Zoobooth: A portable, open-source and affordable approach for repeated size measurements of live individual zooplankton

Body size is a crucial characteristic for most organisms, especially pelagic ones where it influences metabolic rates, clearance rates, swimming speeds, and sensory ranges. In limnic pelagic ecosystems, zooplankton community size structure often indicates trophic structure, as per the Size-Efficiency Hypothesis. Size also significantly shapes phytoplankton ecological niches due to correlations with ecophysiological traits. Measuring organism size, while seemingly simple, presents challenges. Immobilizing or killing specimens simplifies measurement but is unsuitable for many research questions, particularly those involving live, repeatedly measured individuals. Existing methods like the Optical Plankton Counter, FlowCam, and ZooScan offer efficient community-level analysis, but lack the ability to track individual organisms over time. Manual measurement under a microscope, while precise, is time-consuming, requires immobile specimens, and introduces observer bias. While image analysis of photographs can mitigate bias, it still necessitates handling, which risks accidents and induces stress, impacting growth and reproduction in some species. Although the effects of handling on Daphnia are not well-documented, frequent handling poses potential risks, especially for delicate species, and the need to remove them from experimental conditions can compromise ecological studies. For studies requiring repeated measurements of live individuals over their lifespan—to study individual growth variability influenced by environmental factors—existing methods proved inadequate. This led to the development of Zoobooth, a portable, affordable, and open-source system based on a Raspberry Pi and camera, using Python and R for image analysis. This addresses the need for a method offering accurate, low-stress size measurements in a portable setup.

Several instruments have been developed to increase the efficiency of obtaining plankton size data. The Optical Plankton Counter provides a quantitative method based on optical detection and automates the collection of abundance and size data of zooplankton. FlowCam and ZooScan use imaging techniques and image analysis to collect data on shape, size, and other physical properties from organisms in micro- and macroplankton samples. These instruments are highly efficient for studies on species distribution and community structure in natural aquatic ecosystems. However, they are less suitable for laboratory studies on a single or few plankton species where tracking individuals is crucial. Traditional methods involve manual measurement of specimens on glass slides under a microscope, either directly or from photographs. This provides precise data but is time-consuming and risks stressing the organism. Image analysis of photographs improves this process but still involves handling and the risk of stress. Existing methods for tracking live Daphnia, such as Multi-DaphTrack and millifluidic chip-based systems, primarily focus on behavior rather than size measurement. Daphniatox and Faerøvig et al.'s approach are more suited for population data. The automated imaging approach of Heuschele et al. is designed for meiobenthic animals and not ideal for pelagic organisms. The high-throughput microtiter well-plate system described in Duckworth et al. seems to require a long image acquisition time. None of these fully met the study's needs for accurate, low-stress, repeated size measurements of individual zooplankton in a portable setup.

The Zoobooth approach consists of three steps: instrument setup, imaging procedure, and video analysis. The instrument setup comprises a Raspberry Pi 3 B+, a Raspberry Pi NOIR camera (v2) with a 6mm lens, two white LEDs, a touchscreen monitor, an external hard drive, and a power bank, all housed in a portable case. A 3D-printed housing accommodates a disposable cuvette holding the zooplankton. A Python script controls the camera. The camera matrix was determined to correct for lens distortion using a checkerboard calibration. The imaging procedure involves pipetting the individual zooplankton into the cuvette, executing the Python camera script (CameraScript.py) to record a 30-second video, and returning the animal to its culture vessel. Video analysis uses a custom Python script (VideoAnalysisScript.py) with OpenCV for object detection. The script extracts size measurements from each frame meeting defined criteria (brightness, minimum size, position) ensuring the object is a zooplankton and in focus. The image is processed (cropping, distortion correction, grayscale conversion, background subtraction, erosion, dilation) before contour finding. The largest object is tested against species-specific criteria (size range, length-to-width ratio, brightness, sharpness, distance from cuvette edge). If the criteria are met, an ellipse is fitted, and the major axis is recorded as length, converted to millimeters using a pre-determined conversion factor. The results are saved in *.detailedsizedata.csv and *.sizedata.csv (the latter contains the optimal percentile estimate). To find the optimal video size estimate, manual measurements were compared with percentiles from the distribution of video estimates. The percentile that minimized the absolute difference between manual and video measurements was selected as the optimal estimate for each species (e.g., 93rd percentile for D. magna). Assessment data were collected on D. magna (cultured and from a large experiment) and five other freshwater zooplankton species (collected from the field), comparing manual and Zoobooth measurements. Manual measurements were done using microscope photographs and ImageJ software.

The Zoobooth system demonstrated high accuracy in size estimation. For D. magna, the correlation between Zoobooth and manual measurements was 0.97 (Pearson's r). The average absolute difference between methods was low, ranging from 0.10mm for individuals <2mm to 0.32mm for individuals >3mm, with average relative percentage differences below 9%. Testing on other zooplankton species showed varied accuracy depending on the species morphology. Species with Daphnia-like morphology or simple shapes produced good results. However, species with uneven body transparency or prominent appendages (like Polyphemus, Scapholebrius, and Heterocope) required adjustments. The long-term experiment on D. magna (320 individuals, one year, average 26 recordings/animal) resulted in a very low mortality rate (7 out of 320), suggesting minimal stress. The lifespans of the D. magna were comparable to that of healthy individuals, further supporting minimal stress. The Zoobooth method, although less accurate than manual measurement, provided sufficient accuracy for many applications, especially growth modeling. Picture files automatically saved during video analysis allow easy validation. The Zoobooth setup is portable (33 x 28 cm, 2.5 kg), low-cost (approximately 330 euros), and time-efficient (approximately 1 minute per measurement).

Zoobooth offers a cost-effective, robust, accurate, and efficient method for obtaining repeated size measurements of individual zooplankton, surpassing existing methods in several aspects. The accuracy is adequate for many ecological studies, especially parameterizing growth models. The method's low-stress nature, portability, and ease of use are crucial advantages for long-term experiments and fieldwork. While the original design focused on D. magna, the adaptability to other species with slight modifications to the filtering criteria broadens its applicability. The need for adjustments for species with differing morphologies highlights the ongoing development potential. Future improvements could involve more sophisticated image processing techniques or deep learning algorithms for enhanced species identification and filtering. The comparison with alternative approaches showed that Zoobooth is superior in terms of accuracy, cost-effectiveness, portability, and time efficiency. The limitations of other methods, such as long image acquisition times, high costs, or inability to track individuals, further underscore the value of Zoobooth.

The Zoobooth system provides a significant advancement in obtaining repeated size measurements of live individual zooplankton. Its accuracy, low cost, portability, and ease of use make it a valuable tool for ecological studies and education. The open-source nature promotes further development and adaptability to diverse species and research questions. Future work could focus on improving image processing to handle more morphologically diverse zooplankton and integrating advanced image analysis techniques to extract additional data beyond size, like coloration.

The accuracy of Zoobooth is lower than manual measurements, although it is still suitable for many purposes. The method's performance varies across zooplankton species, with species having complex morphologies or uneven body transparency requiring script adjustments. The initial setup requires some technical skills and programming knowledge.