Coral reefs are biogenic three-dimensional structures that provide habitats for many species to thrive and evolve (Graham, 2014; Graham and Nash, 2012; Sanna et al., 2023). The structural complexity (hereafter complexity) of these habitats mediates important ecological functions and services through its influence on ecological processes such as herbivory, predation, and nutrient cycling (Helder et al., 2022; Shantz et al., 2015; Syms and Jones, 2000; Verges et al., 2011). Furthermore, through its impact on diversity, complexity is increasingly seen as a key indicator of reef resilience (Dang et al., 2020), with application to reef restoration efforts (Yanovski and Abelson, 2019).

As the primary architects of reefs, stony corals play an important role in the variation of complexity. Thus, coral richness, cover, and morphology have been found to be significantly related to the complexity and extent of biotic interactions within a reef. For instance, marine protected areas that promote coral cover and richness have more complex reefs compared to non-protected areas (Montero-Serra et al., 2019). In contrast, global and local disturbances are responsible for reef flattening through impact on corals, which can be observed at all spatial scales from local to regional (Alvarez-Filip et al., 2009; Bauman et al., 2022). Chronic stressors such as marine activities (swimming, snorkeling, and scuba diving) have been shown to have negative effects on reef complexity (Chen and Dai, 2021). Similarly, coral bleaching can have pervasive effects on complexity due to gradual erosion of calcium carbonate skeletons after coral death (Roff et al., 2020). On the opposite end, other catastrophic events such as typhoons often result in sudden and dramatic loss of complexity after corals have been extirpated from benthic communities (Pascoe et al., 2021; Yuval et al., 2023). Because of their disproportionally high contribution to the complexity and vulnerability to climate-related stressors in tropical areas, stony corals have eclipsed the numerous abiotic and biotic drivers responsible for the spatiotemporal variations in reef complexity.

Environmental factors play pivotal roles in shaping benthic composition and contributing to observed benthic heterogeneity along both disturbance and natural gradients (Lin et al., 2024; Reverter et al., 2022; Roberts et al., 2015). In turn, communities can vary in complexity depending on the organisms that compose them (Richardson et al., 2017; Rovellini et al., 2024; Tebbett et al., 2020). For example, in tropical areas regularly exposed to typhoons, resilient reefs may exhibit low complexity characterized by a dominance of coral with flattened morphologies (Ribas-Deulofeu et al., 2021). Similarly, in habitats considered marginal to shallow tropical reefs, such as mesophotic coral ecosystems, it is common to observe taxa other than scleractinians that contribute heavily to the overall complexity (Kahng et al., 2017). The same is true for non-reef environments such as coralligenous habitats where octocorals can be at the origin of a complex vertical stratification (Palma et al., 2018). Other canopy-forming organisms of Marine Animal Forests are further well-known as structurally analogous to scleractinian corals but thrive in other environmental conditions (Rossi et al., 2017).

Quantifying the overall complexity of a habitat is elusive due to the multifaceted nature of complexity, which manifests at different spatial scales (Yanovski et al., 2017) and spans a gradient from micro- to macro- scale complexities (Ribas-Deulofeu et al., 2021). Similarly, complexity is scale-dependent, making it challenging to summarize with a single metric (Ribas-Deulofeu et al., 2021). In benthic habitats, biotic and abiotic features all contribute to the different facets of complexity that distinctively influence biodiversity and ecosystem productivity interactions (González-Rivero et al., 2017). Typically, sessile benthos tend to increase the availability of microhabitats (Sarà, 1986), and associate with an abundance of some small and specialized fish species (e.g. Munday et al., 1997). Geologic features primarily shape large-scale topographic features (Helder et al., 2022; McCarthy et al., 2022), which can lead to higher biomass in large, high-trophic level fish species (Aburto-Oropeza et al., 2015). High complexity can capture both aspects and enhance fish biomass and abundance while structuring food webs and boosting productivity (Rogers et al., 2014). On coral reefs, the chain method has been traditionally used to measure ‘linear rugosity’ (Risk, 1972), typically capturing complexity at small to medium scales according to the size of the link. Additionally, the scale of complexity determined from depth gauge profiles (Dustan et al., 2013) can be adjusted using polynomial functions on rugosity indices (Ribas-Deulofeu et al., 2021). On the other end, visually scoring complexity of the seascape remains common (Graham et al., 2015; Polunin and Roberts, 1993; Wilson et al., 2007), while relying on suggestive perceptions of micro- and macro-rugosity in the observer (Ribas-Deulofeu et al., 2021). Besides linear rugosity, numerous metrics can be applied to estimate the complexity of larger reef areas (Fukunaga and Burns, 2020; Mills et al., 2023). Digital elevation models (DEMs) generated from three-dimensional (3D) benthic reconstructions using Structure-from-Motion (SfM; Westoby et al., 2012) now allow the simultaneous extraction of multiple metrics over large reef areas. This ensures that complexity assessment does not rely solely on single and small-scale indices. This approach has now imposed itself upon the characterization of the multiple facets (Urbina-Barreto et al., 2022) and benthic drivers (McCarthy et al., 2022) of complexity. Most studies in coral reefs have focused on changes in complexity metrics after catastrophic events such as coral bleaching and other disturbances (Burns et al., 2016; Ferrari et al., 2016), or along gradients of disturbances (Abdurrachman et al., 2024; Chen and Dai, 2021; Lyons et al., 2015) negatively affecting overall coral cover (Alvarez-Filip et al., 2009; Graham, 2014) or particular coral morphologies (Magel et al., 2019). Less well known, however, is the natural variation in complexity and its relationship with benthic composition along environmental gradients, as observed in the transition zones between tropical and temperate latitudes. Indeed, many physical and biological factors such as temperature, light, aragonite saturation, nutrient availability, larval dispersal, and competition, can affect the composition of the benthic assemblages across latitudes (Abrego et al., 2021; Lin and Denis, 2019; Smale et al., 2010) which could ultimately affect reef complexity.

To address these knowledge gaps, we investigated the interaction between benthic composition and complexity along a latitudinal gradient where benthic communities shift from a dominance of scleractinian corals (tropical) to a dominance of crustose coralline algae (subtropical). Here, we first identified key metrics to capture changes in complexity along latitudes and between study regions. We then typified the differences in complexity between tropical and subtropical zones, emphasizing the contribution of taxa other than scleractinians to the complexity in transition and non-coral reef areas. The results have improved the understanding of the interplay between benthic composition and complexity. Furthermore, the knowledge gained from this study will be fundamental to better predict the ecological consequences of reef habitat decline in tropical areas and their expansion in subtropical areas.

This study investigates the biotic drivers of benthic complexity in a transition area spanning tropical and subtropical assemblages. A set of structural metrics is selected to capture variations in benthic complexity across latitude. A decrease in fine-scale complexity with increasing latitude is associated with an increased benthic contribution of zoanthids and CCA, along with a loss of corals characterized by intricate —mainly branching— morphologies. The presence of unstable substrate negatively influences fine-scale complexity, but was positively related to coarse-scale complexity, which is linked to regional specificities in topographic features and the presence of large boulders at some sites. Overall, this study sheds light on the variations in complexity along latitudinal gradients and highlights the benthic organisms that play central roles in driving these changes. Importantly, these shifts in complexity may be partly related to the distinct environmental conditions in tropical and subtropical regions, making it crucial to differentiate them from changes in complexity observed along disturbance gradients.

Many metrics are relevant for measuring reef complexity (Fukunaga and Burns, 2020; Remmers et al., 2024). The choice of metrics is often subjective, with studies typically focusing on a limited number of metrics and scales (e.g. Asner et al., 2021; Chen and Dai, 2021). These metrics can describe changes in complexity along spatial gradients of disturbance (Chen and Dai, 2021; Magel et al., 2019) or along temporal shifts from coral to turf algae dominance, as observed following coral bleaching and subsequent mortality (Graham et al., 2006; Roth et al., 2018). Selecting appropriate metrics is particularly relevant when anticipating the loss of specific taxa (e.g., stony corals or corals of particular morphologies) that contribute to certain facets of complexity. However, when capturing the multiple aspects of reef complexity arising from various taxa (e.g., Rovellini et al., 2024; this study), careful consideration of the appropriate metrics is key to describing turnover in organisms potentially associated with changes in reef structure. Here, 11 metrics are identified as relevant for describing changes in complexity from tropical coral reefs to subtropical coral communities. Our approach removes redundant information present in certain metrics, and the selected set acknowledges structural complexity at different spatial scales. Thus, fine, medium, and coarse scale metrics are represented, capturing seascape variations affected by changes in benthic taxa abundance and environmental setting.

The variation in structural complexity across latitudes is influenced by both biotic and abiotic factors. On one hand, fine-scale complexity decreases with the decline of intricate coral morphologies and emergence of other sessile taxa at higher latitudes. This pattern, driven by natural shifts in community composition along the latitudinal gradient (Lin et al., 2022), appears to be independent of any loss of coral cover due to disturbances. In our case, this is best captured by the variations in metrics D [1-2 cm] and D [2-4 cm], which indicate a loss of microhabitats along the tropical-subtropical transition. On the other hand, regional topographic features become more pronounced in areas of lower coral cover and accretion, where boulder fields are interspersed with sand. PROC [32 cm] is the metric of choice here to capture this feature, which reveals regional differences but does follow a clear trend in latitudes. This pattern aligns with previous studies identifying biotic components as the main source of fine-scale rugosity, while coarse-scale rugosity is predominantly shaped by geological settings (McCarthy et al., 2022; Ribas-Deulofeu et al., 2021).

Healthy tropical reefs typically host a rich diversity of coral species with various morphologies, among which branching forms often dominate shallow waters (Tamir et al., 2019) and contribute to high fine-scale complexity (Rovellini et al., 2024). The decline of these branching forms, which are particularly sensitive to heat stress and bleaching, is associated with structural flattening and the global degradation of coral reefs. However, in tropical systems exposed or regularly impacted by typhoons, reefs may naturally appear ‘flat’ due to the dominance of corals with flattened morphologies (Ribas-Deulofeu et al., 2021). Similarly, at mesophotic depths where light attenuation is significant, encrusting to foliose morphologies tend to dominate (Pérez-Castro et al., 2023). In ‘marginal’ reefs at high latitudes (Schoepf et al., 2023), where environmental conditions depart from tropical ones, massive, foliose, and/or encrusting corals often dominate while species possessing intricate morphologies are largely absent (Sommer et al., 2014). Besides the intrinsic limitations of these species in withstanding subtropical environments, other factors such as competition for space with other sessile organisms (Abrego et al., 2021; Reimer et al., 2021) likely play a key role in the shift in structural complexity observed at higher latitudes.

Also influenced by reef exposure (Rabelo et al., 2015; Steneck, 1986), zoanthids and CCA are commonly distributed in subtropical coral assemblages (Lin and Denis, 2019; Lin et al., 2024; Reimer et al., 2021). While CCA can locally contribute to reef accretion and form large algal reefs (e.g., Taoyuan algae reef in Northwest Taiwan; Dai et al., 2009), CCA and other turf algae typically overgrow rocky surfaces, forming thin sheets that follow substrate contours. They contribute little to fine-scale complexity but their presence is associated with an increase in coarse-scale complexity when boulders are present. This feature is also associated with an increase in unstable substrate on the East Coast because sand tends to accumulate around large boulder rocks. Thus, from the perspective of fine-scale complexity, a reef ‘flattening’ is observed with increasing latitude. However, this pattern substantially differs from complexity changes observed along sequences of degradation, as it is driven by a turnover between tropical and subtropical species, with topographical features becoming more prominent as coral accretion diminishes. Although the changes in complexity appear to align with known environmental and natural gradients, we cannot completely disregard the potential influence of land-use drivers, which are known to contribute to widespread coastal degradation in Taiwan (Lin et al., 2024) and as observed in other reefs with human footprint (e.g. Lybolt et al., 2011). For some taxa, their response to stress may be confounded with their biogeography (e.g. branching corals). This emphasizes the importance of disentangling the effects of temperature, light and other environmental factors that are often confounded with latitudinal changes in complexity. To address this, examining changes in complexity across depth gradients can provide valuable insights, as light becomes the primary driver of variation at both the individual (morphology; López-Londoño et al., 2024) and community (composition; López-Londoño et al., 2022) levels.

Understanding the drivers of structural complexity across latitudes is crucial for effective reef management and conservation, especially in the face of climate change. The observed shift in benthic complexity, driven by changes in species composition and regional topography, suggests that reefs in biogeographic transition zones may respond differently to environmental stressors compared to tropical reefs in terms of structural complexity. These findings highlight the importance of tailoring conservation strategies to specific latitudinal contexts (Lin et al., 2024). For instance, the dominance of zoanthids and CCA in subtropical reefs may indicate a natural shift towards more resilient, yet less complex reef structures. However, this ‘flattening’ could reduce the habitat complexity that many tropical species depend on, potentially leading to declines in biodiversity across latitudes as well as different functions and services associated with these marginal systems. Therefore, conservation efforts should consider the unique biotic and abiotic factors shaping reef complexity in these regions, and prioritize actions that maintain or enhance structural complexity, which is vital for supporting local diversity and ecosystem functions. A key question in restoration is whether transplanting species that contribute little to structural complexity is ecologically relevant, especially if core reef functions like growth and habitat creation are not directly influenced by the corals being targeted (Bellwood et al., 2024). Although branching species are often selected for restoration projects for reasons of practicality (Boström-Einarsson et al., 2020), they would be far from ideal for use in subtropical regions as characterized here. Interestingly, branching corals, once transplanted in these conditions appeared to be more sensitive to predation and far less successful than encrusting (Carballo-Bolaños et al., 2024). Additionally, monitoring the transition from tropical to subtropical assemblages will be key to predicting and mitigating the impacts of climate change on reef ecosystems, ensuring the long-term resilience of these critical habitats.

In linking the complexity matrix to reef composition, several non-reef building taxa or categories are associated with specific facets of the structural complexity. Unfortunately, as in many other studies, actual contributions to complexity and indirect associations cannot be distinguished, such as the presence of unstable substrate, which is associated with an increase in coarse-scale complexity. Regional estimates of calcification rates in various reef-building taxa could improve future modeling of reef habitat expansion. Increasing the number of sampling sites and long-term monitoring at some sentinel stations would be a major improvement (Denis et al., 2020), but the manual and detailed annotation of the orthomosaics is a significant limitation. AI-assisted annotation tools, such as TagLab (Pavoni et al., 2022), offer promising prospects for fast and accurate automated semantic segmentation over larger photogrammetric plot extent. This could partially help uncover the drivers behind the portion of unexplained variance that remains to be addressed.

All methods employed were non-invasive. Sites were surveyed with the following permits: No. 1110109623, issued by Taitung County Government (www.taitung.gov.tw) ; No. 1110002597, issued by Kenting National Park Headquarters (https://www.ktnp.gov.tw/en/); No. 1113606425, issued by Fisheries and Fishing Port Affairs Management Office – New Taipei City Government (https://fishery.ntpc.gov.tw); and No. 1110300347, issued by Northeast and Yilan Coast National Scenic Area (www.necoast-nsa.gov.tw).