Abstract :
[en] Large predators can shape ecosystem functioning by structuring food webs and amplifying vertical trophic complexity, yet they are often among the most vulnerable taxa under rapid environmental change. In many ecosystems, this influence is closely tied to body size: large predators tend to occupy the highest trophic positions, undergo pronounced ontogenetic shifts in diet as they grow, and can create strong top-down effects that propagate through food webs. Despite this, most work on large predators and food web structure has focused on birds, mammals and fishes, whereas amphibians have received comparatively little attention, even though a few lineages include exceptionally large, fully aquatic species that may play analogous roles. In river ecosystems in particular, vulnerability is intensified by the dendritic structure of habitat networks, which constrains dispersal, amplifies the impacts of longitudinal fragmentation, and can make populations especially sensitive to both climatic shifts and physical barriers.
The Japanese giant salamander (Andrias japonicus), a fully aquatic, endangered amphibian and one of the world’s largest amphibians, offers a rare opportunity to bring these aspects together in a single model system: (i) identifying and quantifying key threats acting at broad spatial scales, and (ii) testing how extreme body size variation associated with gigantism shapes trophic ecology and community structure. In this thesis, we aimed to (1) improve understanding of the main threats faced by the Japanese giant salamander, focusing on climate change and habitat fragmentation by dams, and (2) characterise its trophic ecology from the individual to the community level, explicitly evaluating gigantism as a potential ecological advantage shaping trophic position, niche relationships and food web organisation.
To address climate-related vulnerability across the species’ range, we aimed to assess current habitat suitability and forecast distribution shifts under climate change using species distribution models (SDMs). We modelled habitat suitability for the Japanese giant salamander and projected future suitability for 2050, 2070 and 2090. In addition, to evaluate habitat fragmentation in river networks at the watersheds levels, we aimed to quantify dam-driven longitudinal disconnection within the distribution range of the Japanese giant salamander. We compiled a detailed barrier inventory for eight watersheds and adapted a Dam Fragmentation Index (DFI), while also characterising barrier density and height classes and measuring the proximity of salamander locations to barriers. To characterise trophic ecology and its size dependence, we aimed to quantify ontogenetic dietary shifts (ODSs) and size-related variation in trophic position by combining stomach contents with stable isotopes (δ¹³C, δ¹⁵N). We then scaled up to the food web level by sampling individuals from other major consumers, modelling trophic positions and isotopic niche metrics, and identifying dominant basal energy sources. Finally, we aimed to test whether toxicity, coupled with aposematic traits, can provide trophic advantages comparable to gigantism by analysing trophic coexistence between the Japanese giant salamander and the fire-bellied newt (Cynops pyrrhogaster) using stable isotopes and Bayesian mixing models.
At the broad scale, climatic variables dominated the SDMs, indicating that the distribution of the Japanese giant salamander is strongly constrained by climate. Suitable areas were associated with moderate precipitation during cold and wet seasons and mild summer temperatures, with moderately steep surrounding environments further favouring suitability, whereas land-use predictors contributed comparatively less. Across future scenarios and time horizons (2050–2090), projections consistently indicated a major contraction of suitable habitat, signalling substantial climate-driven loss of future suitability within the current distribution range. River fragmentation analyses revealed pervasive barrier pressure across watersheds. We identified more than 2,000 river barriers, of which only ~5% were equipped with fishways potentially usable by giant salamanders. We highlighted a high fragmentation, and salamander populations frequently occur in close proximity to structures likely to impede their movement. Together, these findings suggest that fragmentation by dams represents a major constraint on longitudinal connectivity and may reduce the capacity for movements required for breeding and gene flow. Trophic analyses showed that gigantism is tightly linked to a strong reorganisation of diet and trophic role along ontogeny. Trophic position increased nonlinearly with body size from ~3.0 to ~5.1, with a marked inflection point at 39 cm snout–vent length that coincided with a clear dietary transition from predominantly aquatic insects to larger prey such as fish, anurans and freshwater crabs. This indicates that extreme growth enables access to higher trophic levels and prey types largely inaccessible to smaller conspecifics. At the community scale, isotopic analyses revealed pronounced size-structured interactions and a strong effect of large salamanders on food web architecture. Smaller salamanders overlapped trophically with mesopredators (e.g., fish, prawns, turtles), suggesting potential competition at smaller size classes, whereas large salamanders occupied higher trophic positions than all other consumers and escaped niche overlap with the rest of the community. Large individuals expanded the community δ¹⁵N range and contributed substantially to trophic evenness and to the total community isotopic niche space, indicating that the largest salamanders disproportionately increase food web vertical complexity and trophic diversity. Basal source analyses further showed that both aquatic and terrestrial energy pathways supported the food web, highlighting coupled in-stream and riparian contributions to community trophic structure. In the comparative predator system, newts reached high trophic levels despite small body size and exhibited no isotopic niche overlap with salamanders. Instead, the two predators were supported by two largely distinct yet interconnected trophic pathways within the same food web, both drawing on aquatic (periphyton) and terrestrial (leaf litter) sources. These results support the idea that toxicity and aposematism can facilitate access to high trophic positions, potentially by enabling “predator-free” foraging opportunities, and thus represent an alternative route to trophic dominance beyond body size.
By integrating range-wide modelling, network fragmentation assessment, and multi-scale trophic ecology, this thesis shows that the Japanese giant salamander faces converging pressures from climate-driven reductions in habitat suitability and pervasive barrier-driven fragmentation in river networks. At the same time, gigantism emerges as the central driver linking individual ecology to community organisation: it generates a step-change in trophic position and diet composition during ontogeny and, in the largest individuals, produces apex-predator status, reduced niche overlap with other consumers, and a disproportionate contribution to whole-community trophic structure. These findings imply that conservation of the Japanese giant salamander is not only a species-level priority but also a means of maintaining trophic architecture and vertical complexity in river ecosystems by specifically preserving the presence of the largest individuals. In practical terms, effective conservation should combine climate-aware planning (identifying and protecting future suitable areas and likely refugia) with connectivity-focused river management (mitigating, adapting or removing barriers, and implementing passage solutions compatible with salamander biology), because the capacity of populations to persist in a changing climate is likely to be constrained by fragmentation in dendritic river networks.
