Deep learning to extract the meteorological by-catch of wildlife cameras.

Microclimate-proximal climatic variation at scales of metres and minutes-can exacerbate or mitigate the impacts of climate change on biodiversity. However, most microclimate studies are temperature centric, and do not consider meteorological factors such as sunshine, hail and snow. Meanwhile, remote cameras have become a primary tool to monitor wild plants and animals, even at micro-scales, and deep learning tools rapidly convert images into ecological data. However, deep learning applications for wildlife imagery have focused exclusively on living subjects. Here, we identify an overlooked opportunity to extract latent, ecologically relevant meteorological information. We produce an annotated image dataset of micrometeorological conditions across 49 wildlife cameras in South Africa's Maloti-Drakensberg and the Swiss Alps. We train ensemble deep learning models to classify conditions as overcast, sunshine, hail or snow. We achieve 91.7% accuracy on test cameras not seen during training. Furthermore, we show how effective accuracy is raised to 96% by disregarding 14.1% of classifications where ensemble member models did not reach a consensus. For two-class weather classification (overcast vs. sunshine) in a novel location in Svalbard, Norway, we achieve 79.3% accuracy (93.9% consensus accuracy), outperforming a benchmark model from the computer vision literature (75.5% accuracy). Our model rapidly classifies sunshine, snow and hail in almost 2 million unlabelled images. Resulting micrometeorological data illustrated common seasonal patterns of summer hailstorms and autumn snowfalls across mountains in the northern and southern hemispheres. However, daily patterns of sunshine and shade diverged between sites, impacting daily temperature cycles. Crucially, we leverage micrometeorological data to demonstrate that (1) experimental warming using open-top chambers shortens early snow events in autumn, and (2) image-derived sunshine marginally outperforms sensor-derived temperature when predicting bumblebee foraging. These methods generate novel micrometeorological variables in synchrony with biological recordings, enabling new insights from an increasingly global network of wildlife cameras.

Flowers as a reservoir of yeast diversity: description of Wickerhamiella nectarea f.a. sp. nov., and Wickerhamiella natalensis f.a. sp. nov. from South African flowers and pollinators, and transfer of related Candida species to the genus Wickerhamiella as new combinations.

Flowers offer favourable microenvironments for yeast growth, and are increasingly recognised as a rich source of novel yeast species. Independent surveys of yeasts associated with flowers and pollinators in South Africa led to the discovery of 38 strains of two new species. Physiological profiles and analysis of the internal transcribed spacer and the D1/D2 domains of the large subunit rRNA gene showed that they represent two novel species that belong to the Wickerhamiella clade. We describe the species as Wickerhamiella nectarea f.a. sp. nov. (type strain EBDCdVSA11-1T, CBS 14162T, NRRL Y-63791T) and W. natalensis f.a. sp. nov. (type strain EBDCdVSA7-1T, CBS 14161T, NRRL Y-63790T). We extend the known range of flower-associated Wickerhamiella species to South Africa and discuss the ecology and phylogenetic relationships of the clade in relation to its host species and biogeography. Examination of growth characteristics supports that the Wickerhamiella clade exhibits a high degree of evolutionary lability, and that specialisation to different niches may occur rapidly. We review the current status of floral yeast biodiversity and nectar as a reservoir of species diversity, and the importance of pollinators and biogeography. In addition, 18 species formerly assigned to the genus Candida are reassigned formally to the genus Wickerhamiella.

Metschnikowia drakensbergensis sp. nov. and Metschnikowia caudata sp. nov., endemic yeasts associated with Protea flowers in South Africa.

In a taxonomic study of yeasts recovered from nectar of flowers and associated insects in South Africa, 11 strains were found to represent two novel species. Morphological and physiological characteristics and sequence analyses of the large-subunit rRNA gene D1/D2 region, as well as the actin, RNA polymerase II and elongation factor 2 genes, showed that the two novel species belonged to the genus Metschnikowia. Metschnikowia drakensbergensis sp. nov. (type strain EBD-CdVSA09-2(T) =CBS 13649(T) =NRRL Y-63721(T); MycoBank no. MB809688; allotype EBD-CdVSA10-2(A) =CBS13650(A) =NRRL Y-63720(A)) was recovered from nectar of Protea roupelliae and the beetle Heterochelus sp. This species belongs to the large-spored Metschnikowia clade and is closely related to Metschnikowia proteae, with which mating reactions and single-spored asci were observed. Metschnikowia caudata sp. nov. (type strain EBD-CdVSA08-1(T) =CBS 13651(T) =NRRL Y-63722(T); MycoBank no. MB809689; allotype EBD-CdVSA57-2(A) =CBS 13729(A) =NRRL Y-63723(A)) was isolated from nectar of Protea dracomontana, P. roupelliae and P. subvestita and a honeybee, and is a sister species to Candida hainanensis and Metschnikowia lopburiensis. Analyses of the four sequences demonstrated the existence of three separate phylotypes. Intraspecies matings led to the production of mature asci of unprecedented morphology, with a long, flexuous tail. A single ascospore was produced in all compatible crosses, regardless of sequence phylotype. The two species appear to be endemic to South Africa. The ecology and habitat specificity of these novel species are discussed in terms of host plant and insect host species.

Effects of volatile compounds emitted by Protea species (Proteaceae) on antennal electrophysiological responses and attraction of cetoniine beetles.

Evolutionary shifts in pollination systems within a plant genus are commonly associated with changes in floral scent, reflecting selection mediated through the sensory systems of various pollinators. The most common cetoniine beetle pollinator of grassland Protea species in South Africa, Atrichelaphinis tigrina, previously has been shown to have a strong preference for the fruity floral scent of these plants over the weak scent of their bird-pollinated congeners. However, it is not known which of the many compounds found in the scent of beetle pollinated Protea species play a role for pollinator attraction. Electroantennograms (EAG) from A. tigrina beetles were recorded in response to 15 compounds emitted by Protea flower heads. EAG responses to all 15 compounds were significantly greater than those to the paraffin solvent in which they were diluted. The greatest responses were observed for aromatics (anisole, methyl benzoate, methyl salicylate, benzaldehyde) followed by the monoterpene ß-linalool, which can comprise up to 66 % of fruity Protea scents. Five compounds that elicited EAG responses (benzaldehyde, ß-linalool, (E/Z)-linalool oxide (furanoid), methyl benzoate, and methyl salicylate) were tested in commercially available yellow bucket traps in the field to test their attractiveness to beetles. Traps baited with methyl benzoate, ß-linalool, (E/Z)-linalool oxide (furanoid), and methyl salicylate caught significantly more insects than did those containing paraffin only. Methyl benzoate also was more specifically attractive to A. tigrina than was (E/Z)-linalool oxide (furanoid) and paraffin baited controls. A second field experiment using a combination of linalool vs. paraffin baited yellow or green traps showed that trap color had a significant effect on the number of trapped beetles. Yellow traps yielded a ten-fold higher number of insect catches than did green traps. However, the combination of yellow color and the scent compound linalool yielded the highest number of catches. This study has shown that the cetoniine beetle A. tigrina can detect a variety of floral compounds and is attracted to compounds comprising a large proportion of the blend that makes up fruity Protea scents, adding support for the hypothesis that change in scent chemistry during the shift from bird to cetoniine beetle pollination in this genus were mediated by beetle sensory preferences.

Reproductive Ecology of the Invasive Alien Shrub Pyracantha angustifolia in the Grassland Biome, South Africa.

Knowledge on reproductive traits of problematic invasive alien plants, such as the woody invasive shrub Pyracantha angustifolia of temperate Chinese origin, can help better manage invasive species. To determine factors contributing to its invasion, we investigated floral visitors and pollen loads, self-compatibility, seed set, seed rain, soil seed banks, and seed longevity in the soil. Generalist insects were recorded visiting flowers and all carried pollen loads of high purity (>70%). Floral visitor exclusion experiments showed that P. angustifolia can set seed (66%) without pollen vectors, although natural pollination resulted in higher fruit set (91%). Fruit count surveys and seed set showed an exponentially increased relationship between seed set and plant size with high natural seed yield (±2 million seeds m-2). Soil core samples revealed a high seed density of 46,400 ± (SE) 8934 m-2 under shrubs, decreasing with distance away from the shrub. Bowl traps stationed under trees and fences confirmed that seeds were efficiently dispersed by animals. Buried seeds survived for less than six months in the soil. Due to high seed production, self-compatibility augmented by generalist pollen vectors, and effective seed dispersal by local frugivores, it is difficult to manage the spread manually. Management of this species should focus on the short life span of seeds.

Metschnikowia proteae sp. nov., a nectarivorous insect-associated yeast species from Africa.

A collection of yeasts isolated from nectar of flowers of Protea caffra (Proteaceae) and associated scarab beetles (Atrichelaphinis tigrina, Cyrtothyrea marginalis, Trichostetha fascicularis and Heterochelus sp.) and drosophilid flies in South Africa, contained 28 isolates that could not be assigned to known species. Comparisons of the D1/D2 domains of the large subunit rRNA gene demonstrated the existence of three separate phylotypes with an affinity to the genus Metschnikowia and more specifically to the beetle-associated large-spored Metschnikowia clade. Twenty-six strains that had similar D1/D2 sequences were mixed in all pairwise combinations. They were found to mate and give rise to large asci typical of those in the clade. The name Metschnikowia proteae sp. nov. (type strain EBDT1Y1(T) = CBS 12522(T) = NRRL Y-48784(T); allotype strain EBDC2Y2 = CBS 12521 = NRRL Y-48785) is proposed to accommodate this novel species. The ecology of this novel yeast species is discussed in relation to its potential plant and insect host species. The additional two single strains isolated from Heterochelus sp. represent two novel undescribed species (Candida sp. 1 EBDM2Y3 and Candida sp. 2 EBDM8Y1). As these single strains are probably haploid mating types of Metschnikowia species, their description is deferred until the species are sufficiently well sampled to permit meaningful descriptions.

The relative contributions of insect and bird pollinators to outcrossing in an African Protea (Proteaceae).

PREMISE OF THE STUDY: A useful, but seldom applied, measure of the effectiveness of different pollinators is their contribution to the rate of outcrossing. This measure is particularly useful in facultatively autogamous plants for which seed set cannot be used as a direct measure of pollinator effectiveness. We used selective exclusion experiments to assess the importance of insects for outcrossing in Protea caffra, a facultatively autogamous shrub with scented flowers that are visited frequently by both birds and insects (mainly beetles). METHODS AND RESULTS: Pollen loads on stigmas, pollen tube growth, seed set, seed mass, germination, and early seedling survivorship were similar for vertebrate-excluded and open-pollinated inflorescences. Pollen-supplementation mostly did not increase seed set, revealing resource limitation. Mean multilocus outcrossing rates, estimated using eight polymorphic allozyme loci, were similar for progeny from inflorescences excluded from bird visitors (0.65) and for those visited by both birds and insects (0.59). Wright's fixation indices indicated that the adult population is near Hardy-Weinberg equilibrium but differed markedly for maternal plants (F(IS) = -0.187 ± 0.065) and their early stage progeny (F(IS) = 0.258 ± 0.002). Since seed from self and cross hand-pollinations were equally viable in terms of germination, this discrepancy in F(IS) could be explained by inbreeding depression that occurs between germination and reproductive maturity. CONCLUSIONS: Since outcrossing rates were not reduced when birds were excluded, we infer that insects are effective agents of cross pollination in P. caffra. This helps to explain the evolution of traits associated with insect pollination, such as fruity floral scent, in this species.