Antifouling Paint Test Report 2015
Presented to Jamestown Distributors, by
Dr. Scott Rutherford
Department of Environmental Sciences
Roger Williams University
Bristol, RI 02809
During the 2015 New England boating season, Jamestown Distributors partnered with faculty and students of Roger Williams University to conduct marine antifouling paint testing in southern New England waters. Antifouling paint, often referred to as "bottom paint" by boaters, typically comprises a mix of biocides designed to inhibit the settling and adherence of organisms on boat hulls and other material (e.g., mooring floats) in both marine and aquatic environments. In partnership with Jamestown Distributors, we designed and implemented a comparison test of antifouling paint performance in Narragansett Bay (Rhode Island) and Sippican Harbor (Marion, Massachusetts) during the 2015 boating season.
Test panels were deployed at four locations in Narragansett Bay (Figures 1 and 2): India Point Park in Providence to the north, the northern Sakonnet River (Tiverton) in the east, Warwick Cove to the west, and Wickford Harbor, adjacent to Mill Cove, to the southwest. A fifth raft was deployed at the head of Sippican Harbor in Marion, Massachusetts (Figure 3).
Figure 1. Raft locations in Wickford Harbor (left) near Pleasant Street Wharf, and in Warwick Cove (right) at the Brewer Street Warwick Cove North Marina.
The Narragansett Bay estuary stretches from Providence and the Providence River in the north, to Point Judith and Newport in the south. Salinity in the estuary can be quite variable, ranging from near-ocean salinities in the lower bay, to single digits in the rivers that feed into the head of the estuary, depending on tidal state and season. It is generally considered to be a partially-mixed to well-mixed estuary.
Figure 2. Raft locations in the Providence River at the Providence Community Boating Center (left) and in the Sakonnet River, Tiverton (right).
Figure 3. Raft location at Burr Brothers Boats, Marion, Massachusetts, at the head of Sippican Harbor.
The Providence raft was initially deployed on May 27, 2015, and secured to a floating dock at the Providence Community Boating Center at India Point Park. The location is at the head of the Providence River, adjacent to one of the larger freshwater sources to Narragansett Bay, the Seekonk River. Although the site is exposed to the south and to the summer southwesterlies that predominate in the area, circulation is restricted by the relatively long and narrow reaches of the Providence River. Salinity was measured by refractometer on multiple observation days, and ranged from 12ppt to 19ppt. Due to its proximity to Providence, this site likely has the poorest water quality (e.g., elevated nitrate concentrations and potential for hypoxia) of all five sites. In addition, we observed more floating debris here compared to the other sites.
The Tiverton raft was launched on May 29, 2015, and was secured to a mooring approximately 50 meters from shore in the northern end of the Sakonnet River (Figure 1). The Tiverton site is the most exposed in terms of wave action, particularly given the prevailing summer southwesterlies, and several of the panels were damaged or lost during the summer and fall. It was also the only location that was not in a circulation-restricted area. Salinity measured by refractometer ranged from a low of 24ppt in mid-June, to near 30ppt during the remainder of the season.
The Marion, Massachusetts, raft was secured to the end of a dock at Burr Brothers Boats, at the head of Sippican Harbor on May 29, 2015. Salinity at the site ranged from 29ppt to 32ppt. The site is bordered by intertidal mudflats and saltmarsh to the north and east and the marina to the west.
The Warwick raft was launched on May 28, 2015, and secured to a dock at the Brewers, Greenwich Bay North Marina in Warwick Cove. Warwick Cove is a narrow embayment lined with marinas along a dredged channel with a six-foot controlling depth. The raft was in the middle of the dock complex with vessels on three sides. Salinity at the Warwick location ranged from 26ppm to 28ppm during the season.
Anchored just west of the docks at Pleasant Street Wharf, the Wickford Harbor raft was launched on May 29, 2015. The site is in the inner harbor in approximately four feet of water at mean low water. The raft was anchored from two points and was not allowed to swing with the tidal current, which can be considerable, due to space constraints. Salinity at the site ranged from 25ppt to 30ppt.
Temperature was recorded hourly by HOBO Water Temperature Pro v2 loggers at four of the five sites. The temperature logger at the fifth site, Wickford, was not attached to the raft when the experiment was terminated. Unfortunately, the only location with reliable water temperature data for the entire period is Providence. At the other sites, it appears that the logger did not stay submerged for the duration of the experiment and was recording a combination of water and air temperature as it floated at the surface. That being the case, the daily low temperatures recorded should approximate the water temperature. In fact, all four sites where data were recorded show a similar pattern.
In addition to the Providence temperature data (which is very similar to NOAA data from Providence, not shown), the logger at Marion seems to have recorded reliable data over most of the period and is similar to that of the Fall River NOAA data (not shown). The data recorded at Tiverton is reliable up to July 20th, when the logger began recording large daily temperature variations, presumably because it was no longer submerged. Even with these problems, the general pattern of temperature variability is similar at all sites, and it would be difficult to attribute variations in the fouling community or intensity of fouling observed at the sites to temperature differences (Figure 4). This is further supported by the similarity in the NOAA data in Narragansett Bay and Fall River (http://tidesandcurrents.noaa.gov/ports.html).
Figure 4. Hourly temperature data for four of the five sites. The large diurnal temperature variations are interpreted to be related to the logger not staying submerged as intended. See text for discussion.
Rafts were constructed of untreated lumber frames with capped PVC pipe for flotation (Figure 5). The paint panels were made from 1/4-inch plywood that was secured to 1-inch by 2-inch furring strips. Each panel is 24 inches long by six inches wide and was designed to have ~18 inches submerged. Paint panels were numbered by etching to ensure identification throughout the experiment. Each raft holds 144 panels.
Figure 5. Schematic of raft design showing dimensions of the frame and spacing of the panels. Two of these units were bolted together at each site holding a total of 144 panels.
All wood parts were painted with TotalBoat TotalProtect epoxy barrier coat to minimize water intrusion. Paint panels were subsequently painted with two coats of the various test paints using a XXX roller to achieve a nominal thickness of XXX. We used only blue paints to eliminate the potential effect of paint color. For each test paint, six replicate panels were painted on both sides and all edges. Panels were randomly mounted on the raft with two panels back-to-back at each mounting location. Only the fully exposed side of each panel was examined for fouling.
With 144 panels per raft, the design allowed for the testing of 24 different paints. Two sets of panels were reserved for control paints. One control set was painted with a second coat of barrier coat. The second control set was painted with two coats of blue TotalBoat Wet Edge Topside Paint, a single-part polyurethane, to assess the effectiveness of a blue paint without antifouling properties. With the two control sets accounting for 12 panels, 22 antifouling paints could be tested on each raft.
We attempted to sample each raft approximately every two weeks as time and weather allowed through June, July, and into August (Table 1). After the August sampling, rafts were not sampled again until mid-October. This presents a "worst-case" scenario simulating the effect of a boat not being used for nearly two months.
At each sampling event, panels were removed from the raft, photographed, and an estimate made of the percent of the panel covered by biofilm, soft fouling, and hard fouling. Biofilm is what most boaters would refer to as "slime" and generally consists of bacteria and microalgae (e.g., diatoms) that settle on the surface of the panel but do not strongly adhere. Estimates of percent coverage were made using a string template of one-inch squares (Figure 6). The template provided a reference for estimating percent coverage by counting the number of squares covered.
After observation and photography, each panel was wiped gently with a saturated sponge to simulate the effect of the boat being used. This action serves two purposes:
- It removes a thin biofilm (the slime) from the panel that is not adhered as might happen when the boat is used.
- It exposes new biocide in ablative paints that rely on boat use to ablate the paint surface periodically.
No attempt was made to remove fouling using the sponge, and it was clear that the impact varied by panel. Panels were then placed in buckets of seawater until all panels were examined.
Figure 6. Template used to estimate percent coverage for each panel. Squares are ~1" and the template was designed to mask out the edges of the panel and the waterline. The underside of the frame has stops attached to ensure the template is placed in the same location on each panel.
After all the panels from the site were examined, they were placed randomly back on the raft. In combination with the six replicates used for each paint, this method served to minimize the effect of panel location on antifouling performance.
By using replicate panels, we could perform statistical analyses of paint effectiveness. The appropriate analysis in this case is an Analysis of Variance (ANOVA) with post-hoc follow-up tests. One assumption of ANOVA is normally-distributed data, although ANOVA is quite robust to violations of this assumption if sample sizes are large or equal. In our case, the sample size was typically six (though some panels were damaged or lost at the Providence and Tiverton sites), not large but reasonable, and, importantly, equal, resulting in an ANOVA robust to violations of the normality assumption.
Each paint-site combination was tested for normality using the Shapiro-Wilk Goodness-of-Fit test (α=0.05). The vast majority of the samples (albeit with a sample size of six) were consistent with being drawn from a normal distribution. Exceptions were the control samples, which all exhibited heavy fouling, and often the best performing paints and worst performing paints, when replicates clustered close to zero percent coverage or 100 percent coverage. Because these samples are at the ends of the range, effects of violation of the normality assumption on the Type I error rate are not particularly an issue. In the Connecting Letters Reports below, paint names in gray text indicate that the sample failed the normality test. For those paints at the top or bottom of the range, this is likely not an issue. For those paints in the middle of the range, their exact position should be viewed with caution.
If the ANOVA was significant (p<0.05) indicating at least one paint was different from the others, follow-up testing was done using the Tukey Honest-Significant-Difference (HSD) pairwise-comparisons reported using the conventional Connecting Letters Report. In addition to simple one-way ANOVAs, a two-way ANOVA was performed to investigate possible interaction effects between paint type and site.
Results are presented for two time frames—August and October.
- The August sampling covers the bulk of the spring/summer fouling, and also includes the bi-weekly wiping of the panels.
- The October sampling represents the end of the boating season in New England, and also provides a worst-case scenario where the panels have not been touched in over two months.
Results: August Sampling
Control panels at each site (Figure 7) at this stage (early August), show that Tiverton and Providence had the heaviest fouling dominated by macroalgae. The other three sites exhibited much less fouling overall, and generally, more colonial tunicates (orange in the photos).
Antifouling performance varied across both paint type and site (Figure 8). At the three lower fouling sites, Marion, Warwick, and Wickford, there are clear patterns of paint performance. Better performing paints include TB Test 3 and TB Test 4, TotalBoat Krypton and Interlux Micron CF. At the sites with more intense fouling (Providence and Tiverton), the pattern was generally similar, but there were some significant differences. For example, TB Test 3 and TotalBoat Krypton were in the middle or toward the bottom of the performance scale.
Of additional interest is the range of performance across the six replicate panels for each paint and site as illustrated by the error bars in Figure 8. Overall, Marion and Providence exhibited the greatest variability across all paints (standard deviations [s] of 0.29 and 0.28, respectively), whereas Tiverton and Warwick showed the least (s=0.22 for both).
Even within a particular paint, performance could be quite variable (e.g., Interlux Micron CF at Providence) or very consistent (Interlux Micron CF at Warwick and Wickford).
This suggests that paint performance can be sensitive to local variations in conditions including sunlight, currents, and proximity to other fouled substrates (e.g., docks), which may explain why different boaters using the same paint can observe differences in performance at the same marina.
In contrast, some paints (e.g., TotalBoat Argo) show a tight cluster of performance across the replicate panels.
Figure 7. Representative control panels from each site, with fouling organisms listed.
Figure 8. Summary of paint performance. Bar height indicates the mean biofilm coverage for the August sampling and the error bar indicates the standard deviation.
With these observations in mind, we proceeded to a two-way ANOVA. This analysis compares all paints across all sites and analyzes the effect of the different paints, the effect of the different sites, and paint-site interactions, which would indicate that different paints perform differently at different sites. At this stage, there was only soft coverage (e.g., tunicates, macroalgae) observed on the control panels and on three Interlux Micron CSC panels (Figure 9) at the Marion location and the Pettit Hydrocoat Eco panels at Tiverton. So we focused our analyses on biofilm coverage.
Figure 9. Panel 60, Interlux Micron CSC from Marion, showing soft growth of green algae. Three of the six Micron CSC panels at this location exhibited green algal growth at this time.
The two-way ANOVA results (Table 2) indicate that there are differences among paints, sites, and a paint-site interaction effect. Given the different fouling communities and fouling intensity observed on the control panels at the different locations (Figure 7), this is not a surprising result. The results demonstrate that location matters, and paints perform differently under different fouling conditions. This is illustrated by TotalBoat Krypton, which is one of the best performing paints overall at Marion, Warwick, and Wickford, the three sites with the least fouling, and one of the worst at Tiverton, which was the site with the heaviest fouling.
With a significant paint-site interaction effect, it is instructive to separate the analyses by site, performing an ANOVA on each of the five sites independently. This is probably the most useful approach for the boater whose vessel resides in one of our five test locations or in another location with similar fouling characteristics. At each site, the one-way ANOVA was highly significant (p<0.001). Results of follow-up tests (Tukey HSD) for each site are discussed below.
Providence - August Sampling Results
The Providence site was one of the heavier fouling sites (along with Tiverton) and also was unusual because of the low salinity. Results are summarized in Figure 10. The best performing paint is TotalBoat Argo (Table 3), though it is not statistically different from four other paints in Group F (Interlux Micron 66 and TB Tests 2, 5, and 6).
As discussed above, many paints, such as TotalBoat Argo and Interlux ACT, are very consistent in their performance, whereas others, such as Interlux Micron 66 and Interlux Micron CF have a wide range of performance from panel to panel (Figure 11). This suggests that some paints are more sensitive to variations in sunlight, current, or proximity to other fouled surfaces (the dock, in this case). The worst-performing paint was Pettit Hydrocoat Eco, but it is not statistically different from 16 other paints.
As of early August, there was no soft growth observed on the test panels, whereas the control panel (Figure 7) was nearly 100% covered with soft growth, mainly macroalgae and tunicates. Thus, all paints are effective and the differences amount to differences in biofilm coverage. Photos of all six replicate panels for the best and worst-performing paints are shown in Figure 12, and give a sense of the total range of paint performance at the site.
Figure 10. Summary of August results for Providence showing paint type versus percent coverage. Each dot represents an observation, and the diamond indicates the mean (central bar) and 95% confidence interval. The width of the diamond indicates the sample size. The horizontal grey line represents the mean of all observations for Providence. Here and elsewhere, Interlux ACT with Slimefighter (Boosted) is abbreviated "Slimefighter (Boosted)".
Figure 11. Photos of two Interlux Micron CF panels showing range of performance.
Figure 12. Photos of all six panels of the best-performing paint (TotalBoat Argo, left) and the worst-performing (Pettit Hydrocoat Eco, right) at Providence as of early August.
Tiverton - August Sampling Results
Overall, the Tiverton site had the second-most intense fouling. This site was the most exposed to wave action and currents, and a few of the panels were damaged or lost as a result of the exposure. In addition, because the raft was tied to a mooring, it had to be towed to shore for each sampling event (five total), which may have weakened the panels. TB Test 1 was the most affected (with the loss of three panels), and its performance here should be viewed with some caution. As with Providence, growth on the test panels was limited to biofilm, with only the control panels showing soft growth to 100% coverage, dominated by macroalgae and tunicates (Figure 7).
None of the paints tested were clearly superior (Figure 13) and there is considerable overlap, as illustrated by the Connecting Letters Report (Table 4). Arguably, four paints appearing in Group C, but not in Groups A and B in the Connected Letters Report performed better than the rest of the field.. Those paints are Interlux Micron 66, TotalBoat JD Select, TB Test 3 and TB Test 4. Strictly speaking, however, those paints are indistinguishable from the other paints in Group C. On the other end of the scale, Pettit Hydrocoat Eco is the only paint in Group A but not in groups B and C, and is the poorest performer at Tiverton. In addition, soft growth (macroalgae) was observed on some of the Pettit Hydrocoat Eco panels.
As is the case with Providence, some paints exhibited a tight cluster of performance over the replicates (e.g., Interlux Micron 66), while others showed considerable variation (e.g., Pettit Hydrocoat). Unlike Providence, this raft was tied to a mooring and allowed to swing with the tidal current. Thus, all panels would receive equal degrees of sunlight and shadow and current flow over time. The source of the intra-paint variation is unclear. It could be related to paint application, which could be a characteristic of the paint or the painter, or to location in the raft, but if the latter were the case, we would not expect such strong paint-to-paint differences in variation. In any case, a detailed study of possible effects should be considered.
Figure 13. Summary of August results for Tiverton showing paint type versus percent coverage.
Marion - August Sampling Results
Compared to Tiverton and Providence, Marion had much less fouling overall (Figure 7). The dominant fouling organisms here were macroalgae and tunicates. Like Providence, the raft was secured to a dock and different panels of the same paint would have experienced different conditions.
The results are summarized in Figure 14 and show a wide range of paint performance. Furthermore, as at Providence and Tiverton, some paints show a tight cluster of results (e.g., TotalBoat Krypton), while others show a much broader range across replicate panels (e.g., TB Test 6). As discussed above, Marion had the greatest overall performance variation across all paints. Six paints stand out as the best performers (Group G, Table 5). Those six are TotalBoat Krypton, TB Test 2, 3, 4, and 5, and Interlux Micron CF.
Here are photos of the best and worst paints at Marion.
These photos show that there is little practical difference between the paints at this stage except, perhaps, for the performance sailor. Although there are differences in biofilm coverage at Marion, the biofilm is a thin "slime" that is typically sloughed off easily. This is clearly not the case at the heavier fouling sites, Providence and Tiverton.
Figure 14. Summary of August results for Marion showing paint type versus percent coverage.
Warwick - August Sampling Results
In terms of overall average across all paints, the Warwick panels showed the least fouling of any site, and there is very little practical difference between the best and worst performing paints, at this stage. Choosing a dividing line between "best" paints and the rest of the group is difficult because there is considerable overlap in performance (Figure 15 and Table 6). Based on the Connecting Letters Report (Table 6), one could argue that Interlux Micron CF and TB Test 3 are the best (they are only in Group I), but they are not statistically different from eight other paints also in Group I.
There also continue to be paints with tightly clustered results (e.g., TB Test 3) and widely ranging results (e.g., Interlux ACT) across the replicate panels. Like the Marion and Providence locations, the Warwick raft was tied to a dock at the marina.
Figure 15. Summary of August results for Warwick showing paint type versus percent coverage.
Wickford - August Sampling Results
The average fouling at Wickford was most similar to that of Marion, more than Warwick, but less than Providence and Tiverton. There was, however, less tunicate fouling in Wickford than in Marion with more macroalgal coverage (Figure 7).
Two paints--TB Test 3 and TotalBoat Krypton--stand out as the best performers at Wickford (Figure 16 and Table 7). Although they are not statistically different from TB Test 4 and Interlux Micron CF (all four reside in Group G), TB Test 3 and TotalBoat Krypton are the only paints to reside exclusively in Group G. The remaining paints form a continuum, with overlapping groups identified by the Tukey HSD tests, as at the other sites.
Intra-paint variation at Wickford was the lowest of the five sites (average standard deviation of 0.10) with the next closest being Warwick (0.13). This indicates that the six panels for each paint are most similar to each other at Wickford. The Wickford raft was double-anchored away from any fixed structures. This means it was not in proximity to any fouled surface, and that it was not free to swing with the tidal change. In the current experiment, we have no means to determine the effect of these variables. In the future, it would be instructive to have multiple rafts in the same environment (e.g., Wickford Harbor) to investigate the impact of proximity to fouled surfaces and the movement of the raft with the tidal currents.
Figure 16. Summary of August results for Wickford showing paint type versus percent coverage.
Results: October Sampling
After the early August sampling, the paints were not disturbed until mid-October, essentially the end of the boating season in New England. This represents a worst-case scenario for the ablative paints, because it is as if the boat had not been used. It also allowed time for soft or hard fouling to develop on the panels. This was desirable because as of early August, there were very few test panels that exhibited more than a biofilm.
Between the August and October sampling dates, many panels were damaged or lost in Providence (25), Tiverton (24), and Marion (17). At Tiverton, five of the six Interlux ACT with Slimefighter (Boosted) panels were damaged or missing precluding any statistical analysis. In contrast, there were zero panels missing or damaged in Warwick and only one in Wickford. The reasons for the damaged and missing panels are unclear, but we suggest several issues are at play.
- First, the Providence and Tiverton locations are exposed to the summer southerly winds and probably suffered some damage from that exposure.
- Second, the barrier coat originally applied to the panels may have been poorly done on some panels, allowing water infiltration. If this were the case, however, we would expect that panels at all locations would have suffered similar rates of damage because there was no order to which site the panels were assigned (e.g., panels barrier-coated first would not all be in Providence or Tiverton). Furthermore, by the early August sampling, the panels had been in the water for over two months, yet very little damage was observed (except for Tiverton, which was so broadly exposed to wind-waves).
- Third, the relatively low salinities may have contributed to the panel destruction in Providence. In some of the Providence panels, the paint layer clung to the top of the panel that was emergent, but the submerged part of the wood panel was missing, as if it had dissolved from under the paint.
Unlike the previous sampling events, the October event discovered a remarkable difference between the back and the front of some of the panels, which were mounted in pairs, back-to-back. In many cases, the backs of the panels were heavily fouled with tunicates or poriferans (sponges), whereas the front (the side that was observed for data collection) had minimal fouling (Figure 17). Potential reasons for this include the following:
- The backs of the panels were painted with only one coat of antifouling, which may have not been sufficient.
- The backs of the panels may have rubbed together enough to remove some of the paint and/or biocide.
- The microhabitat between the panels improved the ability of organisms to attach to the panels.
If the third hypothesis is correct, it suggests that, when given no other choice, trapped between the two panels, the fouling organisms can attach to substrates painted with antifouling paint, though they would otherwise avoid the substrate.
Figure 17. Photos of the back (left) and front (right) of panel 10 at Warwick showing poriferan (sponge) and bryozoan growth on the back of the panel, whereas the front of the panel exhibits only biofilm growth, with the sponge wrapping around the panel edge, but attached.
The control sets exhibited extreme fouling at both Tiverton and Providence, particularly Tiverton, and much less overall fouling at the other sites (Figure 18). At this stage, there were marked differences between the fouling communities and intensity of fouling across the sites.
Unlike the August observations, by mid-October many of the test panels showed soft growth (e.g., macroalgae or tunicates). Thus, the observations distinguished between biofilm coverage, soft coverage, and hard coverage [e.g., bryozoans, cirripeds (barnacles), crepidula (slipper shells)]. In addition, each panel was assigned an overall rating. From the coverage estimates, we calculated a very simple coverage index where weights of one, two, or three were applied to the percent coverage of biofilm, soft, and hard growth, respectively. The results are summarized in Figures 19 and 20 along with results for an overall 1-5 rating for each panel, with one being the best (Figure 21).
We again began the analyses with a two-way ANOVA looking for paint, site, and paint-site interaction effects (Table 8). Again, all three effects were highly significant (p<0.0001), and we proceeded to individual ANOVAs with Tukey HSD follow-up tests to assess the paint performance at each site.
Figure 18. Representative control panels from each site in October. A list of the most abundant fouling organisms is shown next to each panel.
Figure 19. Summary of October results using total percent coverage by paint and location.
Figure 20. Summary of October results using our coverage index by paint and location.
Figure 21. Summary of October results using overall rating by paint and location.
In all three cases, total coverage, coverage index, and overall rating, there is a significant paint-site interaction (p<0.005) indicating that a site-by-site analysis would be the most instructive. Results using coverage data are presented here. The data for overall rating suffered from significant deviations from normality and were not analyzed, though comparison of Figures 19-21 suggest similar results, regardless of the measurement metric.
Providence - October Sampling Results
The October sampling at Providence revealed very little difference in paint performance when measured by total coverage (Figure 22 and Table 9) or our coverage index. In addition, with a few exceptions (e.g., TB Test 6 or TotalBoat Spartan), the percent coverage on replicate panels was quite similar and much reduced from the August sampling (average within-paint standard deviation of 0.15 in August and 0.10 in October).
With regard to total coverage, only TB Test 6 stood out as performing better than the others (it is the only paint not in Group A in the Connecting Letters Report). Using our coverage index, TB Test 6 is again the best performer (it alone occupies only Group D, though there are many other paints in Group D), but the weighting of soft and hard coverage separates some of the poor performers; Interlux ACT with Slimefighter (Boosted) (mainly as a result by a single, poorly performing panel), TB Test 3, TotalBoat Krypton, and Pettit Hydrocoat Eco.
Figure 22. Summary of results for the October sampling at Providence.
Tiverton - October Sampling Results
On the control panels, fouling at Tiverton was extremely heavy (Figure 18), and consisted of colonial and solitary tunicates, sponges, barnacles, and macroalgae. The fouling was multilayered, with sponges covering tunicates, for example.
Using total coverage, TB Test 1 is the best performing paint (Figure 23, Table 10). The Tukey HSD test results in TB Test 1 as the only paint to reside only in Group D, although Group D does encompass eight other paints. That said, half the replicated panels for TB Test 1 were damaged or missing, so this conclusion requires considerable caution.
Using the coverage index, where soft and hard fouling receive more weight than biofilm coverage, all paints are statistically the same with the exception of the poorly performing Interlux Micron CSC.
Figure 23. Summary of results for the October sampling at Tiverton.
Marion - October Sampling Results
As mentioned above, there was considerable panel loss at Marion for reasons that are not clear. The raft was tied to a floating dock in a sheltered location. It is unclear why so many panels were lost here (17) whereas so few were lost in Warwick (0) and Wickford (1), when all three locations appear similar with regard to exposure to wind, current, and salinity. Sea Hawk Cukote was the most impacted, with the loss of four panels. It is included in the statistical results but its performance should be viewed with extreme caution. Three panels of Interlux Micron 66 were lost or damaged, and caution interpreting the results is again warranted.
Interlux Micron 66 is arguably the best paint at Marion using total coverage because it is a member of only Group G. It does, however share Group G with five other paints, Interlux ACT with Slimefighter (Boosted), TB Test 4, TotalBoat Argo, Interlux Micron CF, and TB Test 6, indicating that the six are not statistically different.
Using our coverage index, Interlux Micron 66 and TB Test 6 both occupy only Group F, though they share that group with 11 other paints, indicating that those 12 are not statistically different from one another.
Figure 24. Summary of results for the October sampling at Marion.
Warwick - October Sampling Results
The total coverage results from Warwick (Figure 25) are similar to those of Marion in terms of variation. Both sites have an intra-paint average standard deviation of 0.13 and a similar range of mean performance (~75 percentage points; Table 12). These values contrast with those of Tiverton and Providence, which have both smaller intra-paint variation (average standard deviations of 0.12 and 0.10, respectively), and a smaller mean range (~40 percentage points, excepting one paint that would raise the range to 64 percentage points at Tiverton).
Based on total coverage, there are seven paints that are statistically indistinguishable as the best performers (Group H and Table 12). Of those seven, however, only one, TB Test 6, does not reside in another group. Thus it could be argued that TB Test 6 is the best paint at Warwick. It should be noted, however, that fouling was quite light across all panels, and, from best to worst, is a matter of degrees of biofilm fouling with some light macroalgal coverage. This is also illustrated by the very similar results using our coverage index. If all fouling were only biofilm, the total coverage and coverage index results would be identical.
A comparison between the two sets of results indicates that our weighted index increases the penalty for soft fouling over biofilm fouling as the poorly performing paints are rearranged in the Connecting Letters Report, whereas the better paints hold their positions.
Figure 25. Summary of results for the October sampling at Warwick.
Wickford - October Sampling Results
Wickford had the least fouling of all sites and no soft or hard growth on any of the test panels. Thus our total coverage results and our coverage index results are identical. Curiously, intra-paint variation at Wickford was least of all sites for the August sampling, but greatest of all sites for the October sampling (Figure 26). This, along with the fact that all coverage was biofilm, makes it very difficult to choose a best paint. In fact, 13 paints share Groups E and F in the Connecting Letters Report (Table 13). Only one paint, TotalBoat Argo, resides only in Group F, but the difference is so small (2 percentage points) and the standard deviation so large (Figure 26), that there is no support for that conclusion.
Figure 26. Summary of results for the October sampling at Wickford.
For boaters, we recommend focusing on the results for the site that most closely matches their environment. Providence represents the head of an estuary proximal to a major freshwater source with eutrophic/urban conditions. For those boaters whose vessels reside on a mooring in an open estuary, Tiverton would be the most representative site. Warwick and Wickford represent conditions in a marina that resides within a circulation-restricted harbor. Marion would be indicative of the head of an estuary surrounded by saltmarsh and mudflats, and with very little freshwater input.
There is no one best paint, but generally there are better performers across the range of fouling observed. In some locations, choice of paint is not much of an issue as all paints show similar performance (e.g., Providence and Wickford). At other locations, the type of paint is more important.
Within-paint variation on replicate panels varies from paint to paint and site to site, and can make statistical differentiation difficult because so many paints are so similar in their performance. It is very difficult to distinguish differences in mean coverage of two percentage points when the panel-to-panel coverage might vary by five percentage points or more.
To our knowledge, this is the only antifouling paint test done using a quantitative assessment of paint performance and replication to permit statistical analysis. We suspect it is also the only test done with a procedure repeated every two weeks to simulate boat movement through the water.
Finally, there are many variables that were not controlled for in this study. For example, proximity to docks, currents, sunlight, and the location of panels in the raft were not addressed. Additional information may be gained from studies to investigate the impact of these variables.