Digital agriculture services can greatly assist growers to monitor their fields and optimize their use throughout the growing season. Thus, knowing the exact location of fields and their boundaries is a prerequisite. Unlike property boundaries, which are recorded in local council or title records, field boundaries are not historically recorded. As a result, digital services currently ask their users to manually draw their field, which is time-consuming and creates disincentives. Here, we present a generalized method, hereafter referred to as DECODE (DEtect, COnsolidate, and DElinetate), that automatically extracts accurate field boundary data from satellite imagery using deep learning based on spatial, spectral, and temporal cues. We introduce a new convolutional neural network (FracTAL ResUNet) as well as two uncertainty metrics to characterize the confidence of the field detection and field delineation processes. We finally propose a new methodology to compare and summarize field-based accuracy metrics. To demonstrate the performance and scalability of our method, we extracted fields across the Australian grains zone with a pixel-based accuracy of 0.87 and a field-based accuracy of up to 0.88 depending on the metric. We also trained a model on data from South Africa instead of Australia and found it transferred well to unseen Australian landscapes. We conclude that the accuracy, scalability and transferability of DECODE shows that large-scale field boundary extraction based on deep learning has reached operational maturity. This opens the door to new agricultural services that provide routine, near-real time field-based analytics.

One of Australia's most serious weeds, Chrysanthemoides monilifera subsp. rotundata (bitou bush) was recently found for the first time in Western Australia as a well established population in Kwinana, a major port and industrial area south of Perth, the State's capital. This population is remote from other bitou bush infestations in Australia and had escaped detection despite extensive surveys in the same State for the other subspecies that is present in Australia, Chrysanthemoides monilifera subsp. monilifera (boneseed). The main reasons it went undetected are thought to be the tightly controlled access to this area because of mineral processing and port activities, the unusual invasion route via a heavy industrial area and the morphological similarity to a native species when it is not flowering. Two surveys defined the core population of 1038 plants that are spread along the coast over a 25-ha semi-circle with about a 500-m (1640 ft) diameter. Subsequent surveys of first a 500 m buffer zone and later a 1-km (0.621 mi) buffer found four additional plants, indicating that there is considerable potential for dispersal. We concluded that the survey has not delimited the distribution because of the potential and evidence for long distance dispersal. Cooperation by the various land managers has led to all plants being killed, as an initial step to management of this species. Other steps to be undertaken include an awareness campaign in the area that would need to be surveyed for delimitation of the spatial distribution and seed bank assessment to measure potential dispersal both in space and through time. It remains to be determined what is the best strategic response: eradication or containment.

Determining the diet of flying-foxes can increase understanding of how they function as pollinators and seed dispersers, as well as managing any negative impacts of large roosts. Traditional methods for diet analysis are time consuming, and not feasible to conduct for hundreds of animals. In this study, we optimised a method for diet analysis, based on DNA metabarcoding of environmental DNA (eDNA) from pollen and other plant parts in the faeces. We found that existing eDNA metabarcoding protocols are suitable, with the most useful results being obtained using a commercial food DNA extraction kit, and sequencing 350-450 base pairs of a DNA barcode from the internally transcribed spacer region (ITS2), with ~550 base pairs of the chloroplast rubisco large subunit (rbcL) as a secondary DNA barcode. A list of forage plants was generated for the little red flying-fox (Pteropus scapulatus), the black flyingfox (Pteropus alecto) and the spectacled flying-fox (Pteropus conspicillatus) from our collection sites across Queensland. The diets were determined to comprise predominantly Myrtaceae species, particularly those in the genera Eucalyptus, Melaleuca and Corymbia. With more plant genomes becoming publicly available in the future, there are likely to be further applications of eDNA methods in understanding the role of flying-foxes as pollinators and seed dispersers.

Citation: Scott JK, Batchelor KL, Webber BL (2019) Long term monitoring of recruitment dynamics determines eradication feasibility for an introduced coastal weed. NeoBiota 50: 31-53. https://doi. AbstractBitou bush (Chrysanthemoides monilifera subsp. rotundata) is a Weed of National Significance in Australia and has impacted a significant portion of the eastern coastline. Its discovery in Western Australia was, therefore, a cause for concern. Assessment and control of the isolated and well-defined population began in 2012. To assess the feasibility of eradication in Western Australia as a management outcome for bitou bush, we applied a rigorous data-driven quantification and prediction process to the control program. Between 2012 and 2018 we surveyed over 253 ha of land and removed 1766 bitou bush plants. Approximately 97 person-days were spent over the six years of survey. We measured the seed bank viability for five years starting in 2013, with the 2017 survey results indicating a decline of mean viable seeds/m 2 from 39.3 ± 11.4 to 5.7 ± 2.2. In 2018 we found only ten plants and no newly recruited seedlings in the population. No spread to other areas has been recorded. Soil core studies indicate that the soil seed bank is unlikely to persist beyond eight years. Eradication of the population in Western Australia, defined as five years without plants being detected, therefore remains a realistic management goal. The information generated from the documentation of this eradication program provides invaluable insight for weed eradication attempts more generally: novel detection methods can be effective in making surveys more efficient, all survey methods are not entirely accurate and large plants can escape detection, bitou bush seeds persist in the soil but become effectively undetectable at low densities, and migration of seed was unquantifiable, possibly compromising delimitation. Continued monitoring of the Western Australian population will determine how much of a risk these factors represent to eradication as the outcome of this management program. John K. Scott et al. / NeoBiota 50: 31-53 (2019) 32 A peer-reviewed open-access journal NeoBiota