Search

Recognizing the types of pollen grains and estimating their proportion in pollen mixture samples collected in a specific geographical area is important for agricultural, medical, and ecosystem research. Our paper adopts a convolutional neural network for the automatic segmentation of pollen species in microscopy images, and proposes an original strategy to train such network at reasonable manual annotation cost. Our approach is founded on a large dataset composed of pure pollen images. It first (semi-)manually segments foreground, i.e. pollen grains, and background in a fraction of those images, and use the resulting annotated dataset to train a universal pollen segmentation CNN. In the second step, this model is used to automatically segment a large number of additional pure pollen images, so as to supervise the training of a pollen species segmentation model. Despite the fact that it has been trained from pure images only, the model is shown to provide accurate segmentation of species in pollen mixtures. Our experiments also demonstrate that dedicating a model to the segmentation of a subset of the available pure pollen species makes it possible to train a bin pollen class, corresponding to pollen species that are not in the subset of species recognized by the model. This strategy is useful to cope with unexpected species in a mixture

Pear (Pyrus communis) is the main fruit crop in Belgium and ‘Conférence’ accounts for 90% of the pear production. ‘Conférence’ is self-incompatible and thus requires cross-pollination to produce fruits by fertilization. Hives could be settled in the orchards to facilitate pollination and to allow a better fruit size production. However, blooming occurs early in the spring and climatically unfavorable conditions (frost, low temperatures, rain, etc.) could prevent the pollination by insects. To overcome this inconvenience, parthenocarpy induction by spraying gibberellins is a common practice in ‘Conférence’ orchards. However, whether gibberellin treatment affects fruit set and quality of fruits produced by fertilization has not been investigated. We therefore compared the impact of gibberellin spraying on the development of parthenocarpic fruits and fruits obtained by fertilization. Un-pollinated, open-pollinated and hand-pollinated flowers were sprayed or not with 10 mg L-1 GA4/7 at the balloon, full bloom and petal fall stages and the resulting fruit production and quality were assessed. Compatible hand-pollination increased fruit size and fruit weight by respectively 9 and 8% and decreased fruit length by 10% compared to parthenocarpic fruits whatever the gibberellin treatment. Treatment with GA4/7 increased the fruit set at harvest compared to untreated fruits. The time of GA4/7 spraying significantly affected fruit parameters and treatment was more efficient when applied at the balloon stage. Regarding fruits produced by fertilization, GA4/7 treatment decreased seed development by respectively 12, 13 and 36% when applied at the balloon, full bloom and petal fall stages compared to untreated fruits. Consequently, fruit size decreased by 2 and 5% and fruit weight by 8 and 13% when GA4/7 was sprayed at full bloom and petal fall stages compared to untreated fruits. However, treatment at the balloon stage did not affect fruit size and weight. Thus, our results showed

Recognizing the types of pollen grains and estimating their proportion in pollen mixture samples collected in a specific geographical area is important for agricultural, medical, and ecosystem research. Our paper adopts a convolutional neural network for the automatic segmentation of pollen species in microscopy images, and proposes an original strategy to train such network at reasonable manual annotation cost. Our approach is founded on a large dataset composed of pure pollen images. It first (semi-)manually segments foreground, i.e. pollen grains, and background in a fraction of those images, and use the resulting annotated dataset to train a universal pollen segmentation CNN. In the second step, this model is used to automatically segment a large number of additional pure pollen images, so as to supervise the training of a pollen species segmentation model. Despite the fact that it has been trained from pure images only, the model is shown to provide accurate segmentation of species in pollen mixtures. Our experiments also demonstrate that dedicating a model to the segmentation of a subset of the available pure pollen species makes it possible to train a bin pollen class, corresponding to pollen species that are not in the subset of species recognized by the model. This strategy is useful to cope with unexpected species in a mixture.

Pear (Pyrus communis L.) is the most important fruit crop in Belgium and ‘Conférence’ represents 90% of the pear production. ‘Conférence’ is self-incompatible and requires cross-pollination to develop fertilized fruits. Hives could be settled in the orchards to facilitate pollination and allow a better fruit size production. However, blooming occurs early in the spring and climatically unfavorable conditions (frost, low temperatures, rain, etc) could prevent the pollination by insects. In this case, parthenocarpy induction by spraying gibberellins is a common practice in ‘Conference’ orchards. However, whether gibberellin treatment affect fruit set and quality of fertilized fruits has not been investigated. We therefore compared the impact of gibberellin spraying on the development of parthenocarpic and fertilized fruits. Un-pollinated, open-pollinated and hand-pollinated flowers were sprayed or not with 10 mg/L GA4/7 at balloon, anthesis and wilted stages and the resulting fruit production and quality were assessed. Hand-pollination increased fruit size and fruit weight by respectively 9% and 8% but decreased fruit length by 10% compared to parthenocarpic fruits whatever the gibberellin treatment. Treatment with GA4/7 increased the fruit set at harvest (28% vs. 22%) compared to untreated fruits. The time of GA4/7 spraying significantly affected fruit parameters and treatment was more efficient when applied at balloon stage. Regarding fertilized fruits, GA4/7 treatment decreased seed development by respectively 12%, 13% and 36% when applied at balloon, anthesis and wilted stages compared to untreated fruits. Consequently, fruit size decreased by 2% and 5% and fruit weight by 8% and 13% when GA4/7 was sprayed at anthesis and wilted stages compared to untreated fruits. However, treatment at balloon stage did not affect fruit size and weight. Thus, our results showed that GA4/7 treatment might negatively affect the development of pollinated fruits depending on the timi

European pear requires inter-cultivar cross-pollination by insects to develop fertilized fruits. However, some European pear cultivars such as ‘Conference’ naturally produce parthenocarpic seedless fruits. To better understand the hormonal regulation of fruit set and early fruit development in this European pear cultivar, the phytohormone and polyamine profiles in ‘Conference’ flowers and fruits resulting from both fertilization and parthenocarpic processes were analyzed. The expression of genes involved in phytohormone metabolism and signaling were also investigated. Phytohormone profiles differed more at flower stage 3 days after treatment than in 15 day- and 30-day-old fruits in response to fertilization and parthenocarpy. An increase in auxins, abscisic acid, ethylene precursor, and spermine, and a decrease in putrescine were recorded in the fertilized flowers as compared to the parthenocarpic flowers. Fertilization also upregulated genes involved in gibberellin synthesis and down-regulated genes involved in gibberellin catabolism although the total gibberellin content was not modified. Moreover, exogenous gibberellin (GA3, GA4/7) and cytokinin (6BA) applications did not increase parthenocarpic induction in ‘Conference’ as observed in other European and Asian pear cultivars. We hypothesize that the intrinsic parthenocarpy of ‘Conference’ could be related to a high gibberellin level in the flowers explaining why exogenous gibberellin application did not increase parthenocarpy as observed in other pear cultivars and species.

Pear (Pyrus communis) is the most important fruit crop in Belgium and ‘Conférence’ represents 90% of the pear production. ‘Conférence’ is self-incompatible and requires cross-pollination to develop fertilized fruits. We therefore compared seven pear cultivars (‘Dicolor’, ‘Doyenné du Comice’, ‘Joséphine de Malines’, ‘Seigneur Espéren’, ‘Sweet Sensation’, ‘Triomphe de Vienne’, ‘Williams Bon Chrétien’) adapted to Belgian conditions in order to identify valuable pollinizer cultivars for ‘Conférence’. These cultivars were used as male parent in hand pollination experiments with ‘Conférence’ and the resulting number of fruits was assessed. Fruit size and weight were measured at harvest. Moreover, to evaluate the pollination efficiency of these cultivars and their attractiveness for bees, we analyzed their pollen and nectar production and composition. Pollen is one of the prime nutrient resources used for bee larvae and nectar represents the major source of sugars for bees. Regarding fruit production, the mean weight and size reached 103-154 g and 50-61 mm respectively depending on the crossings. The largest fruits resulted from the pollination of ‘Conférence’ with ‘Triomphe de Vienne’ or ‘Doyenné du Comice’. The cultivars differed in their pollen and nectar quantities and compositions. The mean number of pollen grains per anther ranged from 2,200 for ‘Dicolor’ to 4,600 for ‘Williams Bon Chrétien’. Pollen viability was high whatever the cultivar (60-92%). Pollen polypeptide concentration was variable (31-107 µg mg-1) while amino acid concentration was high and stable (346-390 µg mg-1) among cultivars. The nectar sugar content per flower varied between 52 µg for ‘Seigneur Espéren’ to 255 µg for ‘Conférence’. Pear cultivars produced therefore pollen of high quality regarding amino acid content while their nectar production was poor in sugars. Bee attractiveness might differ among cultivars and needs further research to optimize orchard management.

Many plants require animal pollinators for successful reproduction; these plants provide pollinator resources in pollen and nectar (rewards) and attract pollinators by specific cues (signals). In a seeming contradiction, some plants produce toxins such as alkaloids in their pollen and nectar, protecting their resources from ineffective pollinators. We investigated signals and rewards in the toxic, protandrous bee-pollinated plant Aconitum napellus, hypothesizing that male-phase flower reproductive success is pollinator-limited, which should favour higher levels of signals (odours) and rewards (nectar and pollen) compared with female-phase flowers. Furthermore, we expected insect visitors to forage only for nectar, due to the toxicity of pollen. We demonstrated that male-phase flowers emitted more volatile molecules and produced higher volumes of nectar than female-phase flowers. Alkaloids in pollen functioned as chemical defences, and were more diverse and more concentrated compared to the alkaloids in nectar. Visitors actively collected little pollen for larval food but consumed more of the less-toxic nectar. Toxic pollen remaining on the bee bodies promoted pollen transfer efficiency, facilitating pollination.

Pear is the second fruit production in Belgium and the cultivar Conference represents 90% of this production. Pyrus communis is a self-incompatible species and requires inter-cultivar cross-pollination to develop fruits. However, blooming occurs early in the Spring and climatically unfavorable spring conditions (frost, low temperatures, rain,…) could prevent the pollination by insects. In this case, parthenocarpy induction by spraying plant hormones, mainly gibberellins, is a common practice in orchards and allows achieving sufficient yield. No precise study has been carried out to determine the most efficient hormones, the best application time and the required amounts to apply. Our work aims at highlighting pear tree reproduction and focuses on parthenocarpy processes in the Conference cultivar. We investigated if intrinsic parthenocarpy due to endogenous hormones occured in Conference. We also compared the efficiency of gibberellins, auxins and cytokinins in the parthenocarpic fruit induction. Parthenocarpic fruit development was compared to fertilized fruits resulting from hand compatible cross-pollination. Spontaneous intrinsic parthenocarpy occurred in Conference, but pollen deposit (even incompatible) increased the number of fruits initiated. Gibberellins (GA3) increased the fruit set compared to the other hormonal treatments and trees treated with gibberellins and/or cytokinins produced larger fruits. Parthenocarpic fruits remained, nevertheless, smaller than fertilized ones. Gibberellin inhibitor treatments showed that endogenous gibberellins do not act alone to induce parthenocarpy and polyamines quantification revealed a role for putrescine and spermine in parthenocarpy induction in Conference.