利用水下立体视频工具量化鱼类密度的新方法研究进展

The overall goal of this video analysis technique is to more accurately estimate density, mean length and species composition of fish in deep rocky habitats. This method can help answer key questions in the field of fisheries management such as what is the abundance and size distribution of species inhabiting deep-water rocky reefs. The main advantage of this technique is that it provides more accurate density estimates without extracting fish from the environment.

In advance of this procedure, collect field data as outlined in the text protocol. In stereophotogrammetry, knowing the exact relative position of the cameras is important for accurate measurements. Proper calibration is an important step in this process.

After the field study is complete, create a new project folder containing both the video and calibration files. In the stereo measurement software, navigate to Measurement, New measurement file. Set the picture directory by navigating to Picture.

Set picture directory and then choose the folder containing all of the project files. Navigate to Stereo, Cameras, Left, then Load camera file to select and load the appropriate left camera file. Repeat this process choosing Right instead to load the right camera file.

Next, navigate to Picture, Define movie sequence. Select the left camera video file to define the movie sequence for the left video. Click Picture, Load picture to load the left video file into the measurement software.

After this, click Stereo, Picture, Define movie sequence to define the movie sequence for the right video. Load the video file by selecting Stereo, Picture then load video. Navigate to Measurement, Attributes, Edit load species file to load the species list.

Click Measurement, Information fields, Edit field values to open the information field values table. Enter the survey ID information and save the file to create an event measure observations project. If using a UTC timestamp, frame step forward in the left video until the timestamp starts a new second or until a light flash or hand clap occurs.

Frame step the right video forward until the timestamp light flash or hand clap matches the left video exactly. Then, click the Lock button to ensure the videos play together and maintain synchronization. As soon as the lander starts its first rotation, right-click and select Period definitions, Add new start period to define a new sample period.

Enter zero one as the first period name and click OK.While the lander rotates, mark each fish that comes into frame with a 2D point by right-clicking, select Add point and choose the correct species name. Label to the lowest possible taxonomic level and then click OK.Continue marking each new fish until the rotation completes. It's critical to identify and count each fish to get accurate estimates of MaxN.

Repeat this process for an additional lander rotation ensuring that a new period is defined at the start of each. After all rotations have been enumerated, navigate to Measurement, Measurement summaries, Point measurements and save the 2D points as a TXT file. Open this file as a spreadsheet.

Navigate to Insert, Pivot Table to create a pivot table. Selecting Genus and Species for the row label and Period for the column label. Select the camera rotation that has the greatest number of individuals for a given species to choose the MaxN for that species.

For fish identified only to genus, select a genus level MaxN based on the rotation that had the greatest number of individuals identified to species in that particular genus. Next, use the saved 2D points to navigate to the exact same fish for the 3D measurement. Zoom in at least four times to better identify the tip of the fish snout and the edges of the caudal fins.

Manually click on the tip of the snout and then the edge of the tail in the left camera. Repeat the selection in the same order in the right video. Then, right-click, select Add length and select the correct species identification.

If 3D length measurement is not possible, left-click in the same position on the fish in both videos to mark a 3D point. Fill out the information fields leaving the comment Exclude from length measurement. After completing 3D measurements for all fish, navigate to Measurement, Measurement summaries and 3D point and length measurements.

Save the data as a TXT file to export it for further analysis. Then, determine if adequate samples were obtained as outlined in the text protocol. In this study, underwater stereo video tools are utilized to quantify fish density.

There are clear patterns in the detectable range of the species observed which is likely due to the interaction of each species'size, shape and coloration. The 95%Z distance calculations are then performed for two species in particular. For Sebastes wilsoni and Ophiodon elongatus, the 95%Z distance is seen to be 2.65 meters for Sebastes wilsoni and 3.96 meters for Ophiodon elongatus which translates into effective survey areas of 18.6 meters squared and 46 meters squared respectively.

A simple bootstrap analysis confirms that sufficient sample sizes are obtained as the estimate of 95%Z distance for both samples stabilizes when over 50 surveys are sampled. MaxN counts per survey are then converted into densities. For both species, the densities are seen to be significantly greater over high and medium-relief habitats as compared to low-relief habitats.

Density estimates for the pseudo-stationary lander are standardized using reduced areas of coverage. Mean densities obtained by the rotating camera are 18%greater than those obtained with stationary cameras. Additionally, the coefficient of variation is seen to be a 1.8 times greater when using stationary cameras.

Once mastered, this technique can be used to count and measure fish in only a few minutes if properly performed. When performing this procedure, it's important to remember that 95%Z values are tool and survey specific. And specific values should not be used universally.

Following this procedure, a variety of multivariate or ordination statistics can be performed to answer additional questions about species composition across different habitat types. The implications of this technique extend toward improved understanding of the ecology of deep-water rocky reef species as current survey mechanisms provide only a poor understanding of fish length and abundances. Thought this technique can provide insight into deep-water marine habitats, it can also be useful in other systems such as coral reefs and kelp forests.

Generally, individuals new to this method will struggle because it requires an understanding of the geometry of stereo camera systems. Visual demonstration of this technique is helpful because the calculation of MaxN is derived from a variety of data and thus requires many steps in the software.