The functional effects of a dominant consumer are altered following the loss of a dominant producer

Abstract Human impacts on ecosystems are resulting in unprecedented rates of biodiversity loss worldwide. The loss of species results in the loss of the multiple roles that each species plays or functions (i.e., “ecosystem multifunctionality”) that it provides. A more comprehensive understanding of the effects of species on ecosystem multifunctionality is necessary for assessing the ecological impacts of species loss. We studied the effects of two dominant intertidal species, a primary producer (the seaweed Neorhodomela oregona) and a consumer (the shellfish Mytilus trossulus), on 12 ecosystem functions in a coastal ecosystem, both in undisturbed tide pools and following the removal of the dominant producer. We modified analytical methods used in biodiversity–multifunctionality studies to investigate the potential effects of individual dominant species on ecosystem function. The effects of the two dominant species from different trophic levels tended to differ in directionality (+/−) consistently (92% of the time) across the 12 individual functions considered. Using averaging and multiple threshold approaches, we found that the dominant consumer—but not the dominant producer—was associated with ecosystem multifunctionality. Additionally, the relationship between abundance and multifunctionality differed depending on whether the dominant producer was present, with a negative relationship between the dominant consumer and ecosystem function with the dominant producer present compared to a non‐significant, positive trend where the producer had been removed. Our findings suggest that interactions among dominant species can drive ecosystem function. The results of this study highlight the utility of methods previously used in biodiversity‐focused research for studying functional contributions of individual species, as well as the importance of species abundance and identity in driving ecosystem multifunctionality, in the context of species loss.


| INTRODUC TI ON
Global change is driving biodiversity loss worldwide, making it more important than ever to understand the different roles that individual species play in ecosystems (Bellard et al., 2012;Mantyka-Pringle et al., 2012;Valiente-Banuet et al., 2015). Whereas most previous biodiversity research focused on the effects of species loss on one ecosystem function (e.g., productivity; Cardinale et al., 2007), it is important to recognize that species simultaneously mediate multiple functions (Gamfeldt et al., 2008;Hector & Bagchi, 2007).
Quantifying the role of a species in an ecosystem-and understanding the functional consequences of loss-requires evaluating that species' simultaneous contributions to multiple ecosystem functions (e.g., net primary productivity, decomposition, nutrient cycling), also known as "ecosystem multifunctionality" (Manning et al., 2018).
Much of the multifunctionality research conducted to date has focused on the effect of community-level biodiversity on ecosystem functions (Tolkkinen et al., 2013). Community diversity has been shown to strongly influence ecological function, both at the scale of single functions and overall multifunctionality within an ecosystem (Hector & Bagchi, 2007;Zavaleta et al., 2010). Researchers have identified a combination of sampling and species identity effects, by which individual species, rather than the number of species per se, are the primary drivers of the biodiversity-multifunctionality relationship (Brun et al., 2022;Cardinale et al., 2006;Slade et al., 2017).
Individual species, particularly those that are highly abundant in an ecosystem, have emerged as potential key drivers of ecosystem multifunctionality (Fields & Silbiger, 2022;Hillebrand et al., 2008;Lohbeck et al., 2016). Applying methodologies designed for biodiversitymultifunctionality studies (Byrnes et al., 2014) may allow us to further elucidate the functional effects of numerically dominant species. Dominant species may serve as primary drivers of ecosystem function or, if they are weak functional contributors, potentially limit ecosystem multifunctionality Orwin et al., 2014;Wohlgemuth et al., 2016). Dominant species, defined based on their abundance (e.g., >12% relative abundance in community; Mariotte et al., 2015), display a wide variety of forms across ecosystems, from the northern red oak (Quercus rubra) in the forests of the northeastern United States  to red oat grass (Themeda triandra) in the shrublands of South Africa (Cowling, 1983).
The more abundant a species is in an ecosystem, the more likely it is to significantly influence local environmental conditions and overall ecosystem function (Brun et al., 2022;Ellison, 2019;Lohbeck et al., 2016;Tolkkinen et al., 2013;Wohlgemuth et al., 2016). This phenomenon is typified by the "mass ratio hypothesis," which states that the functional traits of dominant species in an ecosystem will strongly influence ecosystem processes (Grime, 1998;Orwin et al., 2014). Understanding how dominant species contribute to ecosystem function, as well as the possibility that they limit overall ecosystem function by crowding out other species (Altieri et al., 2009;Tingley et al., 2002), is critical for understanding how climate change and biodiversity loss will impact ecological function (Giling et al., 2019;Hillebrand et al., 2008;Tolkkinen et al., 2013).
Many ecosystems contain multiple dominant, foundation, and/ or habitat-forming species, and the interactions between these species may affect ecosystem functioning (Angelini et al., 2011;Austin et al., 2021). Altieri et al. (2007) documented interactions between dominant species on cobble-beaches: where cordgrass aggregations and ribbed mussel beds overlap, they interact to produce a shaded, wave-sheltered habitat that supports higher species diversity than the surrounding area. The functional complementarity of some pairs of dominant species, as well as the potential facilitation of one dominant species by another (Angelini et al., 2011), raises the question of how an ecosystem would be affected by the loss of one of multiple dominant species present (Angelini & Silliman, 2014). If the dominant species compete (e.g., for space; Yakovis et al., 2008), have a facilitative relationship (e.g., through complementary nutrient cycling; Aquilino et al., 2009), or exert an interactive effect on the ecosystem (e.g., by forming complex habitat; Altieri et al., 2007), the loss of one species may affect the other dominant species and ultimately ecosystem function. In this study, we investigated the contributions of, and potential interactions between, a pair of dominant species-the algal producer Neorhodomela oregona and bivalve consumer Mytilus trossulus-to critical functions in coastal ecosystems.
Many of the key ecological processes in coastal ecosystems can be grouped into three sets of functions: productivity, nutrient cycling, and effects on water chemistry (Tolkkinen et al., 2013). Primary productivity is the fixation of carbon via photosynthesis and can be measured though oxygen production and related chemical fluxes (Bracken & Williams, 2013). Primary productivity has been strongly associated with the functional traits of dominant species (Bruno et al., 2006;Mouillot et al., 2011), raising the possibility that the association between biodiversity and productivity is predominantly an effect of these abundant, functionally unique species being included more frequently in more biodiverse samples (i.e., sampling effect; Aarssen, 1997;Huston, 1997).
Primary production, itself, can be limited by nutrient availability (Bruno et al., 2006), which positions the cycling of ammonium, nitrate, nitrite, and phosphate as critical to the overall functionality of coastal ecosystems (Bracken & Williams, 2013;Vanni, 2002).
While nitrate and phosphate can reach high concentrations in coastal waters, ammonium-which is typically at low concentrations in seawater due to preferential uptake-often accumulates in tide pools, due to excretion by invertebrates (Aquilino et al., 2009;Bracken & Nielsen, 2004;Bracken & Williams, 2013). Local-scale accumulation of ammonium and phosphate in coastal ecosystems has been directly tied to the abundance of mussels (Asmus et al., 1995;Bracken & Nielsen, 2004), which corroborates findings that nutrientlimited seaweeds are more abundant and grow more rapidly on mussel beds than on other intertidal surfaces (Aquilino et al., 2009;Bracken, 2004). The dominance of different species in otherwise similar communities can lead to divergence in nutrient cycling rates among communities (Bracken & Williams, 2013;Wohlgemuth et al., 2016). Because seaweeds can account for most of the primary productivity in temperate coastal ecosystems (Mann, 1973) and can strongly influence nutrient fluxes in these ecosystems (Bracken & Nielsen, 2004), understanding the contributions of dominant seaweeds to individual ecosystem functions and ecosystem multifunctionality is critical for anticipating impacts of ongoing species loss.
Dominant species in coastal ecosystems may drive changes in other characteristics of water chemistry, with implications for rates of ocean acidification (Aiuppa et al., 2021;Kroeker et al., 2013).
Marine producers can raise seawater pH via photosynthesis (Bracken et al., 2018) as well as increase pH variation over diel cycles, which may help mitigate local-scale acidification in marine ecosystems (Camp et al., 2016;Wahl et al., 2018). However, producers may also reduce pH in the absence of light, when photosynthesis ceases but respiration continues, shifting the balance from a reduction of inorganic carbon in the water column to a net increase and contributing to further acidification (Krause-Jensen et al., 2015;Mahanes et al., 2022;. Producer-driven changes in pH can affect other species in the ecosystem, particularly calcifying species (e.g., mussels and oysters; Semesi et al., 2009;Wahl et al., 2018), which are disproportionately impacted because calcification, the process in which organisms absorb calcium carbonate from the water column to build body structures, can be reduced at low pH (Kroeker et al., 2013). Acidification shifts the chemical equilibrium toward calcium carbonate dissolution, raising the metabolic cost of calcification for organisms or preventing calcification altogether (Andersson & Gledhill, 2013); therefore, robustly photosynthetic species can serve an important function by raising seawater pH.
We assessed the effects of dominant species from different trophic levels on individual ecosystem functions, groups of functions, and overall multifunctionality in coastal systems, both when acting in concert and after simulated species loss. We conducted a removal experiment on the dominant algal producer N. oregona in tide pools where the mussel M. trossulus was also highly abundant, and we subsequently applied a methodology from biodiversitymultifunctionality studies to measurements of 12 ecological functions. Based on the results of past studies on comparable seaweed and mussel species (e.g., Mahanes et al., 2022), we predicted that the dominant producer species would contribute to ecosystem productivity, raise pH, increase calcification, and drive nutrient absorption, while the dominant consumer was expected to increase respiration, reduce pH, increase calcification, and drive nutrient accumulation.

| Study site
We studied effects of the dominant Oregon pine seaweed  (Mahanes et al., 2022). N. oregona is common in tide pools throughout Southeast Alaska, and its range spans the North Pacific from California to parts of Japan and Russia (Lindeberg & Lindstrom, 2016). M. trossulus is a sessile mussel species, generally smaller than its relatives M. californianus and M. galloprovincialis, which can form dense aggregations and is commonly found along the coastline from California to Alaska, USA (Braby & Somero, 2006).

Mytilus trossulus is a dominant species in tide pools at John Brown's
Beach, accounting for >30% of tide pool surface area (Mahanes et al., 2022). The coexistence of these two species provided an opportunity to investigate the effects of and interactions between two numerically dominant species across a set of tide pools which function as individual, largely self-contained ecosystems when isolated during low tide (Sorte & Bracken, 2015). To quantify the degree to which a dominant producer and a dominant consumer drive ecological function, we conducted a species-removal experiment at our study site from July 5 to July 19, 2019.

| Tide pool physical characteristics
We selected 10 tide pools with similar dimensions and tide height (i.e., position within the intertidal zone) for this study. We measured the physical characteristics of the tide pools by: (1) pumping the water from a tide pool into a graduated bucket to assess volume, (2) placing a flexible mesh quadrat with 10 cm × 10 cm squares on the bottom of each tide pool to measure basal surface area (Bracken & Nielsen, 2004;Sorte & Bracken, 2015), and (3) using a sight level and a surveying rod to gauge tide height in meters (above mean lower-low water). We assigned experimental treatments to the tide pools by repeatedly randomizing assignments until volume, surface area, tide height, N. oregona abundance (calculated as percent cover), and species richness (calculated from community survey data, see below) did not vary between treatments (N = 5, removal or control, based on a generalized linear model with threshold of p > .2). The abundance of N. oregona, M. trossulus, and all other species present was assessed via biodiversity surveys following methods used by Bracken and Nielsen (2004; Appendix S1).

| Ecosystem function data collection
We measured the net community productivity and respiration, as well as day and night rates of net ecosystem calcification and pH change, and the fluxes of ammonium, nitrate and nitrite, and phosphate in the experimental tide pools during both day and night. We conducted light/dark productivity trials, as well as time-series water samplings during the day and night, on the unmanipulated experimental tide pools between July 9 and 12, 2019 (for a timeline of the experiment and sampling, see Figure A1). On July 13, we initiated the manipulations and removed N. oregona from the removal treatment tide pools with scissors, cutting as close to the holdfast as possible without damaging any surrounding species. We then repeated the productivity trials and water samplings on the full set of tide pools between July 14 and 16, 2019 ( Figure A1).

| Light/dark productivity trials
To assess impacts of these dominant species on the productivity of the tide pools, we conducted light/dark incubation experiments before and after the removal of N. oregona Noël et al., 2010;Sorte & Bracken, 2015; Figure A1). We took initial dissolved oxygen measurements from each tide pool with a ProDSS Multiparameter Water Quality Meter (YSI). We then covered each pool with an opaque, black tarp for 30 min of dark incubation. We repeated the measurements and then removed the tarps for a 30 min light-incubation period, at the end of which we took a third and final set of measurements.

| Water sample collection
To assess impacts of these dominant species on tide pool water chemistry and nutrient fluxes, we conducted paired time-series samplings (day and night) before and after N. oregona removal ( Figure A1). We sampled across three time points over a ~2.5 h time series following isolation of the tide pools from the ocean, collecting water chemistry samples at each time point  by hand-pumping 250 mL of water from the bottom of the tide pool into a vacuum flask, and then siphoning the water into two 125 mL amber glass sample bottles to minimize gas exchange. We added the remaining water to a 50 mL plastic tube for nutrient analysis.
All containers were rinsed three times with seawater before use.
We immediately added 60 μL HgCl 2 to preserve each 125 mL water chemistry sample and then sealed the sample bottles for later pH and total alkalinity analysis. Nutrient samples were stored on ice while in the field and then frozen at −20°C prior to analysis.
At each time point, we also measured salinity and temperature with a ProDSS Multiparameter Water Quality Meter (YSI) and light intensity with a MQ-210 Underwater Quantum Meter (Apogee) in each pool. Salinity and temperature data were collected for later use in calculating pH values, and light was recorded to document any changes in weather between sampling dates that might affect biological processes. Samples were processed for pH and total alkalinity according to protocols outlined by Dickson et al. (2007) and nutrient concentrations were analyzed using methods of Bracken et al. (2018; Appendix S1).

| Calculated metrics
We calculated rates of change (i.e., slopes) for all water chemistry metrics collected over the three-sample time series, which included pH, ammonium, phosphate, and nitrate + nitrite. We treated day and night rates of change of each function separately because organisms, particularly producers, may affect these factors differently based on the presence or absence of light. We calculated calcification rate using the formula below .
where ∆TA is the change in total alkalinity between the first and third time points in the sampling (mmol kg −1 ), ρ is the density of seawater (1023 kg m −3 ), V is the water volume of the tide pool (m 3 ), SA is the bottom surface area of the tide pool (m 2 ), and t is the time elapsed (h).
The 2 is included because a single mole of CaCO 3 is formed for every two moles of TA.
We used the dissolved oxygen measurements from the light/ dark experiments to calculate net community productivity (NCP) and respiration (R) in the tide pools according to the formulas below (Noël et al., 2010;Sorte & Bracken, 2015).
In the formulas, ∆[O 2 ] is the change in dissolved oxygen concentration (mg O 2 L −1 ), ∆t indicates change in time, and "dark" and "light" correspond to the covered and uncovered incubation periods, respectively.

| Analyses
All statistical analyses were conducted in R (R-version 4.0.4; R Core Team, 2013) using linear models (lm), mixed-effects models (lmer), and the multifunc package (Byrnes et al., 2014). We applied Relationships between the abundances of the dominant producer (the alga Neorhodomela oregona; green, open symbols and dashed regression lines) and consumer (the mussel Mytilus trossulus; blue, closed symbols and solid regression lines) and 12 individual ecosystem functions: (a) net community production; daytime (b) net ecosystem calcification and (c) pH change; (d) community respiration; nighttime (e) net ecosystem calcification and (f) pH change; daytime (g) ammonium accumulation, (h) nitrate + nitrite uptake, and (i) phosphate uptake; and nighttime (j) ammonium accumulation, (k), nitrate + nitrite uptake, and (l) phosphate uptake. Producer abundance was related to two functions: daytime net ecosystem calcification and respiration. Consumer abundance was related to four functions: net community productivity, daytime net ecosystem calcification, respiration. Each data point represents the abundance of producer (green) or consumer (blue) in a single tide pool. Asterisks indicate significance, NS indicates non-significance, and shaded areas are 95% confidence intervals.
tide pool during a phase of the experiment (pre-removal or postremoval). We used the averaging approach on all 12 functions combined as well as subsets of functions, including productivity (net primary productivity and respiration), water chemistry (the rate of pH change and net calcification; both during day and night for four total responses), and nutrient cycling (fluxes of nitrate and F I G U R E 2 Relationships between the abundances of a dominant producer (green) and a dominant consumer (blue) on averaged rates of (a) overall ecosystem functions as well as groups of functions including change in water chemistry during the (b) day and (c) night, (d) productivity, and change in nutrient levels during the (e) day and (f) night. Abundances of neither the producer N. oregona nor the mussel M. trossulus were associated with averaged overall ecosystem multifunctionality (the mean of all 12 standardized function values). Dominant consumer abundance, however, showed a positive association with productivity and a negative correlation with daytime water chemistry. The average function of each pool (N = 10) is represented in each plot by two points, corresponding to the abundance of the dominant consumer (in blue) and the dominant producer (in green). Asterisks indicate significance, NS indicates non-significance, and shaded areas are the 95% confidence interval.
F I G U R E 3 Number of functions exceeded by the (a) dominant producer and (b) dominant consumer based on multiple thresholds to evaluate effects on ecosystem multifunctionality in intact (unmanipulated) tide pools. The abundance of a dominant producer, the alga Neorhodomela oregona, was not related to ecosystem multifunctionality, whereas abundance of a dominant consumer, the mussel Mytilus trossulus, was positively associated with ecosystem multifunctionality across a wide range of thresholds. Each line indicates the relationship between species abundance and the number of ecosystem functions exceeding a threshold value (indicated by color based on the gradient to the right). Asterisks indicate significance and NS indicates non-significance. nitrite, ammonium, and phosphate; each during day and night for six total metrics).
We also used the standardized data to determine the number of functions in each pool which exceeded the set threshold (Zavaleta et al., 2010), as well as expanded that approach to include all possible thresholds from 5% to 99% (Byrnes et al., 2014). In this multiple threshold approach, the output is the range of potential thresholds for which there is a significant effect of the driver-in this case either dominant producer or dominant consumer abundance-on the number functions exceeding the threshold. A strong dominant species effect is indicated when there is a wide range of thresholds at which its abundance is important in determining the degree of multifunctionality (i.e., the number of functions exceeding a threshold) while a narrow band of significance indicates a weak or negligible effect.
In the analyses on individual functions, averaged functions, and multiple thresholds, we assigned directionality to the response metrics to align with the predicted effects of a dominant producer during the day: higher NCP and respiration were indicated by more positive values, as were higher rates of ecosystem calcification, more positive rates of pH change, and greater nutrient uptake (Table A1).
In a second analysis, we repeated the averaging and threshold cal-  Table A3).
We found that ecosystem multifunctionality was associated with dominant consumer abundance, but not dominant producer abundance, in unmanipulated tide pools using the multiple threshold approach (Figure 3). The abundance of the dominant consumer was positively associated with ecosystem function by the multiple threshold approach over two distinct ranges of thresholds (threshold values 51%-56%, 64%-77%; p < .05). In those same tide pools, TA B L E 1 Relationships between the abundances of the dominant producer (the alga Neorhodomela oregona) and consumer (the mussel Mytilus trossulus) and 12 individual ecosystem functions: net community production; daytime net ecosystem calcification and pH change; community respiration; nighttime net ecosystem calcification and pH change; daytime ammonium accumulation, nitrate + nitrite uptake, and phosphate uptake; and nighttime ammonium accumulation, nitrate + nitrite uptake, and phosphate uptake. the dominant producer was not associated with ecosystem multifunctionality (p > .1), though the relationship between producer abundance and multifunctionality tended to be negative across thresholds. Results for identical analyses using the reflected data are shown in Figures A2, A4, and A5.  Table A4).

| Impact of dominant producer removal on the functional effect of the dominant consumer
The relationship between dominant consumer abundance and ecosystem multifunctionality, as assessed using the multiple threshold approach, differed depending on whether the dominant producer was present (Figure 4). In the experimental tide pools, dominant consumer abundance was negatively related to ecosystem multifunctionality over a narrow band of thresholds where the dominant producer was present (threshold values 5%-23%; p < .05), while the relationships between consumer abundance and multifunctionality tended to be positive in the pools where the producer had been removed (NS; p > .2). Results for analyses on the reflected data are shown in Figure A6.

| DISCUSS ION
We found that the relationships between the abundances of each dominant species and individual ecosystem functions, as well as groups of functions, were consistently in opposing directions. This pattern may reflect the differing roles of producers and consumers in supporting overall ecosystem function, in which different trophic levels tend to contribute to certain functions, or types of functions, in specific ways (e.g., producers raising pH during the day or absorbing nutrients; Aquilino et al., 2009;Bracken et al., 2018). However, dominant consumer abundance was related to many of the functions in the direction predicted to be associated with a producer.
This producer-like effect of the dominant consumer may reflect an indirect effect in which the consumer is affecting ecosystem function through facilitation of non-dominant producers (Aquilino et al., 2009), the total abundance of which was found to be positively related to dominant consumer abundance (F 1,8 = 6.12, p = .038). This potential indirect effect on ecosystem function by a sessile, filterfeeding consumer may differ from that of mobile, herbivorous consumers, which may more strongly impact producers via herbivory, or conversely, herbivores may preferentially consume the dominant producer and enable other producers to flourish (Altieri et al., 2009).
The opposing effects of N. oregona and M. trossulus may be more specifically indicative of the well-documented interactions between tide pool algae and mussels, particularly in terms of nutrient cycling (Bracken & Nielsen, 2004;Pfister, 2007). Either way, the nearly uniform counter-directionality of effects between these two dominant species suggests an ecological equilibrium, maintained by the presence of both species, which may be disrupted if one species is lost.
Interestingly, we found that there was a directional change in the relationship between dominant consumer abundance and ecosystem multifunctionality, from positive during the pre-removal sampling to negative in the control pools in the post-removal sampling (i.e., with the dominant producer still present; Figure A1). This direc- We did not find a comparable effect using the averaging method on un-reflected data, either on groups of functions or overall, which may be due to methodological differences between the averaging and multiple threshold approaches: the multiple threshold method is weighted toward consistent baseline levels across functions, rather than exceptionally high levels of individual functions which may elevate the overall average (Manning et al., 2018). The conclusions drawn from the results of either approach may be limited in their scope due to the relatively small sample size of the experiment. Our reasoning for grouping certain functions together is that related functions may be similarly associated with species abundances. Studies have shown, for example, that calcification rates tend to be higher in relatively high-pH conditions (Semesi et al., 2009;Wahl et al., 2018). However, there may be intergroup interactions occurring among ecosystem functions as well: respiration can directly affect pH by modifying CO 2 levels (Krause-Jensen et al., 2015), and productivity and respiration may be intertwined with nutrient cycling due to potential oxygen limitation of nitrification (Joo et al., 2005;Pfister & Altabet, 2019).
We focused on the un-reflected data but included identical analyses on the reflected data in the supplement for additional context ( Figures A2, A4-A6). The rationale for reflecting the data, where necessary, to produce a positive slope with dominant producer abundance in unmanipulated tide pools was to ensure that significant effects, overall or in groups of functions, were not being obscured by opposing effects. We found this to be the case with dominant producer abundance and daytime nutrient fluxes in intact tide pools: both ammonium and phosphate accumulation tended to be more positive in pools with greater dominant producer abundance, while nitrate and nitrite tended to accumulate more slowly in those pools, resulting in an association between dominant producer abundance and daytime nutrient function in unmanipulated tide pools with the reflected data but no corresponding effect in the un-reflected data.
We used approaches designed for evaluating diversitymultifunctionality relationships to focus on the effects of dominant species on multifunctionality in tide pools, but the methods em-  et al., 2015), stabilize food webs (Shao et al., 2016), and impact community composition (Bracken & Low, 2012). Considerable effort has been devoted to identifying species which drive critical functions in ecosystems, including keystone species (Paine, 1966), foundation species (Ellison, 2019;Fields & Silbiger, 2022), and ecosystem engineers (Losapio et al., 2021). Dominant species may have similarly substantial impacts on the ecosystem by virtue of their abundance (Grime, 1998;Orwin et al., 2014),

ACK N OWLED G M ENTS
We thank G. Bernatchez, Z. Danielson, G. Gallaher, S. Mastroni, and R. Rangel for assistance with field work and/or laboratory analyses, as well as members of the Sorte Lab at UC Irvine and J. Martiny for feedback throughout the project. We also thank the Sitka Sound Science Center, the University of Alaska Southeast Sitka Campus, and the US Coast Guard-Air Station Sitka for support during data collection. Funding was provided by the National Science Foundation (OCE-1756173 to CS and MB). We acknowledge that the field portion of this study was conducted on the unceded lands of the Tlingit people.

CO N FLI C T O F I NTER E S T S TATEM ENT
The authors declare that there are no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets used in this study are available on the Dryad Digital

TA B L E A 2
Analysis of the associations between the abundance of a dominant alga (Neorhodomela oregona) and averaged sets of ecosystem functions (all 12 functions, followed by subsets of related functions) in N = 10 unmanipulated tide pools near Sitka, AK. Note: Dominant consumer abundance was negatively associated with daytime water chemistry and daytime nutrient function across both treatments and the effect did not differ between treatment groups. The significant values have been bolded.

F I G U R E A 1
The data used in this project were collected from N = 10 tide pools at John Brown's Beach near Sitka, Alaska, USA during a 14-day time period in July 2019. Water sampling was conducted as close as possible prior to and immediately following the removal of a dominant producer, the alga Neorhodomela oregona, to minimize the possibility for uncontrolled factors (such as changing weather patterns) to influence measurements.

F I G U R E A 2
Relationships between the abundances of a dominant consumer (blue) and a dominant producer (green) on averaged rates of (a) overall ecosystem functions, change in water chemistry during the (b) day and (c) night, (d) productivity, and change in nutrient levels during the (e) day and (f) night, using data that have been reflected to establish positive directionality for the relationship between each function and Neorhodomela oregona abundance. The abundance of a dominant consumer, the mussel Mytilus trossulus, was (a) negatively associated with averaged overall ecosystem multifunctionality, driven by negative relationships with (d) averaged productivity, (b) water chemistry, and (e) nutrient function during the day. The average function of each pool is represented in each plot by a pair of points, corresponding to the abundance of the dominant consumer (in blue) and the dominant producer (in green) in that tide pool. Algal (N. oregona) abundance was (a) associated with averaged overall ecosystem multifunctionality (the mean of all 12 standardized function values) in N = 10 unmanipulated tide pools, driven most strongly by (e) averaged nutrient function during the day (the mean of the standardized daytime function values of the three nutrients responses). Asterisks indicate significance, NS indicates non-significance, and shaded areas represent a 95% confidence interval.

F I G U R E A 3
Relationships between the abundances of a dominant consumer on averaged rates of ecosystem functions including (a) overall function, change in water chemistry during the (b) day and (c) night, (d) productivity, and change in nutrient levels during the (e) day and (f) night, separated by treatment group: control (dominant producer present; blue, circles and solid regression lines) and removal (dominant producer removed; blue, triangles and dotted regression lines). Following removal of a dominant alga (Neorhodomela oregona), the abundance of mussels (Mytilus trossulus) was negatively associated with (b) daytime water chemistry and (e) daytime nutrient function across both treatments. The effect of M. trossulus did not differ between tide pools where N. oregona was present and pools where N. oregona had been removed. Asterisks indicate significance, NS indicates non-significance, and shaded areas correspond to a 95% confidence interval.

F I G U R E A 5
The multiple threshold approach, using data that have been reflected to establish positive directionality between individual functions and Neorhodomela oregona abundance, showed the abundance of a dominant producer to be (a) positively associated with ecosystem multifunctionality in tide pools. The abundance of a dominant consumer, the mussel Mytilus trossulus, was (b) negatively associated with ecosystem multifunctionality using the same method. Each line indicates the relationship between target species abundance in each tide pool and the number of ecosystem functions in that pool which exceed a certain threshold value, with asterisks included to indicate significance.

F I G U R E A 4
Relationships between the abundances of a dominant consumer in two treatment groups, control (dominant producer present; blue, circles and solid regression lines) and removal (dominant producer removed; blue, triangles and dotted regression lines), on averaged rates of ecosystem functions including (a) overall function, change in water chemistry during the (b) day and (c) night, (d) productivity, and change in nutrient levels during the (e) day and (f) night, using data that have been reflected to ensure positive relationships between each function and dominant producer abundance. Following removal of a dominant alga (Neorhodomela oregona), mussel (Mytilus trossulus) abundance was (b) associated with daytime water chemistry, but the effect of M. trossulus did not differ between tide pools where the dominant producer was present and pools where it had been removed. Asterisks indicate significance, NS indicates non-significance, and the shaded areas represent a 95% confidence interval.

F I G U R E A 6
Following the elimination of a dominant alga (Neorhodomela oregona) from the removal tide pools, mussel (Mytilus trossulus) abundance tended to (a) increase multifunctionality in the control group pools (with Neorhodomela oregona still present) but was (b) negatively associated with ecosystem function in the removal group across a small range of thresholds (using reflected data with positive directionality between individual ecosystem functions and N. oregona abundance). These analyses follow the multiple threshold approach, where each line indicates the relationship between Mytilus trossulus abundance in each tide pool and the number of ecosystem functions in that pool which exceed a certain threshold value, with asterisks indicating significance and NS indicating non-significance.