Abstract Invasive ants can cause enormous damage across multiple sectors and are targeted by eradication programmes where infestations are relatively small, to prevent spread and impact. However, ant eradication programmes can fail if detection and control techniques do not target species‐specific behaviours. Existing techniques for detecting invasive ants at low density (new incursions or the ‘mop‐up’ final phase of eradication) assume ants are foraging above‐ground. While tools and devices deployed above‐ground are effective for many species, some ant species forage below‐ground and can be difficult to detect above‐ground at low density using these methods. Darwin's ant ( Doleromyrma darwiniana ), a pest ant species in New Zealand, offers a unique opportunity to investigate the deployment of below‐ground lures for improved species‐specific management. This species appears to forage below‐ground more than other pest ant species currently managed in New Zealand, making eradication difficult. Here, we tested whether below‐ground lures are more effective at detecting Darwin's ant than lures deployed above‐ground. First, we used several lure types that are commonly used as ant attractants (carbohydrate, lipid, protein, combination) and presented these above‐ and below‐ground to compare relative effectiveness in attracting Darwin's ants. Second, we took the most attractive treatment (comprising lure type and presentation method) and tested this against the current best industry practice for monitoring (a commercial non‐toxic protein lure presented above‐ground) in low‐density detection trials. We found the odds of detection with our newly tested treatment (carbohydrate, below‐ground) were 2.5 times better than the above‐ground current best industry practice. Practical implication . We demonstrate that by taking foraging behaviour into account, detection at low density can be improved, thereby enhancing the probability of ant eradication. The new below‐ground deployment lure we have developed for pest ant management could be applied globally to pest ant management programmes, particularly where poor detection at low density is hindering eradication.
Abstract Males and females of sexually dimorphic ungulates exhibit distinct competitive abilities due to differences in body size, digestive function and role in reproduction. In the attempt to enhance the hunting experience or manipulate population dynamics, management may favour increasing the proportion of either sex. Depending on the relationship between the sexes, management practices may have different consequences for individuals and populations. Females are often considered the primary drivers of population dynamics of polygynous ungulates. The role of males is often overlooked, but mounting evidence suggests that the proportion of males within a population may have important effects on population dynamics, either because of impacts on mating systems or intersexual competition. The effect of males within ungulate populations may thus have important implications for management, particularly when intersexual interactions influence females. We used long‐term data on productivity (2007–2022) and body reserve dynamics (2015–2022) to quantify the influence of an increase in male proportion on population dynamics of free‐ranging mule deer Odocoileus hemionus . We examined if an increasing male proportion, in relation to population density, influenced: (i) individual females through a change in body reserve accumulation; (ii) female population growth rate. Higher male:female ratios were associated with reduced seasonal body fat accumulation in females. Similarly, higher male:female ratios were associated with lower female population growth rates in the following year, consistent with a potential carry‐over effect mediated through female body condition. Changes in male proportion affected body fat reserve accumulation and female population growth rate after accounting for climatic variation and density, at least with the levels observed in the study populations. Practical implication: A common shift in management of mule deer over the late 20th century has been an increase in the proportion of mature males in the population (i.e., higher male:female ratio). Our results suggest that such a management strategy may involve demographic trade‐offs, being associated with reduced female body condition and lower population productivity. These findings are increasingly pertinent as mule deer populations appear to be declining in many areas.
Abstract Pre‐ and post‐wildfire management is increasingly guided by fractional vegetation cover (FVC) products, that is, maps parameterized with satellite imagery, that are used to determine where, when and how vegetation can be managed to moderate ecosystem impacts of wildfire. Several recent FVC products have transformed the science and management of wildland fuels and fire by estimating the type and abundance of vegetation across continuous regional areas. Additional guidance can help inform how well FVC products are parameterized for such broad areas and how they can be applied to the smaller scales at which vegetation is most directly managed for wildfire. We asked how inference for pre‐ and post‐wildfire management would differ between the most common FVC products and benchmark field data in four scenarios of sagebrush‐steppe management in the Great Basin, USA. Case studies included herbicide treatments in California and Nevada, a livestock exclusion in Colorado, post‐fire seeding in Utah and wildfire risk assessment in Idaho. FVC products included the U.S. Department of Agriculture Rangeland Analysis Platform (RAP) and the USGS Rangeland Condition Monitoring and Assessment Program (RCMAP), both 30 m pixel resolution. Agreement in vegetation change or variability between FVC products and field data, specifically the variability in vegetation cover attributable to treatments or wildfire risks, was modest and evident mainly in terms of directionality and not magnitude. We found weak support for the concept of greater agreement among data sources across larger sites, when differences in vegetation across time or space were larger, and when the abundance of the vegetation of interest was moderate. Practical implication . Conclusions about management conditions would differ meaningfully if based on the benchmark field data compared to FVC products and also between FVC products. When using FVC products for management, the management question can be matched to the limitations of the data source or multiple lines of evidence can be integrated together.
Abstract Ancient woodlands in the UK face growing pressures from nearby development despite their recognition as ‘irreplaceable habitats’ with high biodiversity potential and cultural value. Effective mitigation is currently hindered by significant knowledge gaps around impacts from urbanisation, land use intensification and associated human activities on ancient woodland extent, condition and biodiversity. We used a multi‐stakeholder, participatory approach to identify high‐priority research questions to inform evidence‐based conservation policy, practice and mitigation strategies for strengthening ancient woodland protection. We developed an online survey to gather questions and invited participation from 143 experts and their wider networks including practitioners, researchers and policymakers in the environment, planning and construction sectors. Submitted questions were discussed and ranked by a subset of experts during two in‐person workshops, based on urgency, actionability, data needs and policy relevance. Qualitative data generated from workshop discussions were also analysed thematically to identify themes underpinning current policy, practice and research. Twelve priority research questions were identified. The highest‐ranked question focuses on the effective width and composition of vegetated ‘buffer’ areas necessary to protect ancient woodlands from impacts while also providing ecological benefits. Three key themes characterise the current research, policy and practice landscape: (i) evidence gaps and data needs, (ii) ecological context dependency (the need for site‐specific approaches) and (iii) challenges to policy implementation. Practical implications . Clearer, evidence‐based guidance on protective vegetative buffers is needed, tailored to different development pressures and woodland contexts. Current evidence provides a partial foundation for guidance, but new research is required to address knowledge gaps. Consistent assessment of ancient woodland loss and deterioration depends on robust, repeatable condition monitoring before and after development. Monitoring and survey protocols should be standardised but site‐adaptable, using appropriate condition indicators. Assessment and ongoing monitoring can also be improved with cross‐organisational data sharing. Together with targeted research on impact pathways and mitigation effectiveness, these measures would substantially strengthen evidence‐based protection of the UK's irreplaceable ancient woodlands.
Abstract Ecological monitoring programmes increasingly rely on automated image recognition to scale beyond what human annotators can process, yet ecologists lack evidence‐based guidance on how much labelling effort is required to deploy these systems reliably at new sites. We present a decision framework that quantifies the trade‐off between labelling effort and recognition accuracy when transferring vision recognition systems across marine habitats; the benchmark spans five datasets across three oceans and three broad organism groups (fish, corals and other benthic invertebrates), from tropical reefs in the Great Barrier Reef and French Polynesia to a temperate Danish fjord. We evaluated four recognition models (DINOv2, CLIP, ResNet‐50, EfficientNet‐B4) under four adaptation strategies (linear probing, low‐rank adaptation, visual prompt tuning, full fine‐tuning) across three protocols: within‐habitat transfer across 20 reef sites (240 runs), cross‐dataset transfer along a difficulty gradient (40 runs) and few‐shot curves with 0–100 labelled target samples per class (648 runs). Frozen self‐supervised foundation model features, using DINOv2 with a linear classifier and only 1538 trainable parameters, generalised to unseen reef sites at least as well as fully fine‐tuned convolutional baselines four orders of magnitude larger. Feature analysis showed why: the foundation models learned category‐discriminative representations that are habitat‐invariant, whereas the convolutional baselines encoded habitat‐specific visual shortcuts that fail at new sites. The labelling budget scales with transfer difficulty. Within taxon and biome, 10–20 labelled images per class sufficed (balanced accuracy 0.90 at ), an order‐of‐magnitude saving; across taxa or biomes the same budget reached only 0.79 and 0.68, still rising at . Solution . Monitoring programmes expanding to new sites can deploy reliable automated image recognition by pairing a frozen, openly available foundation model (DINOv2) with a simple linear classifier and annotating 10–20 images per class, roughly 1–4 h per site, where the new site keeps the same categories within the same taxon and biome; wider moves need 50 or more. The framework lets programmes budget labelling effort against expected accuracy and applies to any monitoring programme that must generalise across sites with limited labelled data.
