Journal of Applied Ecology 2016, 53, 16–24 doi: 10.1111/1365-2664.

12533

The overlooked role of biotic factors in controlling the

ecological performance of artificial marine habitats
Filippo Ferrario1,2, Ljiljana Iveša3, Andrej Jaklin3, Shimrit Perkol-Finkel1,4 and
Laura Airoldi1*
1
Department BiGeA, Alma Mater Studiorum – University of Bologna, Ravenna, Italy; 2Quebec-Ocean, Universite
Laval, Quebec, QC, Canada; 3Ruder
 Boskovic Institute, Center for Marine Research, G. Paliaga 5, 52210 Rovinj,
4
Croatia; and ECOncrete Tech LTD, 4 Yehoshua’ Bin Nun Street, 62643 Tel Aviv, Israel

Summary
1. Artificial structures are sprawling in marine seascapes as a result of burgeoning coastal
populations, increasing development and energy demand, and greater risks from climate
change, storm surges and sea level rise. Interest in designing marine developments that main-
tain vital ecosystems and critical services is growing, but progress requires understanding the
factors that influence the ecological performance of these novel artificial habitats.
2. We combined field observations and experiments along 500 km of the North Adriatic
coastline to analyse the performance of artificial substrata as habitats to support canopy-
forming algae belonging to the genus Cystoseira, among the most ecologically relevant foun-
dation species along rocky Mediterranean coastlines. We aimed to: clarify the underlying fac-
tors controlling the growth of Cystoseira in the artificial habitat; contrast the relative
importance of these factors between artificial and natural habitats; and test the generality of
the results across different sites and species of Cystoseira.
3. We found that: (i) the growth of canopy algae was significantly lower on artificial struc-
tures compared to rocky reefs; (ii) such lower growth of canopy algae was not related to less
favourable abiotic conditions but to higher biotic disturbance from both consumptive and
nonconsumptive interactions on the artificial structures compared to the natural reef; and iii)
this was consistent across different study sites and canopy-forming species.
4. We conclude that biological factors influencing the growth of canopy algae, such as her-
bivory or other nonconsumptive disturbances, can differ substantially between artificial and
natural habitats. The unusually large and previously unreported biotic pressure characterizing
many artificial structures can negatively affect their performance as habitats to support eco-
logically relevant, foundation species.
5. Synthesis and applications. While nearly all considerations to improve the ecological per-
formance of hard marine infrastructures focus on abiotic factors (e.g. construction materials,
surface texture, habitat complexity or water quality), careful consideration of critical biotic
factors is also needed to further progress the green engineering of sprawling marine infras-
tructures.
Key-words: artificial habitats, biotic disturbance, Cystoseira, fucoid algae, grazing, Mediter-
ranean sea, plant–herbivore interactions, rocky reefs, urban sprawl

marine resources and provide alternative energy sources

Introduction
(Dugan et al. 2011). The projected escalation in marine
Artificial structures (comprising seawalls, breakwaters, urbanization and the growing recognition of severe eco-
dykes, groynes, jetties, pilings, bridges, artificial reefs and logical impacts and habitat loss associated with marine
offshore installations) are sprawling globally in marine urban sprawl (Airoldi et al. 2005a; Airoldi & Beck 2007;
seascapes to protect coastal population and assets, exploit Bulleri & Chapman 2010) have raised the need for mitiga-
tion (Airoldi & Bulleri 2011; Browne & Chapman 2011;
*Correspondence author. E-mail: [email protected] Perkol-Finkel et al. 2012; Firth et al. 2014). Interest in

© 2015 The Authors. Journal of Applied Ecology © 2015 British Ecological Society
 Ecology of marine infrastructures 17

designing marine developments that maintain vital ecosys- (Maggi et al. 2009). Due to their ecological importance
tems and critical services is growing (Adams et al. 2014; and sensitivity to anthropogenic stressors (Perkol-Finkel
Dafforn et al. 2015), but progress towards green engineer- & Airoldi 2010; Sales et al. 2011), they are a priority con-
ing requires understanding the factors influencing the eco- servation habitat (Gianni et al. 2013).
logical performance of these novel artificial habitats We analysed the distribution of canopies of Cystoseira
(Chapman & Underwood 2011). spp. (hereafter Cystoseira) at a variety of artificial struc-
One of the most critical ecological functions to preserve tures and natural reefs in the study region, documenting a
in increasingly urbanized seascapes is the provision of consistent scarcity of Cystoseira in artificial habitats com-
habitat to support native biodiversity (Perkol-Finkel et al. pared to natural reefs (see Results). Because artificial
2012; Dafforn et al. 2015). Studies consistently show that structures were sometimes very close to the natural reefs
artificial structures provide habitats that perform poorly (see Study area), therefore allowing for dispersal, and
compared to natural reefs (Perkol-Finkel, Shashar & because previous experiments had shown that Cystoseira
Benayahu 2006; Burt et al. 2009; Miller et al. 2009), and has the potential to successfully settle and grow on artifi-
tend to support assemblages with low species and genetic cial substrata even in the absence of an adult canopy (Per-
diversity (Fauvelot et al. 2009; Airoldi et al. 2015) and kol-Finkel et al. 2012), we hypothesized that the scarcity
dominance by opportunistic and invasive species (Bulleri of Cystoseira in the artificial habitats would be mainly
et al. 2006; Glasby et al. 2007; Airoldi et al. 2015). Eco- related to postsettlement factors, including either adverse
logical processes underlying the different performance of local environmental conditions (e.g. low water quality) or
artificial habitats compared to natural reefs are still not excess biotic pressure (e.g. grazing or other nonconsump-
fully understood (Firth et al. 2014). Current knowledge tive disturbance from handling and clipping).
suggests that divergence of benthic assemblages is not At the natural sites in the study region, Cystoseira is
simply attributable to incomplete succession, but appears mainly limited by poor water quality, high sedimentation
to be a persistent, possibly stable state (Perkol-Finkel, rates or degraded habitat characteristics (Irving et al.
Shashar & Benayahu 2006; Miller et al. 2009). Several co- 2009; Perkol-Finkel & Airoldi 2010; Strain et al. 2014),
occurring factors could contribute to maintaining these while grazing is not a particularly relevant factor. How-
differences, including effects of construction materials ever, previous evidence (Perkol-Finkel et al. 2012) lead us
(Burt et al. 2009), unique habitat characteristics (Vaselli, to hypothesize that at the artificial sites the biotic pressure
Bulleri & Benedetti-Cecchi 2008; Browne & Chapman could be more important than at the natural sites, exert-
2011), unnaturally high anthropogenic disturbances (Air- ing proportionally greater control than abiotic factors.
oldi & Bulleri 2011), fragmentation and isolation (Airoldi We tested this hypothesis by using caging experiments
et al. 2015). It has also been demonstrated that, whether which allowed to: (i) test the relative importance of abi-
or not specifically planned for the scope, artificial struc- otic vs. biotic factors in influencing the growth of Cysto-
tures affect the distribution of fish assemblages by provid- seira at the artificial sites; (ii) identify whether the biotic
ing enhanced nursery grounds, refugia and feeding areas pressure at the artificial sites was mainly related to meso-
(Brickhill, Lee & Connolly 2005; Clynick, Chapman & grazers or macrograzers; (iii) contrast the relative impor-
Underwood 2007). Therefore, assemblages on artificial tance of biotic factors in influencing the growth of
structures could be exposed to altered grazing pressures Cystoseira in artificial and natural habitats; and (iv) test
compared to assemblages on natural reefs. Whilst consid- the generality of the results across different sites and Cys-
erable attention has been dedicated to the effects of physi- toseira species.
cal properties of artificial habitats on the colonizing biota
(Burt et al. 2009; Firth et al. 2014; Perkol-Finkel & Sella
Materials and methods
2014), the role of biotic factors has received surprisingly
little consideration (Bulleri et al. 2006; Ivesa et al. 2010;
STUDY AREA
Marzinelli, Underwood & Coleman 2011; Dafforn,
Glasby & Johnston 2012). The research was carried out at several sites along 500 km of the
We examined the ecological performance of artificial north Adriatic Sea (Fig. 1, Table S1 in Supporting Information).
substrata as habitats for native biota along the North The Italian side of the Adriatic is a prevailingly sedimentary
Adriatic coastline and experimentally identified the eco- coastal system and is heavily urbanized. The natural site along
the Italian side was located at La Vela (Monte Conero) and was
logical factors influencing the different performance of
comprised of maerl and limestone rocks extending to a depth of
artificial habitats compared to natural reefs. We worked
c. 8 m. The artificial sites along the Italian side, Marotta, Cesen-
with canopy-forming macroalgae, the dominant founda-
atico and Punta Marina were located about 40, 120 and 160 km
tion species on many temperate rocky coasts. In the north of La Vela, respectively. These sites were characterized by
Mediterranean Sea, canopies are mostly comprised of breakwaters 100–150 m long, deployed 200–300 m from the
fucales of the genus Cystoseira. They form ecosystems shore, extending 2–3 m in depth and built with large blocks of
functionally analogous to kelps that play a key role in pri- quarried rock (1–3 m across). Descriptions of both artificial and
mary production and nutrient cycling, modify both physi- natural habitats can be found in Perkol-Finkel & Airoldi (2010)
cal and biological factors and sustain rich communities and Perkol-Finkel et al. (2012).

© 2015 The Authors. Journal of Applied Ecology © 2015 British Ecological Society, Journal of Applied Ecology, 53, 16–24
18 F. Ferrario et al.

Fig. 1. Location of the study region (Adriatic Sea) and insert maps showing the position of the study sites (○,natural reef; ●, artificial
reef) along the Italian (upper left) and Croatian coasts (bottom).

The Croatian side of the Adriatic is a prevailingly rocky lime- site, we quantified the percentage cover of Cystoseira in five repli-
stone coast and compared to the Italian side is far less urbanized. cated quadrats. We used a grid of 25 squares in a 25 9 25 cm
Experiments along the Croatian side were run at three natural frame; in each square, the coverage of Cystoseira was scored
rocky sites and three artificial sites (Fig. 1, Table S1, Fig. S1). from 0 to 4, summing up to 100%.
The natural sites (Faborso, Kamenolom and Stari Grad) were
characterized by gently sloping, rocky bottoms densely forested
FACTORS LIMITING THE GROWTH OF CYSTOSEIRA IN
by Cystoseira (Ivesa, Lyons & Devescovi 2009). The three artifi-
cial sites (Valalta, Bolnica and Marina) were interspersed with ARTIFICIAL HABITATS
natural sites, and generally located very close to them. All artifi-
We tested the hypothesis that the observed limited growth of
cial structures were built with boulders of the same type of
Cystoseira at the artificial sites was related to high biotic pressure
quarried rock (Fig. S1).
(e.g. from grazers or other nonconsumptive disturbance) rather
than unfavourable environmental factors by using caging experi-
DISTRIBUTION OF CYSTOSEIRA AT ARTIFICIAL AND
ments. In these first sets of experiments, we used cages of differ-
ent mesh size to explore which type of grazers could affect
NATURAL HABITATS
Cystoseira at the artificial sites. Since Cystoseira is scarce at the
The distribution of species of Cystoseira at some artificial and artificial sites, we used tiles holding 5–6 juveniles of C. barbata
natural sites along the Adriatic Italian coast had been quantified from La Vela (Appendix S1; Fig. S2).
in previous studies (Perkol-Finkel & Airoldi 2010; Perkol-Finkel The first experiment used cages of 10-mm mesh (hereafter MA)
et al. 2012). These observations indicated that Cystoseira rarely to exclude potential macrograzers and 1-mm mesh (hereafter ME)
occurred on artificial substrata in this area. To analyse the gener- to also exclude potential mesograzers. Cages (12 9 12 9 25 cm)
ality/consistency of this pattern, in October 2010, we quantified were built using a plastic-coated iron wire mesh with a 10-mm2
the distribution of Cystoseira at the three replicated artificial and grid. ME cages were obtained by further wrapping with a mos-
natural sites along the Croatian coastline (Fig. 1, Table S1, quito net. In July 2010, at one of the breakwaters at Marotta, we
Fig. S1), where artificial structures were well interspersed with randomly assigned five replicate experimental tiles to each of
rocky reefs. As typical for that time of the year, Cystoseira had three treatments: ME, MA and uncaged (Fig. S3). Both tiles and
started losing branches, but thalli were still fully visible. At each cages were attached to the rock using epoxy putty. We did not

© 2015 The Authors. Journal of Applied Ecology © 2015 British Ecological Society, Journal of Applied Ecology, 53, 16–24
 Ecology of marine infrastructures 19

include a partial-cage control of potential artefacts associated growth of canopy-forming algae in artificial and natural habitats
with caging because at this initial stage, the potential grazers on and test the generality of results across different sites and species
Cystoseira were uncertain making it difficult to design a partial of Cystoseira. The experiment was run in Croatia, where repli-
control treatment that we could ensure was effective. To minimize cated artificial and natural sites were properly interspersed.
any potential artefacts associated with the potential shelter Here, canopy-forming algae were virtually absent at artificial
offered by cages on hydrodynamics, the experiment was run dur- sites (see ‘Results’), but C. compressa had been occasionally
ing 2 weeks of very calm sea conditions and was set up at the reported (L. Ivesa pers. obs.). We hypothesized that responses
sheltered, landward side of the breakwater. C. barbata grows well might vary among different species of Cystoseira, with C. com-
under sheltered conditions (Perkol-Finkel et al. 2012), and we felt pressa showing a greater growth potential in artificial habitats
that the combination of shelter and calm sea was sufficient to compared to other species.
minimize any relevant effects of cages on hydrodynamics. We used 96 experimental tiles (Appendix S1) colonized by
The experiment lasted 13 days, as responses to treatments were juveniles of either C. barbata or C. compressa, for a total of 48
very rapid. We measured the average height of all juveniles at the independent tiles for each species. We fixed eight tiles for each
beginning and at the end of the experiment and calculated the species at each of three replicated natural rocky sites and three
average final height, expressed as the percentage of the average artificial sites. Four random tiles for each species were caged
initial height, as the response variable. Data were analysed by using the same 10-mm mesh size cage described previously
one-way permutational ANOVA with factor Exclusion (fixed, 3 (MA) while the remaining four were left uncaged. We used
levels: MA, ME and uncaged). Three plots were lost during the MA cages, as by the start of this experiment, we knew that
experiment, and the final number of replicates varied from 3 to 5 Cystoseira was affected by a pool of macrograzers of different
per treatment. In this and the following analyses, we used the sta- habits and sizes, including fishes, crabs and hermit crabs. Par-
tistical package PERMANOVA+ for PRIMER (Anderson, Gorley & tial control of the caging treatment would have required a cage
Clarke 2008) to partition the variability and obtain F-statistics on with so many openings (to allow access to the variety of spe-
matrices of Euclidean distances calculated from the original data cies involved) to become virtually nonexistent, thus it was not
and calculated P-values using 9999 random permutations. We included. As an alternative, we deemed it more effective to
used PERMANOVA (as opposed to a classic ANOVA test) due to ease quantify differences in biotic interactions with Cystoseira at
of use with unbalanced design and to avoid the usual normality both artificial and natural sites using videos (F. Ferrario,
assumptions. Data had homogeneous variances [Levene’ (1960) E.M.A. Strain, P. Guidetti and L. Airoldi, unpublished data).
univariate test run using PERMDISP (Anderson, Gorley & Clarke We regularly cleaned the cages and removed fouling organisms
2008)], and there was no need for transformation. to minimize any alteration to light or hydrodynamics.
In August 2010, we ran a second caging experiment to discrim- The plots were sampled at the start of the study, in December
inate between macrograzers of different sizes (the previous experi- 2010, April 2011 and October 2011 by using digital photographs.
ment had shown no relevant role of mesograzers) and to test for Pictures were analysed in the laboratory by superimposing a digi-
the generality of results at replicated artificial sites. The MA tal grid (100 squares) and estimating the percentage cover of
treatment was modified to prevent large macrograzers but allow either species of Cystoseira as the number of quadrats occupied.
access by small macrograzers (hereafter LMA: 10-mm mesh as Due to significant differences in the percentage cover of C. bar-
before, with 6 9 7 cm openings cut on each side and on the top bata and C. compressa on the tiles at the start of the experiment,
of the cage). The ME treatment was unaltered, therefore prevent- we analysed the data from the two species separately. For each
ing all grazers. At each of three replicated artificial sites (Mar- species, we ran a three-way permutational ANOVA including fac-
otta, Cesenatico and Punta Marina), we assigned five tors Habitat (Fixed, two levels: Artificial vs. Natural), Exclusion
experimental tiles to each of three treatments (ME, LMA and (Fixed, two levels: MA vs. uncaged) and Site (Random, three
uncaged) attaching them at the landward side of one breakwater levels, nested in Habitat). Between April 2011 and October 2011,
as previously described. After 15 days, we assessed the status of the last two sampling dates, there was a large discontinuity in the
juveniles. Because the losses in uncaged treatments were even data. To highlight this variation, we analysed percentage cover
more severe than in the previous experiment, we measured data from both dates. Variances were heterogeneous except for
responses as percentage survival rather than length of juveniles. C. compressa data in April 2011. Whenever variances were
Data were analysed by two-way, permutational ANOVA with fac- heterogeneous, we used a conservative test, setting the signifi-
tors Exclusion (fixed, three levels: M = all grazers excluded; cance level at P = 0
01.
LMA = only small macrograzers allowed; and uncaged = access
allowed to all macrograzers) and Site (random, three levels: Mar-
otta, Cesenatico, Punta Marina). Data had homogeneous vari- Results
ances, and there was no need for transformation. The
nonsignificant interaction term Exclusion 9 Site was pooled with
DISTRIBUTION OF CYSTOSEIRA AT ARTIFICIAL AND
residuals following Underwood (1997) and Anderson, Gorley &
NATURAL HABITATS
Clarke (2008).
Along the Croatian coast, stands of C. barbata, C. com-
pressa and C. crinita were present at natural reefs at
COMPARISON OF BIOTIC PRESSURE BETWEEN
Faborso (total cover = 24  2
1%, mean  SE), Stari
ARTIFICIAL AND NATURAL HABITATS AND DIFFERENT
Grad (total cover 73  4
6%) and Kamenolom (total
SPECIES OF CYSTOSEIRA
cover 27
4  5
1%). No Cystoseira was detected at the
In October 2010, we started a larger caging experiment to con- three artificial sites, even if in the past it had been occa-
trast the relative importance of biotic factors in controlling the sionally detected (Ivesa pers obs).

© 2015 The Authors. Journal of Applied Ecology © 2015 British Ecological Society, Journal of Applied Ecology, 53, 16–24
 Ecology of marine infrastructures 19

include a partial-cage control of potential artefacts associated growth of canopy-forming algae in artificial and natural habitats
with caging because at this initial stage, the potential grazers on and test the generality of results across different sites and species
Cystoseira were uncertain making it difficult to design a partial of Cystoseira. The experiment was run in Croatia, where repli-
control treatment that we could ensure was effective. To minimize cated artificial and natural sites were properly interspersed.
any potential artefacts associated with the potential shelter Here, canopy-forming algae were virtually absent at artificial
offered by cages on hydrodynamics, the experiment was run dur- sites (see ‘Results’), but C. compressa had been occasionally
ing 2 weeks of very calm sea conditions and was set up at the reported (L. Ivesa pers. obs.). We hypothesized that responses
sheltered, landward side of the breakwater. C. barbata grows well might vary among different species of Cystoseira, with C. com-
under sheltered conditions (Perkol-Finkel et al. 2012), and we felt pressa showing a greater growth potential in artificial habitats
that the combination of shelter and calm sea was sufficient to compared to other species.
minimize any relevant effects of cages on hydrodynamics. We used 96 experimental tiles (Appendix S1) colonized by
The experiment lasted 13 days, as responses to treatments were juveniles of either C. barbata or C. compressa, for a total of 48
very rapid. We measured the average height of all juveniles at the independent tiles for each species. We fixed eight tiles for each
beginning and at the end of the experiment and calculated the species at each of three replicated natural rocky sites and three
average final height, expressed as the percentage of the average artificial sites. Four random tiles for each species were caged
initial height, as the response variable. Data were analysed by using the same 10-mm mesh size cage described previously
one-way permutational ANOVA with factor Exclusion (fixed, 3 (MA) while the remaining four were left uncaged. We used
levels: MA, ME and uncaged). Three plots were lost during the MA cages, as by the start of this experiment, we knew that
experiment, and the final number of replicates varied from 3 to 5 Cystoseira was affected by a pool of macrograzers of different
per treatment. In this and the following analyses, we used the sta- habits and sizes, including fishes, crabs and hermit crabs. Par-
tistical package PERMANOVA+ for PRIMER (Anderson, Gorley & tial control of the caging treatment would have required a cage
Clarke 2008) to partition the variability and obtain F-statistics on with so many openings (to allow access to the variety of spe-
matrices of Euclidean distances calculated from the original data cies involved) to become virtually nonexistent, thus it was not
and calculated P-values using 9999 random permutations. We included. As an alternative, we deemed it more effective to
used PERMANOVA (as opposed to a classic ANOVA test) due to ease quantify differences in biotic interactions with Cystoseira at
of use with unbalanced design and to avoid the usual normality both artificial and natural sites using videos (F. Ferrario,
assumptions. Data had homogeneous variances [Levene’ (1960) E.M.A. Strain, P. Guidetti and L. Airoldi, unpublished data).
univariate test run using PERMDISP (Anderson, Gorley & Clarke We regularly cleaned the cages and removed fouling organisms
2008)], and there was no need for transformation. to minimize any alteration to light or hydrodynamics.
In August 2010, we ran a second caging experiment to discrim- The plots were sampled at the start of the study, in December
inate between macrograzers of different sizes (the previous experi- 2010, April 2011 and October 2011 by using digital photographs.
ment had shown no relevant role of mesograzers) and to test for Pictures were analysed in the laboratory by superimposing a digi-
the generality of results at replicated artificial sites. The MA tal grid (100 squares) and estimating the percentage cover of
treatment was modified to prevent large macrograzers but allow either species of Cystoseira as the number of quadrats occupied.
access by small macrograzers (hereafter LMA: 10-mm mesh as Due to significant differences in the percentage cover of C. bar-
before, with 6 9 7 cm openings cut on each side and on the top bata and C. compressa on the tiles at the start of the experiment,
of the cage). The ME treatment was unaltered, therefore prevent- we analysed the data from the two species separately. For each
ing all grazers. At each of three replicated artificial sites (Mar- species, we ran a three-way permutational ANOVA including fac-
otta, Cesenatico and Punta Marina), we assigned five tors Habitat (Fixed, two levels: Artificial vs. Natural), Exclusion
experimental tiles to each of three treatments (ME, LMA and (Fixed, two levels: MA vs. uncaged) and Site (Random, three
uncaged) attaching them at the landward side of one breakwater levels, nested in Habitat). Between April 2011 and October 2011,
as previously described. After 15 days, we assessed the status of the last two sampling dates, there was a large discontinuity in the
juveniles. Because the losses in uncaged treatments were even data. To highlight this variation, we analysed percentage cover
more severe than in the previous experiment, we measured data from both dates. Variances were heterogeneous except for
responses as percentage survival rather than length of juveniles. C. compressa data in April 2011. Whenever variances were
Data were analysed by two-way, permutational ANOVA with fac- heterogeneous, we used a conservative test, setting the signifi-
tors Exclusion (fixed, three levels: M = all grazers excluded; cance level at P = 0
01.
LMA = only small macrograzers allowed; and uncaged = access
allowed to all macrograzers) and Site (random, three levels: Mar-
otta, Cesenatico, Punta Marina). Data had homogeneous vari- Results
ances, and there was no need for transformation. The
nonsignificant interaction term Exclusion 9 Site was pooled with
DISTRIBUTION OF CYSTOSEIRA AT ARTIFICIAL AND
residuals following Underwood (1997) and Anderson, Gorley &
NATURAL HABITATS
Clarke (2008).
Along the Croatian coast, stands of C. barbata, C. com-
pressa and C. crinita were present at natural reefs at
COMPARISON OF BIOTIC PRESSURE BETWEEN
Faborso (total cover = 24  2
1%, mean  SE), Stari
ARTIFICIAL AND NATURAL HABITATS AND DIFFERENT
Grad (total cover 73  4
6%) and Kamenolom (total
SPECIES OF CYSTOSEIRA
cover 27
4  5
1%). No Cystoseira was detected at the
In October 2010, we started a larger caging experiment to con- three artificial sites, even if in the past it had been occa-
trast the relative importance of biotic factors in controlling the sionally detected (Ivesa pers obs).

© 2015 The Authors. Journal of Applied Ecology © 2015 British Ecological Society, Journal of Applied Ecology, 53, 16–24
 Ecology of marine infrastructures 21

(a) Cystoseira barbata water quality, high nutrient and sediment loads and stag-
100
nant conditions can be detrimental for many species of
80 Cystoseira (Irving et al. 2009; Perkol-Finkel & Airoldi
Cover (%)

60 2010; Strain et al. 2014), and it is possible that their

40 effects could have been significant over a time frame
20 longer than the one considered in this study.
0 Biotic pressure appeared to be the most relevant factor
Oct 2010 Dec 2010 Apr 2011 Oct 2011 limiting the survival and growth of Cystoseira on artificial
structures. While biotic pressure had limited to no effects
(b) Cystoseira compressa
100 on canopy-forming algae in natural habitats, on artificial
80 structures both C. barbata and C. compressa persisted
Cover (%)

60
only if caging prevented access to grazers. Although we
could not include a proper control for artefacts related to
40
caging (i.e. preventing alteration of environmental vari-
20
ables while effectively excluding grazers), we believe that
0 differences observed between caged and uncaged plots in
Oct 2010 Dec 2010 Apr 2011 Oct 2011
the artificial habitats reflect a true effect of biotic pres-
Time
sure, rather than alterations of other environmental fac-
ARS MA NAT MA
tors. First, on natural rocky reefs, caging had limited to
ARS Uncaged NAT Uncaged
no effect on survival and growth of Cystoseira. Secondly,
Fig. 3. Caging experiments at artificial and natural habitats. Per- observed responses were consistent across all experiments
centage cover of (a) Cystoseira barbata and (b) C. compressa in and sites, thus spanning a variety of hydrodynamic and
different habitats (artificial= ARS or natural= NAT) and caging other environmental conditions. Thirdly, video measures
treatments (Macrograzer exclusion = MA or Uncaged) . Values (F. Ferrario, E.M.A. Strain, P. Guidetti and L. Airoldi,
are averages (1 SE) from three sites for each habitat for Octo-
unpublished data) clearly showed that Cystoseira was sub-
ber 2010, and December, April and October 2011.
jected to a much greater variety and amount of biotic
interactions on artificial structures than on rocky reefs.
difference between habitats was detected by April 2011 as Herbivory is known to shape and regulate the benthic
a significant Exclusion by Habitat interaction (Table 2). algal community in a variety of natural rocky reef systems
Between April and October 2011, both species of Cysto- from temperate to tropical regions (Miller & Hay 1996;
seira decreased in uncaged treatments also at natural reef Verges, Alcoverro & Ballesteros 2009; Taylor & Schiel
sites. Although by October 2011 the effect of caging was 2010), but little is known about its effects on assemblages
significant at both natural and artificial habitats (Table 2), in artificial habitats (Ivesa et al. 2010; Marzinelli, Under-
the difference between caged and uncaged treatments at wood & Coleman 2011). In our experiments, we showed
the artificial sites was twice as large as that at the natural that biotic control on the biota was much stronger in arti-
sites, and both species of Cystoseira still covered >40% of ficial than nearby natural habitats. By using cages of dif-
the space in uncaged reef treatments compared to <10– ferent sizes, we excluded relevant effects by mesograzers
15% in uncaged artificial treatments. (<1 cm), while we observed that both small and large
macrograzers were severely limiting the growth of canopy-
Discussion forming algae. Subsequent video observations clarified
that the biotic pressure on Cystoseira at the artificial sites
Cystoseira was virtually lacking from all the studied artifi- resulted from a suite of both ‘consumptive’ (i.e. a clear
cial structures, which is consistent with observations from feeding behaviour) and ‘nonconsumptive’ (i.e. algal thalli
other artificial sites in the Mediterranean Sea, where the potentially damaged by handling and clipping without
presence of canopy-forming algae is hardly reported as an obvious consumption) interactions from a wide range of
important component of the epibenthos (Falace & Bres- species, comprising both fishes and crabs of different sizes.
san 2002; Vaselli, Bulleri & Benedetti-Cecchi 2008). Previ- Interestingly, most of the interacting species were classi-
ous work had already shown that the poor performance fied as omnivores rather than as true herbivores (F. Fer-
of artificial structures in supporting these ecologically rele- rario, E.M.A. Strain, P. Guidetti and L. Airoldi,
vant, habitat-forming species was not related to habitat unpublished data).
characteristics limiting dispersal or settlement (Perkol-Fin- The observation that algal assemblages in artificial
kel et al. 2012). Here, we further demonstrated that the habitats can be shaped by a stronger biotic pressure than
lack of Cystoseira on artificial structures was not even in natural reefs is new, and warrants further investigation
directly related to the poor environmental conditions typi- with respect to underlying causes. Many of the organisms
cal of many marine urban locations (Dugan et al. 2011). most frequently interacting with Cystoseira on artificial
Indeed, most transplanted juveniles survived perfectly well structures were also common at natural rocky sites but
for longer than a year when protected by cages. Poor exerted different levels of interactions with Cystoseira in

© 2015 The Authors. Journal of Applied Ecology © 2015 British Ecological Society, Journal of Applied Ecology, 53, 16–24
22 F. Ferrario et al.

Table 2. Relative importance of biotic pressure on the total percentage cover of Cystoseira barbata and C. compressa in artificial vs.
natural habitats

Cystoseira barbata Cystoseira compressa

April 2011 October 2011 April 2011 October 2011

Pseudo- Pseudo- Pseudo- Pseudo-

Source d.f. MS† F P d.f. MS† F P d.f. MS† F P d.f. MS† F P

Habitat = Hab 1 8074 16
93 0
09 1 2256 4
92 0
10 1 15 507 39
22 0
00 1 3341 4
23 0
11
Exclusion = Exc 1 19 849 95
54 0
00 1 26 693 26
88 0
01 1 23 062 27
17 0
01 1 34 100 118
19 0
00
Hab 9 Exc 1 15 440 74
32 0
00 1 2758 2
78 0
18 1 7826 9
22 0
04 1 1186 4
11 0
12
Site (Hab) 4 480 1
59 0
20 4 461 1
51 0
23 4 395 0
76 0
55 4 796 2
22 0
09
Site (Hab) 9 Exc 4 206 0
69 0
60 4 1003 3
28 0
02 4 849 1
64 0
18 4 288 0
80 0
52
Residual 34 301 33 306 35 517 32 358

Pairwise comparison MA vs. Uncaged within levels of Habitat

t P t P

Artificial habitat 13
65 0
01 9
54 0
01

Natural habitat 0
77 0
52 1
19 0
36

†
Mean square.
Factors are as follows: Habitat (Fixed with two levels: Natural vs. Artificial), Exclusion (Fixed with two levels: Macrograzers exclu-
sion = MA vs. no exclusion = Uncaged), Site (Random nested in Hab: three natural rocky sites Faborso, Stari Grad, Kamenolom and
three artificial sites at Valalta, Bolnica and Marina). The experiment started in October 2010. The Table reports a separate analysis for
April 2011 and October 2011 (see Materials and methods).
Data were analysed by PERMANOVA with 9999 permutations on a Euclidean distance matrix.
Significant effects (P < 0
05 for C. compressa in April 2011, P < 0
01 in all other cases) are in bold.

the two habitats, interacting more frequently at the artifi- 2015) and for the restoration of damaged canopies
cial than at the natural sites (F. Ferrario, E.M.A. Strain, (Gianni et al. 2013). So far, nearly all ecological consid-
P. Guidetti and L. Airoldi, unpublished data). One possi- erations in the design of hard marine infrastructures have
ble explanation for these different behaviours is that at focused on critical abiotic factors to try to mitigate their
natural sites effects of grazing on fucoid algae could be impacts and recover lost ecosystem services (Burcharth
mitigated by a larger abundance and variety of more et al. 2007). Different construction materials, surface tex-
palatable algal species compared to the artificial habitats. tures and habitat structural complexities have been tested
Similar hypotheses have been recently proposed to explain and are increasingly implemented to help mimic natural
the unusually high grazing effects on some algal restora- habitats, and providing support for desired native biodi-
tion projects (Campbell et al. 2014). Another hypothesis versity (Chapman & Underwood 2011; Firth et al. 2014).
could be that coastal infrastructures represent ‘oases’ of Pools or features mimicking the levels of complexity in
hard substrata set in a soft bottom environment (Airoldi natural habitats are used in the design of artificial struc-
et al. 2005a): such small, isolated, patchy habitats might tures built along rocky coasts to try to facilitate the
experience greater pressure from predators than large growth of rich assemblages of species (Martins et al.
rocky reefs, similarly to other ‘oases’ systems such as sea- 2010; Firth et al. 2013; Coombes et al. 2015). In addition
mounts (Rowden et al. 2010). Intense angling/spear fish- to these abiotic factors, we have shown that the ecologi-
ing on top predators, typical at the artificial structures in cal performance of artificial structures can be strongly
this region (Airoldi et al. 2005b), could further amplify controlled by biotic factors, including biotic pressure but
these effects, leading to greater pressure from herbivores also recruitment and competition, which have also been
or omnivores. Future work should attempt to quantify reported to differ between artificial and natural habitats
differences in grazing pressure at the seascape level, (Bulleri 2005; Ivesa et al. 2010; Marzinelli, Underwood
including habitats of different sizes and locations. & Coleman 2011). For example, the sparse cover of coral
Marine artificial structures are becoming extremely species on artificial reefs in Florida was related to a
common in many coastal areas globally, and they could stronger competition by algal turfs than in nearby coral
and should be designed to reduce ecological impacts reefs (Miller et al. 2009). It is crucial to incorporate such
while maximizing targeted ecosystem services (Dafforn knowledge in the planning, design and operation of cur-
et al. 2015). For example, the artificial seeding of rent and future marine developments. In our study
canopy-forming algae is a promising technique to region, any effort aimed at designing artificial structures
enhance the ecological value of artificial structures (Per- to support or restore important ecosystem processes,
kol-Finkel et al. 2012; Firth et al. 2014; Dafforn et al. functions and services, such as those provided by native

© 2015 The Authors. Journal of Applied Ecology © 2015 British Ecological Society, Journal of Applied Ecology, 53, 16–24
 Ecology of marine infrastructures 21

(a) Cystoseira barbata water quality, high nutrient and sediment loads and stag-
100
nant conditions can be detrimental for many species of
80 Cystoseira (Irving et al. 2009; Perkol-Finkel & Airoldi
Cover (%)

60 2010; Strain et al. 2014), and it is possible that their

40 effects could have been significant over a time frame
20 longer than the one considered in this study.
0 Biotic pressure appeared to be the most relevant factor
Oct 2010 Dec 2010 Apr 2011 Oct 2011 limiting the survival and growth of Cystoseira on artificial
structures. While biotic pressure had limited to no effects
(b) Cystoseira compressa
100 on canopy-forming algae in natural habitats, on artificial
80 structures both C. barbata and C. compressa persisted
Cover (%)

60
only if caging prevented access to grazers. Although we
could not include a proper control for artefacts related to
40
caging (i.e. preventing alteration of environmental vari-
20
ables while effectively excluding grazers), we believe that
0 differences observed between caged and uncaged plots in
Oct 2010 Dec 2010 Apr 2011 Oct 2011
the artificial habitats reflect a true effect of biotic pres-
Time
sure, rather than alterations of other environmental fac-
ARS MA NAT MA
tors. First, on natural rocky reefs, caging had limited to
ARS Uncaged NAT Uncaged
no effect on survival and growth of Cystoseira. Secondly,
Fig. 3. Caging experiments at artificial and natural habitats. Per- observed responses were consistent across all experiments
centage cover of (a) Cystoseira barbata and (b) C. compressa in and sites, thus spanning a variety of hydrodynamic and
different habitats (artificial= ARS or natural= NAT) and caging other environmental conditions. Thirdly, video measures
treatments (Macrograzer exclusion = MA or Uncaged) . Values (F. Ferrario, E.M.A. Strain, P. Guidetti and L. Airoldi,
are averages (1 SE) from three sites for each habitat for Octo-
unpublished data) clearly showed that Cystoseira was sub-
ber 2010, and December, April and October 2011.
jected to a much greater variety and amount of biotic
interactions on artificial structures than on rocky reefs.
difference between habitats was detected by April 2011 as Herbivory is known to shape and regulate the benthic
a significant Exclusion by Habitat interaction (Table 2). algal community in a variety of natural rocky reef systems
Between April and October 2011, both species of Cysto- from temperate to tropical regions (Miller & Hay 1996;
seira decreased in uncaged treatments also at natural reef Verges, Alcoverro & Ballesteros 2009; Taylor & Schiel
sites. Although by October 2011 the effect of caging was 2010), but little is known about its effects on assemblages
significant at both natural and artificial habitats (Table 2), in artificial habitats (Ivesa et al. 2010; Marzinelli, Under-
the difference between caged and uncaged treatments at wood & Coleman 2011). In our experiments, we showed
the artificial sites was twice as large as that at the natural that biotic control on the biota was much stronger in arti-
sites, and both species of Cystoseira still covered >40% of ficial than nearby natural habitats. By using cages of dif-
the space in uncaged reef treatments compared to <10– ferent sizes, we excluded relevant effects by mesograzers
15% in uncaged artificial treatments. (<1 cm), while we observed that both small and large
macrograzers were severely limiting the growth of canopy-
Discussion forming algae. Subsequent video observations clarified
that the biotic pressure on Cystoseira at the artificial sites
Cystoseira was virtually lacking from all the studied artifi- resulted from a suite of both ‘consumptive’ (i.e. a clear
cial structures, which is consistent with observations from feeding behaviour) and ‘nonconsumptive’ (i.e. algal thalli
other artificial sites in the Mediterranean Sea, where the potentially damaged by handling and clipping without
presence of canopy-forming algae is hardly reported as an obvious consumption) interactions from a wide range of
important component of the epibenthos (Falace & Bres- species, comprising both fishes and crabs of different sizes.
san 2002; Vaselli, Bulleri & Benedetti-Cecchi 2008). Previ- Interestingly, most of the interacting species were classi-
ous work had already shown that the poor performance fied as omnivores rather than as true herbivores (F. Fer-
of artificial structures in supporting these ecologically rele- rario, E.M.A. Strain, P. Guidetti and L. Airoldi,
vant, habitat-forming species was not related to habitat unpublished data).
characteristics limiting dispersal or settlement (Perkol-Fin- The observation that algal assemblages in artificial
kel et al. 2012). Here, we further demonstrated that the habitats can be shaped by a stronger biotic pressure than
lack of Cystoseira on artificial structures was not even in natural reefs is new, and warrants further investigation
directly related to the poor environmental conditions typi- with respect to underlying causes. Many of the organisms
cal of many marine urban locations (Dugan et al. 2011). most frequently interacting with Cystoseira on artificial
Indeed, most transplanted juveniles survived perfectly well structures were also common at natural rocky sites but
for longer than a year when protected by cages. Poor exerted different levels of interactions with Cystoseira in

© 2015 The Authors. Journal of Applied Ecology © 2015 British Ecological Society, Journal of Applied Ecology, 53, 16–24
24 F. Ferrario et al.

engineering considerations when designing coastal defence structures. Perkol-Finkel, S., Ferrario, F., Nicotera, V. & Airoldi, L. (2012) Conser-
Coastal Engineering, 87, 122–135. vation challenges in urban seascapes: promoting the growth of threat-
Gianni, F., Bartolini, F., Airoldi, L., Ballesteros, E., Francour, P., Guide- ened species on coastal infrastructures. Journal of Applied Ecology, 49,
tti, P., Meinesz, A., Thibaut, T. & Mangialajo, L. (2013) Conservation 1457–1466.
and restoration of marine forests in the Mediterranean Sea and the Rowden, A.A., Schlacher, T.A., Williams, A., Clark, M.R., Stewart, R.,
potential role of marine protected areas. Advances in Oceanography and Althaus, F. et al. (2010) A test of the seamount oasis hypothesis: sea-
Limnology, 4, 83–101. mounts support higher epibenthic megafaunal biomass than adjacent
Glasby, T.M., Connell, S.D., Holloway, M.G. & Hewitt, C.L. (2007) slopes. Marine Ecology-an Evolutionary Perspective, 31, 95–106.
Nonindigenous biota on artificial structures: could habitat creation facil- Sales, M., Cebrian, E., Tomas, F. & Ballesteros, E. (2011) Pollution
itate biological invasions? Marine Biology, 151, 887–895. impacts and recovery potential in three species of the genus Cystoseira
Hobbs, R.J., Higgs, E.S. & Hall, C. (2013) Novel Ecosystems: Intervening (Fucales, Heterokontophyta). Estuarine Coastal and Shelf Science, 92,
in the New Ecological World Order. Wiley-Blackwell, Oxford. 347–357.
Irving, A., Balata, D., Colosio, F., Ferrando, G. & Airoldi, L. (2009) Strain, E.M.A., Thomson, R.J., Micheli, F., Mancuso, F.P. & Airoldi, L.
Light, sediment, temperature, and the early life-history of the habitat- (2014) Identifying the interacting roles of stressors in driving the global
forming alga Cystoseira barbata. Marine Biology, 156, 1223–1231. loss of canopy-forming to mat-forming algae in marine ecosystems. Glo-
Ivesa, L., Lyons, D.M. & Devescovi, M. (2009) Assessment of the ecologi- bal Change Biology, 20, 3300–3312.
cal status of north-eastern Adriatic coastal waters (Istria, Croatia) using Taylor, D.I. & Schiel, D.R. (2010) Algal populations controlled by fish
macroalgal assemblages for the European Union Water Framework herbivory across a wave exposure gradient on southern temperate
Directive. Aquatic Conservation-Marine and Freshwater Ecosystems, 19, shores. Ecology, 91, 201–211.
14–23. Temmerman, S., Meire, P., Bouma, T.J., Herman, P.M.J., Ysebaert, T. &
Ivesa, L., Chapman, M.G., Underwood, A.J. & Murphy, R.J. (2010) Dif- De Vriend, H.J. (2013) Ecosystem-based coastal defence in the face of
ferential patterns of distribution of limpets on intertidal seawalls: experi- global change. Nature, 504, 79–83.
mental investigation of the roles of recruitment, survival and Underwood, A.J. (1997) Experiments in Ecology: Their Logical Design and
competition. Marine Ecology Progress Series, 407, 55–69. Interpretation Using Analysis of Variance. Cambridge University Press,
Levene, H. (1960) Robust tests for equality of variances. Contributions to Cambridge.
Probability and Statistics (eds S.G. Ghurye, I. Olkin, W. Hoeffding, Vaselli, S., Bulleri, F. & Benedetti-Cecchi, L. (2008) Hard coastal-defence
W.G. Madow & H.B. Mann), pp. 278–292. Stanford University Press, structures as habitats for native and exotic rocky-bottom species. Mar-
Stanford, CA. ine Environmental Research, 66, 395–403.
Maggi, E., Bertocci, I., Vaselli, S. & Benedetti-Cecchi, L. (2009) Effects of Verges, A., Alcoverro, T. & Ballesteros, E. (2009) Role of fish herbivory
changes in number, identity and abundance of habitat-forming species in structuring the vertical distribution of canopy algae Cystoseira spp.
on assemblages of rocky seashores. Marine Ecology Progress Series, in the Mediterranean Sea. Marine Ecology Progress Series, 375, 1–11.
381, 39–49.
Martins, G.M., Thompson, R.C., Neto, A.I., Hawkins, S.J. & Jenkins, Received 31 March 2015; accepted 28 August 2015
S.R. (2010) Enhancing stocks of the exploited limpet Patella candei Handling Editor: Tadeu Siqueira
d’Orbigny via modifications in coastal engineering. Biological Conserva-
tion, 143, 203–211.
Marzinelli, E.M., Underwood, A.J. & Coleman, R.A. (2011) Modified Supporting Information
habitats influence kelp epibiota via direct and indirect effects. PLoS
ONE, 6, e21936. Additional Supporting Information may be found in the online version
Miller, M.W. & Hay, M.E. (1996) Coral-seaweed-grazer-nutrient
of this article.
interactions on temperate reefs. Ecological Monographs, 66,
323–344.
Miller, M.W., Valdivia, A., Kramer, K.L., Mason, B., Williams, D.E. & Appendix S1. Details of preparation of experimental tiles.
Johnston, L. (2009) Alternate benthic assemblages on reef restoration
structures and cascading effects on coral settlement. Marine Ecology Table S1. Details of study sites.
Progress Series, 387, 147–156.
Perkol-Finkel, S. & Airoldi, L. (2010) Loss and recovery potential of
marine habitats: an experimental study of factors maintaining resili- Figure S1. Representative photographs of natural rocky reefs and
ence in subtidal algal forests at the adriatic sea. PLoS ONE, 5, artificial habitats.
e10791.
Perkol-Finkel, S. & Sella, I. (2014) Ecologically active concrete for coastal
Figure S2. Examples of experimental tiles used.
and marine infrastructure: innovative matrices and designs. From Sea to
Shore - Meeting the Challenges of the Sea (eds W. Allsop & K. Burgess),
pp. 1139–1149. ICE Publishing, London. Figure S3. Representative photographs of experimental treatments.
Perkol-Finkel, S., Shashar, N. & Benayahu, Y. (2006) Can artificial reefs
mimic natural reef communities? The roles of structural features and
age. Marine Environmental Research, 61, 121–135.

© 2015 The Authors. Journal of Applied Ecology © 2015 British Ecological Society, Journal of Applied Ecology, 53, 16–24