Temporally stacked bee forage species distribution modeling for flower abundance mapping.

Predicting spatial distribution of flowering forage availability is critical for guiding migratory beekeeping decisions. Species distribution modelling (SDM) is widely used to predict the geographic distribution or species ranges. Stacked distributions of multiple species (S-SDM) have been used in predicting species richness or assemblages. Here, we present a method for stacking SDMs based on a temporal element, the flowering phenology of melliferous flora species. First, we used presence-only data for thirty key forage species used for honey production in Western Australia, combined with environmental variables for predicting the geographic distribution of species, using MaxEnt software. The output distribution grids were then stacked based on monthly flowering times of each species to develop grids representing the richness of flowering species by grid cell. While designed for modelling flowering forage availability for a migratory beekeeping system, the approach can be used for predicting temporal forage availability for a range of different fauna that rely on melliferous flora. •How to use temporally stacked species distribution modelling for generic distribution of flowering availability using presence-only data.•A procedure for developing flowering richness and availability grids.

Fire reduces eucalypt forest flowering phenology at the landscape-scale.

Plant phenology describes the timing of reproductive events including flowering and fruiting, which for many species are affected by fire disturbance. Understanding phenological responses to fire provides insights into how forest demographics and resources may shift alongside increasing fire frequency and intensity driven by climate change. However, isolating the direct effects of fire on a species' phenology and excluding potential confounders (e.g. climate, soil) has been difficult due to the logistical challenges of monitoring species-specific phenological events across myriad fire and environmental conditions. Here, we use CubeSat-derived crown-scale flowering data to estimate the effects of fire history (time since fire and fire severity over a 15-year time span) on flowering of the eucalypt Corymbia calophylla across a Mediterranean-climate forest (814km2) in southwest Australia. We found that fire reduced the proportion of flowering trees at the landscape-scale, and flowering recovered at a rate of 0.15 % (±0.11% SE) per year. Further, this negative effect was significant due to high crown scorch fires (>20% canopy scorch), yet there was no significant effect from understory burns. Estimates were obtained using a quasi-experimental design which identifies the effect of time since fire and severity on flowering by comparing proportional flowering within target fire perimeters (treatment) and adjacent past fire perimeters (control). Given the majority of fires studied were managed fuel reduction burns, we applied the estimates to hypothetical fire regimes to compare flowering outcomes under more or less frequent prescribed burning. This research demonstrates the landscape-scale effects of burning on a tree species' reproduction, which could broadly impact forest resiliency and biodiversity.

The cost of not acting: Delaying invasive grass management increases costs and threatens assets in a national park, northern Australia.

Globally, invasive grasses are a major threat to protected areas (PAs) due to their ability to alter community structure and function, reduce biodiversity, and alter fire regimes. However, there is often a mismatch between the threat posed by invasive grasses and the management response. We document a case study of the spread and management of the ecosystem-transforming invasive grass, Andropogon gayanus Kunth. (gamba grass), in Litchfield National Park; an iconic PA in northern Australia that contains significant natural, cultural and social values. We undertook helicopter-based surveys of A. gayanus across 143,931 ha of Litchfield National Park in 2014 and 2021-2022. We used these data to parametrise a spatially-explicit spread model, interfaced with a management simulation model to predict 10-year patterns of spread, and associated management costs, under three scenarios. Our survey showed that between 2014 and 2021-22 A. gayanus spread by 9463 ha, and 47%. The gross A. gayanus infestation covered 29,713 ha of the total survey area, making it the largest national park infestation in Australia. A. gayanus had not been locally eradicated within the Park's small existing 'gamba grass eradication zone', and instead increased by 206 ha over the 7-year timeframe. Our modelled scenarios predict that without active management A. gayanus will continue spreading, covering 42,388 ha of Litchfield within a decade. Alternative scenarios predict that: (i) eradicating A. gayanus in the small existing eradication zone would likely protect 18% of visitor sites, and cost ∼AU$825,000 over 5 years - more than double the original predicted cost in 2014; or (ii) eradicating A. gayanus in a much larger eradication zone would likely protect 86% of visitor sites and several species of conservation significance, and cost ∼AU$6.6 million over 5 years. Totally eradicating A. gayanus from the Park is no longer viable due to substantial spread since 2014. Our study demonstrates the value of systematic landscape-scale surveys and costed management scenarios to enable assessment and prioritisation of weed management. It also demonstrates the increased environmental and financial costs of delaying invasive grass management, and the serious threat A. gayanus poses to PAs across northern Australia.

Land conversion and pesticide use degrade forage areas for honey bees in America's beekeeping epicenter.

A diverse range of threats have been associated with managed-bee declines globally. Recent increases of two known threats, land-use change and pesticide use, have resulted from agricultural expansion and intensification notably in the top honey-producing state in the United States: North Dakota. This study investigated the dual threat from land conversion and pesticide use surrounding ~14,000 registered apiaries in North Dakota from 2001 to 2014. We estimated the annual total insecticide use (kg) on major crops within 1.6 km of apiary sites. Of the eight insecticides quantified, six showed significant increasing trends over the time period. Specifically, applications of the newly established neonicotinoids Chlothianidin, Imidacloprid and Thiamethoxam, increased annually by 1329 kg, 686 kg, 795 kg, respectively. Also, the use of Chlorpyrifos, which was well-established in the state by 2001 and is highly toxic to honey bees, increased by ~8,800 kg annually from 6,500 kg in 2001 to 115,000 kg in 2014 on corn, soybeans and wheat. We further evaluated the relative quality changes of natural/semi-natural land covers surrounding apiaries in 2006, 2010 and 2014, a period of significant increases in cropland area. In areas surrounding apiaries, we observed changes in multiple indices of forage quality that reflect the deteriorating landscape surrounding registered apiary sites due to land-use change and pesticide-use increases. Overall, our results suggest that the application of foliar-applied insecticides, including pyrethroids and one organophosphate, increased surrounding apiaries when the use of neonicotinoid seed treatments surged and the area for producing corn and soybeans expanded. Spatially, these threats were most pronounced in southeastern North Dakota, a region hosting a high density of apiary sites that has recently experienced corn and soybean expansion. Our results highlight the value of natural and semi-natural land covers as sources of pollinator forage and refugia for bees against pesticide exposure. Our study provides insights for targeting conservation efforts to improve forage quality benefiting managed pollinators.