Protection and Restoration of Resources

Research focused upon the prototype development of technology, tools, and effective approaches to restoration, as well as bio-geographical characterizations, intended to enable improvements in defining and protecting components of marine protected areas and restoring habitats and populations. A wide range of problems are addressed from removing contaminants to providing new materials and techniques to protect underwater cultural resources.

The following are two representative projects that falls under this theme:

  1. Development and Evaluation of New Technology for the Remote Identification and Enumeration of Larval Fish by R.K. Cowen and C. Guigand (UM/RSMAS); G. Tsechpenakis (UM/CCS); J. Hare (NOAA/NEFSC)

  2. Monitoring Shoreline Fish Assemblages of Biscayne and Florida Bays by D. Johnson, B. Teare and L. Visser (UM/CIMAS); J. Luo (UM/RSMAS); J. Serafy (NOAA/SESFC)

Representative Projects


Development and Evaluation of New Technology for the Remote Identification and Enumeration of Larval Fish

R.K. Cowen and C. Guigand (UM/RSMAS); G. Tsechpenakis (UM/CCS); J. Hare (NOAA/NEFSC)
 

Long Term Research Objectives and Strategy to Achieve Them:

Objectives: To support the development of imaging technology software for the remote enumeration of larval fish. To evaluate the ISIIS technology in the context of ongoing efforts to improve stock assessments.

Strategy: Since the imaging system produces very high resolution imagery at very high data rates necessitating automated image analysis. Our approach aims at the detection of multiple regions (organisms) of interest automatically, while filtering out noise and out-of-focus organisms, and the classification of the detected organisms into pre-defined categories using shape and texture information. Last year we conducted an initial evaluation of ISIIS as an assessment tool for Atlantic Herring on Georges Bank. Atlantic herring larvae are a dominant component of the ichthyoplankton in the late fall / early winter on Georges Bank and have a unique shape compared to other ichthyoplankton present in the area at that time of year. These factors would,
we thought, maximize the ability of ISIIS to image and automatically classify herring larvae and our comparison of traditional and ISIIS sampling is confirming that assumption.

 

ISIIS is a towed digital imaging system capable of quantifying larval fish in situs, via high volume imaging (~70 l s-1) at high resolution (effect <70 μm per pixel). The basic design includes: shadowgraph lighting, a combination of various state-of-the-art digital imaging and computer technologies (i.e. corporating machine vision technology with line scan cameras), and fiber-optic data communication. The towed vehicle also carries sensors that measure
temperature, salinity and depth. Preliminary deployments off the east coast of Florida demonstrate that ISIIS takes clear, identifiable and quantifiable images of larval fish. The major challenge to broad applicability of ISIIS is the development of software that automates target recognition, identification and enumeration. Our first objective addresses this challenge. A second challenge involves evaluating the utility of ISIIS in a fisheries management application. This challenge is addressed by our second objective.

Figure 1. ct 2009 NOAA cruise tract overlain on Ocean Color image of study area.

Preliminary sample analyses – During Oct 2008, we used ISIIS in conjunction with a NOAA cruise collaborating with Dr. Jon Hare (NOAA/NMFS). We conducted two ~ 40 km transect south of Rhode Island (SW of Woods Hole), and conducted replicate BONOG tows to collect comparison ichthyoplankton samples. All plankton samples were sorted and identified, and counted. Similarly, all images were manually and automatically counted (using image analysis). Figure 1 shows a satellite (color) image of the study area with the blow-up designating where each ISIIS transect was run. Note the
transect moved inshore-offshore and thereby transited from high nutrient to lower nutrient waters. Along each transect, we undulated ISIIS form the surface to ca. 40 m depth or within 10 m of the bottom (when bottom depth was less than 50 m). Profiles of tempera-ture, fluorometry and salinity reveal both vertical and horizontal structure (see Figure 2).

Figure 2. Vertical profiles of fluorometry, temperature and salinity for each of two transects. Note the undulation pathways depicted by the dotted lines in the top panels.

Actual locations of individual larval fish observed by ISIIS are depicted in Figure 3, overlain on the fluorometry. This visualization gives unprecedented detail on where in the water column different species of larval fish are located.

Figure 3. Species-specific, depth-specific occurrences of larval fish along transect, overlain on fluorometry. Individual species are color-coded.

To compare the ability of ISIIS to capture both numerically and with sufficient resolution to identify larval fish at comparable rates as the standard BONOG net method, we compared actual catches of the two methods (Table 1). Overall, while some small larvae were difficult to identify in the imagery, the taxonomic distribution and relative number of each taxon were comparable. The total estimated density, however, was lower with ISIIS. Initially we were not sure what to make of this, but then we realized that the BONGO nets were sampling deeper in the water column than ISIIS – so we examined the trend in catch (density) with depth (Figure 4), and found that it increased with depth such that our interpretation is that had we taken ISIIS into deeper water (i.e. as we did with the BONGO net) we would have collected overall comparable densities.

Table 1. Comparison of catch/imagery by fish taxon for BONGO net and ISIIS.

Figure 4. Analysis of number of fish larvae seen by depth stratus in ISIIS imagery. Upper left panel is total number observed by depth, upper right panel
is volume of water sampled by depth. Lower panel combines above panels to estimate larval density by depth.

 

Monitoring Shoreline Fish Assemblages of Biscayne and Florida Bays

D. Johnson, B. Teare and L. Visser (UM/CIMAS); J. Luo (UM/RSMAS); J. Serafy (NOAA/SESFC)
 

Long Term Research Objectives and Strategy to Achieve Them:

Objectives: Shoreline Fish Community Visual Assessment (SFCVA) monitoring component is part of the REstoration, COordination and VERification (RECOVER) program of the Comprehensive Everglades Restoration Plan (CERP). Specific objectives of the SFCVA monitoring component are: (1) to continue the seasonally-resolved, 12-year visual fish monitoring effort that, for the most part, has focused on southern Biscayne Bay; (2) to expand this effort spatially to include sites in northern Biscayne Bay, Card Sound, Barnes Sound and northeastern Florida Bay; (3) to perform data analyses that evaluate variability in these fish communities before, during, and (ultimately) after CERP-related changes to freshwater flow (and salinity) are implemented; and (4) to correlate changes in salinity with changes in the shoreline ichthyofauna. These objectives are being met via calculation of the minimum numbers of samples required to detect change, review of historical literature and existing datasets, collection of new data, and analyses of the “baseline condition” of shoreline fish assemblages at both the community and taxon-specific levels. Its purpose is to provide long-term baseline data and to evaluate the CERP-related impacts on bay systems which are likely to be the strongest and most easily discerned along the mangrove-lined shorelines of South Florida’s mainland.

Strategy: Maintain long-term data monitoring program and develop fish habitat suitability index models with an emphasis on revealing abundance-salinity relationships, through analysis of existing empirical data collected from Biscayne Bay and adjacent systems.

 

1. Annual trends in Taxon-Specific Abundances: Densities of all taxa have been found to be relatively stable over the time series when plotted against
time. All of the taxa examined showed some level of seasonal variation in density and frequency of occurrence and concentration tended to track each other. No clear annual trends emerged within shoreline segments. Densities of the four major taxa were higher at the leeward key shorelines than the mainland shorelines. Along the Mainland shoreline, 75% of the fish community metrics estimated for 2009 fell within the 95% confidence intervals of the historical mean, 25% were higher, and 8% were lower.

2. Habitat Suitability Models: We used a delta approach to generate a triad of habitat suitability index (HSI) models per species. The approach allowed for the testing of three HSI models per combination because three “abundance metrics” are considered: frequency of occurrence, concentration (density when present, exclusive of zeros) and “delta-density” (occurrence x concentration). In the present project, and provide results in both graphic and mathematical form. This was achieved using the 11.5 years of visual census fish monitoring effort.

We detected statistically-significant trends across salinity gradients in one or more abundance metrics six taxa. Where observations under hypersaline conditions were available, most of the statistically-significant salinity trends for individual taxa showed abundance declines beyond 36 psu. The metrics tended to show linear or parabolic relationships with salinity for Biscayne Bay fishes.

3. Community Analyses: We calculated average taxonomic richness across years for the composite and subdivided mainland shoreline and the leeward key shoreline. We examined seasonal and annual variation of yearly indices of taxonomic richness and dominance using multivariate regression and analysis of variance. We found that richness was higher along the Leeward Key where the environment was more stable than along the Mainland Shoreline.

We did not find annual or seasonal differences in richness along the Leeward Key. We found that dry season differences in samples were correlated with temperature, salinity, depth, and the interactions of year x temperature and salinity x depth, while wet season differences were only related to salinity. We found no significant annual differences in richness in the dry season along the Mainland Shoreline, but there were differences in the wet season. We found that dry season differences were primarily due to salinity, depth, dissolved oxygen, temperature, and the interaction of depth and salinity. Wet season differences were correlated with salinity, depth, dissolved oxygen, and the interaction of salinity and depth.

4. Canal Analyses: We found more taxa near low-flow canals than near high-flow canals, especially in the wet season, suggesting that the quantity of freshwater flow has a negative effect. We found impacts from highflow canals as far as 2000 m in the wet season.

Figure 1. Relative abundance ranked using quartiles of selected wet season Biscayne Bay species sampled in visual census surveys in relationship with distance from high-flow and low-flow canals. Black is in the 75% quartile (high density), gray 25-75% (medium density), and white 25% (low density). Low
density bins that are zero are indicated. ACE is a group of silvery fishes composed of athernids, clupeids, and engraulids.

Figure 2. Relative abundance ranked using quartiles of selected dry season Biscayne Bay species sampled in visual census surveys in relationship with distance from high-flow and low-flow canals. Black is in the 75% quartile (high density), gray 25-75% (medium density), and white 25% (low density). Low density bins that are zero are indicated. ACE is a group of silvery fishes composed of athernids, clupeids, and engraulids.