O R I G I N A L P A P E R
Combining gut fluorescence technique and spatial
analysis to determine Littorina littorea grazing dynamics
in nutrient-enriched and nutrient-unenriched littoral mesocosms
Eliecer Rodrigo Dı´az• Patrik Kraufvelin• Johan Erlandsson
Received: 17 June 2011 / Accepted: 15 December 2011 / Published online: 18 January 2012 Ó Springer-Verlag 2012
Abstract Spatiotemporal distribution patterns in relation to feeding behavior of herbivorous gastropods have been studied extensively, but still knowledge about small-scale patterns is limited in relation to eutrophication. This experimental study aimed to describe the small-scale dis- tribution of Littorina littorea in nutrient-enriched and nutrient-unenriched mesocosms in a merely atidal region and relate the distribution to food abundance and possible competing organisms, while checking simultaneously for feeding activities. The latter part was accomplished through the ‘‘gut fluorescence technique’’ GFT (which, to our knowledge, has not previously been used for benthic grazers) to estimate per capita grazing rates and the former part through monitoring of spatial heterogeneity of L. lit- torea and co-variation with sessile organisms (using semivariograms and cross-semivariograms, respectively).
After 5 months of nutrient addition, the abundance and biomass of L. littorea had increased in enriched systems, which also had significantly higher total biomass of green algae. Gut pigment content was higher in L. littorea from
enriched mesocosms, and gut depletion rate was higher in L. littorea from unenriched mesocosms. Spatial analysis showed that L. littorea exhibited generally random patterns (suggesting feeding activities) but sometimes (often in the morning) spatial patchiness (clumped distribution) in both enriched and unenriched conditions. There was mainly positive co-variation between L. littorea and biofilm, while different nutrient conditions exhibited contrasting co-vari- ation between L. littorea and barnacles (positive co-varia- tion in enriched and negative co-variation in unenriched mesocosms). The study offered insights into how feeding behavior and spatial distribution of a species may interact with community components differently under different nutrient regimes. The applied methodology can be useful for purposes of faster examination of grazing effects among different regions and also to compare grazing intensities and interactions between grazers and the benthic communities in disturbed (including pollution and nutrient enrichment) and non-disturbed systems, as well as in up- welling versus non-upwelling areas.
Introduction
A principal challenge for experimental ecology is to develop techniques that allow fast, but reliable, assess- ments of ecosystem variables, such as primary productiv- ity, diversity, and trophic interactions. In this study, we present a combination of two techniques (the gut fluores- cence technique, GFT, and spatial analysis) that can help to disentangle trophic dynamics and species distribution pat- terns in benthic systems. Nutrient enrichment and changes in grazer populations often interact to shape diversity and biomass of benthic macroalgal assemblages and primary consumers (Lubchenco and Gaines1981; Hillebrand2003;
Communicated by F. Bulleri.
E. R. Dı´az P. Kraufvelin (&) J. Erlandsson ARONIA Coastal Zone Research Team, A˚´ bo Akademi University and Novia University of Applied Sciences, Raseborgsva¨gen 9, 10600 Ekena¨s, Finland
e-mail: patrik.kraufvelin@abo.fi; pkraufve@abo.fi P. Kraufvelin
Environmental and Marine Biology, A˚ bo Akademi University, Artillerigatan 6, 20520 Turku/A˚ bo, Finland
Present Address:
J. Erlandsson
Vattenmyndigheten, Va¨sterhavets Vattendistrikt, Vattenva˚rdsenheten, La¨nstyrelsen i Va¨stra Go¨talands la¨n, 403 40 Go¨teborg, Sweden
DOI 10.1007/s00227-011-1860-y
Gamfeldt et al.2005; Eriksson et al.2009) and thereby also the ecosystem functioning (McQuaid 1996; Paine 2002;
Worm et al. 2002; Griffin et al. 2010; Kraufvelin et al.
2010). Although spatial and temporal distribution patterns of herbivorous gastropods in relation to their feeding behavior have been studied extensively on rocky shores (Hawkins and Hartnoll1983; Little1989; Little et al.1991;
Gray and Naylor1995; Johnson et al.1997; Coleman et al.
2006; Diaz et al. 2011), knowledge about small-scale pat- terns and species interactions is still limited in relation to eutrophication. Changes in benthic primary productivity, community composition, and species abundance caused by eutrophication, however, are generally well known (e.g., Cloern 2001; Burkepile and Hay 2006; Kraufvelin et al.
2006b,2010; Russell and Connell 2007; Masterson et al.
2008). Macroalgal-dominated littoral communities possess a high structural and functional resistance against excessive nutrient availability as long as the communities are not seriously affected by other chemical, physical, or biologi- cal processes (Connell1985; Thompson et al.2002; Bokn et al.2003; Worm and Lotze2006; Kraufvelin et al.2006b, 2010). Part of this resistance to mass occurrences of opportunistic macroalgae has been explained by grazing macroinvertebrates such as the common periwinkle Litto- rina littorea (L.) (Kraufvelin et al.2002), other molluscs (Russell and Connell 2007) as well as the amphipod Gammarus locusta L. (Kraufvelin et al.2006a) buffering eutrophication effects by exerting strong top-down control.
The understanding of the feeding behavior and ecology of gastropods, such as L. littorea, is instrumental for understanding the community structure of the shores that they inhabit (McQuaid1996; Carlson et al.2006). L. litto- rea is found on rocky shores at the East and West Atlantic coasts, preferentially at low shore levels (Norton et al.1990;
Carlson et al.2006; Perez et al.2009). Although the feeding preferences of L. littorea are mainly restricted to early successional stages of perennial macroalgae, diatoms, and ephemeral green algae (Norton et al.1990; Wilhelmsen and Reise1994; Jaschinski and Sommer2011), the species has also been characterized as an omnivorous grazer (Chang et al.2011) that can even feed on barnacle larvae (Wahl and So¨nnichsen1992; Buschbaum2000). The feeding activity of littorinids seems to be influenced by several factors such as body mass, water temperature, and tides (Newell et al.
1971; Norton et al. 1990). This leads to the question of whether L. littorea has an endogenous rhythm preferring feeding during the night to avoid predators and desiccation.
Normally, L. littorea feeds when substrates are damp or when it is submersed (Norton et al.1990) and during that time the gastropods exhibit less aggregation and their guts contain more food. When littorinid snails are inactive (e.g., on dry substrates), they tend to group in crevices, among mussels and barnacles or other architecturally complex
microhabitats forming clumps at different shore levels (Raffaelli and Hughes 1978; Chapman and Underwood 1996; Kostylev et al. 1997; Chapman 2000; Diaz et al.
2011; Erlandsson et al. in prep.). In situations with nutrient enrichment, primary productivity in terms of biofilms and/
or green algae will generally be enhanced, which may imply increased food availability or more nutrient rich food. This could, in turn, shorten the browsing distances and periods of L. littorea, only having to move a few centimeters away from the aggregations to reach feeding spots or being able to spend far less time foraging.
Our examination of the relationships between feeding activities and spatial aggregation of L. littorea in nutrient- enriched and nutrient-unenriched mesocosms comprised the study of its feeding (e.g., ingestion rates), the estimation of its spatial heterogeneity, and analysis of its relationships with other species components in the community. The feeding activity of a herbivorous gastropod could either be investi- gated by indirect or by direct methods. The indirect methods are studying gastropod movements over time or grazer exclusion by cages (Newell et al.1971; Underwood 1980;
Underwood and Jernakoff 1984; Boaventura et al. 2002;
Hutchinson and Williams 2003; Coleman et al. 2006), whereas the ‘‘gut fluorescence technique’’ (GFT) represents a direct method. GFT is one of the most broadly used methods in pelagic systems and it takes gut content, time of digestion, and defecation processes directly into account. It has been successfully used to estimate grazing activity of zooplankton in a variety of aquatic habitats (Mackas and Bohrer 1976;
Bernard and Froneman 2005). The principle behind the technique is that algal pigments ingested can be quantita- tively extracted from the animal using organic solvents (Ba˚mstedt et al.2000). The main benefit of the technique is that it is possible within 24 h to obtain data about how much a particular species consumes. GFT estimates grazing activity through the quantification of the ‘‘daily ingestion rate,’’ which contains three variables that can be experimentally obtained:
(1) integrated gut pigment, (2) gut depletion rate, and (3) gut pigment destruction. In spite of the simplicity of GFT, it has, to our knowledge, not been tested previously in benthic sys- tems. Following estimation of the temporal feeding activities of L. littorea snails by GFT, geostatistical tools were used to assess their spatial aggregations for the very same time periods. Within this process, semivariograms and fractal dimension were first used to distinguish between spatial patchiness and randomness (see Diaz and McQuiad 2011;
Diaz et al.2011for distribution of grazers) and then cross- semivariograms were used to examine the co-variation between L. littorea and different community components (barnacles, biofilm, and macroalgae).
The central aims of the present study were to document responses to nutrient enrichment and determine grazing dynamics of L. littorea and its co-variation with the rocky
shore community under controlled nutrient-enriched and nutrient-unenriched conditions. The following hypotheses were tested: (1) The abundance and biomass of L. littorea and the biomass of green macroalgae will be stimulated after 5 months of nutrient enrichment. (2) The combination of GFT (feeding) with semivariograms (spatial distribution) will show that L. littorea feeds (and disperses) differently during the day/night cycle and at different nutrient condi- tions. (3) There will be spatial co-variation (negative or positive) between L. littorea and other dominant organisms (barnacles, biofilm, and macroalgae), and this co-variation may be expressed differently at different nutrient levels. All parts of the study sum up to provide novel information about responses to nutrient enrichment, the trophic dynamics of L. littorea on nutrient-enriched and nutrient-unenriched temperate rocky shores as well as about interactions between the gastropod and its surrounding community.
Materials and methods
Solbergstrand rocky littoral mesocosms
All measurements were made in eight rocky littoral meso- cosms at Marine Research Station Solbergstrand by the Oslofjord (59°370N, 10°390E) in SE Norway. Each meso- cosm had a length of 4.75 m, a breadth of 3.65 m, and a maximum depth of 1.35 m (Fig.1). Throughout this study, the systems were kept non-tidal, since natural shores in the region are basically atidal (tidal amplitude ca 0.35 m) and we wanted specifically to control for all other factors to ascertain that the observed effects were due to the nutrient treatments. The water volume was 12 m3, and each flow- through mesocosm received water from 1 m depth from the Oslofjord at a rate of 5 m3h-1and with a short mean water residence time of 2–3 h. A wave machine generated con- stant waves (18 strokes per minute) with 11 cm amplitude corresponding roughly to a wind force of up to 5 m/s (Kraufvelin et al.2010). The entire mesocosm facility was covered with a transparent black net in order to reduce the light and UV effects (by approximately 50%) down to the bottoms of the mesocosms, where sugar kelp, Laminaria saccharina (L.) J.V. Lamouroux was grown within a sep- arate scientific project run simultaneously. With regard to macroalgae and L. littorea, including gastropod behavior (Kraufvelin et al. unpubl.), the mesocosm conditions resembled very closely natural conditions on semi-sheltered concrete walls and rock pools on shaded shores right outside Solbergstrand, that is, in the middle parts of the Oslofjord.
At the time of these experiments, the history of the rocky littoral communities of individual Solbergstrand mesocosms dated back [12 years. Rocky shore assem- blages were introduced in 1996 by transplanting small
boulders with attached macroalgae and associated animals, onto concrete steps in each mesocosm, and with time, mesocosm communities have been corresponding well with natural rocky shores in the region. (Bokn and Lein 1978;
Bokn et al.2003; Kraufvelin et al. unpubl.). The mesocosm experiments, which this study is part of, were run from April to September 2008 and comprised nutrient addition to four mesocosms, while the remaining four served as background controls receiving only fjord water. Before the start of the experiments, all mesocosms were evened out as described in Kraufvelin (2007), that is, the amount and type of macroalgae and, for example, the abundance of L. lit- torea were registered in each mesocosm and from meso- cosms where they were in excess, individuals were moved into mesocosms, where the occurrences were lower. These measures also ensured that there were no carry over influences from previous experimentation.
Enriched mesocosms were treated with 32 lM inorganic nitrogen (N) and 2 lM inorganic phosphorus (P) above the background levels in the Oslofjord (for which monitoring data was provided by Norwegian Institute for Water Research) continuously in the period April–September 2008. These nutrient addition levels are similar to con- centrations recorded in eutrophic areas locally (Kristiansen and Paasche 1982) and globally (Cloern 2001), and cor- responding nutrient addition levels have been utilized as
‘‘highs’’ during previous experiments in these mesocosms (Bokn et al.2002,2003; Kraufvelin et al.2006a,b,2010).
Nutrients were added as a mixture, which consisted of 14.3 mol N as NH4NO3and 0.9 mol P as H3PO4 and an N/P mol ratio of 16/1. The actual nutrient concentrations of the mesocosm water were not analyzed on a regular basis, but it could be ascertained from day to day that the desired nutrient concentrations were achieved thanks to constant monitoring of the amounts of nutrients that were auto- matically pumped up from separate trays for each enriched mesocosm. The two nutrient treatment levels were inter- spersed among the mesocosms, but in a systematic way instead of randomly to avoid the risk of getting too many parallel treatments at one end of the mesocosm facility.
Determination of abundance and biomass of L. littorea and green algae
In September 2008, the abundance and biomass of L. lit- torea on the northern (sunny) walls and the total amount of green algae were estimated in each mesocosm. The number of L. littorea was counted in 60 frames of a size of 5 9 5 cm within a fringe of 5 cm below the mean water level repeatedly every fourth hour over 24 h. Only the mean abundance per mesocosm (number given for 100 cm2) was used for the further analysis, in which four enriched mesocosms were compared to four unenriched
mesocosms. L. littorea biomass was estimated from the average size of the snails in three randomly chosen frames from each mesocosm. The size of L. littorea was measured from the apex to the opposite point of the operculum, and these length measurements were transferred to dry weights using the equation by Asmus (1987):
Length (cm)¼ 2:37 þ 0:33 ln DW(g) ð1Þ The average dwt of L. littorea was multiplied with the average abundance to get the biomass per 100 cm2on the wall, and this data were analyzed by a two-way nested ANOVA with factors nutrient (fixed, two levels) and basin (random, four levels, nested in nutrient) and the three frames as replicates. The cover of green macroalgae was estimated layer by layer (so that the total cover theoretically could exceed 100%) on the mesocosm steps, walls, bottom, and on the wave machine, using a 40 9 40 cm grid containing 25 smaller 8 9 8 cm quadrats. Cover values were transferred to biomass from wet weights of known surface areas of the algal species in the mesocosms (Kraufvelin et al.2010). The total biomass of green algae is hereafter referred to as total biomass of Ulva spp. (due to the dominance of Ulva lactuca with minor contribution from Ulva intestinalis) in contrast to the green turfs that were separately estimated inside the small 5 9 5 cm quadrats within the same fringe on the walls in which L. littorea were estimated above. This latter data set consists of a mixture of Cladophora spp. and Ulva intestinalis and is referred to as green turfs on the walls.
As for L. littorea abundance above, only the total biomass of Ulva spp. per mesocosm was used for the further analyses, in which four enriched mesocosms were compared to four unenriched mesocosms. For these variables, the differences between unenriched and enriched mesocosms were analyzed by one-way ANOVA, while differences for L. littorea biomass were analyzed by a two-way nested ANOVA.
Before the analyses, it was checked for normality with Kolmogorov–Smirnov’s test and homogeneity of variances by Cochran’s test. Total biomass of Ulva spp. was transformed by the square root, and biomass of L. littorea was transformed by ln(x ? 1) to meet the assumptions of parametric tests.
Determination of L. littorea ingestion rates and grazing impacts using the gut fluorescence technique (GFT)
For the GFT-work, individual snails of L. littorea were collected from the north-western corners of all mesocosm walls in order to avoid disturbing L. littorea on the northern walls, where the spatial analyses were carried out. After this, the biomass of L. littorea was determined as described by Asmus (1987) above. By the use of tweezers, the organisms were immersed in chloridric acid (HCl, 7%) for 5 s in order to destroy the amount of chlorophyll-a remaining on the shell and while taking care that the operculum was not covered with acid. Then, the shell was dried using paper, and the animal was crushed and Fig. 1 Schematic view over one Solbergstrand mesocosm. Most sampling for this article took place on the northern wall to the upper right
immediately immersed in a vial containing 8 ml methanol (80%) for 24 h in the darkness at 4°C.
Ingestion rates (I, unit lg Chl-a g ind-1 day-1) of L. littorea were estimated using the equation of Mackas and Bohrer (1976), which also has been used in pelagic studies (Perissinotto1992; Bernard and Froneman2005):
I¼ K G0= 1ð b0Þ ð2Þ
where K(h-1) is the gut depletion rate, G0(lg g-1 ind-1) is the integrated gut pigment, and b0 is the non-dimensional index of pigment destruction.
Integrated gut pigment G0(lg g ind-1)
Three individuals of L. littorea were collected from each mesocosm at intervals of 4 h over a 24 h period, and the pigments were extracted as the animals were being col- lected. One-way repeated measures (RM) ANOVA was used to test for differences in gut contents of L. littorea between nutrient conditions at different times of the day, where nutrient level was the main factor and time was the within subject factor. The assumptions of normality and homoscedasticity were checked and if they were violated, ln(x ? 1) transformations were used. The assumption of sphericity was violated (Mauchly’s test), and therefore, the P values were adjusted using the Greenhouse–Geisser criterion (Scheiner and Gurevitch1993).
Gut depletion rate K(h-1)
To determine the time necessary for the algal food to pass through the gut of L. littorea, 33 individuals were collected from each nutrient level and the pigment concentration of three individuals at intervals of 25 min (11 intervals = 4 h and 58 min) was determined. The concentration of chlorophyll-a was plotted versus time, and a non-linear regression equation was calculated. The significance of the regression was tested. The slope of the equation corre- sponded to the rate of pigment evacuation from the gut over time, which was compared between enriched and unenriched mesocosms using comparison of slopes (Sokal and Rohlf1995).
Gut pigment destruction (b0)
In order to investigate the loss of chlorophyll-a to non- fluorescent derivatives, a ‘‘two-compartment budget approach’’ used for pelagic organisms was adapted. The loss of pigment into non-fluorescent components in the digestion process, which represented the non-dimensional variable b0, was thereby estimated. Our modification applied to benthic ecology consisted of the use of ceramic plates, 7.5 by 7.5 cm, containing a known amount of
microalgae, instead of a volume of water containing a known concentration of phytoplankton. A total of eight replicate ceramic plates were prepared (one per mesocosm) and placed out for microalgal colonization during 4 months (May to September 2008). After this period, 58 individuals of L. littorea were removed from enriched and unenriched mesocosms and placed into separate aquaria (24 each from enriched and unenriched mesocosms and additionally 10 individuals were used to estimate the initial/basal content of chlorophyll-a in their guts, five from enriched and unenriched mesocosms, respectively). The aquaria con- tained individual compartments for each L. littorea to prevent the snails from feeding on the shells of each other.
The L. littorea specimens remained in isolation for 24 h with constant air and water flow before the experiment.
At the start of the experiment, half of the ceramic panel was cut and submerged into a petri dish containing three L. littorea specimens, leaving them feeding for 5 h. The other half of the ceramic panel was left submerged in another petri dish without L. littorea. Once the feeding period ended, the individuals were removed and their gut contents were determined. Similarly, the concentration of chlorophyll-a on both ceramic panels were determined. The loss of pigment into non-fluorescent derivatives (b0) was expressed as percentage and estimated using the equation:
b0¼ ½Cf t ðGtþ PtÞ=Ctg 100 ð3Þ where Ctis the concentration of chlorophyll-a in the con- trol panel (without L. littorea), Gt and Pt are the concen- tration of chlorophyll-a in the guts of L. littorea specimens and on the treatment panel at the end of the incubation, respectively. Gut pigment destruction estimates between mesocosms were compared using the Mann–Whitney test, due to the non-normality nature of the data.
Analysis of spatial patchiness of L. littorea on the mesocosm walls using semivariograms and fractal dimension
The spatial distribution patterns generated by the behavior of L. littorea were analyzed by counting individuals in a fixed transect (length: 3 m) within a fringe of 5 cm below the mean water level on the northern (sunny) walls of each mesocosm. The transects were sampled using contiguous quadrats of 5 9 5 cm, which allowed a sample size of 60 quadrats per transect with a minimum lag of 5 cm, defined as the distance between centers of two adjacent quadrats.
L. littorea individuals were counted in every quadrat and every 5 h during 24 h in every mesocosm. Additionally, estimation of percent cover of barnacles, green algal turfs (i.e., the mixture of Cladophora spp. and Ulva intestinalis L.), Hildenbrandia rubra (Sommerfelt) Meneghini, and biofilms on the wall was done by taking digital
photographs of each quadrat, which were later analyzed using the program Image Tool 3.0.
Spatial heterogeneity (patchiness) over time was esti- mated using geostatistical tools, and the fractal dimension D. Semivariogram analysis was used to determine spatial variability and spatial dependence in the distribution of L. littorea at different scales. The relationship between semivariance and lag was analyzed in order to be able to determine the spatial patterns at different times. The semivariance (Y(h)) was estimated using the equation:
YðhÞ¼ 1 2NðhÞ
XNðhÞ
i¼1
ðZiþh ZiÞ2 ð4Þ
where N is the total number of data points; N(h) is the number of pairs of data points separated by the lag h; Ziand Zi?hare the values of the studied variable at points i and i ? h (Dale2000). Fractal scaling analysis was used as an estimation of the heterogeneity of spatial distributions over the range of small scales (0.05–3 m). The fractal dimension (D) was calculated from the logarithmic semivariogram (log–log plot of semivariance and lag), using the equation:
D¼ ð4 mÞ=2 ð5Þ
where m is the absolute slope of the regression between semivariance and spatial lag (see e.g., Schmid 2000).
Fractal dimension is a non-integer measure of heteroge- neity. Values of D lower than 1.5 indicate spatial trends in the distribution, for example, environmental gradients, while values larger than 1.5 indicate patchy distributions (Kostylev and Erlandsson 2001). Simulations of distribu- tions along a transect have shown that data generated randomly produce spatial patterns with D values between ca 1.97 and 2 (Erlandsson et al. 2005). This indicates independence of the variance from the spatial lag (the slope of the regression in the semivariogram is not significantly different from 0), that is, random distribution patterns or homogeneity (Dale2000).
Lags up to half of the transect length were included in the regression analysis of the semivariogram. In order to make the analysis statistically robust, the minimal sample size used to analyze the variance at different lags was 30.
This is because semivariances do not represent variation between all data points at lags larger than half of the transect length (Schmid 2000; Erlandsson and McQuaid 2004; Erlandsson et al. 2005), as at each successively lar- ger scale, the number of comparisons decreases by one (from 59 pairs of combinations at lag 0.05 m to 30 pairs at lag 1.5 m).
Different fractal dimensions ‘‘D’’ can be estimated for each scaling region, and to detect these scaling regions, three conditions/steps need to be achieved (Kostylev and Erlandsson 2001): (1) detection of scaling breaks using
residual analysis, (2) significant linear regression of the suggested scaling sub-relationship, and (3) significant dif- ference between the slopes of successive scaling regions (see Kostylev and Erlandsson 2001; Erlandsson and McQuaid 2004; Erlandsson et al. 2005 for more details).
Sequential table-wise Bonferroni tests (Hochberg 1988) were applied for all the regression analyses to adjust the P values into accordance with the number of tests performed.
Analysis of spatial co-variation between L. littorea and community components using cross-semivariograms
In order to describe the relationship between the spatial patterns of L. littorea and spatial patterns of barnacles and algae across different spatial scales (from 0.05 to 1.5 m lags), cross-semivariogram analysis was used. The cross- semivariance was estimated by the equation:
YxzðhÞ¼ 1 2NðhÞ
XNðhÞ
i¼1
ðXiþh XiÞðZiþh ZiÞ ð6Þ
where N is the total number of data points; N(h) is the number of pairs of data points separated by the distance or lag h; Xiand Xi?h, and Zi and Zi?hare the values of two different variables (e.g., density of L. Littorea and barnacle cover, respectively) at points i and i ? h (Dale 2000;
Erlandsson and McQuaid 2004; Erlandsson et al. 2005).
The studied community components were as follows:
barnacles, biofilm, green algal turfs, and H. rubra.
A positive or a negative cross-semivariance value at a certain lag indicates a positive or a negative co-variation, respectively, at that scale. A cross-semivariance value approaching zero indicates no co-variation between vari- ables at that scale. To test whether cross-semivariance values were significantly different from 0, the distributions of pairs of variables along each transect were randomized 1,000 times and cross-semivariance was calculated at each scale for each random permutation. Each randomized value was compared with the appropriately observed cross- semivariance value. Then, the probability of each observed cross-semivariance value being higher (positive relation- ship) or lower (negative relationship) than by chance alone was calculated, and an alpha level of 0.05 was applied. The analyses were carried out using Matlab 7.0.1.
The significant co-variation detected was categorized into three groups of spatial scales: (1) microscales com- prising lags from 5 to 50 cm, (2) mesoscales comprising lags from 50 to 100 cm, and (3) macroscales comprising lags between 105–150 cm. The microscale has been defined as the scale where the organisms interact, for example, L. littorea intraspecifically, interspecifically, and with their food item (Underwood and Chapman 1996).
The mesoscale is the scale where the assemblage can reveal a patchy structure. The macroscale represents up to half of the total length of the transect. The frequencies of signifi- cant co-variation among the three groups of scales were compared using goodness of fit.
Results
Background nutrient concentrations, abundance, and biomass of L. littorea and total biomass of Ulva spp.
Background nutrient levels in the Oslofjord were quite high during the experimental period or around 0.45 lM P and 17 lM N (measurements from Norwegian Institute for Water Research from surface water in May and in August 2008). Since the nutrient dosing worked perfectly throughout the experimental period, this meant that enri- ched mesocosms on average had 5.4 times higher P levels and 2.9 times higher N levels than unenriched mesocosms.
A number of significant differences between the enri- ched and the unenriched mesocosms had occurred after 5 months of experimentation, and among the ones of direct relevance for this study, both the abundance and biomass of L. littorea on the northern wall as well as the total biomass of green Ulva spp. in the mesocosms were stimulated by nutrient enrichment. There was almost 60% higher abun- dance (F1,6= 6.01, P \ 0.05) (Fig. 2a) and almost 100%
higher biomass (F1,6= 13.67, P \ 0.01) (Fig.2b) of L. littorea on the walls of enriched systems compared to the unenriched ones in September, despite the original numbers and biomass of adult L. littorea being equal in the mesocosms in April (data not shown). The total biomass of Ulva spp. was in September seven times higher in enriched mesocosms than in unenriched mesocosms (F1,6= 17.07, P\ 0.01, Fig.2c), despite equal levels in April.
L. littorea ingestion rates and grazing impacts using GFT
The gut depletion rate (K) was higher (steeper slope in the regression) in L. littorea in unenriched than in enriched mesocosms (slope test: F4,15= 6.6, P \ 0.05; Table1, Fig.3), while the integrated gut pigment (G0) was higher in L. littorea from enriched than from unenriched mesocosms (F1,22= 7.9, P \ 0.01; Table1, Fig.4). There were also differences in the integrated gut pigment between times of the day (F6,132= 6.09, P \ 0.001) in such a way that G0 tended to be higher in the evening/night at 20.00 and 00.00 than in the morning/day at 04.00, 08.00 and 12.00. There was no interaction between nutrient input and time of the day (Fig.4). No significant differences were found in ingestion
rates (I) and gut pigment destruction rates (b0) for L. littorea between enriched and unenriched mesocosms (Table1).
Spatial patterns of L. littorea using semivariograms
Spatial structure/heterogeneity in the distribution of L. lit- torea (dependence between variability in the number of L. littorea and lag), that is, indicating clumping (not feeding), was often observed in the morning (ca half of all
Ulvaspp. biomass g wwt
0 100 200 300 400 500 600 700 800 900 1000
Unenriched Enriched
B A
Littorinaabundance per 100 cm2
0 1 2 3 4 5 6 7 8 9
Unenriched Enriched
B A
A
B
C
Enriched
Littorinabiomass g dw per 100 cm2
Unenriched
A
B
0 0,5 1 1,5 2 2,5 3 3,5
Fig. 2 aAverage abundance b Average biomass of L. littorea ? SD per 100 cm-2on the walls in enriched and unenriched mesocosms.
c Total biomass in g wwt of Ulva spp. ?SD in enriched and unenriched mesocosms. Significant differences are denoted by letters above the bars
morning transects) both in nutrient-enriched and nutrient- unenriched mesocosms (regressions significant, D \ 1.97, Table2). Most other distributions of L. littorea in the noon, evening, and at midnight (both treatments; 21 of 24 tran- sects) showed spatial independence (random patterns, non- significant regressions) indicating feeding and, overall, very few transects showed significant multiple scaling regions (Table2). The spatial structure observed was always a patchy/aggregated distribution.
Spatial relationship between L. littorea and community components using cross-semivariograms
The distribution of different community components (bar- nacles, biofilm, green algal turfs, H. rubra) on the meso- cosm walls can be seen in Fig.5as background data to the
investigation of co-variation between L. littorea and com- munity components. Among these, the higher amount of green turfs in unenriched mesocosms is an unexpected result, which should not be mixed up with the total biomass of green Ulva spp. in the mesocosms, which was higher in the enriched mesocosms (Fig.2b). The relationships between spatial variability of L. littorea distributions and the community components did not vary much over 24 h.
However, differences in the sign of the spatial co-variation were detected for some relationships:
Spatial co-variation between barnacles and L. littorea
Significant negative spatial co-variation between L. littorea and barnacles dominated in the unenriched mesocosms, but not in the enriched mesocosms. Most of the significant neg- ative co-variation were at the largest lags 105–150 cm, (v2= 27.12, P \ 0.001 in unenriched mesocosms). There was only one significant positive spatial relationship between L. littorea and barnacles in unenriched mesocosms. In con- trast, there were more significant positive relationships in enriched mesocosms distributed equally among micro-, meso-, and macro scales (v2= 2.71, P = 0.25) (Fig.6a).
Spatial co-variation between biofilm and L. littorea
The co-variation between biofilm and L. littorea exhibited predominantly positive relationships at both nutrient levels in terms of the number of significant lags found. Most positive co-variation was found at meso- and macro lags Table 1Variables obtained to determine daily ingestion rates of
L. littorea in nutrient-enriched and nutrient-unenriched meso- cosms ± SD: (1) integrated gut pigment (G0), (2) gut depletion rate
(K) (K does not have a SD because it was calculated from the slope of gut content versus time, see methods), (3) gut pigment destruction (b0), (4) daily ingestion rate
Treatment Integrated gut pigment, G0
(P
lg Chl-a. g-ind-1)
K(h-1) b0 Daily ingestion rate
(I, lg Chl-a. g-ind-1day-1)
Enriched mesocosms 66.78 0.264 0.36 ± 0.4 24.78 ± 2.8
Unenriched mesocosms 37.54 0.381 0.14 ± 0.2 16.88 ± 9.7
Fig. 3 Non-linear regressions of gut depletion rate (time in h) versus chlorophyll-a content in enriched systems (black diamonds; R2= 0.44;
y = 7.52e-0.264x) and in unenriched systems (open squares; R2= 0.34;
y = 2.48e-0.382x)
Time (hours)
µg Chla.gind-1
0 2 4 6 8 10 12 14 16 18 20
20 pm 0 am 04 am 08 am 12 pm 16 pm 20 pm Enriched mesocosms Unenriched mesocosms
Fig. 4 Integrated gut pigment for L. littorea. Average in the amount of chlorophyll-a (?SD) contained in the gut of individuals that were inhabiting enriched and unenriched mesocosms at each time during 24 h
Table 2 Regression exponents of the logarithmic semivariograms, and fractal dimensions (D) for the spatial lags of L. littorea distribution in the different mesocosms and transects: (a) unenriched and (b) enriched mesocosms
Mesocosm and Time Lags (m) Slope R2 P Fractal
dimension (D)
Spatial pattern
a. Unenriched
1: 7 am 1 0.05–1.5 0.17 0.57 0.000001 1.917 Dependence—patchy
1: 12 am 0.05–1.5 0.15 0.61 0 1.925 Dependence—patchy
1: 18 pm 0.05–1.5 0.04 0.06 ns 1.979 Independence—random
1: 00 pm 0.05–1.5 0.03 0.03 ns 1.987 Independence—random
1: 7 am 2 0.05–1.5 -0.03 0.02 ns 1.984 Independence—random
2: 7 am 1 0.05–1.5 0.08 0.14 0.045 1.962 Patchy
2: 12 am 0.05–1.5 0.11 0.40 0.0002 1.944a Dependence—patchya
Multiple scaling regions 0.05–1.05 0.12 0.46 0.0007 1.940 Dependence—patchy
1.10–1.5 -0.91 0.51 0.03 1.545 Patchy
2: 18 pm 0.05–1.5 -0.03 0.02 ns 1.984 Independence—random
2: 00 pm 0.05–1.5 -0.02 0.01 ns 1.991 Independence—random
2: 7 am 2 0.05–1.5 -0.01 0.01 ns 1.994 Independence—random
3: 7 am 1 0.05–1.5 0.07 0.25 0.0045 1.964 Patchy
3: 12 am 0.05–1.5 -0.03 0.04 ns 1.984 Independence—random
3: 18 pm 0.05–1.5 -0.02 0.003 ns 1.992 Independence—random
3: 00 pm 0.05–1.5 0.05 0.07 ns 1.974 Independence—random
3: 7 am 2 0.05–1.5 -0.04 0.04 ns 1.982 Independence—random
4: 7 am 1 0.05–1.5 0.11 0.35 0.0005 1.945 Dependence—patchy
4: 12 am 0.05–1.5 0.11 0.27 0.0031 1.947 Patchy
4: 18 pm 0.05–1.5 0.06 0.04 ns 1.972 Independence—random
4: 00 pm 0.05–1.5 0.11 0.19 0.015 1.947 Patchy
4: 7 am 2 0.05–1.5 0.09 0.38 0.0003 1.954 Dependence—patchy
Total 7 am 1 2–4 transects show spatial structure
Total 12 am 2–3 transects show spatial structure
Total 18 pm 0 transects show spatial structure
Total 00 pm 0–1 transect shows spatial structure
Total 7 am 2 1 transect shows spatial structure
b. Enriched
1: 7am 1 0.05–1.5 0.05 0.08 ns 1.973b Independence—randomb
Multiple scaling regions 0.05–0.6 0.19 0.66 0.0013 1.905 Dependence—patchy
0.65–1.5 -0.31 0.29 0.022 1.845 Patchy
1: 12 am 0.05–1.5 0.04 0.10 ns 1.978 Independence—random
1: 18 pm 0.05–1.5 0.02 0.01 ns 1.989 Independence—random
1: 00 pm 0.05–1.5 0.06 0.11 ns 1.968 Independence—random
1: 7 am 2 0.05–1.5 0.1 0.27 0.0034 1.951 Patchy
2: 7 am 1 0.05–1.5 0.07 0.11 ns 1.967 Independence—random
2: 12 am 0.05–1.5 0.05 0.09 ns 1.975 Independence—random
2: 18 pm 0.05–1.5 -0.04 0.04 ns 1.981 Independence—random
2: 00 pm 0.05–1.5 0.06 0.13 0.049 1.969 Patchy
2: 7 am 2 0.05–1.5 0.01 0.01 ns 1.993 Independence—random
3: 7 am 1 0.05–1.5 -0.02 0.01 ns 1.988 Independence—random
3: 12 am 0.05–1.5 -0.05 0.12 ns 1.974 Independence—random
3: 18 pm 0.05–1.5 0.0002 0.00 ns 1.999 Independence—random
3: 00 pm 0.05–1.5 0.02 0.01 ns 1.991 Independence—random
3: 7 am 2 0.05–1.5 0.17 0.48 0.00002 1.917 Dependence—patchy
4: 7 am 1 0.05–1.5 0.1 0.36 0.0005 1.949 Dependence—patchy
(enriched: v2= 7, df = 2, P \ 0.05, and unenriched:
v2= 59.27, df = 2, P \ 0.001). Significant negative co- variation was scarcely present in unenriched mesocosms, but more abundant in enriched mesocosms. This negative co- variation was distributed evenly through the micro-, meso-, and macrolags at both nutrient levels (v2= 6, df = 2, P[ 0.05 and v2= 5.8, df = 2, P [ 0.05) (Fig.6b).
Spatial co-variation between H. rubra and L. littorea
On the mesocosm walls, H. rubra was only present in one enriched mesocosm. Here, the relationship between H. rubra and L. littorea was only negative, which was significant at meso- and macrolags (v2= 34.86, df = 2, P\ 0.001) (Fig.7a).
Spatial co-variation between green algal turfs and L. littorea
Along the analyzed wall transect, only two unenriched mesocosms exhibited green turfs. The spatial co-variation
was predominantly negative, but there were also a few positive relationships. While the negative relationships were distributed evenly through the lags (v2= 2.33, df = 2, P[ 0.05), the positive co-variation was more abundant at macrolags (v2= 12, df = 2, P \ 0.01) (Fig.7b).
Discussion
Mostly similar overall responses to nutrient enrichment as in previous Solbergstrand mesocosm experiments were found in macroalgal and macrofaunal community structure, and this was also true for the stimulation of L. littorea and total biomass of Ulva spp. (Fig. 2a,b,c), supporting hypothesis 1, but the occurrence of green algal turfs only on the walls of two unenriched mesocosms was an exception (Fig.5). A higher total abundance of L. littorea in nutrient-enriched systems was also found by Kraufvelin et al. (2002) and higher total biomass of Ulva spp. in enriched systems by, for example, Bokn et al. (2003), Karez et al. (2004), Kraufvelin (2007), Kraufvelin et al.
(2006b,2010). For L. littorea, the abundance stimulation was probably due to a much higher recruitment (higher nativity, higher survival, lower mortality) in the enriched systems during summer, since most individuals present in September were juveniles. In addition, the average size of L. littorea was higher in enriched systems causing a sig- nificantly higher biomass of the grazer and revealing that also the growth of the juveniles had been enhanced. These results for L. littorea thus reflected processes that were taking place in the mesocosms during the 5 months the experiments lasted and that were under the influence of the nutrient treatment, such as a stimulation of total biomass of Ulva spp. (Fig.2c). This probably implied increased food availability, increased food nutrient richness, and more
0 20 40 60 80 100 120
green turf barnacles biofilm Hildenbrandia
percentage of cover Enriched mesocosms Unenriched mesocosms
community components
Fig. 5 Mean ? SD of the cover of community components on the northern walls of the mesocosms
Table 2continued
Mesocosm and Time Lags (m) Slope R2 P Fractal
dimension (D)
Spatial pattern
4: 12 am 0.05–1.5 0.03 0.03 ns 1.984 Independence—random
4: 18 pm 0.05–1.5 0.03 0.02 ns 1.986 Independence—random
4: 00 pm 0.05–1.5 0.13 0.31 0.0013 1.936 Dependence—patchy
4: 7 am 2 0.05–1.5 0.30 0.71 0.000000 1.849 Dependence—patchy
Total 7 am 1 1–2 transects show spatial structure
Total 12 am 0 transects show spatial structure
Total 18 pm 0 transects show spatial structure
Total 00 pm 1–2 transect shows spatial structure
Total 7 am 2 2–3 transects show spatial structure
Significant P values after a sequential Bonferroni correction are in bold face
a Scaling break at the lag 1.05 m
b Scaling break at the lag 0.6 m
favorable species composition of the food for L. littorea, as previously also was demonstrated for G. locusta by Kraufvelin et al. (2006a). Apparently, these initial effects of nutrient enrichment, like a stimulation of fast-growing green algae and certain grazers, seem to be quite universal and to take place also in the absence of tides and at lower light intensity. These results are also in agreement with corresponding findings from field investigations on natural temperate shores (e.g., Worm and Lotze 2006; Eriksson et al.2006,2009; Masterson et al.2008). The lack of green turfs on the walls of enriched systems may be due to several reasons, among others a higher abundance and biomass of grazing L. littorea, since green turfs are among the preferred food items for this species and could thus be rapidly grazed away by L. littorea (Wilhelmsen and Reise
1994) or with the help of other dominant grazers such as G. locusta (Kraufvelin et al.2006a). Hence, from a whole- mesocosm perspective, the grazers do not seem to be able to control total biomass of Ulva spp., but it seems that, at least on the walls, the amount of green algal turfs is grazer mediated.
The present study shows that the gut fluorescence technique (GFT) also works for benthic grazers such as L. littorea (Fig.3), and the preferred feeding on green filamentous and sheet-like algae, rich in chlorophyll-a pigments, makes this grazer appropriate for the technique.
One of the main criticisms of the technique is that it pro- vides only a measure of the herbivorous activity of an organism and fails to consider the possibility that organ- isms are consuming alternative carbon sources, including
A B
Fig. 6 Number of significant lags that exhibited statistical signifi- cance in the spatial co-variation between L. littorea and an overall present community component. a Spatial relationship between L. littorea and barnacles. This analysis showed positive spatial co-variation only in enriched mesocosms, while negative spatial
co-variation was predominant in unenriched mesocosms. b Spatial relationship between L. littorea and biofilms. This showed a dominance of positive spatial co-variation in both enriched and unenriched mesocosms
A B
Fig. 7 Number of significant lags that exhibited statistical signifi- cance in the spatial co-variation between L. littorea and a community component that was not present in all mesocosms. a Significant negative spatial co-variation between L. littorea and Hildenbrandia sp
was present in the enriched mesocosm where the algae occurred.
bSignificant spatial co-variation between L. littorea and green turfs in the two unenriched mesocosms was mainly negative
detritus or heterotrophic carbon sources (Boyd et al.1980;
Ba˚mstedt et al. 2000). However, many rocky shore gas- tropods, such as L. Littorea, are predominantly herbivo- rous, feeding mainly on the thin film of epiphytic algae or diatoms, on green algae as well as on brown algal germ- lings (Wilhelmsen and Reise1994), which means that the GFT may be very suitable to estimate the grazing activity in these benthic animals.
Pakhomov and Froneman (2004) suggested that inges- tion rates of pelagic animals can either be affected by food availability or by changes in feeding behavior related to seasonal variation. Although the daily ingestion rate was not significantly different between L. littorea-inhabiting enriched and unenriched mesocosms, the integrated gut pigment (G0), indicating how much algal pigment (food) is contained in the gut in 24 h, showed lower values at low nutrient levels, whereas the gut depletion rate (K) was faster in L. littorea-inhabiting unenriched mesocosms. This suggests that L. littorea consumes and retains more food in enriched environments and that the depletion rate and integrated gut pigment were sensitive to food availability.
However, the increased snail density and biomass under enriched conditions may have enhanced competition for food leading to a situation, where snails were similarly resource limited irrespective of enrichment level. In addi- tion, it is also possible that these results largely reflect both qualitative and quantitative effects within the algae (Jaschinski and Sommer 2011). In this sense, see also Kraufvelin et al. (2006a), where a path analysis showed that indirect effects on G. locusta density from nutrients via green algae were 50% bigger than direct nutrient treatment effects on gammarid abundance.
The reasons why the pigment destruction variable did not differ between L. littorea-inhabiting environments with different nutrient levels and that the variability was so great within the estimations are not easy to determine. The vari- ability could have been caused by (1) variability in the amount of algae consumed by individual L. littorea due to the experimental condition (experimental stress), and/or (2) spatial heterogeneity in the abundance, species composition, and nutrient content of algae colonizing the ceramic plates.
However, the results suggest that food availability does not affect the ingestion rate of L. littorea. Recently, Durbin and Campbell (2007) argued that pigment destruction should not be estimated to calculate the daily ingestion rate, since assimilation and destruction of pigments in the gut passage (b0) are already estimated and present in the calculation of gut depletion rate. Under this view, recalculating the values, non-significant differences in the daily ingestion rates of L. littorea between enriched and unenriched conditions were still observed (15.9 ± 2 and 9.8 ± 8.3 lg Chl-a g ind-1 day-1, respectively) and there was still a high feeding vari- ability within unenriched mesocosms.
Nevertheless, the opposite magnitudes in G0 and K between nutrient levels and no differences in ingestion rate are in agreement with the premises of Optimal Foraging Theory, which argue that animals should be capable of adjusting gut passage time depending on both food avail- ability (Taghon1981; Penry and Jumars1986) and/or quality of ingested food (Pakhomov and Froneman2004). It may therefore be concluded that the response of L. littorea to high food availability is to slow down the gut depletion rate (K) and the reverse at lower food availability.
At both nutrient levels, spatial heterogeneity in L. lit- torea could be found (especially in the morning), although random distribution patterns dominated (Table2), indi- cating that there may be certain times when snails are clumped (e.g., resting) and other times when they are dispersed (e.g., feeding) during the day/night cycle. There has been a considerable debate about when intertidal spe- cies are feeding and about the relationship between feeding and the tidal regime and day/night periods (Hawkins and Hartnoll1983; Little et al.1991; McQuaid1996; Chapman 2000). Interestingly, with continued experimentation using replicated days and nights, the applied techniques would have allowed us to formally test when these grazers were actually consuming algae through the evaluation of the gut contents (integrated gut pigment variable, G0) over time and to relate these values to their spatial distribution. In the present experiment, the results for G0show that the con- centrations of algal pigments in the guts of L. littorea varied between different times over the studied 24 h and that some of these values seemed to fit with their spatial distribution patterns (hypothesis 2 partly confirmed, Fig.4, Table2). It has been suggested that intertidal grazers such as L. littorea mainly feed during the night as an adaptive response to avoid visual predators (Carlson et al.2006) and desiccation (Newell et al.1971; Chapman and Underwood 1996). Nevertheless, the variability in the activity of L. littorea was great with some individuals being found in patches and some dispersed, regardless of nutrient condi- tion and time of the day.
As our sampling was carried out on homogeneous sur- faces without crevices (on concrete walls), our study shows that clumping behavior can be determined by other biotic factors than, for example, crevices and shelter on the rock surface (e.g., Underwood and Chapman 1996; Erlandsson et al. in prep), such as the comprising community (Fig.5), especially the barnacles, although we do not want to underestimate the potential effect of the complexity of the substratum (Skov et al.2010). Here, the space in between barnacles and mussels can thus be important for the abundance of Littorina sp. depending on the size of the littorinid species or morph/ecotype (Kostylev et al. 1997).
Furthermore, the spatial relationship of L. littorea and barnacles in the present study did not change during the
day, but it differed with regard to nutrient levels (Fig.6a).
Thus, L. littorea from enriched mesocosms preferred to inhabit spots where barnacles were present (mainly posi- tive co-variation), while L. littorea from unenriched mes- ocosms avoided patches with barnacles (mainly negative co-variation) suggesting that the nutrient level of the sys- tem may drive this relationship and providing support for hypothesis 3. However, earlier or unpublished field studies indicate that barnacle cover affects the distribution of L. littorea negatively, just as could be seen in the unen- riched mesocosms (Fig.6a), while rough periwinkles (e.g., Littorina saxatilis Olivi and Littorina arcana Hannaford- Ellis) are affected positively by barnacles (Kostylev et al.
1997; Erlandsson et al. in prep.), and in Littorina sitkana Philippi, the size of the snail also affects this association (Jones and Boulding 1999). In our study, the clearly dif- ferent L. littorea preferences for barnacles at the different nutrient levels are not easily explained, but one reason may be that L. littorea in the enriched mesocosms, due to higher competition and less green turfs on the walls, also had to feed on epiphytes on the barnacles, which was not the case in the unenriched mesocosms. Indications that L. littorea could be capable of feeding on barnacle larvae would further complicate these interactions (Wahl and So¨nnich- sen1992; Buschbaum2000) and may be another reason for the positive co-variation between L. littorea and barnacles found in the enriched systems. On the other hand, the differences in spatial co-variation between L. littorea and barnacles in enriched and unenriched mesocosms can also be due to the higher abundance of L. littorea in the enri- ched mesocosms showing positive co-variation with bar- nacles regardless of its dispersion.
With regard to other interactions with community components, that is, biofilm, green algal turfs, and H. rubra on the wall, some additional interesting findings were made (Figs.6b, 7a,b). The co-variation between L. littorea and biofilms also differed between enriched and unenriched mesocosms, being only positive in unenriched mesocosms and both positive and negative in enriched mesocosms, suggesting that feeding on biofilms, which is in agreement with their expected diet (Norton et al. 1990; Hillebrand et al. 2000; Skov et al. 2010), was more important in unenriched environments. This is slightly in contrast to the situation for L. littorea co-variation with barnacles above but may be due to complex preference patterns among the snails such as differences in reactions to nutrient enrich- ment levels for the various potential food resources, that is, green turfs, biofilm, epiphytes on barnacles, etc. (Karez et al.2004; Kraufvelin et al.2006a). Some differences in community structure between mesocosms may also have been caused by the higher abundance and biomass of L. littorea, and/or thereby higher grazing rates in the enri- ched mesocosms. An increased grazing pressure and
decreased amount of green turfs on the walls may have promoted the domination of biofilms and eventually the presence of the encrusting alga Hildenbrandia sp. (Bertness et al. 1983). H. rubra was observed in one out of four enriched mesocosms, while this species was absent from unenriched mesocosms. It has been reported that Hilden- brandia sp. uses antifouling chemical defense to inhibit settlement of foliose algae and microalgae (Madikiza2005).
This could cause the inhibition of food searching in L. lit- torea. The lack of H. rubra seemed to be compensated for by the presence of green algal turfs on the walls of unen- riched mesocosms. A general increase in the primary pro- ductivity in enriched mesocosms could also in itself have facilitated the development of macroalgae that rapidly were consumed by grazers (Kraufvelin et al.2002,2006a), in turn promoting the cover of biofilms. In unenriched mesocosms, on the other hand, some spots of green algal turfs on the walls could sustain the low abundance of grazers.
The realism of mesocosm studies may always be ques- tioned; see for example Perez (1995) and Kraufvelin (1999) regarding mesocosms in general and Kraufvelin et al. (2006b, 2010) regarding the Solbergstrand meso- cosms specifically. Nevertheless, with regard to this study, there were a number of undisputable advantages with using the mesocosm approach compared to visiting many dif- ferent field sites. Among these, there were controlled nutrient levels and equal substrate material, topography, wave action (both wave height and direction), water cur- rents, water temperatures, light conditions (both intensity and timing), and predator abundance (constantly low).
Most importantly for this study, there were four replicated
‘‘shores’’ of each nutrient level available within a few meters and these shores/mesocosms could be accessed by the same researchers within a few seconds enabling repe- ated ‘‘simultaneous’’ sampling. A similar study could not have been done in the field by the same amount of resources and man-power. In that sense, the possible restrictions imposed by the mesocosm enclosure, for example, lower predator levels and thereby a possibly altered gastropod behavior (see Coleman et al. 2006;
Coleman 2010) and lower wave exposure and thereby lowered dilution of nutrients (see Valdivia and Thiel2006), should not be more serious than site to site differences (context dependency) out in the field (Burkepile and Hay 2006; Connell and Irwing2009; Wahl et al.2011; Bulleri et al. unpubl.).
To summarize, this study offers insights into feeding behavior and spatial distribution of L. littorea and how the species interacts with community components through the consumption of certain algal groups and then promotion of the recruitment of other components in the community differently under different nutrient regimes (possible interaction between top-down and bottom-up effects).
