Melissa Thomas

Identifying insect pests, whether as adults, larvae, or eggs, is critical in pest management. Computational learning algorithms have demonstrated strong potential in achieving high identification performance, but these methods typically require large, balanced, and well-annotated datasets. This creates a challenge for insect pest identification, as rare species, despite often being the most damaging to crops, are underrepresented in available datasets. Moreover, annotating large-scale datasets is both costly and labour-intensive. To address this issue, we develop a semi-supervised learning approach, Cost-Focal FixMatch, which extends the widely used FixMatch framework by integrating class-aware reweighting and focal loss to better handle class imbalance. Specifically, we introduce a simple yet robust method for applying class weighting in cross-entropy and focal loss functions. The proposed method generates higher-quality pseudo labels compared to the baseline, ensuring better learning. We evaluate our approach using a repurposed IP102 dataset, which comprises four primary insect life stages, and a mixed IP102 dataset, where the class labels jointly represent insect species and their corresponding life stages. Our method considerably improves the classification of minority classes, achieving a notable increase in recall for the Larva class from 64% under the baseline FixMatch to 82% using MobileNetV3Small backbone. On the Mixed IP102 dataset, our approach achieves almost 9% better improved average recall than the baseline FixMatch built upon the EfficientNetV2S network.

Insect pests pose a significant risk to agriculture and biosecurity, reducing crop yields and requiring effective management. Accurate identification of early life stages is often required for effective management but is generally reliant on expert evaluation, which is both costly and time-consuming. To address this, we use a deep learning-based approach for insect species and life-stage classification from digital images. We repurposed the IP102 dataset by adding detailed annotations for four life stages — egg, larva, pupa, and adult — alongside the original species categories. Two deep learning models, based on ResNet50 and EfficientNetV2M, were tested for classification accuracy in this dual-layered identification task. Although both models accomplished the task well, the EfficientNetV2M model performed slightly better than the ResNet50, achieving 72.4% precision, 72.1% recall, and an F1-score of 72.0%. Our results demonstrate the potential of deep learning for automated insect species and life-stage classification, providing a high throughput and efficient solution towards agricultural monitoring and pest management.

Interactions between natural and sexual selection have been integral to the development of sexual selection theory, yet the role of natural selection in sexual trait evolution has received far less empirical attention than the role of sexual selection, and the extent to which natural and sexual selection are aligned remains an important empirical question.

Insect cuticular hydrocarbons (CHCs) provide a good model for exploring interactions between natural and sexual selection because of their dual roles in sexual signalling and desiccation resistance.

We used the geometric framework for nutrition to explore the potential for, and magnitude of trade-offs between male attractiveness, desiccation resistance and the abundance and blend of CHCs in the cricket Teleogryllus oceanicus.

Attractiveness and desiccation resistance were both maximized on a diet that was rich in carbohydrate relative to protein. Although maximum expressions occurred at significantly different locations in nutritional space, the difference in angles between maximum attractiveness and desiccation resistance were small.

Attractiveness and desiccation resistance were both associated with CHCs of moderate, rather than maximum, abundance and with a blend that was rich in two shorter chained alkenes relative to a single longer chained alkane. Partial correlations between fitness traits suggested that CHCs contribute indirectly to male attractiveness via their role in desiccation resistance.

Our findings are consistent with the idea that natural and sexual selection on CHCs can be broadly aligned, facilitating rapid adaptation to ecological conditions and promoting speciation.

Biosecurity activities primarily include pre-border and border quarantine, post-border surveillance and post-border eradication. Budget allocated to quarantine and surveillance activities ultimately influence the expenditure and success rate of eradication campaigns. Optimal portfolio allocation examined in previous research is susceptible to potential severe uncertainties existing in ecology and in the behaviour of invasive species itself. These uncertainties, together with a limited budget, make it difficult for decision makers to allocate the total management budget to each biosecurity activity in a robust manner.

Info-gap decision theory is applied to model the severe uncertainty in invasive species management, and robust optimize the total management cost.

This research shows that using a combination of pre-border and border quarantine (to reduce the incursion probability) and post-border surveillance (to enable early detection and rapid response), enables decision makers to be more robust to potential uncertainty.

Further, it is reported that investment in quarantine that is more cost-effective should outweigh that in surveillance, in line with precautionary principle.

Increasing the estimated population threshold for surveillance detection also gains more robustness.

Synthesis and applications: Portfolio allocation options developed in this research provide decision makers with a way to manage the invasive species spatially, cost-effectively, and confidently by allocating the total management budget in a robust manner. The methods outlined in this research can not only be applied to invasive species, but also the conservation of endangered species that are constrained by severe uncertainty in ecological modelling and limited resources.

1. Biosecurity activities primarily include pre-border and border quarantine, post-border surveillance and post-border eradication. Budget allocated to quarantine and surveillance activities ultimately influence the expenditure and success rate of eradication campaigns. Optimal portfolio allocation examined in previous research is susceptible to potential severe uncertainties existing in ecology and in the behaviour of invasive species itself. These uncertainties, together with a limited budget, make it difficult for decision makers to allocate the total management budget to each biosecurity activity in a robust manner.

2. Info-gap decision theory is applied to model the severe uncertainty in invasive species management, and robust optimize the total management cost.
This research shows that using a combination of pre-border and border quarantine (to reduce the incursion probability) and post-border surveillance (to enable early detection and rapid response), enables decision makers to be more robust to potential uncertainty. Further, it is reported that investment in quarantine that is more cost-effective should outweigh that in surveillance, in line with precautionary principle.

3. Increasing the estimated population threshold for surveillance detection also gains more robustness.

4. Synthesis and applications: Portfolio allocation options developed in this research provide decision makers with a way to manage the invasive species spatially, cost-effectively and confidently by allocating the total management budget in a robust manner. The methods outlined in this research can not only be applied to invasive species, but also the conservation of endangered species that are constrained by severe uncertainty in ecological modelling and limited resources.

Grassland alvar is a rare plant community that occurs throughout North America and northern Europe, and may require control of encroaching vegetation to be maintained or restored. We evaluated the hypothesis that restoration techniques used to restore the alvar ecosystem do not lead to declines in small mammal abundance. More specifically, we used a BACI design to compare how two methods of vegetation control, prescribed burns and mechanical removal, affected small mammal populations. The restoration was conducted beginning in 2019 on Pelee Island, Ontario, Canada. Live trapping of small mammals and associated vegetation sampling were conducted before and after on treatment and control locations. The only small mammal species to be caught during the study was the white-footed mouse (Peromyscus leucopus (Rafinesque, 1818)), which had an observed decrease across all treatment sites and the control site. Generalized linear mixed effects models demonstrated that the main effects of treatment and year best explained mouse abundance at the site level. Interannual variability appeared to explain more variation in mouse abundance than treatment effects. Our study did not provide strong evidence that the vegetation control measures we employed might limit white-footed mouse abundance.

Global declines in insect abundance are of significant concern. While there is evidence that climate change is contributing to insect declines, we know little of the direct mechanisms responsible for these declines. Male fertility is compromised by increasing temperatures, and the thermal limit to fertility has been implicated as an important factor in the response of insects to climate change. However, climate change is affecting both temperature and hydric conditions, and the effects of water availability on male fertility have rarely been considered. Here we exposed male crickets
Teleogryllus oceanicus
to either low or high‐humidity environments while holding temperature constant. We measured water loss and the expression of both pre‐ and postmating reproductive traits. Males exposed to a low‐humidity environment lost more water than males exposed to a high‐humidity environment. A male's cuticular hydrocarbon profile (CHC) did not affect the amount of water lost, and males did not adjust the composition of their CHC profiles in response to hydric conditions. Males exposed to a low‐humidity environment were less likely to produce courtship song or produced songs of low quality. Their spermatophores failed to evacuate and their ejaculates contained sperm of reduced viability. The detrimental effects of low‐humidity on male reproductive traits will compromise male fertility and population persistence. We argue that limits to insect fertility based on temperature alone are likely to underestimate the true effects of climate change on insect persistence and that the explicit incorporation of water regulation into our modeling will yield more accurate predictions of the effects of climate change on insect declines.
The thermal limit to fertility has been implicated as a factor in the response of insects to climate change. However, climate change is also affecting hydric conditions. We demonstrate strong effects of hydration stress on mating behavior and male fertility that will adversely affect insect reproduction under climate change.

Harmful and often catastrophic damages from the introduction of non-indigenous species (NIS) are widely acknowledged. While preventing the introduction of NIS through border and pre-border measures has been the first line of defence, post-border surveillance has recently attracted considerable attention as it increases the likelihood that small invasive populations will be found and eradicated quickly before they become widespread. We develop a novel and practical optimal surveillance model across space and a number of surveillance techniques for four different invasive pests, determining where and how to best allocate resources to detect and eradicate these pests. Our focus is on Barrow Island, a Class A Nature Reserve in Australia, home to thousands of native plants and animals, many endemic to the island, where limited industrial activity and environmental protection coexist. It is also home to the world's largest non-government quarantine and surveillance system. Our results provide a unique platform that finds the maximum net benefit from post-border surveillance expenditures across species, locations, and surveillance methods, thus protecting a key environmental asset.

Invasive species eradication campaigns often fail due to stochastic arrival events, unpredictable detectability and incorrect resource allocation. Severe uncertainty in model parameter estimates may skew the eradication policy results. Using info-gap decision theory, this research aims to provide managers with a method to quantify their confidence in realizing successful eradication of particular invasive species within their specified eradication budgets (i.e. allowed eradication cost) in face of information-gaps. The potential introduction of the Asian house gecko Hemidactylus frenatus to Barrow Island, Australia is used as a case study to illustrate the model. Results of this research demonstrate that, more robustness to uncertainty in the model parameters can be earnt by (1) increasing the allowed eradication cost (2) investment in pre-border quarantine and border inspection (i.e. prevention) or (3) investment in post-border detection surveillance. The combination of a post-border spatial dispersal model and info-gap decision theory demonstrates a novel and spatially efficient method for managers to evaluate the robustness of eradication policies for incursion of invasive species with unexpected behaviour. These methods can be used to provide insight into the success of management goals, in particular the eradication of invasive species on islands or in broader mainland areas. These insights will assist in avoiding eradication failure and wasteful budget allocation and labour investment.

The often complex cocktails of hydrocarbon compounds found on the cuticles of insects can serve both naturally and sexually selected functions, contributing to an individual's ability to withstand water loss and attract mating partners. However, whether natural and sexual selection act synergistically or antagonistically on a species' cuticular hydrocarbon (CHC) profile remains unclear. Here, we examined the ontogeny of the CHC profile in a species of cricket, Teleogryllus oceanicus, while manipulating humidity during development. We predicted that juvenile crickets should produce only those compounds that contribute to desiccation resistance, while those compounds contributing specifically to male attractiveness should be produced only at sexual maturity. Further, if attractive CHCs come at a cost to desiccation resistance as predicted by some models of sexual selection, then males reared under low humidity should be constrained to invest less in attractive CHCs. Crickets reared under low humidity produced more long-chain methyl-branched alkanes, alkenes and alkadienes than did crickets reared under high humidity. The abundance of n-alkanes was unaffected by humidity treatment. Sexual dimorphism in the CHC profile was not apparent until adult emergence and became exaggerated 10 days after emergence, when crickets were sexually mature. Males produced more of the same compounds that were increased in both sexes under low humidity, but the humidity treatment did not interact with sex in determining CHC abundance. The data suggest that CHC profiles which protect crickets from desiccation might have synergistic effects on male attractiveness, as there was no evidence to suggest males trade-off a CHC profile produced in response to low humidity for one associated with sexual signalling.