Ecklonia radiata is the dominant habitat forming kelp supporting temperate reef ecosystems in Australia (Wernberg et al., 2019a). E. radiata is a foundation species on Australia’s Great Southern Reef, which is a renowned biodiversity hotspot (Bennett et al., 2016) that is currently declining (Wernberg et al., 2013; Vergés et al., 2016). Fertile E. radiata plants (n = 5) were collected from Muttonbird Island (30°18’S 153°08’E) on 11^th of November 2020 and returned to the laboratory in a moist calico bag for processing. Fronds were cleaned of epiphytes by gently scraping with a knife and steeped in an iodine/povidone Betadine Antiseptic (Faulding Pharmaceuticals) solution (2% in 1 µm filtered, UV-sterilised seawater (FSW)) for 5 seconds to prevent contamination of cultures (Alsuwaiyan et al., 2019). After drying at 20°C for one hour, visibly fertile tissue was submerged in 2L FSW at 20°C and gently stirred. The number of spores released was counted using a compound microscope and a hemocytometer counting chamber, and the final density was 21 250 spores mL^-1 of seawater. Of this spore solution, 20 mL was used to inoculate experimental dishes for experiment 1. For experiment 2, stock cultures of E. radiata gametophytes of mixed origin from the Solitary Islands (Coffs Harbour region, NSW, Australia), which were stored at 20°C in permanent culture and produced in a similar manner, were used.

Micrograzers present in turf surrounding patches of E. radiata were collected from Charlesworth Bay (30°16’S 153°08’E) on 22^nd of January 2021 for experiment 1, and on 12^th of June 2021 for experiment 2. The gastropod Anachis atkinsoni was selected as the grazer in the experiments as this species commonly occurs in turf habitats along Australia’s east coast. This marine gastropod is endemic to Australia and occurs widely within tropical and temperate systems (Wilson, 1994). For both experiments, individual A. atkinsoni all of similar size were assigned to different temperature treatments (20°C, 23°C and 26°C) and were acclimatized from the collection temperature over a period of 7 days by adjusting temperatures each day until treatment levels were achieved. While acclimatizing, gastropods were fed ad libitum Sargassum spp.

A full factorial design was used to test the effect of the presence or absence of micro grazers on gametophytes under three temperatures, resulting in a total of six treatments. Five independent replicates were available for each treatment at each planned sampling time (n = 5), resulting in a total of 150 experimental replicates. To establish experimental gametophyte cultures, petri dishes (n = 150) containing one 20x20 mm coverslip as a surface for settlement were inoculated with 20 mL of the spore solution. Spores were left to settle overnight at 20 °C, after which the media was refreshed with FSW and quarter strength F growth media (AlgaBoostTM 2000×, AusAqua Pty Ltd., Wallaroo, SA, Australia) and Petri dishes were assigned to a treatment temperature of 20°C, 23°C or 26°C (n = 50 dishes per temperature). The 20°C treatment was established via the ambient room temperature, which was controlled by a thermostat, while the 23°C and 26°C treatments were established using replicate heating mats (Lerway 21W Heating Incubator Mat) (n = 3 mats per temperature treatment), which were slowly increased to desired temperatures over a 7-day period. Gametophytes were cultured in the experimental treatments for two months before grazers were added. In this time they germinated, and the gametophytes had grown considerably (average length over all treatments 146 µm). Temperatures were checked daily to ensure stability among treatments (20.4 ± 0.07, 23.2 ± 0.03 and 25.8 ± 0.04, respectively; mean ± SE). Light levels were also measured among experimental replicates daily (36.5 ± 0.8 µmol/photons/second;, mean ± SE) to ensure consistency among treatments and to confirm light levels were within the optimal range (~15-42 µmol/photons/second) for growth of E. radiata gametophytes (Novaczek, 1984a; Mabin et al., 2013). Heating mats and the ambient temperature treatment were interspersed along laboratory benches so there was no systematic bias in light levels or any other factor. Throughout the duration of the experiment, cultures were refreshed every week with FSW and quarter strength F growth media (AlgaBoost™ 2000×, AusAqua Pty Ltd., Wallaroo, SA, Australia).

Gametophytes were sampled before grazing exposure to establish a baseline with which to compare recovery after grazing. This was done by taking three random photographs of coverslips (n = 5) from each treatment at x200 magnification using a MIchrome 20 Color Microscope camera mounted on a stereo microscope (Olympus BX53). Total coverage [total surface area (µm²) of gametophytes per field of view (FOV=3.2mm²)], mean individual length [averaged over 10 individuals (µm)], mean individual surface area [averaged over 10 individuals (µm²)] and abundance (number of individuals per FOV) of gametophytes was quantified using ImageJ 1.53e imaging software and averaged over three photos per replicate. Temperature acclimated snails were then assigned at random to half the gametophyte cultures (n = 1 per dish) and allowed to graze for 48 h after which they were removed, and their length (2.25 mm ± 0.06) and survival were recorded using an MIchrome 20 Color Microscope camera mounted on a dissecting microscope and ImageJ 1.53e imaging software. This grazing time period was sufficient to detect an effect of grazing without risk of all gametophytes being grazed, as determined by a prior pilot study. Immediately after the removal of grazers (i.e day 1 post-grazing) sampling of gametophytes was done using the aforementioned photographic method. To track post-grazing recovery of gametophytes, cultures were maintained for 32 days and sampled 7, 18 and 32 days after grazers were removed. Individual replicates were available for each sampling time, so dishes were never re-sampled. The cultures sampled at day 32 were affected by fouling so were not included in the final analysis of gametophyte recovery post-grazing.

We also tested whether grazing rates of A. atkinsoni differed between temperature treatments independent of gametophyte growth. A. atkinsoni were collected near Charlesworth Bay (n = 30) on 9^th of August 2021 and assigned to different temperature treatments (19°C, 23°C and 26°C, n = 10 per temperature). A. atkinsoni were allowed to graze on plates of dental wax (Thompson et al., 1997) for 3 days following acclimatization. Temperature measurements were taken daily to ensure temperature stability (19.2 ± 0.1, 23.2 ± 0.1, 26.0 ± 0.2; mean ± SE). After removal of A. atkinsoni, the total grazed surface area (µm²) of wax plates was measured using ImageJ 1.53e imaging software.

To investigate if gametophytes survive ingestion by A. atkinsoni, we assessed regrowth of gametophytes from grazer faecal matter and undigested but highly aggregated gametophytes in grazer mucus. A. atkinsoni were collected near Charlesworth Bay (n = 12) on 12^th of June 2021 and acclimated to 19°C for 48 hours in aerated holding tanks. Gametophytes taken from stock cultures kept at 19°C were blended using a hand-held stick blender to small (approximately 5-10 cells) equal fragments and left to settle in 50mL glass beakers (n = 4) with 40 mL FSW and containing 3 coverslips for 48 hours. A. atkinsoni were added to the beakers and left to graze for 5 days, which allowed sufficient faeces to appear. Both faecal pellets (ingested gametophytes, n = 24) and undigested but highly aggregated gametophytes in mucus of A. atkinsoni (n = 12) were taken from the beakers and cultured in separate multi-well plates at 19°C (19.0 ± 0.2 °C and 19.6 ± 0.4 µmol/photons/second, respectively; mean ± SE).

Sampling of faecal matter and clumped gametophytes was performed at 7, 14 and 21 days after subculturing. Individual replicates for each sampling time were available, so that at every timepoint 8 dishes with faecal pellets and 4 dishes with clumped gametophytes were examined. For every piece of faecal matter, it was noted if visible gametophytes and/or sporophytes were present and whether they were dead (unpigmented) or alive (pigmented). The clumped gametophytes were photographed at 3 random fields at x200 magnification using a MIchrome 20 Color Microscope camera mounted on a stereo microscope (Olympus BX53). Total coverage [total surface area of gametophytes per field of view (µm²)] of gametophytes was measured using ImageJ 1.53e imaging software.

Presumably due to the small volume of culture media in dishes, approximately 50% of grazers died just before the grazing period (48H) had concluded. However, we made sure that grazing did occur in each dish and photos that did not contain faeces, a grazing trail or evidence that the grazer did not graze prior to removal or death were removed from the dataset to represent the effects of grazing as accurately as possible. This resulted the loss of 31 experimental replicates.

Two-factor analyses of variance (ANOVA) were used to test for an effect of temperature and grazing on the total coverage, average individual length, average individual surface area and abundance of gametophytes at the conclusion of the grazing period. A single-factor ANOVA was used to test for an effect of temperature on the size of A. atkinsoni grazing marks on wax plates. A three-factor ANOVA was used to test for an effect of grazing, temperature and time post-grazing on the recovery of gametophytes. Individual experimental replicates were used at every sampling time, allowing the inclusion of ‘time’ as an independent explanatory variable. Response variables analysed to assess gametophyte recovery were total coverage, average individual length, average individual surface area and abundance of gametophytes quantified from day 1 to day 18. Baseline data were used to compare gametophyte recovery. A single-factor ANOVA was used to test for an effect of time on the coverage of clumped gametophytes following A. atkinsoni grazing. All ANOVAs utilised type III sums of squares as appropriate for unbalanced data (Shaw and Mitchell-Olds, 1993; Queen et al., 2002), which resulted from grazer mortalities.

The assumption of homogeneity of variance was assessed using a Levene’s test. Normality of data was assessed visually using QQ-plots and tested using a Shapiro-Wilk test. Gametophyte abundance, coverage, length and surface area were log-transformation prior to analysis on data from experiment 1 in order to not violate the assumption of normality. Significance was set at alpha level 0.05 and pair-wise comparisons were undertaken when significant fixed effects were detected. Data were analysed and plotted using the software R (R Core Team, 2020).

The presence of grazers significantly reduced the total coverage of gametophytes (µm²) after the 48-hour grazing period (F _1,23 = 5.97, p=0.02), but there was no effect of temperature on the cover of gametophytes (F _2,23 = 1.96, p=0.16) and no interaction between grazers and temperature during this 48-hour period (F _2,23 = 1.32, p=0.29, Figures 1A ; Table S1 ). Similarly, the presence of grazers significantly reduced the abundance of gametophytes (F _1,23 = 7.1, p=0.01) after a 48-hour grazing period but there was no effect of temperature on abundance (F _2,23 = 0.5, p=0.61) nor an interaction between grazing and temperature (F _2,23 = 0.10, p=0.91; Figures 1B ; Table S1 ). There was no effect of grazing or temperature on mean individual length or mean individual surface area of gametophytes ( Table S1 ). Temperature did not have a significant effect on the average surface area of A. atkinsoni grazing marks in dental wax (F _2,27 = 0.24, p=0.79).

The presence of grazers significantly reduced the total coverage of gametophytes (F _1,71 = 13.74, p<0.01, Figure 2 and Tables 1 , S2 ). The total coverage of gametophytes was significantly reduced at 18 days relative to day 1 and day 7 (F _2.71 = 4.66, p=0.01, pair-wise test among time points; Figure 2 and Tables 1 , S2 ). Moreover, total gametophyte coverage was also significantly lower at 26°C than at 23°C, while no difference was found in coverage between 20°C and the other temperature treatments (F _2,71 = 3.56, p=0.03, pair-wise test among temperatures; Figure 2 and Tables 1 , S2 ). While there were no interactions (F _4,71 = 0.51, p=0.73) between these factors ( Figure 2 and Tables 1 , S2 ), gametophyte coverage decreased considerably over time in the grazed treatments at 23°C and 26°C. There was a final decrease in coverage of 73% and 83% for the grazed treatments, respectively, compared to 42% and 60% in non-grazed treatments (Fig 2.). Additionally, after 18 days only the non-grazed treatment at 20°C saw gametophyte coverage recover to control conditions ( Figure 2 ).

Exposure to grazers significantly reduced the mean individual length of gametophytes (F _1,71 = 5.01, p=0.03, Figure 3 and Tables 1 , S2 ). The mean length of gametophytes was significantly smaller at 18 days, while they were similar size at day 1 and day 7 (F _2,71 = 9.34, p<0.01, post-hoc test among time; Figure 3 and Tables 1 , S2 ). The mean length was also lower at 26°C and 20°C than 23°C (F _2,71 = 4.13, p=0.02, pair-wise test among temperature; Figure 3 and Tables 1 , S2 ). While there were no significant interactions between these factors on gametophyte length (F _4,71 = 0.58, p=0.68), the decrease in length appeared steeper in grazed compared to non-grazed treatments. This was particularly apparent at 23°C and 26°C, which were associated with declines in gametophyte length of 51% and 70% in the grazed treatments, respectively ( Figure 3 ), compared to 26% and 44% in non-grazed treatments. Additionally, both the grazed and non-grazed treatments did not recover to control levels ( Figure 3 ), with length decreasing in all treatments during the post-grazing period.

The presence of grazers significantly reduced the abundance of gametophytes (F _1,71 = 22.25, p<0.01). Moreover, the mean abundance of gametophytes at 18 days was significantly greater than abundance at 7 days (F _2,71 = 7.47, p<0.01, pair-wise test among time; Figure 4 and Table 1 , S2 ). However, temperature did not have a significant effect on gametophyte abundance (F _2,71 = 0.61, p=0.55) and there were no interactions (F _2,71 = 0.38, p=0.82) among factors ( Figure 4 and Table 1 , S2 ).

Gametophyte abundance increased by just over 50% from day 1 to day 18. The abundance of gametophytes recovered to be greater than control levels in both grazed and non-grazed treatments ( Figure 4 ). As no extra spores were added to the dishes, the increased abundance can either come from the growth of gametophyte fragments too small to notice in earlier counts or the fragmentation by grazers.

Live (pigmented) gametophytes were repeatedly found in A. atkinsoni faecal matter, with the proportion of total faecal matter containing live gametophytes increasing during 21 days of culturing ( Figure 5 ). Throughout the 21-day culturing period, dead (unpigmented) gametophytes were present in 50% of faecal matter. However, live gametophyte appeared in 13% of faecal matter on day 7, which increased to 40% at subsequent dates. Live gametophytes were identified as full filaments ( Figure 6B ) as well as spores ( Figure 6C ) in faecal pellets of A. atkinsoni.

No significant difference was found in the coverage of clumped undigested gametophytes over time (F _2,9 = 0.99, p=0.41), signifying gametophytes remained aggregated in A. atkinsoni mucus over a 3-week period post-grazing. In one replicate of clumped gametophytes, juvenile sporophytes appeared after 21 days, while these never appeared in experimental treatments in experiment 1 ( Figure 6A ).

Turfing algal expansion is thought to be facilitated by suppressed recruitment of kelp microscopic stages (O’Brien and Scheibling, 2018), and micrograzers present in turfs may contribute to this continued inhibition by consuming kelp gametophytes (Dayton et al., 1984). We found that A. atkinsoni, a turf associated grazer, significantly reduced both the abundance and coverage of E. radiata gametophytes in laboratory experiments after 48H. Presumably in natural settings when grazers continually graze gametophytes within turf, this impact may be much larger. These results provide novel evidence that a gastropod micrograzer (<2.5 mm) commonly present in turfing algae can impact the microscopic stages of a kelp gametophyte and could inhibit kelp recruitment and transitions from turf to kelp forest. Studies on such trophic interactions of gametophytes are sparse (Veenhof et al., 2022), but those that exist also suggest that gastropods do graze on microscopic stages of kelp. For example, the limpet Margarites helicinus was found to graze gametophytes of kelps Alaria escuelenta, Laminaria digitata and Saccharina latissima, which could result in reduced sporophyte recruitment (Zacher et al., 2016). Similarly, Macrocystis pyrifera gametophytes were grazed by two species of gastropod, reducing their abundance (Henriquez et al., 2011). While studies on micrograzers are not common, meso-grazers have been shown to impact macro algae (Hay et al., 1987). For instance, selective grazing of meso-grazers can decrease growth rates and biomass in kelps (Poore et al., 2014; Gutow et al., 2020) and facilitate spore release in red seaweed (Buschmann and Santelices, 1987). In addition, juvenile stages of meso- and other larger grazers such as sea urchins and gastropods could fulfil the same ecological role, but this remains untested.

Trophic interactions between algae and grazers can be impacted by increases in temperature consistent with projected levels of ocean warming (Provost et al., 2017; Christie et al., 2019; Miranda et al., 2019). Increases in temperature have been found to both enhance grazing rates (e.g. Miranda et al., 2019) and inhibit grazing (e.g. Provost et al., 2017), and population models show that herbivore abundances can be unaffected by increased temperatures (O’Connor et al., 2011). Our temperature treatments did not influence grazing by A. atkinsoni as indicated by no difference in grazing on dental wax and no interactions between temperature and grazing on gametophyte abundance and coverage. This is most likely due to the fact that experimental temperatures fell within the range of temperatures A. atkinsoni naturally experiences throughout its wide distribution, which spans tropical to temperate habitats (Wilson, 1994). Given the suite of micrograzers present in turfing algae, increasing temperatures may differentially affect grazing rates among different species present, specifically those with narrower distributions. To understand how ocean warming and grazing rates may influence kelp gametophyte persistence within turfs, a larger range or even entire communities of micrograzers need to be explored.

Interactions with abiotic factors other than temperature can also change the outcomes of ecological interactions and may be at play within turfs. For example, the clearing of sediment by grazers has been found to enhance gametophyte survival (Zacher et al., 2016), which may indicate that micrograzers can counteract negatives effects of accumulated sediment in turfs on gametophyte recruitment. Conversely, the physical structure of turfing algae may shelter gametophytes from grazers (Harris et al., 1984; Camus, 1994) and counteract negative impacts of micro-grazing on survival of gametophytes. These interactions can thus change the impact of micrograzers alone on gametophyte survival in turfs. While these factors were not taken into account in this study, the finding that grazers present in turf can reduce gametophyte cover and abundance demonstrates a biological mechanism that may contribute to the strong hysteresis observed when turf cover is maintained over kelp forests.

The ability of gametophytes to recover post-grazing differed depending on which variable was measured. Abundance increased steadily post-grazing, while coverage and length decreased over time. While abundance was lower overall in grazed compared to non-grazed treatments, it increased over time to recover to eventually exceed initial abundances at the start of the experiment. This could signify the continual germination of spores that were too small to record at earlier sampling times. These spores could have germinated and grown larger over time and were noted as small gametophytes at later sampling points, also accounting for the reduction in the length of gametophytes over time. Alternatively, grazing may have fragmented gametophytes creating a higher abundance of smaller individuals. Given that an increase in abundance was seen in both grazed and non-grazed treatments the former hypothesis is more likely. In natural systems, filamentous gametophytes may persist on the benthos for 3 to 7 months, as a bank of microscopic growth forms that can recruit into adults when conditions are suitable (Carney et al., 2013; Akita et al., 2019; Schoenrock et al., 2020; Veenhof et al. 2022, in press). In culture, spores and gametophytes can remain dormant in a vegetative state for decades (Veenhof et al., 2022; Barrento et al., 2016, in press). While it is not known how long such seed banks can persist, it is possible that there exists a bank of diverse ages and sizes of spores and gametophytes that could withstand periodic grazing pressure.

While there was no significant effect of temperature on post-grazing recovery, we did detect an overall temperature effect on gametophyte length and coverage, which is consistent with the major influence temperature has on general gametophyte growth and survival (Veenhof et al., 2022; tom Dieck, 1993, in press). Specifically, gametophyte cultures at 23°C had greater length and coverage than cultures at 20°C and 26°C, irrespective of grazing. Ocean temperatures of approximately 26°C are representative of the upper thermal limits for populations of E. radiata, and growth is optimal between 20 and 23°C for Western Australian E. radiata populations (Novaczek, 1984b; Mohring et al., 2013; Mohring et al., 2014; Alsuwaiyan et al., in review). In situ temperature has been found to correlate with temperature tolerance of E. radiata gametophytes (Mohring et al., 2014) and our results are consistent with this result. The E. radiata cultured here is from sporophytes inhabiting the warmest edge at the tropical-temperature transition zone off eastern Australia where average annual temperatures are 23°C. As ocean warming is projected to continue, however, E. radiata populations will come under considerable temperature stress especially at the warm, equatorward edge of their distribution (Martínez et al., 2018; Davis et al., 2021), where temperatures often exceed 26°C.

Our results reveal that E. radiata gametophytes can survive consumption by the common turf grazer A. atkinsoni. This is consistent with previous studies on other kelp species. For instance, spores of the kelp Lessonia nigrescens were found to survive ingestion by the urchin Tetrapygus niger (Santelices et al., 1983) and could grow following cultivation of faecal pellets. Similarly, microscopic stages of algae were found in the gut content of several molluscs and were able to grow from fragmented pieces and spores in faecal pellets (Santelices and Correa, 1985). In culture, gametophytes that are shredded have the ability to regrow quickly and become sexually mature and this is a commonly used culture technique. Passing through the digestive track of grazers might thus increase biomass through shredding and regrowth, which in turn can increase abundance. In addition, the aggregation of gametophytes in the mucus trails of A. atkinsoni appeared to benefit stage transitions. Aggregated gametophytes in mucus appeared to be healthy and even produced the only juvenile sporophytes observed in this study. How this may translate into field settings is unknown but an increased density of Lessonia berteroana juvenile sporophytes was seen in patches where herbivores were present (Oróstica et al., 2014). Additionally, Henriquez et al. (2011) found recruitment pulses of M. pyrifera after annual die-back of adult canopy despite a high presence of limpets, and Zacher et al. (2016) found an increased number of microscopic sporophytes in some of their experimentally grazed treatments. While both these studies hypothesized that cleared substrate by grazers give the gametophytes a chance to recruit, we suggest that the ingestion and aggregation by gastropod grazing might also contribute to the observed recruitment pulses. Fertilization in seaweeds and kelps can only take place if the male and female gametophytes are in close enough proximity (Reed, 1990) which can be influenced by water motion, substrate rugosity and other environmental factors (Amsler et al., 1992). We hypothesize that some of these barriers to fertilization could be overcome by aggregation of gametophytes in mucus trails of gastropod grazers. While the densities of spores, gametophytes and adult plants necessary for recruitment has been researched extensively (e.g. Reed et al., 1991; Graham, 2003; Layton et al., 2020), these models do not consider the possibility of aggregation via grazing. The relative contribution of grazers to kelp recruitment is not known but based on our results could prove an integral component in recruitment models. For instance, if gametophyte density is low due to turf expansion or limited spore settlement and germination in turfs, micrograzers such as A. atkinsoni could increase the chances of successful recruitment through shredding and dispersal, allowing gametophytes to increase their biomass, as well as increasing chances of fertilization by aggregating gametophytes in their mucus trail. In this way, micrograzers may serve a similar ecological role to pollinators in land plants. Certainly, both large and small herbivores can influence the patterns of recruitment post-disturbance (Oróstica et al., 2014), indicating they can shape recovery of kelp forests.

In summary, micrograzers can have both positive (dispersal, increased fertilization, and sporophyte recruitment) and negative (reduced coverage and abundance after grazing) impacts on gametophytes highlighting the complexity of trophic interactions in turf-kelp dynamics. Negative consequences of grazing were expected as herbivory by grazers such as fish and larger gastropods is known to maintain turf cover, particularly at species range edges (Bennett et al., 2015; Zarco-Perello et al., 2021). Moreover, the lack of recovery in gametophyte coverage post-grazing is consistent with global increases in turf cover (Filbee-Dexter and Wernberg, 2018; Feehan et al., 2019; Pessarrodona et al., 2021). Higher temperatures consistent with future warming also decreased gametophyte coverage, but had no impact on grazing rates, suggesting that E. radiata gametophytes may be increasingly challenged under climate change. However, the survival of gametophytes and possible enhanced recruitment of sporophytes following grazer ingestion may indicate beneficial effects of grazing. This highlights the complexity of ecological interactions that take place on a microscale. Gastropod micrograzers are both high in abundance and biomass and thus likely to influence kelps through grazing, and while their influence is poorly understood (Barnes, 2019) they may represent a potential underexplored driver of change in kelp-turf dynamics.