•A method to measure the colour in an objective and contactless way is presented.•Two methods to calculate colour information from reflectance spectra are compared.•Accurate results were achieved to ...measure the colour of vine tomatoes.•An R2 over 0.9 was achieved to predict a* and hue.•Visualization of the colour variability of individual tomatoes is made possible.
As consumers buy with their eyes, colour is considered one of the most important quality parameters of food products. Traditionally, this is defined by human inspection, or measured using a colorimeter or a spectrophotometer. As the first is subjective and prone to factors like fatigue, this is not ideal for industrial use. The second only measures a small area of the food product, making it difficult to get a clear overview of the colour of the whole sample. To overcome these limitations, hyperspectral imaging has been used in this research to measure the postharvest colour of vine tomatoes. Two methods to calculate the colour based on hyperspectral images are compared. The first is the use of a direct method to calculate the colour from the spectra in terms of CIELab-values, while the second method is a soft modelling approach involving multivariate statistics. The soft modelling method was found to achieve the best results (R2L*=0.86; R2a*=0.93; R2b*=0.42, R2Hue=0.95, R2Chroma=0.51), but its applicability is limited to the range of products on which the models have been trained. The direct method is more generally applicable, but was found to lack robustness against intensity variations due to the curvature and glossiness of the tomatoes.
Analysing the estuarine use patterns of juveniles of marine migrant fish species is vital for identifying important sites for juveniles as well as the basic environmental features that characterize ...these sites for different species. This is a key aspect towards understanding nursery function. Various estuarine systems along the Portuguese coast (Minho, Douro, Ria de Aveiro, Mondego, Tejo, Sado, Mira, Ria Formosa and Guadiana) were sampled during Spring and Summer 2005 and 2006. Juveniles of commercially important marine fish species
Solea solea,
Solea senegalensis,
Platichthys flesus,
Diplodus vulgaris and
Dicentrarchus labrax, predominantly 0-group individuals, were amongst the most abundant species and had distinct patterns of estuarine use as well as conspicuous associations with several environmental features. Juvenile occurrence and density varied amongst estuaries and sites within them, and differed with species. Sites with consistently high juvenile densities were identified as important juvenile sites (i.e. putative nursery grounds). Through generalized linear models (GLM), intra-estuarine variation in occurrence and density of each of the individual species was largely explained by environmental variables (temperature; salinity; depth; percentage of mud in the sediment; presence of seagrass; importance of intertidal areas; relative distance to estuary mouth; macrozoobenthos densities; and latitude). Decisive environmental factors defining important sites for juveniles varied depending on the system as a result of different environmental gradients, though there were common dominant features for each species regardless of the estuary considered. Analysed environmental variables in the GLM also accounted for inter-estuarine variation in species' occurrence and density. In several estuaries, the identified important juvenile sites were used by many of these species simultaneously and may be of increased value to both management and conservation. Overall, the variability in site features amongst estuaries highlighted the tolerance of these species to different available environmental conditions and provided fundamental information for future spatially explicit modelling of their distribution. This should ultimately enable the prediction of species response to habitat alterations.
One of the longest marine monitoring programs in the North Sea has been the spatiotemporal surveying of subsurface plankton since 1931. During this period a regime shift was detected in the late ...1980s culminating in marked changes in phytoplankton, zooplankton and in the fisheries of horse mackerel. Here we used the phytoplankton colour index, a visual biomass estimate, from 1948 to 2010 and total diatom abundance from 1958 to 2010 to test whether the well-documented regime shift could have been anticipated by the recently developed Early-Warning Signals for critical transitions (EWS). We estimated EWS, namely autocorrelation and standard deviation, within moving windows along the time series prior to the regime shift. We found that both statistics increased revealing that the North Sea ecosystem was becoming progressively unstable prior to the regime shift. Moreover, this high-resolution time series permitted us to test for robustness, error and significance of the EWS. We did that by dividing the time series into independent blocks and estimating EWS after bootstrapping and randomising the blocks. This alternative approach confirmed the robustness of the EWS with limited associated errors. In particular, we found that the warning was significantly evident years before the onset of the regime shift. We conclude that EWS may provide robust and timely warning for upcoming regime shifts depending on the quality and quantity of recorded data in marine ecosystems.
Predicting stability from current ecosystem performance is theoretically difficult, but early statistical warning signals (EWS) may enable the anticipation of regime shifts. However, little is known ...regarding the behavior of EWS in shifts with cyclic dynamics. In this study, we use indicators to assess the stability of a three-species system in a competitive loop similar to a rock–paper–scissors (RPS) hierarchy. In two scenarios, the RPS is simulated using a 3-D automaton whose input matrix combines probabilities of pair-wise dominance with differential reaction frequencies. The first scenario uses the data of a microbial experiment in which the RPS hierarchy is characterized by incomplete dominance within species pairs and differences in reaction frequency between species pairs. The input of the second model was chosen to generate a contrasting scenario: a symmetric RPS interaction gradually subjected to a stressor. The reaction frequency of one species pair was modeled to decay linearly over time. The relative species abundances are monitored spatiotemporally. In the first scenario, abundances oscillate stably despite initial large swings, whereas in the second scenario, one species gradually dominates, eventually resulting in transitivity. In both scenarios, species cluster spatially in patches of single species. In scenario 1, the average patch size remains constant throughout the iterations and possibly contributes to the overall stability; however, in the second scenario, a further homogenization takes place. In the first scenario, EWS reflected the system's stability with species abundances settling into a stable basin. In scenario 1 only one of the EWS indicators detected consistently the loss of resilience. Sensitivity analysis revealed excessive variability in dominance resulted in immediate loss of the RPS hierarchy.
Plums are widely consumed, both fresh and processed. During harvest, handling, or transportation, they are exposed to static and dynamic compression forces exceeding the critical stress for tissue ...damage. Compression-related damage typically develops further as the fruit ripens and softens, facilitating the occurrence of rot, which might cause significant losses in the supply chain. Early detection of these damages is crucial to sorting the damaged fruit out and deviating it to processing, thus preventing food waste. However, early-stage bruises or damages on plums are not visible, especially not in dark-skin cultivars. Therefore, this study aimed to explore the potential of hyperspectral imaging in the 430 to 1 000 nm range and deep learning algorithms to detect these invisible bruises at an early stage. To this end, 'Presenta' plums were impacted at three different levels to simulate varying degrees of damage. Images of both bruised and non-bruised plums were taken immediately after bruising and 24 and 48 h after bruise induction. Three distinct CNNs were trained to analyze the images. Two of these networks were implemented using transfer learning (ResNet and HSCNN), while the third was custom-designed for this specific purpose. The most informative wavelengths were identified as inputs for the CNNs employing PCA. F1 scores over 81% were obtained in all cases, and almost 100% accuracy was obtained in classifying the bruised plums with the highest impact energy of 0.50 J. Thus, detecting and classifying bruised plums using only three wavelengths is possible, paving the way for in-line sorting with multispectral cameras in packing houses.
•Invisible bruises can be detected using VIS-NIR hyperspectral imaging.•F1 scores higher than 87% were obtained for detecting invisible bruises.•Wavelength reduction to 3 bands obtained results higher than 86%.
The density of macrobenthos was evaluated in the nursery grounds for flatfish of six estuarine systems along the coast of Portugal by comparison with adjacent non-nursery areas. The dry weight and ...density of macrobenthic fauna were significantly higher in the nursery grounds. Polychaeta, Bivalvia, Oligochaeta and Isopoda were found to be significantly more abundant in the nursery than in the non-nursery grounds. For Isopoda and Bivalvia, total dry weight was also significantly higher in the nursery areas. Correspondence analysis based on density showed that the nursery areas of the different estuaries grouped together relative to non-nursery sites, with a relative similarity in the abundance of Oligochaeta, Spionidae,
Amage spp.,
Scrobicularia plana and
Cerastoderma edule. Taking into consideration the generally opportunistic feeding ecology and low dietary selectivity of the flatfish species, the results indicated higher prey availability in the nursery grounds of the Portuguese estuaries, an important factor contributing to the quality of these areas. The relatively high macrobenthic productivity in the nursery areas might be linked to physical and biological interactions favouring the recruitment and maintenance of the communities.
Retinal Vein Occlusion (RVO) is a common retinal vascular disorder which may cause severe loss of vision. Retinal cannulation appears to be the most effective treatment, but given the small diameter ...of a retinal vein, it is too difficult and risky for a surgeon to perform this procedure manually. This work reports on the development of an innovative robotic manipulator to assist vitreoretinal surgeons during this procedure using a co-manipulation control strategy. The robotic manipulator features a new Remote-Center-of-Motion mechanism with four degrees of freedom. This mechanism is particularly interesting for applications in minimally invasive surgery where an instrument needs to be manoeuvred in a highly confined space around a fixed incision point. The developed manipulator is shown to be a great asset in improving the quality of retinal cannulations compared to the manual procedure. This is shown by cannulation experiments performed on a custom made eye model and an injectable retina model that effectively simulate real retinal cannulations.
► Study of the influence of sampling effort on metrics of a fish-based index. ► Pseudo-random samples were generated from data of four Portuguese estuaries. ► For metrics on percentages, the means ...levelled off with less than 20 hauls. ► Metrics on species richness required much larger samples ► solutions alternative to increasing the sampling effort are suggested.
Multimetric fish-based indices have been increasingly gaining importance in Europe, as the Water Framework Directive (WFD) requires fish fauna, and particularly its composition and abundance, to be taken into account in the assessment of the ecological quality of continental surface waters, including transitional waters. These indices are composed of several metrics, mostly related with structural and functional characteristics of fish communities, such as species richness, the role of nursery areas, or trophic web structure. Therefore, ecological quality assessments should ensure that these structural and functional characteristics of fish communities were covered by the sampling methods used. In the present work, the influence of sampling effort on several metrics of the Estuarine Fish Assessment Index (EFAI) was studied. Pseudo-random samples were generated from data of four Portuguese estuaries and bootstrap cycles were performed, in order to obtain metrics’ means and standard deviations per number of hauls analysed. The number of hauls necessary for the means to level off differed with the metrics considered. Generally, for metrics on percentages (percentage of marine migrants, percentage of estuarine residents and percentage of piscivores) the curve levelled off with less than 20 hauls, both for the estuary as a whole and for different estuarine salinity zones. On the other hand, metrics on species richness required much larger samples. In order to decrease to −5% the current estimated bias of metrics, the WFD sampling costs would have to be more than 3 times higher than they currently are. The findings in the present study are of great importance for an effective assessment of estuarine ecological quality and particularly in the context of the WFD, as the metrics studied are common to other Member State indices.
Resilience research often includes time series analysis in search of trend shifts. Recently statistical signals including changes in variance and autocorrelation, were proven to be universal ...indicators of stability. In this paper the suitability of those early warning signals was tested on two case studies: oysters production of France (case study 1) and catch rates of the native oyster in the Solent, UK (case study 2). First, trend analyses were performed and their association to the evolution of the North Atlantic Oscillation (NAO) index was assessed. For the French oyster production, two sinusoidal waves were found in the first and second order residuals with periods of 33 and 8 years. In the Solent, the trend depiction showed after an initial increase, the catch rates of oysters declined over time from West to East. A positive relationship was apparent between NAO and both the production and population data, with a significant correlation in case study 2. Furthermore, a high and significant spectral coherency in the 8 year period of case study 1 revealed a near phase opposition between the influence of NAO and the French oyster production. For the first and second case study, the early warning signals calculated were lag-1 autocorrelation and variance, respectively. In case study 1, the stability of the low frequency wave of the first order residuals could not be assed, the trend of lag 1 autocorrelation was thus not conclusive. However the high but stable lag-1 autocorrelation of the second residuals, revealed a probable underlying stability. In case study 2 significant increase in variance reflected the instability prior to the decline. It is discussed that in periodic signals the number of residual fluctuations needs to be sufficient to allow the generation of an autocorrelation trend within the limits of the sliding window, used for its calculation. The importance of time scale and appropriate de-trending is also questioned when calculating and interpreting early warning signals as indicators of underlying (in)stability. Finally besides their scientific use, also the utility of early warning signals in managerial settings are discussed.