Fruit-feeding butterfly diversity varies by height and season in dry forest

In the dry, thorny forests of northeastern Brazil, the story of butterfly diversity has just gained a striking new chapter. A year-long study conducted in the Serra da Jacobina, part of the northern Espinhaço Range in Bahia, has revealed that fruit-feeding butterflies in a tropical dry forest divide themselves not only by season but also by height—occupying distinct vertical layers of the forest in ways that scientists say could reshape how these threatened ecosystems are surveyed and conserved. The research, published in The Science of Nature, recorded 2,166 individual butterflies belonging to 51 species of the family Nymphalidae, making it one of the most complete portraits yet of how butterfly communities are organized in the Caatinga, the largest tropical dry forest region in South America and one of the most heavily disturbed by human activity.

The team, led by Bianca Santana Dias Nascimento of the State University of Feira de Santana together with colleagues from the Federal University of the São Francisco Valley and the Federal University of Minas Gerais, set out to answer a deceptively simple question: do canopy and understory communities of fruit-feeding butterflies differ in a dry forest, the way they famously do in rainforests? Despite decades of research on vertical stratification in humid tropical forests, dry forests had been largely overlooked, particularly in mountainous regions where seasonal gradients are extreme. The Caatinga presented an ideal test case. Annual rainfall in the region ranges from roughly 477 to 1,129 millimeters, mean temperatures hover between 25 and 29 degrees Celsius, and the landscape sits inside Brazil’s so-called drought polygon, where high temperatures, low humidity, and recurrent droughts define life for plants and animals alike.

The sampling design was meticulous. Across one year, from October 2022 to September 2023, the researchers established twenty fixed plots measuring 25 by 4 meters, spaced at least 100 meters apart to ensure independence. In the center of each plot they hung two Van Someren-Rydon traps, a standard tool in fruit-feeding butterfly research: one positioned in the understory at about 1 meter above the ground and another in the canopy at 6 meters or higher. All 40 traps were baited with a fermented mixture of banana and sugarcane juice prepared at a 6:1 ratio and allowed to ferment for 48 hours. Traps were deployed monthly and left in the field for two days at a time, accumulating a total of 576 hours of sampling per trap over the year. Alongside the butterfly counts, the team measured temperature, humidity, light intensity, and rainfall each month, and tagged 1,094 plants to track fruiting phenology using Fournier’s semi-quantitative intensity method, counting only ripe fruits to capture the true availability of food resources.

What the traps revealed was a community split cleanly along vertical lines. Community composition differed significantly between strata, as confirmed by a non-metric multidimensional scaling ordination and a PERMANOVA test on Bray-Curtis dissimilarities (F = 10.37, p = 0.001). Nine species were found exclusively in the understory and eight exclusively in the canopy, while 34 species occurred in both layers. The overall champion was Hamadryas februa, a Biblidinae butterfly that accounted for 371 individuals, most of them captured near the ground. In the canopy, however, a different species reigned: Biblis hyperia nectanabis, with 122 individuals. Species richness itself did not differ significantly between strata, but abundance told a sharper story: the understory averaged 108.9 individuals per sampling unit, roughly double the canopy’s 53.2, a statistically robust difference. The researchers attribute this imbalance to the understory’s more stable microclimate and its role as a repository for decomposing fruits and organic material—a continuous food base that the hotter, windier, and more radiation-exposed canopy cannot match. Forest cover in the understory also buffers extremes such as heavy wind and rain, which can damage the delicate wings of small butterflies and hamper their flight.

The temporal patterns proved equally surprising, and in some respects they ran counter to what ecologists have come to expect from wetter tropical systems. In tropical ecosystems such as the Atlantic Forest, the Cerrado, and many Neotropical dry forests, butterfly diversity is usually positively associated with humidity and precipitation. Here, the opposite held for abundance: rainfall and humidity had negative effects on the number of butterflies captured, while only temperature showed a positive association with species richness. Species richness peaked between December and February, the hottest stretch of the year, reaching a maximum of 30 species. Abundance, by contrast, peaked in September—the driest month—with 308 individuals captured, roughly two months after the rainfall peak in May and June. The researchers interpret this lag as a cascade of ecological timing: early rains stimulate host plants to flush leaves, caterpillars feed and develop, and the resulting wave of adult butterflies emerges weeks later, just as the dry season sets in. Temperature, meanwhile, is known to accelerate insect metabolism, development, and reproduction, which may explain why the warmest months produced the richest communities.

Just as intriguing was what the data said about food. Despite the butterflies’ reputation as fruit feeders, the density of zoochoric fruiting plants had no measurable effect on either abundance or richness. Fruiting in the forest was sparse throughout the year—only 7.3 percent of tagged individuals bore ripe fruits, with little temporal variation. This scarcity fits with what is known about tropical dry forests, where fleshy-fruited, animal-dispersed plants make up only 4 to 13 percent of the flora, a stark contrast to humid forests. The team suggests that most of the butterflies recorded are dietary generalists, switching among nectar, tree sap, plant exudates, minerals, feces, and decaying animal matter as availability shifts. Species usually classified as nectar feeders, including several Archaeoprepona, Dynamine, and Siproeta species, were observed feeding on fruits and other unconventional resources, demonstrating the dietary flexibility that likely allows these insects to persist in a habitat where preferred foods appear only briefly.

To test whether subfamilies and tribes share the same seasonal rhythms, the researchers turned to circular statistics, a technique that converts linear time into angular data by treating each month as a 30-degree interval on a 360-degree circle. The method calculates a mean angle indicating the average month of activity and a vector length indicating the strength of seasonality. The analysis showed that most groups had non-random, predominantly bimodal temporal distributions—and, crucially, that the major groups were staggered across the calendar rather than overlapping. Biblidinae, which together with the Satyrini tribe makes up more than 70 percent of the community, peaked at different times than Charaxinae, Nymphalinae, Brassolini, and Morphini. The community-level patterns even diverged between strata: the understory showed a single activity peak between July and September, while the canopy exhibited two peaks, in December and September. Only the tribe Morphini, represented by a single species, was strictly unimodal, peaking in July during a period of relatively high rainfall. Taken together, the staggered peaks point to temporal niche partitioning—a strategy that may reduce competition among groups by separating their periods of reproduction and activity through the year.

The implications reach beyond pure natural history. The 51 species recorded exceed the richness reported in most other Caatinga dry forest studies, which have documented between 12 and 42 species, and the list includes Caatinga endemics such as Fountainea halice moretta and Hypna clitemnestra forbesi, as well as species rarely found in dry forests, such as Archaeoprepona demophon demophon and A. demophon thalipus. The Serra da Jacobina complex, a mountain chain of roughly 250 kilometers, has few protected areas and faces chronic pressures from land-use change, past deforestation, historic gold mining, and desertification. The authors argue that because butterfly communities are structured along both vertical and seasonal axes, biodiversity assessments that sample only one stratum or only one season risk seriously undercounting the diversity these forests hold. Fruit-feeding butterflies are widely used as ecological indicators because of their sensitivity to microclimatic conditions, making them a practical barometer for conservation planning.

The study is also, according to the authors, the first medium-to-long-term comparison of vertical stratification in fruit-feeding butterflies in a Brazilian tropical dry forest, and it fills a conspicuous gap in a biome that remains among the least studied in South America. With the Caatinga losing habitat at an accelerating pace and its endemic species facing elevated extinction risk, findings like these carry a practical urgency: understanding where and when species occur—in the canopy or the shade below, in the wet months or the dry—may determine which parts of this fragile ecosystem can be protected before they are gone. The shimmering, fruit-baited traps of Serra da Bananeira have shown that even a forest that drops its leaves and waits for rain holds a hidden architecture of diversity, written in two layers and a full turn of the seasons.