Little evidence that lowering the pH of concrete supports greater biodiversity on tropical and temperate seawalls

9)

However, compared to natural rocky shores, artificial structures tend to support lower species 12 diversity and/or abundances (e.g., Moschella  Masters of the tiles were created following Loke and Todd's (2016) protocol, using silicone 1 rubber moulds (Freeman Bluesil™ V-340). Tiles were then cast from the moulds using 2 cement/aggregate ratio = 1/3 and water/cement ratio = 3/5. Pre-drilled holes were set in the 3 centre of the concrete tiles for installation on seawalls. Trials were conducted using concrete coupons (5 cm × 5 cm × 2 cm) to determine the best 12 carbonation conditions (wet or dry), and the duration of curing (2, 6, 12, 20 days) and 13 carbonation (7, 22, 29 days) required to reduce the pH of the tiles. Concrete coupons were 14 split in half using a tile saw and the surface and cross section of the split tiles were stained 15 with two pH indicator dyes: (1) Phenolphthalein and (2)  transitions from yellow to light blue from pH 6 to 7, becoming dark blue for pH values above 21 8 (Guilbeau et al. 2003). When the stained carbonated tiles were colourless (phenolphthalein) 22 and light blue (bromothymol blue), it indicated that the external front-facing surface of the 23 carbonated tiles had a pH estimated to be between 7 and 8 ( Fig. 2A). 24 After several trials were conducted, it was found that the tiles were more rapidly carbonated 1 when dry as opposed to wet, and when they were left to cure for longer before being exposed 2 to CO2. Carbonation duration (>28 days), however, was the most important variable to 3 achieve a pH of less than 8 ( Fig. 2A). A sub-sample of the final batch of tiles were assessed 4 using the indicator dyes, which showed that the surface of the carbonated concrete tiles was 5 no more than than pH 8. 6 Attempts were also made to quantify the pH of the concrete tiles using a pH meter, but there 7 has been a longstanding lack of a standardised protocol for measuring the pH of pore fluid in 8 concrete (Alonso et al. 2012). Additionally, while the method is often used to test for internal 9 concrete pH, it does not give an accurate measurement of surface pH. Therefore, this method 10 was only used to confirm the differences in internal pH between treatments at the 6-month 11 time point (Fig. S1, Table S1). All tiles were prepared in Singapore before half were sent to 12 the UK. 13 At each site, 24 of each tile treatment (carbonated and non-carbonated) were installed along 10 seawalls at mid-shore height, approximately 1.5 m above chart datum, and spaced at least 0.5 11 m apart. Six replicates of carbonated and non-carbonated tiles were removed randomly at 3, 12 6, 9 and 12 months. However, due to unforeseen temporary restricted access to Pulau Hantu, 13 collection for the 9-month time point could not be carried out, hence we included a 15-month 14 time point instead for that site. 15

Study sites
Prior to removal of the tiles, fast-moving organisms were picked and placed into self-sealing 16 plastic bags. The tiles were then photographed (for subsequent algal cover analysis) before 17 being removed from the seawall and placed into larger self-sealing plastic bags.  Table 1). 22 After algal removal from the smooth surface, the tiles were placed into the freezer (-20 °C) 1 for subsequent sorting, counting and identification using a dissecting microscope. All 2 specimens were identified to species or morphospecies level except for polychaetes, which As tiles were lost due to wave action, there was an unequal number of replicates for some 6 sites and treatments (Table S2), but there were at least four replicates per treatment per site 7 per time point. Data were first examined for the presence of outliers, heterogeneity, non-8 normality and overdispersion (Zuur et al. 2010). We then tested for differences in total 9 abundance and species richness using generalised linear models (GLMs). Models with 10 Poisson error were first constructed separately for the two countries with treatment, site, and 11 month (categorical) as fixed effects, but models with negative binomial error were 12 subsequently used to analyse abundance due to over-dispersed data. 13 With differences in sample numbers between sites at some time points (described above) and 14 significant differences in abundance and species richness between months and sites, we 15 removed interaction terms (Table S3)  We used permutational distance-based multivariate analysis of variance (PERMANOVA; 1 Anderson 2001) to test for differences in community composition between treatments (we 2 removed 15 th month data as they were un-replicated in time; please see the Methods section 3 for more information). As both countries hosted no overlapping species, analyses were 4 conducted separately for temperate and tropical systems. The abundances were log(X+1)-5 transformed and the full resemblance matrix was calculated on Bray-Curtis similarities and p 6 values were generated using 9999 unrestricted random permutations of residuals. 7 PERMANOVA revealed significant differences in community composition among months, 8 but did not reveal significant differences among treatments; canonical analysis of principal 9 coordinates (CAP) plots were then used to examine these temporal differences. All 10 multivariate analyses were performed using the PRIMER v7 with the PERMANOVA add-on 11 (Anderson et al. 2008). 12

Abundance and species richness 14
A total of 78,114 individuals of 68 species/morphospecies were collected and identified from 15 experimental tiles across both countries. Of these, 13 were temperate species from Plymouth, 16 and 55 were tropical species from Singapore. Although there were more unique species found 17 on carbonated tiles than non-carbonated tiles at both sites in Plymouth, this was not observed 18 in Singapore (Table 2; further details in Table S5,  both of which were found on both treatments at both sites in their respective countries. 22 GLMs showed a significant effect of month on abundance and species richness in both 1 Plymouth and Singapore. There was also a significant effect of site on abundance and species 2 richness in Singapore (Table 3), with lower rates of colonisation at Pulau Hantu (Fig. 3). 3 There was, however, no significant effect of treatment in either country (Table 3, S7). 4 Site-and month-specific GLMs revealed that there were significant effects of carbonation at 5 some months and sites, but they were not ubiquitous and none occurred in the final 12-month, 6 time point (Table 4; further details in Table S8, S9). Carbonated tiles had greater total 7 abundance than non-carbonated tiles at Cremyll at the 9-month time point, and at Pulau 8 Hantu at the 6-month time point (Table 4). In Singapore, species richness was greater on 9 carbonated tiles than non-carbonated tiles at the 3-month time point at Pulau Seringat, and at 10 the 6-month time point at Pulau Hantu. There were no other significant effects of carbonation 11 detected from site-and month-specific GLMs. Singapore respectively; Table 5), but none between treatments regardless of country or month 18 (Table 5). Despite significant results for the interaction term (site × treatment × month) in 19 Singapore, no significant differences were detected when pair-wise comparisons were 20 conducted between treatments within sites and months. 21 In Plymouth, barnacle A. modestus, dominated the surfaces of all tiles (Fig. 4). Despite 22 having higher percentage cover on carbonated tiles than non-carbonated tiles at the 3-month 23 time point, there was no observed difference at the final 12-month time point. In Singapore, 1 biofilm which dominated at 3-month and 6-month time points was succeeded by barnacles 2 and encrusting algae by the 9-month time point (Fig. 4). However, mean barnacle cover fell 3 from 31% to 18% between 9-month and 12-month time points (Fig. 4). Although there 4 appears to be marginal differences between treatments at the 9-month time point, with higher 5 barnacle percentage cover than algae on non-carbonated tiles, this was not observed at the 6 final 12-month time point (Fig. 4). 7

DISCUSSION 8
Findings from our bilateral one-year study indicate that lowering the pH of concrete did not 9 significantly increase the abundance and species richness of intertidal benthic organisms on 10 retro-fitted enhancement tiles, and did not significantly alter the community composition they 11 support. Concrete is generally considered damaging to the environment, yet it remains one of Hantu at the 6-month time point was also mainly due to a single snail species, N. undata, a 16 microalgal feeder (Underwood 1984). Concrete carbonation, however, had little or no effect 17 at sites which had low algal growth generally, such as at Cremyll and Turnchapel in the UK 18 (Fig. 4). 19 Even though there might be some early differences in abundance and species richness 20 between tile treatments in Singapore, the effects of carbonation did not persist. Biofilm 21 formation can strongly influence the settlement of macrofouling taxa such as barnacles, 22 serpulids and mussels (reviewed by Almeida & Vasconcelos 2015), but the lack of significant 23 differences between treatments beyond six months suggests that, even if there were 24 differences in initial microalgal attachment, it was not enough to influence subsequent 1 successional species. Additionally, the surface pH of non-carbonated tiles in Singapore 2 appeared to have reduced to <8 by month 6 ( Figure S1). This is in line with findings by 3 Dooley et al. (1999) who suggested that the pH of concrete surface will approach seawater 4 pH after three to six months in marine environments. As such, colonisers may not experience 5 major differences in concrete pH between tiles of different treatments after a few months of 6 seawater exposure. 7 Substrate alkalinity is also unlikely to affect primary or secondary consumers during low tide, . In this study, many barnacles died in Singapore after initial colonisation, 2 which then served as microhabitats for smaller organisms such as the crab Nanosesarma 3 minitum, snails Zafra spp. and polynoids (Fig. 5). non-carbonated tiles at Cremyll at the 9-month sampling point (Table 3, Fig. 3). 20 To gain a more comprehensive understanding on the effects of concrete pH, future studies 21 can take regular measurements of the tile pH as well as the seawater pH in the water-22 retaining pits of the tiles. There is also a lack in standardised protocol for testing the pH of 23 other hard substrates such as granite, limestone and other naturally occurring rocks (Aho & 24 Weaver 2006), which would be useful for investigating the role of substrate pH in influencing 1 marine biodiversity. Nevertheless, this study provides some insight to the potential effects of 2 pH on marine benthic colonisation from an ecological engineering perspective. 3 As the demand for urban coastal development rises in response to the threats of sea level rise 4 and increasing coastal populations, it is important to consider engineering solutions that can 5 maximise the ecological functioning of artificial structures. However, the influence of 6 substrate pH on benthic colonisation is relatively understudied with little evidence to support 7 the hypothesis that lowering concrete pH can increase species richness or abundance of 8 organisms. Our experiment indicates that the effects of pH on benthic colonisation is non-9 significant and we suggest that manipulation of the physical structure of habitat enhancement 10 units, such as increasing topographical complexity and adding water-retaining features, is a 11 more effective eco-engineering approach to enhancing the ecological value and species 12 diversity on seawalls. 13

Acknowledgements 14
We thank members of the Experimental Marine Ecology Laboratory and Tropical Marine 15 Science Institute and Richard Ticehurst for their assistance in the field. We also thank Tan   25   iii) Pre-proof verion of Hsiung, A.R., Tan, W.T., Loke, L.H., Firth, L.B., Heery, E.C., Ducker, J., Clark, V., Pek, Y.S., Birch, W.R., Ang, A.C. and Hartanto, R.S., 2020. Little evidence that lowering the pH of concrete supports greater biodiversity on tropical and temperate seawalls. Marine Ecology Progress Series, 656, 193-205. 27 tiles that dried for the same amount of time (control) stained with phenolphthalein (left) and 1 bromothymol blue (right).