Interest and growth in marine aquaculture are increasing around the world, and with it, advanced spatial planning approaches are needed to find suitable locations in an increasingly crowded ocean. Standard spatial planning approaches, such as a Multi-Criteria Decision Analysis (MCDA), may be challenging and time consuming to interpret in heavily utilized ocean spaces. Spatial autocorrelation, a statistical measure of spatial dependence, may be incorporated into the planning framework, which provides objectivity and assistance with the interpretation of spatial analysis results. Here, two case studies highlighting applications of spatial autocorrelation analyses in the northeast region of the United States of America are presented. The first case study demonstrates the use of a local indicator of spatial association analysis within a relative site suitability analysis – a variant of a MCDA – for siting a mussel longline farm. This case study statistically identified 17% of the area as highly suitable for a mussel longline farm, relative to other locations in the area of interest. The use of a clear, objective, and efficient analysis provides improved confidence for industry, coastal managers, and stakeholders planning marine aquaculture. The second case study presents an incremental spatial autocorrelation analysis with Moran’s I that is performed on modeled and remotely sensed oceanographic data sets (e.g., chlorophyll a, sea surface temperature, and current speed). The results are used to establish a maximum area threshold for each oceanographic variable within the online decision support tool, OceanReports, which performs an automated spatial analysis for a user-selected area (i.e., drawn polygon) of ocean space. These thresholds provide users guidance and summary statistics of relevant oceanographic information for aquaculture planning. These two case studies highlight practical uses and the value of spatial autocorrelation analyses to improve the siting process for marine aquaculture.
Remote Sensing and GIS
Mangroves provide many ecosystem services including a considerable capacity to sequester and store large amounts of carbon, both in the sediment and in the above-ground biomass. Assessment of mangrove above-ground carbon stock relies on accurate measurement of tree biomass, which traditionally involves collecting direct measurements from trees and relating these to biomass using allometric relationships. We investigated the potential to predict tree biomass using measurements derived from unmanned aerial vehicle (UAV), or drone, imagery. This approach has the potential to dramatically reduce time-consuming fieldwork, providing greater spatial survey coverage and return for effort, and may enable data to be collected in otherwise hazardous or inaccessible areas. We imaged an Avicennia marina (grey mangrove) stand using an RGB camera mounted on a UAV. The imaged trees were subsequently felled, enabling physical measurements to be taken for traditional biomass estimation techniques, as well as direct measurements of biomass and tissue carbon content. UAV image-based tree height measurements were highly accurate (R2 = 0.98). However, the variables that could be measured from the UAV imagery (tree height and canopy area) were poor predictors of tree biomass. Using the physical measurement data, we identified that trunk diameter is a key predictor of A. marina biomass. Unfortunately, trunk diameter cannot be directly measured from the UAV imagery, but it can be predicted (with some error) using models that incorporate other UAV image-based measurements, such as tree height and canopy area. However, reliance on second-order estimates of trunk diameter leads to increased uncertainty in the subsequent predictions of A. marina biomass, compared to using physical measurements of trunk diameter taken directly from the trees. Our study demonstrates that there is potential to use UAV-based imagery to measure mangrove A. marina tree structural characteristics and biomass. Further refinement of the relationship between UAV image-based measurements and tree diameter is needed to reduce error in biomass predictions. UAV image-based estimates can be made far more quickly and over extensive areas when compared to traditional data collection techniques and, with improved accuracy through further model-calibration, have the potential to be a powerful tool for mangrove biomass and carbon storage estimation.
Offshore aquaculture of giant kelp (Macrocystis pyrifera) has been proposed by the US Department of Energy for large scale biofuel production along the west coast of California. The Southern Californian Bight provides an ideal area for offshore kelp aquaculture as the upwelling and advection of cool, nutrient-rich waters supports the growth of vast native giant kelp populations. However, concentrations of nutrients vary greatly across space, can be limiting for kelp growth over seasonal to interannual time scales, and inputs of nutrients to surface waters may be subject to local circulation processes. Therefore, it is important to understand both the spatiotemporal variability of seawater nitrate concentrations and the appropriate scale of observation in order for offshore kelp aquaculture to be successful. Here, we use a combination of satellite sea surface temperature imagery, in situ measurements, and modeling to determine seawater nitrate fields across multiple spatial and temporal scales. We then combine this information with known giant kelp physiological traits to develop a kelp stress index (KSI) for the optimal siting of offshore kelp aquaculture over seasonal to decadal scales. Temperature to nitrate relationships were determined from in situ measurements using generalized additive models and validated with buoy data. Summer and winter relationships were significantly different, and satellite-derived products compared well to buoy validations. Surface nitrate patterns, as derived from satellite temperature products, reveal the spatial variability in nitrate concentrations, and indicate areas that that may cause nutrient stress seasonally and during the negative phase of the North Pacific Gyre Oscillation. As the spatial scale of the surface nitrate product decreased, the negative bias increased and fine scale spatial variability was lost. Similarly, the averaging of daily nitrate concentration determinations over longer time scales increased the negative bias. We found that daily, 1 km spatial resolution nitrate products were most sufficient for identifying localized upwelling and areas of consistently high surface nitrate concentrations, and that areas in the northern and western-most portions of the Southern California Bight are the most suitable for sustained offshore kelp aquaculture.
Environmental DNA (eDNA) analysis is a rapid, non-invasive method for species detection and distribution assessment using DNA released into the surrounding environment by an organism. eDNA analysis is recognised as a powerful tool for detecting endangered or rare species in a range of ecosystems. Although the number of studies using eDNA analysis in marine systems is continually increasing, there appears to be no published studies investigating the use of eDNA analysis to detect sea turtles in natural conditions. We tested the efficacy of two primer pairs known to amplify DNA fragments of differing lengths (488 and 253 bp) from Chelonia mydas tissues for detecting C. mydas eDNA in water samples. The capture, extraction, and amplification of C. mydas eDNA from aquaria (Sea World, San Diego, CA, United States) and natural water (San Diego Bay, CA, United States) were successful using either primer set. The primer pair providing the shorter amplicon, LCMint2/H950g, demonstrated the ability to distinguish cross-reactive species by melt curve analysis and provided superior amplification metrics compared to the other primer set (LTCM2/HDCM2); although primer dimer was observed, warranting future design refinement. Results indicated that water samples taken from deeper depths might improve detection frequency, consistent with C. mydas behaviour. Overall, this pilot study suggests that with refinement of sampling methodology and further assay optimisation, eDNA analysis represents a promising tool to monitor C. mydas. Potential applications include rapid assessment across broad geographical areas to pinpoint promising locations for further evaluation with traditional methods.
An understanding of marine ecosystems and their biodiversity is relevant to sustainable use of the goods and services they offer. Since marine areas host complex ecosystems, it is important to develop spatially widespread monitoring networks capable of providing large amounts of multiparametric information, encompassing both biotic and abiotic variables, and describing the ecological dynamics of the observed species. In this context, imaging devices are valuable tools that complement other biological and oceanographic monitoring devices. Nevertheless, large amounts of images or movies cannot all be manually processed, and autonomous routines for recognizing the relevant content, classification, and tagging are urgently needed. In this work, we propose a pipeline for the analysis of visual data that integrates video/image annotation tools for defining, training, and validation of datasets with video/image enhancement and machine and deep learning approaches. Such a pipeline is required to achieve good performance in the recognition and classification tasks of mobile and sessile megafauna, in order to obtain integrated information on spatial distribution and temporal dynamics. A prototype implementation of the analysis pipeline is provided in the context of deep-sea videos taken by one of the fixed cameras at the LoVe Ocean Observatory network of Lofoten Islands (Norway) at 260 m depth, in the Barents Sea, which has shown good classification results on an independent test dataset with an accuracy value of 76.18% and an area under the curve (AUC) value of 87.59%.
The sensitivity to changes in water quality inherent to seagrass communities makes them vital for determining the overall health of the coastal ecosystem. Numerous efforts including community-based coastal resource management, conservation and rehabilitation plans are currently undertaken to protect these marine species. In this study, the relationship of water quality parameters, specifically chlorophyll-a (chl-a) and turbidity, with seagrass percent cover is assessed quantitatively. Support Vector Machine, a pixel-based image classification method, is applied to determine seagrass and non-seagrass areas from the orthomosaic which yielded a 91.0369% accuracy. In-situ measurements of chl-a and turbidity are acquired using an infinity-CLW water quality sensor. Geostatistical techniques are utilized in this study to determine accurate surfaces for chl-a and turbidity. In two hundred interpolation tests for both chl-a and turbidity, Simple Kriging (Gaussian-model type and Smooth- neighborhood type) performs best with Mean Prediction equal to −0.1371 FTU and 0.0061 μg/L, Root Mean Square Standardized error equal to −0.0688 FTU and −0.0048 μg/L, RMS error of 8.7699 FTU and 1.8006 μg/L and Average Standard Error equal to 10.8360 FTU and 1.6726 μg/L. Zones are determined using fishnet tool and Moran’s I to calculate for the seagrass percent cover. Ordinary Least Squares (OLS) is used as a regression analysis to quantify the relationship of seagrass percent cover and water quality parameters. The regression analysis result indicates that turbidity has an inverse relationship while chlorophyll-a has a direct relationship with seagrass percent cover.
Image classification allows to profess a great number of images and identify specific objects within these images...
We present thorough this review the developments in the field, point out their current limitations, and outline its timelines and unique potential. In order to do so we introduce the methods used in each of the advances in the application of deep learning (DL) to coral research that took place between the years: 2016–2018. DL has unique capability of streamlining the description, analysis, and monitoring of coral reefs, saving time, and obtaining higher reliability and accuracy compared with error-prone human performance. Coral reefs are the most diverse and complex of marine ecosystems, undergoing a severe decline worldwide resulting from the adverse synergistic influences of global climate change, ocean acidification, and seawater warming, exacerbated by anthropogenic eutrophication and pollution. DL is an extension of some of the concepts originating from machine learning that join several multilayered neural networks. Machine learning refers to algorithms that automatically detect patterns in data. In the case of corals these data are underwater photographic images. Based on “learned” patterns, such programs can recognize new images. The novelty of DL is in the use of state-of-art computerized image analyses technologies, and its fully automated methodology of dealing with large data sets of images. Automated Image recognition refers to technologies that identify and detect objects or attributes in a digital video or image automatically. Image recognition classifies data into selected categories out of many. We show that Neural Network methods are already reliable in distinguishing corals from other benthos and non-coral organisms. Automated recognition of live coral cover is a powerful indicator of reef response to slow and transient changes in the environment. Improving automated recognition of coral species, DL methods already recognize decline of coral diversity due to natural and anthropogenic stressors. Diversity indicators can document the effectiveness of reef bioremediation initiatives. We explored the current applications of deep learning for corals and benthic image classiﬁcation by discussing the most recent studies conducted by researchers. We review the developments in the field, point out their current limitations, and outline their timelines and unique potential. We also discussed a few future research directions in the ﬁelds of deep learning. Future needs are the age detection of single species, in order to track trends in their population recruitment, decline, and recovery. Fine resolution, at the polyp level, is still to be developed, in order to allow separation of species with similar macroscopic features. That refinement of DL will allow such comparisons and their analyses. We conclude that the usefulness of future, more refined automatic identification will allow reef comparison, and tracking long term changes in species diversity. The hitherto unused addition of intraspecific coral color parameters, will add the inclusion of physiological coral responses to environmental conditions and change thereof. The core aim of this review was to underscore the strength and reliability of the DL approach for documenting coral reef features based on an evaluation of the currently available published uses of this method. We expect that this review will encourage researchers from computer vision and marine societies to collaborate on similar long-term joint ventures.
Underwater gliders have become widely used in the last decade. This has led to a proliferation of data and the concomitant development of tools to process the data. These tools are focused primarily on converting the data from its raw form to more accessible formats and often rely on proprietary programing languages. This has left a gap in the processing of glider data for academics, who often need to perform secondary quality control (QC), calibrate, correct, interpolate and visualize data. Here, we present GliderTools, an open-source Python package that addresses these needs of the glider user community. The tool is designed to change the focus from the processing to the data. GliderTools does not aim to replace existing software that converts raw data and performs automatic first-order QC. In this paper, we present a set of tools, that includes secondary cleaning and calibration, calibration procedures for bottle samples, fluorescence quenching correction, photosynthetically available radiation (PAR) corrections and data interpolation in the vertical and horizontal dimensions. Many of these processes have been described in several other studies, but do not exist in a collated package designed for underwater glider data. Importantly, we provide potential users with guidelines on how these tools are used so that they can be easily and rapidly accessible to a wide range of users that span the student to the experienced researcher. We recognize that this package may not be all-encompassing for every user and we thus welcome community contributions and promote GliderTools as a community-driven project for scientists.
Tomorrow’s smart lakes and oceans will be able to, among other things, predict changes in the water environment and produce information critical to proper management and planning. Smart Lake Erie – a proof of concept – will integrate data from distributed sensors using resilient networks to feed adaptive, predictive analytics that define and perhaps even perform necessary management actions. This paper presents the Smart Lake Erie pilot as a series of steps that include convening innovation competitions, engaging stakeholders, securing the core observation system, and designing then building a sustainable early warning system for harmful algal blooms. The pilot’s data platform will show what is needed to serve new data contributors, service providers, stakeholders and consumers of the data and information service paradigm. Lessons learned from the early implementation of the pilot will be applicable to the overall Great Lakes region, other regional associations within the U.S. Integrated Ocean Observing System (IOOS), and the Global Ocean Observing System (GOOS).