Table 1.
Mean values and results of 1-way ANOVAs of nutrient concentrations for ‘Ambient’ and ‘Enhanced’ nutrient conditions on experimental patch reefs.
Fig 1.
Kelp collapse and recovery at different urchin densities over time.
Percentage cover of canopy-forming algae at a range of sea urchin densities (individuals m-²) on patch reefs initiated as the ‘kelp bed’ (a) and (b), and ‘barrens’ state (c) and (d) in northern Port Phillip Bay, Nov. 2012 to Dec. 2013. Plots (a) and (c) represent reefs with ambient nutrient levels, and (b) and (d) show results for reefs with enhanced nutrient levels.
Table 2.
Results of 1-way fixed effects model I Analysis of Covariance testing the differences in cover of canopy-forming macroalgae at different urchin densities (covariate) and nutrient conditions (ambient vs. enhanced) for the 4 periods of a priori interest across a quarter (3 months), half (6 months), three quarters (9 months) and a full year (13 months).
Fig 2.
Abundance (mean ± SE) of E. radiata recruits on reefs above and below the critical urchin density (4 urchins m-2) for kelp recovery after 13 months (‘Kelp’ and ‘Barrens’ refers to the initial states of reefs). Nutrient enhancement did not influence recruitment and therefore data for ‘nutrient enhanced’ and ‘ambient nutrient’ reefs were pooled for display.
Fig 3.
Dependence of algal species richness and diversity on canopy cover.
Panel a) details cover of canopy-forming algae, b) shows macroalgal species richness, and c) macroalgal species diversity, against sea urchin densities at the end of the 13 month experimental period. Reefs with initial states of ‘kelp bed’ and ‘urchin’ barrens’ are indicated by a grey circle and a white triangle respectively; data are means ± SE. Reefs with enhanced nutrients have been pooled with reefs with ambient nutrient conditions since the addition of fertiliser did not influence response variables (see Tables 2 and 3). Note that species present at the start of the experiment (E. radiata for kelp bed reefs and encrusting red algae for all reefs) were excluded from the analysis. Arrows in (a) show responses to experimental manipulation of sea urchin biomass in kelp beds (thick grey arrows = forward-shift ‘collapse’ from kelp to urchin barrens) and on sea urchin barrens (thin black arrows = reverse-shift ‘recovery’ from urchin barrens back to kelp beds).
Table 3.
Results of 2-way fixed effects model I Analysis of Covariance testing the significance of differences in responses of species richness (a) and Shannon diversity (b) across different urchin densities (covariate), dependent on initial ‘reef state’ (kelp vs. barrens) and nutrient conditions (ambient vs. enhanced) after 13 months of treatment.
Fig 4.
Percentage cover of turf-forming algae versus cover of canopy-formers.
Open circles display reefs with ambient nutrient conditions and filled circles are reefs with enhanced nutrient levels across all experimental patch reefs at the conclusion of the 13 months experiment (the treatment of ‘reef state’ is not indicated). Fitted line represents treatments pooled across ambient and enhanced nutrient reefs (R² = 0.75, y = -0.056x + 4.16, values derived from linear regression with transformation ln(Y)) because a 1-way ANCOVA showed no significant difference in relationships between reefs with enhanced nutrients and those with ambient nutrient levels (homogeneity of slopes: F1,24 = 0.85, P = 0.36; test between treatments after factoring for the covariate, F1,25 = 1.00, P = 0.33).