Calibration of total nitrogen content in seaweeds (Ulva sp.) by NIRS

INTRODUCTION
Sea lettuce (Ulva sp.) is a common seaweed on French seashore. In some conditions it can become invasive and cause environmental problems. Their valorization in animal feeding – especially fish – is limited by their low protein content (around 10%DM). In the framework of studies on protein enrichment of algae, there was a need for tools for monitoring the kinetics of protein accretion when cultivated in fertilized waters. Non-destructive direct measurement by NIRS on fresh samples was tested to evaluate the potential of this tool. In addition, NIRS measurements on dried samples were performed in the laboratory to evaluate NIRS for more precise measurements.

EXPERIMENTAL
Algae were collected in the coastal ponds around Palavas (southern France) and cultivated in sea water tanks with addition of ammonia, phosphate and with or without iron and carbon (CO2). They were sampled at several times over three months, leading to a total of 183 samples. Samples were then rinsed with freshwater and, for most of the samples, excess water was removed in a spinner. NIR spectra were collected with ASD Labspec Pro spectrometer (350-2500nm) equipped with a reflectance probe. Samples were then dried and ground (1mm sieve). Dry matter (DM) content, total nitrogen (N) and total minerals (ash) were determined in the laboratory. Spectra on dried and ground samples were collected on FOSS NIRSYSTEM 6500. All calibrations were performed with PLS (partial least squares) regression. Several mathematical pretreatments of spectra were tested to identify the most adequate. A validation was performed by randomly removing 25 samples before calibrating, and applying the calibrations obtained on these samples.

RESULTS AND DISCUSSION
The average N content was 5.39% DM and varied widely, from 1.61% to 10.28% with standard deviation (SD) of 4.16%. Variation was mostly related to the culture duration and conditions (higher N content obtained without CO2 input). DM content was 18.3% DM on average (range 5.5%-31.2%) and ash content was 19.6% DM on average (range 5.4%-33.9%). NIRS calibrations on fresh samples led to a cross-validation error (SECV) and RPD (=SD/SECV) values of 1.44% and 3.0 for DM, 2.87% and 1.7 for ASH and 0.37% and 4.4 for N respectively. Validation showed that DM and N calibrations were robust since SEP values (4.71% and 0.41% respectively) were close to SECV. However this was not the case for ash which presented a high SEP value (5.04) showing that the calibration was not stable.
Calibration for N on dried samples was more precise than for fresh samples, with SECV and SEP values of 0.20 and 0.31% respectively. Ash was also better calibrated but the SEP was still high (4.1%).
The results obtained allow a routine use of NIRS on fresh samples to monitor DM and N content of algae during cultivation for N enrichment. This is a very interesting tool since there is no use keeping the algae in cultivation as soon as they reach a plateau of N concentration around 7%DM which corresponds to a protein content around 35%.