Data analysis was carried out in R Studio (R Core Team, 2025). Rarefied sampling curves used the R package “iNEXT” (Chao et al., 2016). All Generalized Linear Mixed Models (GLMMs) and Generalized Linear Models (GLMs) used the R package “lme4” (Bates et al., 2015), while post-hoc Tukey’s tests for these models used the R package “emmeans” (Lenth, 2023). Aspect was not detected to have any correlation with dependant variables, likely due to the prevailing ocean current swinging in opposite directions between summer and winter (Montaño et al., 2023), and was therefore removed from statistical analyses.

The Chasing M2 Pro ROV did succeed in collecting sufficient footage for an extensive analysis of the targeted reefs, but it seemed that the vehicle was at the limit of its suitability for this type of project. The power supply and internal sensors failed while it was underwater, which led to repeat visits to the reefs and wear and tear on the vehicle. Another difficulty was encountered with the recording of positioning data. A small ultra-short baseline (USBL) acoustic positioning system was investigated to enable more accurate positioning of the ROV underwater but unfortunately, this equipment did not cooperate well with DML’s other hardware and was not deemed suitable for the project. Despite these difficulties, the Chasing M2 Pro ROV was a cost-effective alternative to the larger, more robust commercial ROVs (not available for this study) which increase substantially in cost and complexity.

Varying camera to substrate angles resulted in most images being at oblique angles. The issue this raises is that oblique angle images introduce parallax error, geometric distortion and scaling error, making any measurement taken from a laser scale inaccurate whether that be measurement of sponges or quadrat area (Lesser and Slattery, 2021). Therefore, no measurements could be made using the laser scale which ruled out the use of some common measurements for abundance and species density data, such as biomass. This also affects percentage cover estimation, as features closer to the camera will be over-represented while features further away are under-represented. However, the level of parallax error was limited by restricting the photoquadrat to the immediate field of view and was appropriate for the purposes of identifying coarse categories and analyzing SSA.

Analysis of ROV footage revealed highly diverse sponge communities inhabiting all three of the surveyed mesophotic reefs. Species richness was relatively similar amongst reefs, with 47 species recorded at Te Poroiti, 40 at Nukutai, and 46 at Motuhaku. Rarefied sampling curves show that species richness was still increasing in the last photoquadrats analyzed on Nukutai and Te Poroiti Reef but had plateaued at Motuhaku Reef (Figure 5). This suggests that further sampling in future would capture greater species richness at Nukutai and Te Poroiti reefs (Figure 5). Hence, findings presented in this study represent a conservative estimate of diversity.

A total of 53 different sponge species were identified from ROV footage (Table 1), this being a conservative number to ensure certainty of classifications as discussed earlier. Sponges were predominantly from the class Demospongiae, but some calcareous sponges were also seen (Table 1). Of those identified, six (11% of all species recorded) were possibly new or undescribed species (Figure 5, Table 1).

This study has updated the current biogeographic knowledge of the Bay of Plenty. Fifteen sponge species identified (22%) presented new occurrence records for the region and some species’ known depth ranges were extended (Table 2). The sponges identified in this study represent a highly mixed community made up of multiple well connected and diverse assemblages (Figure 6, Table 1). The total list of 53 species comprised of species with northern warm-water and southern cold-water affinities relative to the BoP (22% and 12%, respectively), as well as a range of temperate (27%) and generalist species (39%) (Figure 6, Table 1). Temperate represents species found only in temperate waters, both to the North and to the South of the BoP. Generalist represents species found from the tropics or subtropics all the way through to cold-temperate waters.

The main species seen to have necrotic tissue were all large, ‘tough’ sponges with robust, dense spicular skeletons (subclass Heteroscleromorpha, orders Haplosclerida and Tetractinellida). The species that commonly had visible necrotic tissue on Motiti’s mesophotic reefs were Ecionemia alata, Petrosia (Petrosia) cf. hebes, Stelletta conulosa, Stelletta crater, Stelletta maori, Geodia regina, as well as two unidentified morphotypes; ‘orange meandering sponge’, and ‘black and orange cup’.

Depth alone did not appear to have an effect on the probability of seeing sponge necrosis across all reefs (p >0.05, z = -1.03) (Figure 7B), but Nukutai Reef and its interaction with depth were significant predictors of sponge necrosis (p < 0.01, z = 2.76, and p < 0.01, z = -2.87, respectively) (Figure 7A) (Supplementary Table 6). Tukey-adjusted pairwise comparisons of the means revealed that the probability of seeing necrosis at Nukutai Reef was significantly lower at deeper depths compared to Motuhaku and Te Poroiti (Figure 7A) (Supplementary Table 7). For example, the GLMM model predicts that at 50 m depth, there is a 0.38% probability of seeing sponge necrosis at Nukutai Reef with an uncertainty (SE) of 0.54% (Supplementary Table 7). In contrast, the predicted probability of seeing sponge necrosis at 50m depth at Motuhaku Reef is 17% (SE ± 4%), and 14% (SE ± 5%) at Te Poroiti Reef (Supplementary Table 7).

This means that overall, the probability of seeing sponge necrosis in the mesophotic zones of the Motiti reefs was the same regardless of depth (Figure 7B). However, whatever is driving sponge necrosis on these reefs (e.g., environmental stressors) has a weaker influence in the deeper waters on Nukutai Reef (Figure 7A), as evidenced in the interaction term (Supplementary Table 6).

Overall, SSA increased with depth (t = 8.24, p < 0.01), and was significantly greater on Nukutai Reef compared to Motuhaku (z ratio = -3.35, p < 0.01) (Supplementary Tables 8, 9). This means that as the reef gets deeper, larger areas of the reef are smothered with deposited sediment (Figure 8B). However, although this trend holds true for all reefs sampled, the baseline level of SSA was greater at Nukutai Reef than Motuhaku (Figure 8A). In addition, many of the cup sponges photographed had layers of fine sediment covering their surfaces, showing the recent occurrence of a heavy sedimentation event (Figure 9). Sediment deposition was determined to be recent because fine sediment was sitting in a layer on top of many Ecionemia alata purple cup sponges (Figure 9B). This species does not tolerate high levels of sedimentation and will shed mucus to clear sediment that is sitting on it’s surface (Ayling, 1983; Battershill and Bergquist, 1990; Bell et al., 2015; Schönberg, 2021). In Figure 9B stringy/webbed mucus that the sponge is producing to get rid of sediment that has deposited on top of it is evident.

From November 2021 to December 2022, the BoP experienced a marine heatwave which lasted for 381 days (Figure 10). The heatwave varied from strong to moderate intensity throughout the event, with a maximum sea surface temperature deviation of 2.5°C above the historic average (Figure 10). Since surveys for this study took place in early August of 2022, this means that the Motiti reefs had been continuously experiencing a marine heatwave for the past eight months at the time ROV surveys were carried out.

Mesophotic rocky reefs are known to be biodiversity hotspots and the Motiti mesophotic archipelago has proven to be no exception (Bo et al., 2012; Donald, 2021; Rossi, 2013; Bart et al., 2021; Bell et al., 2022). Through a comprehensive examination of sponge biodiversity on Motiti’s mesophotic reefs, this study conservatively identified fifty-three sponge species from ROV footage. The still-increasing rarefied sampling curves indicate that this study has only partially captured the true diversity of these mesophotic sponge communities. Greater sampling effort using equipment capable of finer-scale surveys would enable a more comprehensive record of species richness in future monitoring. Although species richness is difficult to compare between studies (Clarke and Warwick, 1998) it remains valuable information, providing a benchmark in an area with limited existing knowledge.

Over a quarter (28%) of the sponge species recorded in this study constitute novel occurrence records in terms of either their geographical distribution or depth range. Some of these newly documented occurrences, such as Homaxinella erecta, bridge previously known distributions both to the north and south of the BoP (Kelly, 2022; WoRMS Editorial Board, 2025). The diversity in sponge taxa found in this study provides valuable insights into the influence of currents within the BoP’s mesophotic zone. The composition of sponge communities in the Motiti mesophotic reefs and the broader BoP region reflects an amalgamation of warm-water (northern) and cold-water (southern) affinities, indicating the complex interplay of oceanic currents within this area. Notably, the proportion of species with demonstrated warm-water preferences (22%) was nearly double that of the species extending into colder regions further south (12%), indicating the pronounced influence of the warmer East Auckland Current (EAUC) in the BoP. Considering the significant influence of the EAUC in the BoP, as indicated by this study and supported by Donald (2021), an important question arises: how will the increasing frequency and intensity of marine heat waves transported from the tropics to New Zealand (Behrens et al., 2022) affect precious mesophotic ecosystems within the Bay of Plenty? These discoveries highlight the existence of a highly mixed sponge assemblage characterized by taxonomically diverse and interconnected communities. Their associations with oceanic currents make them a crucial area for further investigation and conservation efforts.

Among the 11% of new or undescribed species identified from ROV footage, the tumbleweed sponge emerged as an intriguing subject of investigation. Donald (2021) identified the tumbleweed sponge as a ‘probably new species’ due to its highly unusual morphology for a branching sponge in New Zealand, with no comparable species previously observed. It is most likely a Callyspongia species (Battershill, personal communication, June 2025). Its abundant presence was noted on Nukutai Reef between 40–50 m depth, in February 2021 (see https://youtu.be/7jfiwkn_W8M for original video transect footage from Donald, 2021). However, this study in the same region and including the same reef, conducted only a year and a half later, revealed a starkly contrasting scenario. The tumbleweed sponge was seldom encountered and when it was, specimen appeared markedly degraded (see Supplementary Figures 5T-U). In fact, they were only recorded as ‘possibly tumbleweed sponge’, due the obscurity and rarity of sightings in the ROV footage, the absence of physical samples, and limited prior records. There were only a few sightings of it and a full identification has not been possible (it has not been collected), hence confirmation of the taxonomy and ecology of this species remains to be carried out. There is however a possibility that the tumbleweed sponge originally identified by Donald (2021) has completely disappeared from these reefs. It is noted, however, that there are many sponge species that can have a rather ephemeral life history (Perkins et al., 2025) and we do acknowledge that the repeat surveys were not carried out along exactly the same transects, though they covered a much greater area of the Nukutai Reef than the preliminary survey in Donald (2021). Nonetheless, mesophotic sponge reefs are renowned as long-lived communities (Teixidó et al., 2009; Bell et al., 2014; Micaroni, 2022) and combined with the degree of sponge necrosis in other species, it would appear that these reef systems have been under significant stress. The size of the sponges alone, for example the large patches of encrusting sponges seen in ROV footage, indicates that the individuals observed in this study are likely to be decades old (Ayling, 1983). Consequently, this rapid shift in the tumbleweed sponge population, whether they have completely disappeared or there is still a remnant population, is a matter of considerable ecological significance.

The precise cause behind this shift is yet unknown, but a key initial observation was that all sightings except one, were exclusively at the deepest depths exceeding 80 m. Heat stress may be contributing factor to consider, as it is possible that the remnant population of the tumbleweed sponge persists only in deeper, cooler areas on these reefs (Idan et al., 2018). In addition, we observe that this community change coincides with a large marine heatwave event together with sedimentation effects caused by a major cyclonic event, Cyclone Gabriel (Harrington et al., 2023). It cannot be assumed that this combination of climatic events was the cause, nor that rapid fluctuations in these benthic species abundance are abnormal. However, it is important to note that these community shifts are happening elsewhere in this region, for example along shallow coastal reefs (Schiel et al., 2024), and predicted tropicalization of temperate waters (Wernberg et al., 2016). We propose that there is an opportunity for further monitoring at this site to examine longer term trends.

The widespread necrosis of sponges seen in this study provides evidence for sponge disintegration, morphology disruption and mortality under heat stress, which has been described as sponge rot by Ayling (1981), as sponges “melting” off the reef by Bell et al. (2023), and observed also as growth abnormalities in response to storm stress by Battershill and Bergquist (1990). New Zealand’s coastal waters have experienced unprecedented marine heat waves in recent years (Bell et al., 2023; Salinger et al., 2023). As previously mentioned, the BoP experienced a prolonged and intense marine heatwave from November 2021 to December 2022. The timing of this marine heatwave, in line with the consistent presence of necrotic sponges seen in this study across all reefs and depths, supports recent studies in pointing towards heat stress as the cause of large-scale sponge necrosis (Perkins et al., 2022; Botté et al., 2023; Bell et al., 2023, 2024).

A key observation is that sponges exhibited visible signs of necrosis across all surveyed depths, from 30 to 80.5 m. The importance of this ecologically is the permeation of heat stress down to at least 80.5 m deep, despite a possible thermocline, likely to occur during severe and prolonged cyclonic storm events. Depth is therefore not acting as a thermal refuge for these mesophotic sponge communities in times of increasingly severe weather events. Perkins et al. (2022) came to the same conclusion for a mesophotic cup sponge morphospecies on the east coast of Tasmania, which they surveyed at depths down to 83 m. This finding is important in the context of other studies that have recorded temperate sponge necrosis as, except for Perkins et al. (2022), most have only surveyed shallow ecosystems down to 20 or 25 m (Ayling, 1981; Bell et al., 2023, 2024).

Observations of healing sponges suggest that, despite the marine heatwave continuing from summer into winter 2022, recovery was facilitated by cooler winter sea temperatures. Healing sponges showed an estimated 2–3 months of recovery, which can be seen in photos where hard lines are present between white tissue and pigmented tissue as evidence of pinacoderm regrowth over scarred or lost sponge matrix (Figure 3A), as opposed to fuzzy, sloughing white tissue indicative of deterioration (Figure 3B) (Ayling, 1981; Cerrano et al., 2001).

Rapid sponge death may have occurred earlier in the marine heat wave and therefore the full impacts on benthic communities were missed by the time ROV surveys were undertaken. Images from ROV footage show tissue regression on extremely degraded sponges which are not included in the aforementioned list of common, ‘tough’ species (e.g., Figure 3C). The difference in tumbleweed sponge population between this footage and Donald (2021) also provides evidence that some sponges had died off rapidly before ROV footage was captured on our revisitation to these reefs. Because of this, the full extent of marine heatwave damage to benthic marine diversity and TME function is yet unseen for the Motiti reefs.

An often-overlooked factor is the compounding, or synergistic, effect that sedimentation (from land run-off) and heat stress can have on marine sponges (Strand et al., 2017; Scanes et al., 2018). Sedimentation and heat stress both significantly shape the Motiti mesophotic sponge assemblages, as evidenced by the widespread necrosis and sediment smothered rocky substrate. The greater SSA on Nukutai Reef is understandable as it is the northern side of Motiti Island, part of the same reef shelf that the island sits on, and therefore exposed to sediment run-off from Motiti Island itself (Crawshaw, 2022). By contrast, Te Poroiti and Motuhaku are both rocky pinnacles situated further away from Motiti Island. SSA increased with depth across all reefs, which is a natural trend that aligns with oceanic physical processes. Sediment gets lifted into suspension by wind-driven mixing in shallow waters, but falls out of suspension and smothers greater extents of benthic substrate on deeper reef areas, especially where it accumulates on flat areas (Airoldi, 2003). The resilience of sponges on these deeper reefs with high sedimentation levels largely depends on their species-specific ability to shed or tolerate sediments while under additional (heat) stress. This species-specific response invokes consideration of sponge morphology, physiology, the ability to ‘back-flush’ or ‘sneeze’ (Kornder et al., 2022), and elements of the sponge metabolism and biochemistry in response to increased energy demand coupled with less food and heightened pathogenic attack (Bannister et al., 2010, 2012).

Sedimentation also impacts the survival of sponges disintegrating under heat stress, by inhibiting the recruitment process of sponge propagules. Deposited sediment smothers benthic communities and as little as 10 mg cm^-2 can exclude sponge larval recruitment (Battershill, 1987; Abdul Wahab et al., 2019). The ability for a number of species of sponge to rapidly re-organize cellular integrity such that the morphology becomes more ‘fluid’ (melts), appears to be associated with a response to unfavorable conditions, such as heat stress, and is a mechanism that results in fragmentation (Ayling, 1981) and generation of asexual propagules (Battershill and Bergquist, 1990). This latter phenomenon heightens chances of survival of the fragmenting sponge, as the relatively large size of propagules can re-settle into available space, re-attach and grow (Battershill and Bergquist, 1990). However, if sediment is settling and smothering vacated reef space faster than sponges can re-attach, recruit and grow, for example following fragmentation and generation of asexual propagules from sponge rot, recovery of sponge assemblages after a die-off or disturbance event like a marine heatwave would be severely hindered. This compounding stress of marine heatwaves and heavy sedimentation could cause recruitment of new individuals to become limited to niches and slopes where minimal sediment deposition occurs.

Given the projected increase in the intensity and severity of future weather events (IPCC, 2023; Noll and Andrews, 2023), although heavy sedimentation may not currently constitute the prevailing conditions, it is likely a frequent and escalating stressor for Motiti Island reef communities. The compounding effect of these stressors is likely to play a role in the impacts of the marine heatwaves and will significantly alter the species composition of benthic communities, as we are starting to see in this study.