The Dietary Importance of Kelp-Derived Detritus to Pelagic and Benthic Consumers along the West Coast of Vancouver Island, Canada

Stable isotope analysis was used to determine the relative dietary importance of kelp-derived detritus to plankton and benthic organisms along a gradient of kelp abundance driven by recovering sea otter populations along the west coast of Vancouver Island (WCVI), Canada. The study used region-specific kelp isotope values (δ 13 C and δ 15 N) and season-specific phytoplankton isotope values to model dietary contributions of kelp-derived detritus (KDD). In general, KDD contributions were moderate to high in most plankton size fractions during the summer and decreased during the winter, particularly in the kelp sparse region. Hypothesized regional and spatial (distance from the coast) differences in kelp detritus contributions to zooplankton were not evi-dent. Modeled estimates o f the KDD contribution to benthic invertebrates were high (>40%) and independent of the organism size, among regions and between seasons, with the exception of Astraea gibberosa in the kelp abundant region. Local oceanography, natural kelp isotope signature variation, and significant overlap between kelps’ and blooming phytoplankton isotope values led to a large uncertainty in the assessed KDD contributions in benthic organisms. These results highlighted the importance of the KDD as a widespread and stable year-round food source in coastal kelp populated regions.


Introduction
Identifying drivers of nearshore ecosystem productivity, diversity and stability contribute to our understanding of food web, community and ecosystem ecology. Detritus, as an autochthonous and allochthonous subsidy, is an important bottom-up driver of trophic structure and food web dynamics by influencing energy, carbon and nutrient budgets [1] [2]. Generally, the biomass of pelagic and benthic organisms is a function of local water column productivity [2].
However, the flow of subsidies across habitats can create diverse and subsidized food webs where local productivity is low [2]. The dispersal and concentration of subsidies can be affected by local surface and bottom currents [3], regional productivity, and the ratio of habitat perimeter to area [4].
Coastal marine consumers feed directly, or indirectly, on kelp-and/or phytoplankton-derived organic matter. Kelps (Phaeophycea: Order Laminariales) are large fleshy macroalgae that occupy low intertidal and shallow subtidal rocky reef habitats of temperate coastal marine ecosystems [5]. Kelps contribute substantially to nearshore primary productivity [6], with estimates as high as 3000 g C m −2 ·yr −1 [6] [7] [8] [9]. In British Columbia, bull kelp (Nereocystis luetkeana) sporophytes are able to assimilate 1400 g C m −2 during summer [10] and giant kelp (Macrocystis pyrifera) productivity estimates range from 460 to 1300 g C m −2 ·yr −1 [6] [7] [11]. Most of the ocean-based photosynthesis occurs along coastal margins, which occupy only 0.1% of the total ocean area but average phytoplankton productivity in upwelling regions ranges from 200 to 973 g C m −2 ·yr −1 [12] [13]. Phytoplankton concentrations along the west coast of Vancouver Island (WCVI) reach a maximum during late spring to mid-summer when solar radiation is high and nutrients are readily available from winter mixing of the water column and upwelling [14] [15]. The relative contribution to particulate organic matter (POM) of these two main primary production sources (phytoplankton and macrophytes) can vary spatially and temporally [16] [17] [18] [19].
However, the year-round presence of kelp-derived detritus (KDD) could be important in the food web dynamics, and system stability, and have considerable effects on trophic structure and biodiversity [20].
With over 80% of the annual kelp production dispersed to the surrounding shallow and deep-water ecosystems through both KDD and dissolved organic matter (DOM) [21] [22] [23] [24], kelp production is important to kelp associated communities [16] [25] [26]. It may also be transported great distances from source populations as suspended POM [17] [18] [27]. Kelp-derived detritus (KDD) can be detected at great depths [3] [24] and is known to subsidize island and nearshore communities [4] [16] [28]. However, some studies have called into question the suitability of kelp as a food source because of high C:N B. C. Ramshaw [29], high concentrations of secondary metabolites [30], and structural rigidness [31]. Bacterial degradation of KDD can lower C:N ratios and secondary metabolite content [32], thereby increasing its nutritional quality. Relative to phytoplankton, kelp-derived carbon can contribute similarly or more to the diets of nearshore benthic invertebrate filter-feeders with estimates as high as 90% [16] [33] [34]. More importantly, KDD can increase consumer growth rates by 2 to 5-fold, relative to pure phytoplankton diets [33], and may have important consequences for population dynamics and ecosystem productivity [20].
Carbon and nitrogen stable isotope analysis is a commonly used and powerful tool in food web ecology that quantifies the incorporation of potential food sources (e.g. phytoplankton and kelp) into consumers and estimates the average trophic level of consumers [35] [36]. Variability and uncertainty in source isotope values complicate this method and have possibly led to the overestimation of KDD contribution to organisms in previous studies [37]. In marine ecosystems, δ 13 C values typically increase by 0‰ to 1‰ per trophic level [35] [36] [38] [39]. Changes in δ 15 N values with each trophic level transfer vary with tissue type [40] and feeding mode [39]. The average increase of δ 15 N values with trophic level ranges from 1.4‰ for consumers of invertebrates [39] to 4‰ for filter-feeders [34].
The research aims of this study were twofold: first, to assess the KDD contributions to the tissues of plankton and benthic invertebrates near and offshore in a ~20-fold kelp productivity gradient resulting from sea otter population recovery along the WCVI; second, to investigate seasonal variability in the KDD contributions to pelagic and benthic invertebrates along the WCVI.

Study System
The research was conducted along the west coast of Vancouver Island (WCVI), British Columbia, Canada (Figure 1), situated at the northern end of the Cali-  [45]. In the Open Journal of Marine Science presence of sea otters, sea urchins are heavily preyed upon which is followed by re-establishment of extensive kelp forests that support diverse and productive ecosystems [5] [46] [47]. As a result of the North Pacific maritime fur trade, sea otters were extirpated from the coast of British Columbia by 1929 [48]. However, between 1969 and 1972 eighty-nine animals were reintroduced to the northwest coast of Vancouver Island from western Alaska [49]. This population has increased rapidly and expanded north-and southwards along the WCVI [50] [51]. At the time of sampling, sea otters were distributed from the northern tip of Vancouver Island southward to an area just south of Clayoquot Sound ( Figure 1) and have dramatically increased kelp populations along this coastline [52] [53].

Experimental Design
Three research cruises were conducted off the WCVI between July 2009 and July 2010. The study region included three large sounds that were characterized by large differences in kelp biomass corresponding to the sea otter occupation time [52] [53]. A "space-for-time substitution" approach [54] was used to take advantage of sea otter reintroduction and range expansion along the WCVI. Kyuquot Sound (Figure 1), in the north (nearest to where sea otters were first rein-sparse" and "kelp abundant" regions summer sampling occurred between July 18th and 23rd, 2009; summer sampling off the kelp intermediate region was conducted between July 27th and July 28th, 2010. Winter sampling only took place off the kelp abundant and kelp sparse regions between January 23rd and 31st, 2010 due to inclement weather. It was assumed that sampled organisms were feeding on KDD produced within their respective sounds. Regional kelp and phytoplankton isotope value estimates (δ 13 C and δ 15 N), see In the laboratory, all samples were thawed, randomly sub-sampled in triplicates to identify and enumerate major plankton groups within each size fraction.
Samples were then dried in a Fisher Scientific oven at 50˚C for at least 24 hours, ground into a powder and packaged into 5 × 8 mm tin capsules for stable isotope analysis. KDD was visible microscopically, especially in the 63 -125 µm and 125 -250 µm size fractions, but was difficult to quantify and is therefore not included.

Benthic Organism Stable Isotopes
Benthic dredging was performed with a 60 cm by 30  To test for possible lipid extraction (i.e. defatting) effects on red turban snail stable isotope values, five whole snails were split in half and one half was defatted, and the other remained untreated, following the method of Bligh and Dyer [58] and then lightly rinsed with distilled water on 0.7 μm Whatman GF/F filters with light suction. Samples were re-dried and ground into a powder. Additionally, C:N ratios were used to check for excess lipids in both cases.

Data Analyses
Size-fractionated plankton: Linear regression was used to assess the relationship between distance from the kelp forest and plankton δ 13 C values.
Benthic organisms: Where assumptions were met, paired t-tests were used to test for effects of acidification and lipid extraction of benthic organisms. In cases of non-normality or heterogeneous variance, Wilcoxon Matched pairs tests were used. Linear regression was used to assess relationships between stable isotope values and organism size (length or width). All statistical analyses were performed in R [59].
MixSIR: We used the isotope mixing model program MixSIR (version 1.0, [60]) to determine percent KDD contribution to plankton and benthic organisms. MixSIR is a graphical user interface that performs Bayesian analysis of stable isotope mixing models using Hilborn sampling-importance-resampling (SIR) algorithm [61]. The benefits of a Bayesian approach to stable isotope mixing models include: 1) accounting for uncertainty in source stable isotope values, 2) accounting for uncertainty in the estimates of source contributions as there is underlying uncertainty in the mixture and source isotope values, 3) determining a unique solution when more than two sources are present [62] [63]. Outputs of the program include a median percent contribution of all sources included and their respective 95% confidence intervals. The degree of overlap between confidence intervals defines the probability that two estimates are the same. Source δ 13 C fractionation and fractionation standard deviations were set to 0‰ for summer and winter plankton. Source δ 15 N fractionation and fractionation standard deviations were set to 1.47‰ ± 0.39‰, while for winter it was 2.32‰ ± 0.46‰. For other consumers the values used were 0.5‰ ± 0.13‰ for δ 13 C and 2.2‰ ± 0.3‰ (for primary consumers), 2.13‰ ± 0.18‰ (for omnivores and detritivores) and 3.3‰ ± 0.26‰ (for higher consumers) for δ 15 N [64]. These values will be justified in the "Results" section.

Surface Plankton Nitrogen Stable Isotopes
Summer: With respect to plankton δ 15

Trophic Level Isotope Fractionation
Although summer 63 -125 μm plankton δ 13 C was more enriched than the 125 -Open Journal of Marine Science step-wise enrichment with size fraction. The mean 13 C fractionation (±SD) was −0.34‰ ± 1.51‰ (i.e. values were more depleted with increasing trophic level) between 63 -125 μm and these two fractions. There was no difference among winter plankton δ 13 C fractions. Considering these factors, source δ 13 C fractionation and standard deviation were "0" for summer and winter.
Because the summer and winter plankton δ 15

Kelp-Derived Detritus Incorporated by Plankton
Summer: During the summer in the kelp abundant region, the median KDD contributions to plankton ranged from 21% to 81% for all size fractions ( Figure   3). Median KDD contributions to this fraction remained high and constant with distance from the kelp forest. The two largest fractions were similar in this regard with the exception of a few distances where the median KDD contribution was much higher, such as 1 km along the 250 -500 µm fraction. For example, for the 250 -500 µm fraction at 1 km the median contribution was 81% (Figure 3).
Within a size fraction there was large overlap in confidence intervals with distance and between transects at the same distance except for the 500 -2000 µm fraction where there was greater variability with distance but the two transects were generally similar.
In the kelp intermediate region, median KDD contributions to plankton varied between 16% and 55% for all size fractions. It was the highest in 63 -125 μm plankton with the median KDD contributions ranging from 47% to 55% and remained moderately high and constant with distance from the kelp forest ( Figure 3). Median KDD contributions slightly decreased with increasing fraction size. Samples greater than 500 µm were not collected for this region.
In the kelp sparse region, median percent KDD contribution to plankton was highly variable with distance and between transects for all size fractions of plankton ranging from low (16%) to very high (81%) (Figure 3). Median con- Winter: In the kelp abundant region, the median KDD contributions to plankton ranged from 4% to 68% for all size fractions (Figure 4). Median KDD contributions to the 63 -125 μm fraction remained high and constant with distance from the kelp forest with large overlap in confidence intervals with distance and between transects. For the 125 -250 µm plankton median KDD contributions decreased moderately with distance and had relatively low overlap in confidence intervals between 0 and 10 km, indicating that there is a low probability that the contributions were similar between these two distances. The 250 -500 µm had smaller medians (13% -29%) relatively to the two smaller fractions and only decreased slightly with distance within the fraction. The 500 -2000 µm fraction was highly variable (6% -41%) with distance and had confidence intervals that did not overlap at certain distance.
During the winter in the kelp sparse region, KDD median contributions to plankton were from 4% to 80% for all size fractions ( Figure 4). The 63 -125 µm fraction had a strong contrast with respect to KDD contribution between transects as transect 3 medians decreased, from 80% to 27%, with distance and transect 4 medians increased (from 6% to 47%, Figure 4). In the 125 -250 μm plankton, the median KDD contribution decreased with distance and had no overlap in confidence intervals between 0 and 10 km, while along the transect 4, KDD contributions remained low and constant with distance. The 250 -500 μm plankton showed a similar decline in contribution with distance. The precipitous decline (from 73% to 20%) was observed between 0 and 0.5 km. The 500 -2000 µm fraction had no trends and was highly variable with distance. The median KDD contributions were from 15% to 79% (Figure 4).

Kelp-Derived Detritus Incorporated by Benthic Organisms
Astraea: Paired defatted and untreated Astraea gibberosa samples showed no differences in δ 13 C and δ 15  There was considerable within region δ 13 C variation among sampling sites (df = 7, F = 9, p < 0.001) which did not allow samples from within the same region to be pooled for modeled KDD contribution estimates because it led to bimodal posterior probability distributions. This variation did not exist for δ 15 (Figure 7), which was within the confidence intervals for most summer size-classes. In the kelp sparse region, the

Discussion
Contrary to expectation, regional differences in kelp contribution to primary consumers were not detected despite a 20-fold difference in the giant kelp (Macrocystis pyrifera) abundance between regions. Furthermore, no consistent decrease in kelp contribution to primary consumers with increasing distance from source kelp forests within any region was evident. Nevertheless, during the winter, the KDD contribution to plankton was lower than during the summer with pronounced variability between regions and along transects away from kelp forests. The KDD contribution to benthic organisms was also high in all regions during both summer and winter. The exception was Astraea in the kelp abundant region where KDD contribution was on average more than 3-fold lower than the kelp sparse region.

Plankton δ 13 C Variability
In general, some size fractions of plankton along a particular transect during both summer and winter showed significant decreases in δ 13 C values with the distance from kelp forests. This was expected in the kelp abundant region because of higher kelp production, hence lateral transport and faster dissipation of While, this may reflect the inability of small plankters of this fraction to feed on kelp-derived production, other explanations are likely in play. For example, accumulation of lipids that are 13 C depleted [66] may affect their isotopic ratios [67]. Plankton δ 13 C from the kelp sparse region was more depleted along one transects oceanward for all plankton smaller than 500 µm. This coincides with their food source (surface POM) being more depleted with distance along this transect [27]. On the other hand, along the nearby transect, plankton remained relatively enriched in 13  KDD can support higher densities of consumers at higher trophic levels [4] [20] [52]. In this study, only δ 15 N differences within plankton size fractions were found during the summer between the kelp abundant and absent regions. For example, the 500 -2000 µm size fraction within the kelp abundant region was 1.9‰ enriched compared to <250 µm size fractions. However, this is likely not biologically significant, nor is it prevalent enough to conclude whether or not kelp abundant region plankton are feeding at higher trophic levels. There were no within-region differences among δ 15 N values of plankton size fractions, which inhibits the conclusion that larger size fractions of plankton were feeding at higher trophic levels [74].

Incorporation of Kelp-Derived Detritus by Plankton
KDD contribution to majority size fractions of zooplankton was moderate to high during the summer in all regions. Contrary to our predictions but in agreement with the surface POM, no regional and along transects differences in kelp contribution were evident. The two largest zooplankton size fractions (250 -500 and 500 -2000 µm) showed the largest spatial variability. Furthermore, uncertainties in the KDD contribution estimates were greatest for the kelp sparse region. This, however, could be misleading to how important KDD is in this region because the high uncertainty in the modeled estimates likely resulted from the inclusion of Nereocystis being the major, and possibly the only, kelp species contributing to POM. This species had a mean δ 13 C value of −18.31‰, which was the closest to blooming phytoplankton values of −18.91‰ than in any other region [27]. Similarity in source isotope values leads to uncertainty (i.e. large confidence intervals) in modeled contribution estimates [62]. In general, for a given distance and region, median KDD contributions to plankton were similar to median KDD contributions to POM [27].
During winter, there was a greater spatial variability within regions and pronounced differences in detritus contributions between transects. In general, the KDD contribution was similar during both seasons. Nevertheless, during the winter, kelp contribution was patchier, especially in the kelp sparse region.
These results indicate that the moderate to high KDD contribution to plank- the shelf and upwelled water and particles with depleted 13 C being in the nearshore area depleting the surface POM δ 13 C. Although KDD production may be relatively high in this region it is possible that the retention of this production in the region is reduced by regional oceanography and/or diluted by upwelled, 13 C depleted POM [27].

Incorporation of Kelp-Derived Detritus by Benthic Invertebrates
KDD incorporated by the snail Astraea gibberosa differed greatly within regions and among sites within the kelp abundant region and therefore could not be combined for mixing model estimates. Individuals from Kamils Island had the highest KDD (median 27%) contribution compared to other sites (≤5%). Surprisingly, the KDD contribution to the other two regions was high with a median contribution of 68% -97%. One explanation for these results may be that abundant kelp populations in the kelp abundant region is also associated with abundant understory red algae in the kelp abundant region [78]. Red algae are highly depleted in 13 C (~−30‰; [79]) and could be an abundant food source in the kelp abundant region leading to the low contribution of kelp-derived production to these organisms, or, as some have suggested, there is superiority in mixed diets compared to single foods [80] [81] [82]. Second, the individuals in the kelp abundant region were approximately 2 cm smaller and thus more 13 C depleted. Sea otter populations at equilibrium density are thought to be food limited [83] and changes in prey availability will influence sea otter foraging behavior [84]. Sea otters preferentially feed on sea urchins when available [46] [84]. However, in areas where otters have been established for quite some time large urchins are absent [46] and otters consume less preferred prey that has become more energetically profitable [85] [86]. It is therefore plausible that in the kelp abundant region large Astraea heavily preyed upon and reduced in abundance. Finally, smaller Astraea may eat less kelp tissue but consume diatoms and bacteria on Open Journal of Marine Science kelp blades that are relative 13 C depleted [33] [87] [88]. Also, smaller Astraea could be structurally limited due to smaller or weaker mouth parts and not be able to consume, or possibly even digest kelp, due to its rigidness [31]. In general, modeled estimates of KDD contribution to benthic invertebrates were similar, among regions and between seasons. The incorporation of KDD by Calliostoma was similar and high (>60%) for all regions. This was not the case for Chlamys spp. as kelp sparse region scallops had higher contributions of KDD The surprisingly high contribution of KDD to consumers from the kelp sparse region during the summer and winter may be explained as follows. First, although giant kelp abundance is 20-times higher in the kelp abundant region, there is still an ample standing stock of other kelp species present in the kelp sparse region [78] contributing to this region's POM pool. Second, the δ 13 C values of the most abundant Nereocystis in this region appeared similar to the values of blooming phytoplankton resulted in highly uncertain modeled KDD contributions. Altering the kelp proportions in this region from 100% Nereocystis to include Macrocystis would enrich the kelp δ 13 C value used in the mixing model and potentially decrease the maximum possible contribution of the giant kelp KDD to organisms within this region. Estimated uncertainty was higher in both regions during the winter because kelp values used in the mixing model had larger within region variability. The presence or absence of sea otters indirectly alters kelp forest species composition and grazer-kelp interactions through successional processes [53] [89], and therefore, may affect consumer diets and isotope values. Drift kelp and KDD are known to accumulate on gentle slopes and depressions as well as in areas of low current velocities suggesting importance of local hydrodynamics [90]. Biber [91] reports that areas with low standing stock of drift macroalgae and high currents can still have significant fluxes of drift pass through. These processes can lead to shifts in consumer spatial distributions in response to food availability [90] leading to isotope enrichment and potentially obscuring predicted patterns.
In conclusion, it appears that the local oceanography, kelp forest species composition, kelp species isotope variation, and kelp's similarity with blooming phytoplankton isotope values led to highly variable and at times counterintuitive results leading to uncertainty in the modeled KDD contributions. Nevertheless, in regions with dynamic hydrodynamics (i.e. fast currents, upwellings/downwellings, filament and eddy formation) that characterize narrow shelf regions, the KDD Open Journal of Marine Science contribution to the various pelagic and benthic ecosystems is modestly high irrespective of the sea otter occupation. This is largely driven by the redundancy in the kelp species composition when multi-year Macrocystis species can be replaced by the annual species such as Nereocystis. Our study could uncover the inter-annual variability in the KDD contributions, which may hypothetically be more pronounced in the absence of Macrocystis kelp forests [22]. Ultimately, our findings speak to how widespread or limited the indirect effects of otters (e.g., kelp forest community succession), and other sources of ecosystem change, can be and are helpful in applications of ecosystem based coastal management.