{NIR}}{R_{Red}} - 1}{\sqrt{\tfrac{R_{NIR}}{R_{Red}} + 1}} &\quad \text{CIG} &= \frac{R_{NIR}}{R_{Green}} - 1 \\[6pt] \text{DVI} &= R_{NIR} - R_{Red} &\quad \text{CIRE} &= \frac{R_{NIR}}{R_{RE}} - 1 \\[6pt] \text{RVI} &= \frac{R_{Red}}{R_{NIR}} &\quad \text{NDRE} &= \frac{R_{NIR} - R_{RE}}{R_{NIR} + R_{RE}} \\[6pt] \text{SAVI} &= \frac{1.5 \, (R_{NIR} - R_{Red})}{R_{NIR} + R_{Red} + 0.5} &\quad \text{MTVI2} &= \frac{1.5 \left[ 1.2(R_{NIR} - R_{Green}) - 2.5(R_{Red} - R_{Green}) \right]} {\sqrt{(2R_{NIR}+1)^2 - (6R_{NIR} - 5\sqrt{R_{Red}}) - 0.5}} \\[6pt] \text{OSAVI} &= \frac{1.16 \, (R_{NIR} - R_{Red})}{R_{NIR} + R_{Red} + 0.16} &\quad \text{MCARI} &= \left[(R_{RE} - R_{Red}) - 0.2(R_{RE} - R_{Green})\right] \frac{R_{RE}}{R_{Red}} \\[6pt] \text{TSAVI} &= \frac{0.08 \, (R_{NIR} - 0.5R_{Red} - 0.08)}{R_{Red} + 0.5R_{NIR} + 0.08} &\quad \text{PRI} &= \frac{R_{Green} - R_{RE}}{R_{Green} + R_{RE}} \\[6pt] \text{EVI} &= \frac{2.5 \, (R_{NIR} - R_{Red})}{R_{NIR} + 6R_{Red} - 7.5R_{Blue} + 1} &\quad \text{SIPI} &= \frac{R_{NIR} - R_{Blue}}{R_{NIR} - R_{Red}} \\[6pt] \text{EVI2} &= \frac{2.5 \, (R_{NIR} - R_{Red})}{R_{NIR} + 2.4R_{Red} + 1} &\quad \text{NDWI} &= \frac{R_{NIR} - R_{Green}}{R_{NIR} + R_{Green}} \\[6pt] \text{mND705} &= \frac{R_{750} - R_{705}}{R_{750} + R_{705} - 2R_{445}} &\quad & \end{aligned} \]
The index with the highest R² against above-ground biomass is NDRE. NDRE is analogous to NDVI but replaces the red band (~665 nm) with a red-edge band (typically ~705 nm): The advantage is reduced saturation at high biomass. Around 665 nm, chlorophyll-a absorption quickly reaches an asymptote, so reflectance in the red band changes little as biomass increases and NDVI tends to plateau. In the red-edge region, however, reflectance remains sensitive: as biomass/chlorophyll increases, reflectance near ~705 nm continues to decrease (with a concomitant shift of the red-edge to longer wavelengths). Consequently, NDRE preserves sensitivity in dense canopies and correlates more strongly with biomass than NDVI, resulting with an R² square slightly higher than for the NDVI (R²_NDVI = 0.65, R²_NDRE = 0.67)… -_-’
Bivalve diversity & spatial distribution
The student quantified the density of all bivalve species present in the sediment while sorting each core, and all taxonomic identifications were subsequently verified by a taxonomist.
Figure 7 shows the mean density of each bivalve species across the 60 cores. In total, nine distinct species were identified. The most common was Scrobicularia plana (209 individuals per m², on average, present in about 77% of cores), whereas Ruditapes decussatus was the least common (0.95 individuals per m², present in about 1% of cores).
Bivalve diversity by core
Figure 8 shows per-core diversity for the two areas. Each dot is a core (colored by site) and panels display Shannon diversity (H′), species richness (S), and Pielou evenness (J′). Cores from area H generally sit higher for H′ and S, with less separation for J′.
Bayesian comparisons (reported as H–L) back this up. Shannon diversity is higher in H: mean difference +0.365 with a 95% credible interval [0.155, 0.579] and \(P(H\!-\!L>0)=1.00\). Very little posterior mass falls within a small ±0.10 ROPE, so the increase is not only credible but also practically meaningful. Richness is also higher in H: +1.23 species on average, 95% CrI [0.17, 2.31], \(P>0=0.988\). About one-third of the posterior lies within a ±1 species ROPE, suggesting a positive but modest gain in the number of species per core. For evenness, the difference is small and uncertain: +0.116, 95% CrI [−0.020, 0.247], \(P>0=0.956\), with ~15% of the posterior inside a ±0.05 ROPE.
Overall, area H hosts richer and more diverse bivalve assemblages than area L, while evenness is at most slightly higher in H and may be similar between areas.
Relationship between bivalves and seagrass biomass
Figure 9 shows the relationship between the density of bivalve and the above-ground biomass of seagrasses.