Pollination is a key ecological process sustaining biodiversity and food security, yet global patterns of plant–pollinator specialization remain unresolved. Using a global dataset of quantitative plant–pollinator networks (>3,400 networks, >110,000 interactions), we show that specialization is latitudinally structured, but not as a single directional pattern. We find no consistent monotonic increase or decrease in specialization towards the Equator. Instead, network-level and pollinator specialization peak at low northern latitudes near the tropical–subtropical boundary, whereas plant specialization is hemispherically asymmetric. Climate is most often the best-supported predictor of specialization compared with latitude, biodiversity and environmental productivity. Network-level specialization declines with increasing mean annual temperature, with strong group-specific differences. Precipitation exerts nonlinear effects: plant specialization peaks at intermediate precipitation but is lower at the wettest sites, whereas pollinator specialization is lower at intermediate precipitation, although often in interaction with temperature. Pollinator groups diverge strongly, especially in contrasting precipitation-related patterns between birds and insects. These findings show that global plant–pollinator specialization is structured by latitude and climate in complex and taxon-specific ways, with implications for uneven responses of pollination networks to future climatic change.

Semi-arid ecosystems are shaped by frequent wildfires, making the management of landscape disturbances a major conservation challenge in this region. Ant–plant nectar interactions, driven by mutualisms, contribute to ecosystem functioning and biodiversity, with interaction networks a poorly understood aspect of ecosystem recovery from disturbance. In this study we characterised ant interactions with nectar resources in the Scotia mallee region of Australia across three post-fire ages: 35 years (long unburnt), 10 years (old-intermediate) and 6 years (young-intermediate). We did this by describing the number and diversity of interactions between ants and nectar resources and quantifying different ecological network indices. Long unburnt sites were characterised by a higher number of distinct ant–plant interaction combinations, as well as a higher network modularity (Q) and species specialization (d′), reflecting increases in the diversity of plant nectar resources with time since fire. In contrast, the connectance (C) was higher in young-intermediate sites due to frequently repeated interactions between generalist ants and nectar resources, primarily honeydew-producing hemipterans. A comparatively higher incidence of ant–nectar interactions was also observed in young-intermediate sites despite lower vegetation cover, likely due to habitat openness and the dominance of opportunistic ants. Our findings suggest that greater plant diversity in long unburnt sites supports more complex ant-interaction networks indicating higher stability and ecological resilience. Overall, the study highlights the importance of understanding ant–plant interaction networks for assessing biodiversity and informing better management decisions.

Background and Aims Species with small geographical ranges provide insights into adaptation, speciation and genetic drift, while also presenting clear conservation challenges. Homoranthus A.Cunn. ex Schauer (Myrtaceae), an Australian genus with many narrow endemics, offers a model for understanding how ecological and spatial factors drive diversification. We examined a regional hotspot with a high number of Homoranthus narrow endemics to assess patterns of genetic diversity and inform both evolutionary understanding and conservation planning.

Methods We generated genome-wide single-nucleotide polymorphism data using DArTseq for 282 individuals across 13 Homoranthus species (40 % of the genus), including 10 narrow endemics, to assess population genetic structure and diversity.

Key Results All species showed strong genetic isolation, even over a few kilometres, with populations diverging within hundreds of metres. Homoranthus lunatus includes two highly divergent, non-sister lineages, suggesting taxonomic revision. Inbreeding was common but unrelated to range size, and heterozygosity remained moderate, indicating intrinsic buffering of genetic diversity. Genome sizes were notably small relative to other angiosperms.

Conclusions Ecological isolation, life-history traits, and limited dispersal drive both speciation and extinction risk in Homoranthus. Diversification and endemism are linked to spatial isolation, highlighting the need for conservation strategies that address ecological connectivity in addition to species protection.

Weeds pose a significant challenge in Australia, costing the livestock industry an estimated AUD 2,409 million annually due to control measures and lost production. The application of computer vision for weed detection in agriculture is an expanding area of research, offering the potential to develop autonomous, site specific weed management (SSWM) solutions that minimize herbicide use and reduce the labour needed for weed control. Using images collected by autonomous ground vehicles in Southeast Australian pastures infested with a range of weeds, several computer vision techniques were considered.

In one study, a pipeline was developed using local binary patterns, histograms of oriented gradients, and colour features, and an extreme learning machine, to produce a fine, patch-based segmentation of images containing serrated tussock, thistle species, white horehound, and Bathurst burr, yielding a mean Intersection over Union (mIOU) of 87.6, 81.6, 79.5 and 87.1 respectively. The fine-grained segmentation produced by this model enabled more accurate localization of individual weed plants, which should enable a greater precision in herbicide application.

In another study, a dual-purpose network for plant segmentation and spray point detection was developed using a novel convolutional neural network (CNN) architecture incorporating ConvNeXt modules. This network was tested on a dataset of serrated tussock and achieved a mean Intersection over Union (mIOU) of 80.7 for segmentation and an F1-score of 79.6 for spray point detection, even under field conditions not encountered during training. These results outperformed state-of-theart dual-purpose networks. Furthermore, when a novel normalization technique, termed HistMatch, was applied prior to input, the network’s performance improved to an mIOU of 85.4 and an F1-score of 80.6, respectively.

The dual-purpose network was then tested on a multispecies dataset of serrated tussock, white horehound, variegated thistle and stinging nettle, yielding mIOU values of 88.8 and 82.6 for the plant and spraypoint tasks respectively. This demonstrated that the network was able to generalize well to the situation in which multiple species of weed are present in a field.

The studies presented in this thesis have advanced the understanding of computer vision techniques for weed localization in pastures in several ways. Firstly, the ELM-based segmentation approach demonstrated greater precision in weed localization compared to previous methods that relied on classification or bounding box techniques. Additionally, the development of a dual-purpose network for plant segmentation and spray point localization in pastures represents the first of its kind, enabling more flexible treatment options depending on whether herbicide is applied to the entire plant or just its center. HistMatch normalization proved to be a superior technique compared to other normalization methods, particularly in scenarios where illumination and growing conditions in the test set differ from those encountered during training. Further testing of this dual-purpose network showed that it remains robust even in complex scenarios where multiple weed species are present in the field.