Optimizing pollencounter for high throughput phenotyping of pollen quality in tomatoes

Abstract

The macro “PollenCounter” in ImageJ was initially developed to assess pollen viability in grapevine. We set out to see if PollenCounter could be used to assess pollen number and viability in tomatoes. We tested different optimization scenarios by adjusting the pollen size (100–900, 200–900 pixel 2 ) and circularity of pollen grains (0.4–1, 0.5–1, and 0.6–1) on 31 microscopic images of stained tomato pollen. Both total pollen number and proportion of viable pollen were positively and significantly correlated with the outputs from manual counting. The scenario with 100–900 pixel 2 pollen size and 0.4–1 circularity had the highest association for pollen number ( r = 0.99) and pollen viability ( r = 0.86). PollenCounter is 32-fold faster than manual counting. We added a command to the macro to automatically save the outputs containing the number of total and viable pollen, avoiding transcription errors inherent to manual counting. We successfully applied the optimized PollenCounter to discriminate tomato genotypes based on pollen number and pollen viability under heat stress. Our results show that PollenCounter, as an open-access macro, can be customized and improved to meet users’ needs. The use of PollenCounter can save time and money in pollen quality assessment. We outline the steps to optimize the macro for other samples or crop species. The optimized macro could allow efficient screening of a large germplasm collection for pollen thermo-tolerance and selection of best thermo-tolerant individuals in breeding programs.

Floral Biology and Pollination Efficiency in Yam (Dioscorea spp.)

“…The pollen counter can determine the pollen number in less than 60 s compared to 5-68 min required for a manual count [57]. A similar trend was reported by Ayenan et al [51] as the semi-automated pollen counter was 32-fold faster than manual counting. The counting method by Kakui et al [58] allowed shortening the required time to 3 min per flower (more than five times faster than the counting chamber method).…”

“…The percentage viability of a pollen sample is determined, considering the number of viable grains with respect to the total number of grains found in each microscopic field. This counting can either be manual, semi-automated or automated [51]. Manual counting consists of counting stained and unstained pollens observed in fields, under compound microscopes.…”

“…Manual counting consists of counting stained and unstained pollens observed in fields, under compound microscopes. Automated and semi-automated devices for pollen count and viability assessment use software to process a series of images captured by the camera associated with the microscope [51]. Features of this automated pollen counter, its advantages and the prerequisites for its use in yam breeding programs are described in Section 3.5.…”

“…Specific information on optimizing image-based software depends on the crop (its pollen size and circularity). The association between manual count and this morphometric software is high (r > 0.75) if the optimization is done correctly [51,55].…”

“…The manual pollen counting method used in previous yam studies is laborious, time-consuming and limits the number of accessions and samples that can be evaluated [26,48,49]. It is also prone to various errors, including colours' appreciation, counting, and data recording errors [51]. An alternative method for pollen counting and viability assessment is, therefore, necessary.…”