Synergies among stressors drive unanticipated changes to alternative states, yet little has been done to assess whether alleviating one or more contributing stressors may disrupt these interactions. It would be particularly useful to understand whether the synergistic effects of global and local stressors could be alleviated, leading to slower change or faster recovery, if conditions under the control of local management alone were managed (i.e. nutrient pollution).
We utilized field-based mesocosms to manipulate CO2 (i.e. forecasted global concentrations) and nutrients (i.e. local pollution) to test the hypothesis that, where synergies exist, reducing one contributing stressor would limit the effect of the other. In testing this hypothesis, we considered the response of turfing algae, which can displace kelp forests on urbanized coastlines.
Initial manipulations of CO2 and nutrient enrichment produced an anticipated synergistic effect on the biomass of turfing algae.
Following exposure of algal turfs to a combination of enriched nutrients and CO2, a subsequent reduction in nutrients was able to substantially slow further increases in turf growth. Despite this substantial effect, the historical legacy of previous nutrient enrichment was evident as greater turf was maintained relative to ambient conditions (i.e. ambient CO2 and nutrients). Such legacies of past stressors may be stubborn (e.g. persist as intergenerational change) where the alternative state (i.e. turf algae) has substantial resilience to restorative actions.
Synthesis and applications. As stressors accumulate across global to local scales, some combine to produce synergistic effects which cause changes of disproportionate ecological magnitude. While strong synergies attract heavy scrutiny, there remains substantial merit in assessing whether their influence can be ameliorated by managing a contributing stressor. Of note, we show that by reducing a locally determined stressor (nutrients), its synergistic effects with a globally determined stressor (CO2 enrichment) on a key taxon (turf algae) may be substantially reduced. These results suggest that in the face of changing climate (e.g. ocean acidification), the management of local stressors (e.g. water pollution) may have a greater contribution in determining the dominant state than current thinking allows.
Novel environmental conditions created by human activities are, with increasing frequency, transforming ecosystems into new, non-historical habitats (Scheffer et al. 2001; Hobbs, Higgs & Harris 2009 and examples within). Many of these seemingly abrupt shifts are prompted as altered environmental conditions push systems over a threshold (or unstable equilibrium) that marks the border between the ‘basins of attraction’ of alternative states (Holling 1973; Scheffer et al. 2001). The newly established habitats, or alternative states, typically comprise species, interactions and functions perceived to be of less ‘value’ to human societies than those of the system they replaced (Dudgeon et al. 2010). Subsequent ecological restoration commonly aims to mitigate the numerous abiotic changes which led to transformations (i.e. stressors) such that resilience (i.e. the basin of attraction of the novel state) is reduced and systems are pushed back towards their historical arrangement (Suding, Gross & Houseman 2004; Hobbs, Higgs & Harris 2009; Lotze et al. 2011; Morecroft et al. 2012). Defining the success of such restoration is, however, complicated as definitions of ‘historical’ and ‘non-historical’ habitats are rarely clear due to natural variability in ecosystems and the pervasiveness of human influence (Connell et al. 2008; Hobbs, Higgs & Harris 2009). Where human experiences and values lead to the selection of a non-pristine baseline, reaching this management target does not necessarily mean that the system has been restored, but merely that it is improved relative to its novel configuration (Connell et al. 2008). Recent experimental work indicates that while restorative actions can enable recovery to the defined historical habitats in some degraded systems, others are resilient to restoration efforts as they have either undergone a shift to an alternative persistent state with a strong basin of attraction or the dynamic equilibrium between alternative states has shifted (Beisner, Haydon & Cuddington 2003; Suding, Gross & Houseman 2004; Lotze et al. 2011 and examples within). It is the existence of these alternative states and their resilience to restorative actions which underlie many of the difficulties in anticipating the potential for re-establishment of historical habitats (Beisner, Haydon & Cuddington 2003). The permanence of novel habitats and effectiveness of post-disturbance management in restoring historical habitats are therefore poorly understood areas of applied ecology, representing major challenges for research and management.
The approaches that have most successfully returned systems to their defined historical habitats are typically those which simultaneously ameliorate the multiple stressors that caused the initial transition such that the alternative state's basin of attraction is lessened (Scheffer et al. 2001; Beisner, Haydon & Cuddington 2003; Lotze et al. 2011). The management of stressors modified over particularly large spatial and temporal scales, however, may be impossible or infeasible (Solomon et al. 2009), meaning that only a subset of altered conditions can be effectively managed (Falkenberg et al. 2010; Morecroft et al. 2012). Due to the irreversibility and persistence of global climate change, these conditions are anticipated to be the backdrop against which any local-scale change, either degradation or restoration, occurs. Interactions between, and among, global and local stressors created by human activities influence transitions to novel habitats (Hobbs, Higgs & Harris 2009; Pettorelli 2012). Further complicating this process is the potential for interactive effects which range from additive (where the response can be predicted based on the effects of individual stressors) to synergistic or antagonistic (where the response is greater or lesser than would be predicted from adding the independent effects of stressors, respectively) (Crain, Kroeker & Halpern 2008; Darling & Côté 2008). While the role of interactive effects in determining transitions to novel habitats has been considered for a number of combinations of stressors and systems (reviewed in Crain, Kroeker & Halpern 2008), less attention has been given to the influence of such interactions on recovery to the historical habitat. As these interactions are anticipated to drive phase-shifts to alternative states that themselves have substantial resistance to change (i.e. a strong basin of attraction), removing one stressor alone may be insufficient to force a transition back to the original state (Scheffer et al. 2001). It is coming to be recognized, however, that restoring a system to its historical state may not require reversal of change which drove the initial transition (Suding, Gross & Houseman 2004). In terms of synergies, it has been proposed that the effects of these interactions may be disrupted by restoring a subset of the altered conditions (Russell et al. 2009). If such disruption of synergistic interactions is possible, it would indicate the potential for effective local management to facilitate a return to the historical ecosystems despite the irreversibility and persistence of altered global stressors.
Degradation and replacement of historical habitats has occurred in many coastal ecosystems influenced by local-scale human activities, including wetlands, seagrass beds, coral reefs and kelp forests (Bellwood et al. 2004; Lotze et al. 2006; Airoldi & Beck 2007). Temperate coastlines of southern Australia are typically dominated by canopies of long-lived, topographically complex algae (Connell & Irving 2008). In comparison with this baseline condition, which was widely observed until the 1970s when coastal development was accelerated and is still observed in regional areas where human impacts are negligible, sites in southern Australia impacted by nutrient enrichment are characterized by comprehensive loss of kelp canopies and their replacement by mats of turfing algae (Connell et al. 2008; Gorman, Russell & Connell 2009). Such change manifests owing to elevated nutrients which enable the normally ephemeral turfs to persist among fragmented canopies and compete against kelp for space such that recruitment is inhibited (Gorman & Connell 2009) and kelp forests are displaced (Connell & Irving 2008). This persistence and expansion of turfs may be further facilitated by future atmospheric enrichment of CO2 (Connell & Russell 2010; Falkenberg et al. 2010), with the simultaneous enrichment of nutrient and CO2 pollution anticipated to enable a synergistic increase in the spatial cover and biomass of turf (as identified in Russell et al. 2009; Falkenberg, Russell & Connell 2012). Conditions that enhance the potential for turfs to become spatially dominant and reduce the area available for kelp recruitment following its removal by storm events promote transitions from the historical kelp-dominated habitat to the novel one associated with mats of turfing algae. Therefore, where the combination of enriched local (nutrients) and global (CO2) conditions facilitates a synergistic increase in turfs, restoration of the historical habitat will require feedbacks between turf and these stressors to be broken such that its dominance is reduced and space is again available for kelp recruitment.
The aim of this study was therefore to assess whether the alleviation of a local stressor under local governance could effectively slow or reverse the increase in abundance of taxa that are forecasted to dominate under future climate conditions. Specifically, we wanted to determine whether, following exposure to enriched nutrient and CO2 conditions, the change in biomass of a species which characterizes the novel habitat on temperate coastlines would be dampened when nutrients were reduced, but CO2 enrichment was maintained. We hypothesized that where enriched nutrients and CO2 combined to drive a synergistic increase in turf algae, this effect would be limited by reducing elevated nutrients alone.
Materials and methods
Experimental site and set-up
Experimental mesocosms were moored in an open boat harbour protected from the predominant swell by a breakwall adjacent to the Gulf St. Vincent at Outer Harbour, South Australia (34·473395° S, 138·292184° E). The 250-L experimental mesocosms (L × W × H: 0·5 × 0·5 × 1 m; A-cast brand transparent acrylic, Asia Poly, Kuala Lumpur, Malaysia, see Russell, Passarelli & Connell 2011; for spectral properties) were filled with natural seawater pumped directly from the harbour; therefore, initial seawater chemistry (i.e. before experimental manipulation) was characteristic of these waters. During the experimental period, one-third of the seawater was removed from each mesocosm and replaced with fresh seawater weekly to maintain water quality.
This experiment was conducted using turf-forming algae (see Appendix S1 in Supporting Information for the definition of turf used here) initially collected attached to their natural substratum from rocky reef at Horseshoe Reef, Gulf St. Vincent, South Australia (35·13757° S, 138·46266° E). Samples of turf approximately the same size (5 × 5 cm) were placed in field-based holding mesocosms for 8 weeks before the experiment commenced to enable acclimation to mesocosm conditions.
The experiment had two key components which together ran for 12 months from August 2009–August 2010. First, we assessed whether enriched CO2 and nutrients would have a synergistic effect on the biomass of turf algae. To do this, turf algae were subjected to nutrients (ambient: current concentrations adjacent to natural catchments of 0·013 ± 0·001 mg L−1 NOX vs. elevated: concentrations adjacent to urban catchments of 0·232 ± 0·032 mg L−1 NOX; Gorman, Russell & Connell, unpubl. data) and CO2 (current: current ambient of 280–380 ppm vs. future: IS92a model scenario for 2050 of 550–650 ppm) in a crossed design. Three replicate mesocosms were used per treatment combination, with ten turf samples randomly assigned to each experimental mesocosm following the initial acclimation period. Conditions were then gradually altered (as described in the ‘Experimental treatments: nutrient and CO2 addition’ subsection below) over a further 2-week period until they reached pre-designated experimental levels. These initial experimental conditions were maintained for 6 months between August 2009 and February 2010.
In the second phase of the experiment, we assessed whether the change in biomass of algae exposed to the combination of elevated nutrients and future CO2 would be limited if the local-scale factor of nutrients was reduced, while CO2 enrichment was maintained. To do this, the specimens initially exposed to elevated nutrients with future CO2 were reallocated, either to elevated nutrients with future CO2 (i.e. nutrients maintained) or to ambient nutrients with future CO2 (i.e. nutrients reduced). In addition, the control treatment of ambient nutrient with current CO2 was continued to provide a contemporary baseline for biomass under ‘ambient’ conditions. Three replicate mesocosms were used per treatment combination, with replicate specimens of algal turfs in each mesocosm (n =5). These experimental conditions were then maintained for a further 6 months between February and August 2010.
In addition, to determine how closely change in turf cover in the control mesocosms matched that occurring in the field, we compared turf cover in the contemporary control mesocosms with that in the field throughout the experimental period (details in Appendix S2).
Experimental treatments: nutrient and CO2 addition
Nutrients were enriched to concentrations similar to those experienced in waters off the coast of metropolitan Adelaide (target NOX: mean ± SE, 0·232 ± 0·032 mg L−1; measured in laboratory; 0·3796 ± 0·0255, see Table S3 for further detail) using Osmocote Plus® (Scotts, Australia) controlled release fertilizer (6-month release: 15, 5, 10 N-P-K). Osmocote pellets (10 g per mesocosm) were placed in a nylon mesh bag (1-mm mesh size) and attached to the bottom of each appropriate mesocosm (i.e. those of elevated nutrients; in the ambient and reduced nutrient treatments, nutrients were simply not added). The concentration of supplied nutrients was quantified by regularly collecting water samples using 25-mL sterile syringes, which were filtered (0·45-μm glass fibre) and immediately frozen. Samples were later analysed on a Lachat Quickchem 8500 Flow Injection Analyser (Hach, CO, USA) for ammonia, phosphate and nitrite + nitrate (NOX) (for results, see Appendix S4 and Table S3, S4). Additionally, to quantify the effect of elevated nutrients in the absence of biota, a trial was conducted whereby 10 mesocosms identical to the field mesocosms were established in the laboratory and maintained for 5 weeks between March and April 2011. Using the same method as in the field, 10 g of Osmocote was added to half of these mesocosms, with water samples regularly collected and analysed from all mesocosms (for results, see Appendix S4 and Table S3, S4).
Target CO2 was based on the current ambient (current: 280–380 ppm) and the IS92a model scenario for atmospheric CO2 concentrations in the year 2050 (future: 550–650 ppm). The pH of mesocosms exposed to the future CO2 treatment was reduced from ambient (mean ± SE: 8·17 ± 0·02) to the experimental level (target: 7·95; measured: mean ± SE; 7·94 ± 0·01, see Table S3). The CO2 concentration of seawater within mesocosms was maintained by directly diffusing CO2 gas into the water column as required to maintain the experimental level and was controlled using temperature-compensated pH probes and automatic solenoid controllers (Sera, Heinsberg, Germany). Total alkalinity (AT) of seawater in mesocosms was measured weekly using colorimetric titration (Hanna Instruments, Woonsocket, RI, USA). Concentrations of pCO2 and bicarbonate (HCO3−) were then calculated from measured AT, pH, salinity and temperature using CO2SYS for Excel (Pierrot, Lewis & Wallace 2006) with constants from Mehrbach et al. (1973), as adjusted by Dickson & Millero (1987) (results summarized in Appendix S3 and Table S3, S4).
Change in percentage cover of turf was calculated for the first experimental phase (i.e. final – initial percentage cover; August 2009–February 2010) by quantifying the percentage cover of turf at both time points by overlaying a 2·5 × 2·5 cm quadrat over each algal specimen, within which the percentage cover was visually estimated to the nearest 5 per cent (Drummond & Connell 2005). In addition, we also quantified the final mass following the initial experimental period (i.e. February 2010) by carefully scraping all turf biomass from a standard area of the individual specimens (1 × 1 cm) using a razor into a pre-weighed aluminium tray, rinsing with fresh water to remove excess salt and dried to a constant weight at 60 °C for 48 h before weighing.
In order to quantify the change in turf biomass over time following nutrient reduction, we measured the change in fresh weight from the time treatments were altered at the start of phase two (February 2010) until the end of the experimental period (August 2010). Change in fresh weight was quantified by gently patting the specimens dry and then weighing them to the nearest 0·01 g using an electronic balance.
The response of algal turfs to experimental conditions was analysed using analysis of variance (anova). A two-way anova was used to test the effect on the change in percentage cover and dry mass of turf algae following the initial enrichment, while a repeated-measures (mixed split-plot design) anova was used to test the change in fresh weight of turfs over time following reallocation of samples and implementation of the nutrient reduced treatment. The water column physicochemical parameters were also analysed using anovas. Where significant treatment effects were detected, post hoc comparison of means was used to determine which factors differed (details of the specific anovas and post hoc comparisons are provided in Appendix S3).
Following enrichment, the greatest change in turf cover and dry mass was observed when elevated nutrients and future CO2 were experienced in combination (Fig. 1a and 1b; Table S1a and S1b). This treatment caused turf cover and biomass to increase synergistically, that is, by a greater magnitude than would be anticipated by adding their isolated effects (Fig. 1a and 1b). Specifically, the detectable effects of nutrients in the absence of future CO2 (i.e. elevated nutrients – ambient nutrients under current CO2 = 14%, 0·004 g) and the effects of CO2 in the absence of elevated nutrients (i.e. future CO2 – current CO2 under ambient nutrients =2%, 0·001 g) do not add to their combined effect (i.e. future CO2 and elevated nutrients – current CO2 and ambient nutrients =37%, 0·009 g), because their combined effect is multiplicative (i.e. 131 and 80% greater than their additive effects, for change in percentage cover and dry mass, respectively).
When the nutrient reduction treatment was implemented, the change in turf biomass (fresh weight) was significantly affected by a treatment × time interaction (Fig. 2; Table S2). Initially, the change in fresh weight was not significantly different between the treatments (day 22: ambient nutrients, current CO2 = reduced nutrients, future CO2 = elevated nutrients, future CO2; P >0·05, Tukey HSD), but within 2 months all treatments had diverged (day 51: ambient nutrients, current CO2 < reduced nutrients, future CO2 < elevated nutrients, future CO2; P <0·05, Tukey HSD). The fresh weight in the elevated nutrients, future CO2 treatment continued to increase rapidly throughout the experimental period, with the change in this treatment significantly greater than the other two at all but one subsequent measurement times (the exception was day 108, elevated nutrients, future CO2 = reduced nutrients, future CO2, P =0·067, Tukey HSD). While the change in fresh weight was not as substantial in the other treatments (i.e. nutrient reduced, future CO2 and ambient nutrients, current CO2), they also increased consistently, with the fresh weight of turf in the nutrient reduced, future CO2 treatment tracking higher than that of the ambient control (Fig. 2).
Comparison to field conditions
The percentage cover of turf algae in the mesocosms was similar to that quantified in the field at each time point (Fig. S1). In both mesocosms and the field, percentage cover of turf gradually increased throughout the experimental period (Fig. S1).
Interactions among stressors created by human activities have the potential to drive transitions to novel habitats. Observed examples of shifts between alternative states include the switch in terrestrial deserts from perennial vegetation to bare soil with ephemeral plants, lakes from clear water with submerged vegetation to turbid phytoplankton-dominated waters and tropical marine reefs from corals to fleshy macroalgae (Scheffer et al. 2001 and references within). Stressors often interact synergistically to influence key taxa, and also ecosystems, more strongly than would be anticipated based on the addition of their isolated effects (Crain, Kroeker & Halpern 2008). For example, temperature, salinity and ultraviolet radiation combine to increase the embryonic mortality of gastropods (Przeslawski, Davis & Benkendorff 2005), potentially limiting their ability to continue providing their current ecosystem function which is the removal of key primary producers (Lotze, Worm & Sommer 2001; Russell & Connell 2007). Similarly, where these stressors alter the occurrence of key habitat-forming taxa that are involved in strong interactions, their combined effects can hasten transitions to novel habitats (Hobbs, Higgs & Harris 2009). The results of our initial manipulation provide evidence that such interactions may also be prevalent in rocky temperate coastlines when nutrients and CO2 are simultaneously enriched, with the abundance of turf algae increasing synergistically (as identified in Russell et al. 2009; Falkenberg, Russell & Connell 2012). As algal turfs characterize the novel habitat of this system, such an increase would be anticipated to promote transitions away from the historical habitat dominated by kelp canopies.
The influence of multiple stressors on transitions from historical to novel habitats has been heavily discussed (e.g. the influence of overfishing, declining water quality and climate change on the shift from coral to macroalgae; Bellwood et al. 2004). The same is not true, however, for the influence of interactions, especially synergies, on the converse transitions from novel to historical habitats (e.g. the reverse shift from macroalgae to coral, but see Dudgeon et al. 2010). It has been suggested that reversing shifts between habitat states may be particularly difficult where only one of the conditions that drove the initial shift can be restored, as the effect of such restorative actions may be insufficient to weaken the ‘basin of attraction’ of the alternative state. Consequently, we assessed the results of our initial enrichment to gain an indication of the extent to which locally modified nutrient conditions determined the synergistic response of turf algae, and whether nutrient reduction had the potential to limit the expansion of turf even where CO2 enrichment was maintained (c.f. restorative actions after the shift has occurred). The results indicate that while enriched CO2 and nutrients combine to produce a synergistic increase in abundance of turfs, future CO2 would have little effect in the absence of elevated nutrients (as identified in Russell et al. 2009; Falkenberg, Russell & Connell 2012). Specifically, we found that elevated nutrients prompted an increase in turf under both current and future CO2 conditions. In contrast, although future CO2 facilitated an increase in turfs under elevated nutrients, it had little effect where nutrients were maintained at their ambient level. Such a response is characteristic of incremental co-limitation of nutrients and CO2, whereby processes determining growth are firstly restricted by the primarily limiting resource until it is in adequate supply, in this case nutrients, at which time the limiting resource switches to be a second factor, in this case CO2 (Davidson & Howarth 2007; Allgeier, Rosemond & Layman 2011). Based on the results of our initial enrichment, we anticipated therefore that nutrient reduction could disrupt the synergistic effects of preceding nutrient and CO2 enrichment.
The response of turfs to restorative management of nutrient concentrations under maintained enrichment of CO2 supports our suggestion that where these stressors have synergistic effects on turf algae, it will be possible to substantially reduce their effects by restoring a subset of those conditions that were initially altered. Following reduction in nutrients and continuation of CO2 enrichment, the increase in turf biomass was limited relative to that which occurred where both nutrient and CO2 enrichment were maintained. Such a response was likely observed as the reduction in nutrients caused limitation to an extent that continued enrichment of CO2 could not maintain an elevated growth response. This key result indicates that where the factors of nutrients and CO2 interact to influence turfs in a synergistic manner, appropriate local management of nutrients can disrupt the feedbacks that maintain this novel composition, even following the establishment of globally altered CO2 conditions. This result demonstrates the key role of local environmental conditions in determining the response of systems to forecasted global stressors, and highlights the potential for local management, including that which prevents nutrient inputs, to reduce the effect of irreversible global climate change.
While reduction in local stressors following the establishment of altered global conditions could disrupt synergies and limit further change to the system, delayed action may not be as effective as a proactive approach that precludes these interactions. We show that a reduction in nutrients following establishment of enriched CO2 conditions substantially reduced the rate of increase in turf biomass. Of concern, however, is that there appeared to be a legacy from the historical conditions of enriched nutrients because biomass did not reduce to levels quantified under ambient conditions (i.e. ambient nutrients, current CO2). This legacy may represent a positive feedback whereby the greater biomass established under combined nutrient and CO2 enrichment was self-sustaining. Such an effect may be long-lasting and produce intergenerational change where the novel state (e.g. mats of turf algae) has substantial resilience to restorative actions (Scheffer et al. 2001; Beisner, Haydon & Cuddington 2003). Our results suggest, however, that although further expansion of turf algae may be prevented where local management reduces nutrient pollution under future CO2 conditions, the initial expansion of turfs may be avoided where future climates manifest under good local water quality. Consequently, the establishment of effective management of local conditions, such as nutrients, may be most beneficial before forecasted climate conditions become established.
Forecasting the potential effects of anticipated change occurring at both global and local scales currently requires the use of mesocosms that enable manipulation of environmental factors impossible to modify in the field. Conditions within such mesocosms are, however, an imperfect approximation of those in natural ecosystems (Carpenter 1996). As such, limitations are typically placed on the interpretation of results from such experimental approaches (Wernberg, Smale & Thomsen 2012). We had concerns that the sheltered conditions within our mesocosms could minimize the removal of turf associated with water movement and that turf expansion would be greater in mesocosms than in the field. Our data reveals, however, that turf cover in control mesocosms (i.e. ambient nutrients, current CO2) increased at a similar rate and magnitude to turfs in the adjacent Gulf St. Vincent. Therefore, the conditions that strongly influenced turf growth in our mesocosms were likely to be representative of conditions throughout the Gulf St. Vincent during the experimental period. Our mesocosm experiment therefore not only enables comparisons to be made between the responses of turfs under ambient conditions with that of their future counterparts, but also allows confidence in predictions regarding the magnitude of change in turf cover under future management scenarios.
The striking ecosystem shifts that occur where human-driven stressors combine to produce synergistic effects often provide the impression that prevention of further change, or its reversal, will be difficult to achieve (Lotze et al. 2011). To date, effective management of such change has typically involved amelioration of a broad suite of stressors (Lotze et al. 2011). Our results indicate, however, that disrupting the effects of synergies (e.g. between nutrients and CO2 on turf algae) may not actually require all stressors to be alleviated, but rather the local-scale stressor(s) that strongly drives the initial interaction (e.g. nutrients). These findings empower local policy makers (e.g. Department of Environment, Water and Natural Resources) and managers of water quality (e.g. SA Water) who are implementing policy initiatives to decrease nutrient pollution. South Australia is a global leader in the use of ecological sciences to inform policy initiatives that aim to reduce nitrogen loads (i.e. by 75% in coastal waters) in line with improving coastal ecology. Measures implemented to achieve these targets include the increased effectiveness of wastewater treatment plants (i.e. Environment Improvement Plan, SA Water) and development of infrastructure for water recycling (i.e. 27 000 ML per year, or 30% of total wastewater flows). We caution, however, that disrupting the effects of synergies will be far more difficult to achieve where the key stressors result due to human activities over large spatial and temporal scales that are not easily reversed, such as anticipated levels of enriched CO2 (Solomon et al. 2009; Pettorelli 2012).
In conclusion, as environmental conditions are altered across global to local scales, some will combine in synergistic ways to cause change of disproportionate ecological magnitude. We show that by reducing a locally determined stressor (i.e. nutrient pollution), its synergistic effects with a globally determined stressor (i.e. CO2 enrichment) on turf-forming algae may be substantially reduced. Consequently, management actions to reduce the total load of nutrients released into marine systems (e.g. improved wastewater treatment and recycling) may be of particular importance in determining the occurrence of this taxon and the ecological structure of coastal marine systems in southern Australia. The detectable legacy effect where nutrients were removed following the establishment of a synergy with enriched CO2 suggests that proactive management strategies which prevent such interactions may be more effective than approaches to disrupt them. Importantly, these results suggest that in the face of changing climate (e.g. ocean acidification), effective management of local stressors (e.g. water pollution) may have a greater contribution in determining natural habitats than currently anticipated.
We thank volunteers, including members of the Southern Seas Ecology Laboratories, who assisted with the collection of algae and assembly and maintenance of experimental mesocosms. Financial support for this research was provided by an ARC grant to S.D.C. and B.D.R and an APA to L.J.F.