Accuracy and Precision of Low-Cost Echosounder and Automated Data Processing Software for Habitat Mapping in a Large River

Round 1

Reviewer 1 Report

The authors present results from an assessment of a recreational-grade Lowrance HDS echosounder combined with data analysis using cloud-based BioBase EcoSound software for depth, substrate hardness and vegetation in the Saint John River, New Brunswick, Canada. The study was well-conceived, results were of value to those interested in characterization of river and lake ecosystems, and I found the manuscript to be quite well-written. Notwithstanding, I do have a few comments that warrant consideration by the authors.

First of all, the authors present results using a comprehensive set of statistical analyses of direct measurements and hydroacoustically-derived attributes.  Against this backdrop, I found the statement on l.112 that “Transect locations were chosen haphazardly…” to be rather surprising. Why wasn’t a more thoughtful statistically-based sampling scheme adopted? It is difficult to ascertain how representative the sampling ultimately was. Perhaps a map showing the transect and point measurements would be appropriate to at least indicate spatial relationships between sampling locations.

Related in some ways to the above comment, what is the range of depths surveyed? A cumulative distribution function or probability distribution function of depths (ideally comparing echosounder and manual measurements) would help the reader better understand the physical setting for the study. This is also helpful given that a key finding that apparent hardness was found to vary inversely with depth for a specific substrate class (l.321-324 ). Moreover, some additional details are warranted in discussion of bottom hardness in section 4.2. Specifically, hardness is more than simply substrate size class – the water content and bulk density are important attributes. Are any data available about these properties for the finer textured material? Moreover, the authors should also elaborate on the specific acoustic attribute of the bottom echo used within the BioBase EcoSound software to estimate bottom hardness (E2).

The section describing sources of errors (section 4.4) is valuable but would benefit from a few more tangible details and examples. Specifically, the authors should provide the beam angle of the transducer and compare the ensonified areas vs. sampled areas for different depths and sampling locations. While their particular study site apparently had little gradient in slope, some further elaboration on how steeper gradients can influence analysis and interpretation might be useful for those interested in sites with greater variation in depth.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

This is a very well written manuscript with a robust method for testing. I only found three issues that I thought the authors should address. One is the statement on line 357 that "...EcoSound showed changes in biovolume with a smoother pattern than what was observed during manual observation", which references Figure 5. It seems to me that part of this is attributed to how vegetation was measured with the manual method (line 255 says the upper limit [25% used for 1-25% category), which is evident with the graph in Figure 5. I did not see where this was brought up in the Discussion and should as a methodological artifact (unless I misinterpreted). Two, is that I would like to see some information from the study system on turbidity and how this might affect results. Because the authors found submersed aquatic vegetation, I suspect that water clarity was quite high, but also because other readers will want to apply the results of this study to their system, a brief discussion about how turbidity might affect results in other systems would be beneficial. Three, is a brief discussion of how BioBase EcoSound that uses a kriging algorithm that is unknown to the user creates a "black box" system and that more advanced users might be able to tweek a kriging procedure to get better results. The authors allude to this in their last sentence of section 4 (line 483-485), citing a study that examined three interpolation methods, which included kriging. Kriging uses one of several underlying models of spatial dependence and these models can be evaluated against each other to produce different results from a kriging procedure.

Including these three issues would strengthen the paper.

Author Response

Please see the attachment.

Author Response File: Author Response.docx