Aiptasia larvae are phenotypically validated as a model of coral bleaching using high-throughput machine-learning image analysis

Coral reefs are among the most vibrant and vital ecosystems on Earth, supporting a vast array of marine life and protecting coastlines from storms. Yet, these underwater cities are vanishing. A primary driver of this loss is a phenomenon known as bleaching, which occurs when rising ocean temperatures break the delicate partnership between coral animals and the microscopic algae living inside them. These algae provide the coral with food through photosynthesis, but when the water gets too hot, the coral expels them, turning white and starving. To understand how to save these reefs, scientists need to study exactly how and why this partnership collapses. For years, researchers have used a small sea anemone called Aiptasia as a stand-in for coral. While the adult anemones are useful, they are large, opaque, and difficult to study in large numbers without damaging them. This limitation has made it hard to run the rapid, large-scale experiments needed to find solutions.

A team of researchers has now turned to a different stage of the anemone's life: the larva. These tiny, transparent creatures are only about the width of a human hair and are clear enough to see right through. The researchers discovered that these larvae react to heat stress in much the same way adult corals do, losing their internal algae and "bleaching." To prove this, they subjected thousands of larvae to warm water and watched them over several days. They found that within two days of heat stress, the number of healthy larvae dropped significantly, and by four days, most had lost their algae. This confirmed that the tiny, transparent larvae are a perfect, fast-growing model for studying the breakdown of the coral-algae relationship.

However, counting and analyzing these tiny creatures by hand is incredibly slow and tedious. A single microscope image can contain dozens of larvae in different states of health, and determining whether an algae cell is safely inside a larva's tissue or just floating in its stomach requires looking at the image in three dimensions. Doing this for hundreds of images took the researchers more than twenty hours of intense focus. To solve this bottleneck, they built a new computer program called SYMPHONY. This tool uses artificial intelligence to look at the microscope images, identify each individual larva, and determine its health status in a matter of minutes. The program works by flattening the 3D images into a 2D format that preserves the depth information needed to tell if the algae are in the right place. It then uses a pre-trained system, originally designed to spot cells in human tissue, to find the larvae, and a second system to classify them as healthy, bleached, or too difficult to tell.

The results showed that the computer program could match the work of human experts with high accuracy, identifying the correct health status of the larvae about 79 percent of the time. More importantly, the program confirmed the same trends the humans found: heat stress causes the larvae to lose their algae, and the longer the heat lasts, the more severe the loss becomes. While the computer did flag some images as too confusing to judge automatically, it still saved the researchers over twenty hours of labor for this single study. By combining a tiny, transparent animal model with a smart, automated way to count them, the researchers have created a powerful new tool. This approach allows scientists to test thousands of larvae quickly, opening the door to faster discoveries about how corals respond to heat and what might help them survive a warming ocean.