Warming oceans increase feeding pressure on seaweeds that build habitats

A general poleward shift in distribution of marine species due to ocean warming was forecast decades ago^1–4. Since then, numerous studies have documented unusual and increasing occurrences of warm-water species of diverse phyla at higher latitudes^5–8, such as hermatypic corals⁹, echinoderms^10,11, mollusks⁷ and fish^11–14. Range expansions have been accentuated in regions where intensifying warm currents from the tropics accelerate warming at temperate latitudes^11,15. Such currents have increased the arrival of warm-water vagrant species and allowed them to remain at higher latitudes for longer periods of time, attain higher abundances and sometimes establish permanent populations^11,16. As a consequence, the proportion of tropical species increase, resulting in a ‘tropicalization’ of the community. In some ecosystems this change in species composition has led to increased competition and species displacement^17,18, modification of food-webs and ecosystem functioning¹⁹, and regime shifts caused by the arrival of habitat-modifying species, such as herbivorous urchins or fish^8,20.

Increasing abundances of tropical herbivorous fish on temperate reefs will prompt major changes and threaten the services that these ecosystems provide¹¹. Temperate reefs are characterised by high abundances of habitat-forming seaweed, typically dominated by kelp²¹, which support high levels of biodiversity and multiple fishery resources²². Rates of herbivory by fish are generally low in temperate reef systems^23–26, although in some places a few fish species have shown high-intensity but small-scale effects on seaweed communities^27–30. In contrast, herbivorous fish play a critical role in healthy tropical reef ecosystems where they often facilitate corals by maintaining low abundances of seaweeds³¹; herbivorous fish typically attain higher abundances and greater diversity of species and feeding modes in tropical reefs in comparison to temperate reefs^32–34. Increasing abundances of herbivorous fish on temperate reefs can translate into higher rates of herbivory and a reduced abundance of seaweed through direct consumption³⁵ or indirectly by affecting seaweed reproduction³⁶ and recruitment¹⁴. This can have cascading effects that encompass multiple trophic levels, and reduce the resistance and resilience of kelp ecosystems^11,14,37. Moreover, a loss or reduction in habitat-forming seaweeds can also indirectly affect other systems, such as seagrass meadows, that often receive energy and nutrients from detached seaweeds from adjacent reefs^38,39.

Evidence for the impact of tropical herbivorous fish in temperate marine ecosystems has been increasing in recent years^11,40. Tropical parrotfish, which consume seagrass at higher rates than local temperate herbivores, are increasing in abundance in the northern Gulf of Mexico⁴¹. In southern Japan, temperate reefs now host permanent populations of tropical rabbitfish and parrotfish¹³ which have contributed to the decline of extensive kelp forests^42–44. Eastern Mediterranean habitats that used to be dominated by seaweeds have been transformed into barren seascapes after the intrusion and settlement of tropical rabbitfish^45,46 through the Suez Canal⁴⁷. South-eastern Australia has experienced an increase of tropical surgeonfishes⁴⁸ and rabbitfishes³⁵ that have caused the decline of kelp forests on offshore reefs³⁵.

In Western Australia, an exceptional marine heatwave during the summer of 2010–2011^49–51 allowed the poleward migration and population expansion of the tropical rabbitfish Siganus fuscescens in the region^14,52,53. This species is a characteristic herbivore of tropical reefs^54–56, but now is an important consumer of kelp in south-eastern Australia³⁵. Here we investigate if the heatwave, by allowing the expansion of rabbitfish, had indirect effects on kelp forests at latitudes lower than those at which it directly caused physiological impacts on kelp. We present a temporal analysis of changes in abundance of herbivorous fish, kelp and rates of herbivory from years prior to the marine heatwave (2004 and 2007) to the present (2016).

Within the fish community, three species were identified as documented consumers of kelp: the temperate Kyphosus sydneyanus (silver drummer) and Olisthops cyanomelas (herring cale) and the tropical Siganus fuscescens (rabbitfish, Supplementary Table S1). Among these herbivores, K. sydneyanus was the most abundant species in 2004 and 2007 (2.2 ± 1.9 and 3.0 ± 2.6 individuals 125 m⁻² [mean ± SE], respectively), followed by O. cyanomelas (0.3 ± 0.2 and 1.5 ± 0.7 individuals 125 m⁻²). However, abundances varied among reefs (PERMANOVA, pseudo-F_{6, 51} = 2.84, P = 0.013). Cow Rocks had the highest abundance of temperate herbivorous fish and The Lumps had the lowest, while Whitfords Rock and Wreck Rock hosted intermediate abundances. S. fuscescens was not recorded on any reefs in 2004 or 2007 (Fig. 1a). In contrast, S. fuscescens was the most abundant herbivore recorded in 2016 (9.4 ± 2.1 individuals 125 m⁻²; Fig. 1a), representing 19% of all individual fish recorded. The abundance of K. sydneyanus (4.7 ± 2.7 individuals 125 m⁻²) and O. cyanomelas (0.2 ± 0.1 individuals 125 m⁻²) did not change between years and consistently represented ~10% of the total fish abundances recorded (PERMANOVA, pseudo-F_{2, 51} = 1.98, P = 0.12). S. fuscescens was not recorded at The Lumps, which still had the lowest herbivore abundances; however, S. fuscescens was present on all other reefs, where Cow Rocks consistently had the highest abundances of herbivorous fish (Fig. 1a).

Kelp biomass decreased from 2.1 ± 0.6 kg 0.25 m⁻² in 2004 to 0.58 ± 0.17 kg 0.25 m⁻² in 2016, representing a loss of 1.48 kg 0.25 m⁻² or 71% (Fig. 1b; PERMANOVA, pseudo-F_{1, 32} = 6.29, P = 0.017). Kelp declined most at Wreck Rock (84%) and Whitfords Rock (67%). At Cow Rocks, kelp abundance was low in both 2004 and 2016 (but still declined 10%), whereas at The Lumps kelp biomass remained high in both 2004 and 2016, showing no signs of decline (Fig. 1b).

Before the heatwave, there was no consumption of kelp tethers, except at Cow Rocks, where low herbivory rates were recorded (Fig. 1c). After the heatwave in 2016, kelp consumption rates were 30-fold higher than those recorded in 2004 and 2007 (Fig. 1c; PERMANOVA, pseudo-F_{2, 78} = 3246, P = 0.0001). In 2016, 5-day herbivory assays resulted in 88.8% ± 2.6 consumption day⁻¹ (mean ± SE; equivalent to ca. 3.7% hr⁻¹) of kelp biomass across all reefs hosting rabbitfish. Caged kelp tethers did not show any loss of biomass, supporting the inference that kelp loss was due to consumption by fish. These rates far surpassed those measured over similar time periods prior to the heatwave (20.8% ± 0.1 day⁻¹ in 2004 and 20.6% ± 0.1 day⁻¹ in 2007; equivalent to ca. 0.9% hr⁻¹). Additional 4 hr video-filmed herbivory assays in 2016 recorded consumption rates of 28.4% ± 2.9 hr⁻¹ at reefs hosting rabbitfish, suggesting most of the consumption in the 2016 five-day assays occurred at higher rates than when calculated across the entire deployment time. In contrast to The Lumps where no herbivory and no rabbitfish were recorded (Fig. 1c), Cow Rocks had the highest herbivory rates (47.1% ± 0.5 hr⁻¹), followed by Wreck Rock (22.7% ± 0.5 hr⁻¹) and Whitfords Rock (20.3% ± 1.8 hr⁻¹) (Fig. 1c; PERMANOVA, pseudo-F_{6, 78} = 476, P = 0.001). Video-analyses of the herbivory assays confirmed that rabbitfish were responsible for all the consumption at Whitfords Rock and Wreck Rock, while consumption at Cow Rocks was shared between rabbitfish and silver drummers, which consumed approximately 50% each.

There was a significant positive relationship between rabbitfish abundance and herbivory rates (Fig. 2a; PERMANCOVA, pseudo-F_{1, 8} = 98, P = 0.001); whilst there was no relationship between abundance of temperate herbivorous fish and herbivory rates (Fig. 2b; PERMANCOVA, pseudo-F_{1, 6} = 0.1, P = 0.464). In turn, there was a significant negative relationship between kelp consumption rates on tethers and standing kelp biomass on the reefs (Fig. 2c; PERMANCOVA, pseudo-F_{1, 4} = 11, P = 0.03).

Our results document the establishment of rabbitfish (Siganus fuscescens) on temperate reefs off Perth following the 2011 marine heatwave, and show a concurrent steep increase in consumption rates of kelp. The abundance and rates of herbivory by two temperate herbivorous fishes (Kyphosus sydenayus and Olisthops cyanomelas) did not change among surveyed years, indicating that the increase in the consumption rates of kelp, and decrease in kelp abundance, is likely linked to the colonizing tropical rabbitfish — an inference that was supported by the video-analyses. Rabbitfish have not displaced the temperate herbivorous fish, but rather they have been an addition to the guild. Kyphosus sydneyanus consumed kelp at higher rates per individual than S. fuscescens, but kelp consumption by rabbitfish was higher due to their higher abundances. Siganus fuscescens has become now the most important kelp consumer on these reefs.

The broader spatial distribution of rabbitfish in south-western Australia after the heatwave is currently unknown, but we would expect to find similar patterns of abundance and herbivory across the other extensive inshore reefs in the region. The majority of the evidence suggest that S. fuscescens tends to inhabit shallow near-shore reefs adjacent to seagrass meadows. Past studies have found greater abundance closer to the coast^53,57. Although in south-eastern Australia rabbitfish were more common on offshore reefs, their abundance was low (2.2 ± 1.8 individuals 125 m⁻²)³⁵. Our results suggest that S. fuscescens have specific habitat preferences. Despite being <1 km away from reefs hosting abundant rabbitfish populations they were absent at The Lumps, which is a slightly deeper (~7 m vs ~5 m) more contiguous reef environment than the other reefs surveyed. This may be related to the behaviour of tropical herbivorous fish which tend to prefer habitats with lower seaweed cover⁵⁸. Further research is required to clarify these habitat affinities to make accurate predictions of future geographic distribution changes and associated impacts with habitat-forming seaweeds.

Past studies on the tropicalization of herbivore communities have regarded rabbitfish, including S. fuscescens, mainly as grazers^14,45. However, we found that S. fuscescens can also be an important browser in temperate ecosystems. This is in agreement with studies from south-eastern Australia³⁵ and tropical reefs (Great Barrier Reef) that found S. fuscescens as the main consumer of Sargassum^54–56. These differences between studies are not surprising. Although S. fuscescens, and herbivores in general, often feed selectively from the available food sources⁵⁹, they usually feed on a diverse mix of primary producers⁶⁰, allowing them to play different functional roles⁵⁷. We did not measure grazing rates but rabbitfish were observed biting on the substrate and epiphytic algae on seagrass shoots.

This feeding multifunctionality has important implications for the potential impact that these tropical herbivores can have in temperate ecosystems. High rates of grazing by rabbitfish can reduce the recruitment of habitat-forming seaweeds by consuming seaweeds in early-life stages, as they do in tropical reefs to maintain a seaweed-free state^37,61. Past studies have concluded that this was the main impact of rabbitfish on temperate seaweed communities^14,45. However, high rates of browsing by a single species can reduce kelp biomass considerably⁶². Rabbitfish have inhabited south-western Australia for at least three years⁵² and based on our video-filmed herbivory assays they have likely been consuming kelp throughout this period. Our results suggest that the decline in the biomass of kelp on inshore reefs has occurred due to increased consumption by rabbitfish. Wherever we observed rabbitfish we found an associated increase in herbivory and a decrease in kelp biomass; where they were absent, the opposite pattern was found. In addition to the direct reduction in kelp biomass, high levels of browsing can diminish the reproductive output of seaweeds³⁶; although this remains poorly understood. In the case of E. radiata in the region, it produces the highest number of zoospores between April and May⁶³, the same time of the year when our tethered kelps were consumed at high rates. Remaining kelps on the reefs with rabbitfish showed clear signs of heavy pruning by browsers (personal observation, S. Zarco-Perello & T. Wernberg), suggesting that rabbitfish do consume, and reduce, kelp reproductive tissue, potentially causing a decline in the supply of zoospores and abundances of gametophytes that later would become kelp recruits⁶⁴.

It is likely that other factors have acted in synergy with herbivory to cause the decline of kelp. Temperature and UV radiation have been increasing over the last decades in the region^15,65,66 and both can affect survival, growth and recruitment of kelp^67–71. The marine heatwave that impacted the region in 2011 could have caused some mortality⁷², but not at the levels observed in our results at the latitude of our study sites⁸. The slightly deeper environments at The Lumps, the only reef with consistently high abundance of kelp, could have been less affected by these physical factors, although the difference in depth is not likely to be substantial enough to attenuate high temperatures or UV radiation^67,70, pointing to the low herbivory rates and its historical higher levels of productivity as the cause of its higher kelp abundance²⁴.

The present study provides further evidence of the consequences of the globally emerging process of tropicalization of temperate reefs^11,40. As the poleward-migration of marine species is becoming more accentuated due to global warming, novel biological interactions within^17,73 and between trophic levels are emerging^19,74. The alteration of species richness and abundance of consumers impact lower trophic levels, leading to changes in ecosystem functioning and structure modifications⁷⁵. Our results show that extreme marine heatwaves not only can have immediate devastating direct effects on temperate reefs^8,72 but also can have longer-term indirect effects by boosting the immigration of tropical herbivores that increase herbivory rates and reduce the abundance of habitat-forming seaweeds. This contrasts with observations from other tropicalized regions of the world where the process has been more gradual, such as Japan^42,43, the Mediterranean Sea⁴⁵ and south-eastern Australia³⁵. With the ongoing ocean warming^65,76, further range expansion and population increase of S. fuscescens is expected. Together with other environmental disturbances, such as high temperatures, these tropical herbivores pose a growing threat to temperate reefs, increasing the likelihood of regime shifts to vegetation-free states and the associated loss of environmental services of great value to human societies from these ecosystems.

We are grateful for assistance in the field from C. Tuckett, R. Hemsworth, S. Dee and Y. Mulders, and for the help with the stereo video analyses from T. Bond and M. Birt. This work was funded by grants from the Australian Research Council (DP170100023) and The Hermon Slade Foundation (HSF13-13) to TW.

Conceived the study: T.W., S.Z.P. Collected data: S.Z.P., M.V., T.W. Analysed data: S.Z.P., T.L. Wrote the paper: S.Z.P., T.W. with input from all authors. Provided funding: T.W.

Salvador Zarco-Perello, Email: salvador.zarco.perello@gmail.com.

Thomas Wernberg, Email: thomas.wernberg@uwa.edu.au.