If efforts to tackle biodiversity loss and its impact on human wellbeing are to be successful, conservation must learn from other fields which use predictive methods to foresee shocks and preempt their impacts in the face of uncertainty, such as military studies, public health and finance. Despite a long history of using predictive models to understand the dynamics of ecological systems and human disturbance, conservationists do not systematically apply predictive approaches when designing and implementing behavioural interventions. This is an important omission because human behaviour is the underlying cause of current widespread biodiversity loss. Here, we critically assess how predictive approaches can transform the way conservation scientists and practitioners plan for and implement social and behavioural change among people living with wildlife. Our manifesto for predictive conservation recognises that social-ecological systems are dynamic, uncertain and complex, and calls on conservationists to embrace the forward-thinking approach which effective conservation requires.

20 Reliably predicting sustainable exploitation levels for many tropical species subject to hunting 21 remains a difficult task, largely because of the inherent uncertainty associated with estimating 22 parameters related to both population dynamics and hunting pressure. Here, we investigate a 23 modelling approach to support decisions in bushmeat management which explicitly considers 24 parameter uncertainty. We apply the approach to duiker Cephalophus spp., assuming either a 25 constant quota-based, or a constant proportional harvesting, strategy. Within each strategy, we 26 evaluate different hunting levels in terms of both average yield and survival probability, over 27 different time horizons. Under quota-based harvesting, considering uncertainty revealed a 28 trade-off between yield and extinction probability that was not evident when ignoring 29 uncertainty. The highest yield was returned by a quota that implied a 40% extinction risk, 30 whereas limiting extinction risk to 10% reduced yield by 50%-70%. By contrast, under 31 proportional harvesting, there was no trade-off between yield and extinction probability. The 32 maximum proportion returned a yield comparable with the maximum possible under quota-33 based harvesting, but with extinction risk below 10%. However, proportional harvesting can 34 be harder to implement in practice because it depends on an estimate of population size. In 35 both harvesting approaches, predicted yields were highly right-skewed with median yields 36 differing from mean yields, implying that decision outcomes depend on attitude to risk. The 37 analysis shows how an explicit consideration of all available information, including 38 uncertainty, can, as part of a wider process involving multiple stakeholders, help inform 39 harvesting policies. 40Introduction 41 Many studies raise alarm over the present rate of wild meat harvesting as a major cause of 42 population decline and extinction risk for many species [1][2][3]. With wild meat providing a 43 major source of protein and household income to some of the world's poorest people [4][5][6], 44 both subsistence and commercial hunting in West and Central Africa are on the rise [1, 6, 7].45 Bushmeat harvest across the Congo Basin alone is estimated to occur at more than six times 46 the sustainable harvest rate [1]. 47 However, reliably estimating a sustainable harvest level remains problematic. Ecological 48 systems are highly complex and the relevant biological data on mammals in tropical forests is 49 scarce [8]. Information is often collected during short field seasons [9-11], across different 50 spatial scales and in different ecosystems [12, 13], producing point estimates of population 51 parameters and species abundances that vary considerably between studies [14, 15]. As a result, 52 traditional techniques such as monitoring offtakes and correlating them with changes in 53 harvested species dynamics such as abundance and age structure [14, 16] will not accurately 54 assess the sustainability of harvesting. To address this...

Background:Although there have been recent advances in the development of animal-attached 'proximity' tags to remotely record the interactions of multiple individuals, the efficacy of these devices depends on the instrumentation of sufficient animals that subsequently have spatial interactions. Among densely colonial mammals such as fur seals, this remains logistically difficult, and interactions between animals during foraging have not previously been recorded. Results:We collected data on conspecific interactions during diving at sea using still image and video cameras deployed on 23 Antarctic fur seals. Animals carried cameras for a total of 152 days, collecting a total of 38,098 images and 369 movies (total time 7.35 h). Other fur seals were detected in 74 % of deployments, with a maximum of five seals detected in a single image (n = 122 images, 28 videos). No predators other than conspecifics were detected. Detection was primarily limited by light conditions, since conspecifics were usually further from each other than the 1-m range illuminated by camera flash under low light levels. Other seals were recorded at a range of depths (average 27 ± 14.3 m, max 66 m). Linear mixed models suggested a relationship between conspecific observations per dive and the number of krill images recorded per dive. In terms of bouts of dives, other seals were recorded in five single dives (of 330) and 28 bouts of dives <2 min apart (of 187). Using light conditions as a proxy for detectability, other seals were more likely to be observed at the bottom of dives than during descent or ascent. Seals were also more likely to be closer to each other and oriented either perpendicular or opposing each other at the bottom of dives, and in the same or opposite direction to each other during ascent. Conclusions:These results are contrary to animal-attached camera observations of penguin foraging, suggesting differing group-foraging tactics for these marine predators. Group foraging could have consequences for models linking predator behaviour to prey field densities since this relationship may be affected by the presence of multiple predators at the same patch.

During 2021-22 High Pathogenicity Avian Influenza (HPAI) killed thousands of wild birds across Europe and North America, suggesting a change in infection dynamics and a shift to new hosts, including seabirds. Northern Gannets (Morus bassanus) appeared especially severely impacted, but limited understanding of how the virus spread across the metapopulation, or the demographic consequences of mass mortality limit our understanding of its severity. Accordingly, we collate information on HPAIV outbreaks across most North Atlantic gannet colonies and for the largest colony (Bass Rock, UK), provide impacts on population size, breeding success, adult survival, and preliminary results on serology. Unusually high numbers of dead gannets were first noted in Iceland during April 2022. Outbreaks in May occurred in many Scottish colonies, followed by colonies in Canada, Germany and Norway. By the end of June, outbreaks had occurred in five Canadian colonies and in the Channel Islands. Outbreaks in 12 UK and Ireland colonies appeared to follow a clockwise pattern with the last infected colonies recorded in late August/September. Unusually high mortality was recorded at 40 colonies (75% of global total colonies). Dead birds testing positive for HPAIV H5N1 were associated with 58% of these colonies. At Bass Rock, the number of occupied sites decreased by at least 71%, breeding success declined by ~66% compared to the long-term UK mean and adult survival between 2021 and 2022 was 42% lower than the preceding 10-year average. Serological investigation detected antibodies specific to H5 in apparently healthy birds indicating that some gannets recover from HPAIV infection. Further, most of these recovered birds had black irises, suggestive of a phenotypic indicator of previous infection. Untangling the impacts of HPAIV infection from other key pressures faced by seabirds is key to establishing effective conservation strategies for threatened seabird populations, HPAIV being a novel and pandemic threat.

Traditional approaches to guiding decisions about harvesting bushmeat often employ singlespecies population dynamic models, which require species-and location-specific data, are missing ecological processes such as multi-trophic interactions, cannot represent multispecies harvesting, and cannot predict the broader ecosystem impacts of harvesting. In order to explore an alternative approach to devising sustainable harvesting strategies, we employ the Madingley General Ecosystem Model, which can simulate ecosystem dynamics in response to multi-species harvesting given nothing other than location-specific climate data. We used the model to examine yield, extinctions, and broader ecosystem impacts, for a range of harvesting intensities of duiker-sized ectothermic herbivores. Duiker antelope (such as Cephalophus callipygus and Cephalophus dorsalis) are the most heavily hunted species in sub-Saharan Africa, contributing 34%-95% of all bushmeat in the Congo Basin. Across a range of harvesting rates, the Madingley model gave estimates for optimal harvesting rate, and extinction rate, that were qualitatively and quantitatively similar to the estimates from singlespecies Beverton-Holt model. Predicted yields were somewhat greater (around 5 times, on average) for the Madingley model, which would be expected given that the Madingley simulates multi-species harvesting from an initially pristine ecosystem. This match increased the degree of confidence with which we could examine other predictions from the ecosystem model, as follows. At medium and high levels of harvesting of duiker-sized herbivores, there were statistically significant, but moderate, reductions in the densities of the targeted functional group; increases in small-bodied herbivores; decreases in large-bodied carnivores;and minimal ecosystem-level impacts overall. The results suggest that general ecosystem models such as the Madingley model could be used more widely to help estimate sustainable harvesting rates, bushmeat yields and broader ecosystem impacts across different locations and target species.