Academia.eduAcademia.edu

Figure 5 – uploaded by Rupert Ormond

See full PDFdownloadDownload figure
Fig. 5. Cumulative standard deviations (of arcsine square-root transformed data) for randomised hard coral cover for the six methods. Note that for MAP, PHOTP and PHOTS, replicates are single quadrats, whereas for VIDEO they are single frames and for LIT and LPT they are 5 m sections. Assuming MAP to provide the best estimate of benthic cover, the accuracies of the five other methods were evaluated as the differences between cover values estimated by each and the values determined by MAP (Table 1). Overall, among the other methods, VIDEO appears to be the most accurate, followed by PHOTP. Hard coral cover was most accurately estimated by VIDEO and least accurately by LPT, with all photo- graphic methods tending to underestimate hard coral On comparison with MAP, PHOTP appeared to give the most accurate estimates of cover by different hard coral growth forms, followed by PHOTS (Table 3), with the least accurate being obtained by LPT, LIT and VIDEO. PHOTP and PHOTS underestimated the cover of both branching and encrusting coral, and PHOTS the cover of massive corals (Fig. 3). LPT, LIT and VIDEO over- estimated the cover of massive corals and underestimated the cover of encrusting corals. Kruskal—Wallis tests

Figure 5 Cumulative standard deviations (of arcsine square-root transformed data) for randomised hard coral cover for the six methods. Note that for MAP, PHOTP and PHOTS, replicates are single quadrats, whereas for VIDEO they are single frames and for LIT and LPT they are 5 m sections. Assuming MAP to provide the best estimate of benthic cover, the accuracies of the five other methods were evaluated as the differences between cover values estimated by each and the values determined by MAP (Table 1). Overall, among the other methods, VIDEO appears to be the most accurate, followed by PHOTP. Hard coral cover was most accurately estimated by VIDEO and least accurately by LPT, with all photo- graphic methods tending to underestimate hard coral On comparison with MAP, PHOTP appeared to give the most accurate estimates of cover by different hard coral growth forms, followed by PHOTS (Table 3), with the least accurate being obtained by LPT, LIT and VIDEO. PHOTP and PHOTS underestimated the cover of both branching and encrusting coral, and PHOTS the cover of massive corals (Fig. 3). LPT, LIT and VIDEO over- estimated the cover of massive corals and underestimated the cover of encrusting corals. Kruskal—Wallis tests

Related Figures (13)

This study was undertaken at Ras Um Sidd, the southernmost headland within the coastal resort of Sharm El Sheikh, South Sinai, Egypt (Fig. 1). There, reef topography and community are typical of sheltered fringing reefs in the northern Red Sea, with a steep slope and high coral cover, dominated by a mixed assemblage of corals on the upper reef face, and massive Porites Fig. 1. Map of the south-eastern part of the Sinai Peninsula, Egypt, showing the location of the sampling site at Ras Um Sidd, in relation to Sharm El] Sheikh, and the Ras Mohammed National Park (NP).
This study was undertaken at Ras Um Sidd, the southernmost headland within the coastal resort of Sharm El Sheikh, South Sinai, Egypt (Fig. 1). There, reef topography and community are typical of sheltered fringing reefs in the northern Red Sea, with a steep slope and high coral cover, dominated by a mixed assemblage of corals on the upper reef face, and massive Porites Fig. 1. Map of the south-eastern part of the Sinai Peninsula, Egypt, showing the location of the sampling site at Ras Um Sidd, in relation to Sharm El] Sheikh, and the Ras Mohammed National Park (NP).
Differences in percent cover of broad substrate categories between the best estimate provided by the MAP method (column 1) and the estimates obtained using the other five methods
Differences in percent cover of broad substrate categories between the best estimate provided by the MAP method (column 1) and the estimates obtained using the other five methods
Fig. 2. a) and b) Percent cover of the main benthic substrate categories as estimated by the six different sampling methods. a) shows the two most abundant categories, b) the less abundant categories, hence the difference in vertical scales. Rare categories (recruits, other biota, indeterminate substrate) are not shown. Error bars show 95% confidence intervals (except for the category damage, where error bars indicate standard deviations). All sampling methods were initially applied using the maximum sampling effort possible or a very high
Fig. 2. a) and b) Percent cover of the main benthic substrate categories as estimated by the six different sampling methods. a) shows the two most abundant categories, b) the less abundant categories, hence the difference in vertical scales. Rare categories (recruits, other biota, indeterminate substrate) are not shown. Error bars show 95% confidence intervals (except for the category damage, where error bars indicate standard deviations). All sampling methods were initially applied using the maximum sampling effort possible or a very high
*Indicates that the respective category was recorded by one method only. Results of one-way ANOVA and Kruskal-Wallis tests, as well as post- hoc Tukey tests and Kruskal—Wallis pair-wise comparisons for estimates of substrate cover by main benthic substrate categories and by hard coral growth forms, obtained using the different methods
*Indicates that the respective category was recorded by one method only. Results of one-way ANOVA and Kruskal-Wallis tests, as well as post- hoc Tukey tests and Kruskal—Wallis pair-wise comparisons for estimates of substrate cover by main benthic substrate categories and by hard coral growth forms, obtained using the different methods
Differences in percent cover of coral growth forms between the best estimate provided by the MAP method (column 1) and the estimates obtained using the other five methods
Differences in percent cover of coral growth forms between the best estimate provided by the MAP method (column 1) and the estimates obtained using the other five methods
(Zar, 1999; Barker Bausell and Yu-Fang Li, 2002). Powe analysis was carried out using arcsine square-roo transformed data and was only applied for data tha followed a normal distribution after transformation. Prio to carrying out a power analysis, for each of the sp methods cumulative standard deviation curves of rando mised means were plotted for selected substrate types 1 order to determine whether these curves levelled off thereby indicating that the sample size is large enough t¢ give a true measure of the respective standard deviations for use in the power analysis. For most methods standar« deviations showed a reasonable degree of levelling off Another parameter required for the power analysis i Pearson’s r. This value indicates the strength o correlation between paired observations and is therefor basically an expression of the placement error that arise: through repeatedly laying the transect line. For VIDEO Pearson’s r was calculated from two repeated lin placements, and was found to range from 0.5 to 0.9 fo different substrates, being 0.7 for hard coral. For MAI and PHOTS Pearson’s 7 was taken from a study tha repeatedly assessed quadrats on the Ras Um Sidd reef fla and was 0.8 for abundant substrate categories (> 10% cover) (Leujak, 2006). Pearson’s r was assumed to bi similar for PHOTP and was high for all quadrat method: since it is relatively easy to relocate the same quadrats fo repeat sampling by taking underwater laminated copies o either the mapped quadrats or the quadrat photographs For LPT and LIT Pearson’s 7 was assumed to be only 0.5 since the placement error, due to relaying the transect lint along a slightly different route, can be relatively large. Th« “effect sizes” (ES) employed in power analysis wer calculated as desired detectable difference in cover value: Fig. 3. Percent cover of different hard coral growth forms as estimated by the six different sampling methods. Rare forms such as tabular Acropore spp., free-living corals and foliose corals are not shown. Error bars show 95% confidence intervals.
(Zar, 1999; Barker Bausell and Yu-Fang Li, 2002). Powe analysis was carried out using arcsine square-roo transformed data and was only applied for data tha followed a normal distribution after transformation. Prio to carrying out a power analysis, for each of the sp methods cumulative standard deviation curves of rando mised means were plotted for selected substrate types 1 order to determine whether these curves levelled off thereby indicating that the sample size is large enough t¢ give a true measure of the respective standard deviations for use in the power analysis. For most methods standar« deviations showed a reasonable degree of levelling off Another parameter required for the power analysis i Pearson’s r. This value indicates the strength o correlation between paired observations and is therefor basically an expression of the placement error that arise: through repeatedly laying the transect line. For VIDEO Pearson’s r was calculated from two repeated lin placements, and was found to range from 0.5 to 0.9 fo different substrates, being 0.7 for hard coral. For MAI and PHOTS Pearson’s 7 was taken from a study tha repeatedly assessed quadrats on the Ras Um Sidd reef fla and was 0.8 for abundant substrate categories (> 10% cover) (Leujak, 2006). Pearson’s r was assumed to bi similar for PHOTP and was high for all quadrat method: since it is relatively easy to relocate the same quadrats fo repeat sampling by taking underwater laminated copies o either the mapped quadrats or the quadrat photographs For LPT and LIT Pearson’s 7 was assumed to be only 0.5 since the placement error, due to relaying the transect lint along a slightly different route, can be relatively large. Th« “effect sizes” (ES) employed in power analysis wer calculated as desired detectable difference in cover value: Fig. 3. Percent cover of different hard coral growth forms as estimated by the six different sampling methods. Rare forms such as tabular Acropore spp., free-living corals and foliose corals are not shown. Error bars show 95% confidence intervals.
To assess the power and the minimum detectable difference (MDD) of the reef check method (a global coral reef monitoring initiative), LPT was analysed selecting four partly overlapping 20 m sections with 40 points per section. Since the reef-check method is not carried out along permanent transects, the power and Fig. 4. Multi-dimensional scaling plot for hard coral genera for log (x+ 1) transformed abundances as recorded by the six methods. Symbols represent cover data of the 10 replicates for each method. The stress value is a measure of the distortion between the distance of the rankings in the MDS configuration and the analogous rankings in the similarity matrices. The stress value of 0.16 falls within the lower range indicating that the plot represents a useful two-dimensional representation.
To assess the power and the minimum detectable difference (MDD) of the reef check method (a global coral reef monitoring initiative), LPT was analysed selecting four partly overlapping 20 m sections with 40 points per section. Since the reef-check method is not carried out along permanent transects, the power and Fig. 4. Multi-dimensional scaling plot for hard coral genera for log (x+ 1) transformed abundances as recorded by the six methods. Symbols represent cover data of the 10 replicates for each method. The stress value is a measure of the distortion between the distance of the rankings in the MDS configuration and the analogous rankings in the similarity matrices. The stress value of 0.16 falls within the lower range indicating that the plot represents a useful two-dimensional representation.
Results of Kruskal—Wallis tests and Kruskal-Wallis pair-wise comparisons for estimates of substrate cover by hard coral genera, obtained using the different methods *Indicates that the respective category was recorded by one method only.
Results of Kruskal—Wallis tests and Kruskal-Wallis pair-wise comparisons for estimates of substrate cover by hard coral genera, obtained using the different methods *Indicates that the respective category was recorded by one method only.
Number of hard coral genera recorded by each of the different methods along the whole transect (50 m), both for the maximum sampling efforts described in the methods section, and for typically used protocols involving lower sampling efforts % is the percentage of genera detected by each method on comparison with the number recorded by the full MAP method (sample size 45 quadrats as opposed to 10 quadrats only for MAP) method.
Number of hard coral genera recorded by each of the different methods along the whole transect (50 m), both for the maximum sampling efforts described in the methods section, and for typically used protocols involving lower sampling efforts % is the percentage of genera detected by each method on comparison with the number recorded by the full MAP method (sample size 45 quadrats as opposed to 10 quadrats only for MAP) method.
Determination of sampling effort necessary to detect a 20% relative change for abundant substrate types (> 10% cover) and a 50% relative change for rarer substrate types (< 10%) with a power of 80%
Determination of sampling effort necessary to detect a 20% relative change for abundant substrate types (> 10% cover) and a 50% relative change for rarer substrate types (< 10%) with a power of 80%
For each method the transformed (tr) mean percent cover and the actual change to be detected are given (as arcsine square-root transformed data, 20% change equals 12% on the transformed scale, 50% change equals 23%), together with the required sample size (expressed as number of quadrats or points or frames or length of transect in metres), both for =0.05 (N;) and for 7=0.10 (V2). Pearson's r was 0.8 for MAP, PHOTP, PHOTS and 0.7 for VIDEO and 0.5 for LPT and LIT. The required sample size was calculated using the following formula: Table 6 (continued )
For each method the transformed (tr) mean percent cover and the actual change to be detected are given (as arcsine square-root transformed data, 20% change equals 12% on the transformed scale, 50% change equals 23%), together with the required sample size (expressed as number of quadrats or points or frames or length of transect in metres), both for =0.05 (N;) and for 7=0.10 (V2). Pearson's r was 0.8 for MAP, PHOTP, PHOTS and 0.7 for VIDEO and 0.5 for LPT and LIT. The required sample size was calculated using the following formula: Table 6 (continued )
Time required and equipment costs for each of the six sampling methods both as employed, and as required for the sampling effort necessary to detect a 20% relative change in hard coral cover with a power of 80% and a=0.10 We assumed that two dives could be done per day, taking into account the time required to reach the dive site and pre-dive preparation.
Time required and equipment costs for each of the six sampling methods both as employed, and as required for the sampling effort necessary to detect a 20% relative change in hard coral cover with a power of 80% and a=0.10 We assumed that two dives could be done per day, taking into account the time required to reach the dive site and pre-dive preparation.
Lowest numbers imply that the methods performed best. Ranking of accuracies refers to hard coral cover in comparison with the best estimate obtained from MAP using a maximum sampling effort. Ranking of time-efficiency refers to the sample effort required to detect a 20% relative change in hard coral cover with a power of 80% (a=0.10). Ranking of the determination of genera diversity refers to maximum sampling sizes used in this study. Repeatability was defined as the possibility of returning to exactly the same sampling unit in future monitoring. Ranking of sampling methods using selected criteria
Lowest numbers imply that the methods performed best. Ranking of accuracies refers to hard coral cover in comparison with the best estimate obtained from MAP using a maximum sampling effort. Ranking of time-efficiency refers to the sample effort required to detect a 20% relative change in hard coral cover with a power of 80% (a=0.10). Ranking of the determination of genera diversity refers to maximum sampling sizes used in this study. Repeatability was defined as the possibility of returning to exactly the same sampling unit in future monitoring. Ranking of sampling methods using selected criteria
Related topics:
BiologyTotal Hardness of Waterimage resolution
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved