The familiar picture of fertilization is a field of sperm cells behaving as solitary competitors. A 2026 study in Nature Communications shows that this is a poor general model for arthropods. Across insects, spiders, crustaceans, centipedes and their relatives, sperm have repeatedly evolved ways to bind together and travel as units.
R. Antonio Gomez of Syracuse University and colleagues assembled sperm-structure records for 615 arthropod species and six related outgroups, then mapped the traits onto a time-calibrated evolutionary tree. Their open-access paper, published June 3, found sperm conjugation in every living arthropod subphylum and in 34 of 88 sampled orders.
This is one study, not settled consensus. It is a large comparative reconstruction built from decades of published microscopy, not an experiment that directly tested why every sperm group formed.
What counts as a sperm team
The paper defines sperm conjugation as two or more sperm physically bound together. The conjugates form in the male reproductive tract through species-specific changes to sperm shape and extracellular material secreted around them. The researchers call that binding material sperm-associated material, or SAM.
They recognized five broad arrangements. Pairs join two sperm, usually at their heads. Aggregates embed several heads in a cap or plug of secreted material. Rouleaux stack sperm in an orderly, repeating sequence. Spermatostyles are slender rods with sperm attached along them. Ensheathed groups wrap multiple sperm in a common coat.
Not all of these units swim. Ensheathed groups can be immotile until the coat is shed inside the female reproductive tract. That fact alone makes a simple “faster swimmers win” explanation incomplete.
The groups range from two cells to thousands
The range is difficult to fit into one mental image. The paper shows a silverfish with sperm joined precisely in pairs. A solifugid, an arachnid related to spiders and scorpions, stacks 32 plate-shaped sperm in 16 pairs. A sawfly aggregate contains about 240 to 350 sperm. At the far end, some rouleaux hold thousands of cells in ordered files.
These are not accidental clumps produced when sperm happen to collide. The groups form before ejaculation, and their repeated, species-specific arrangements depend on sperm anatomy and binding material made in the reproductive tract.
That distinction also separates conjugation from agglutination, the unwanted sticking together sometimes seen in a sample. One is an evolved reproductive structure. The other can be a pathological or laboratory phenomenon.
The 500-million-year story is one of reinvention
The analysis assigned unconjugated sperm to the common ancestor of arthropods with a posterior probability above 0.99. Grouping came later, then kept coming back. Depending on the model histories sampled, sperm conjugation was independently gained an estimated 34 to 63 times and lost 29 to 77 times. The authors summarize the central estimates as roughly 45 gains and 45 losses.
The oldest inferred origin sits on the branch leading to insects and remipedes, a small class of cave-dwelling crustaceans. The estimate falls between 452.6 and 508.5 million years ago, during the Cambrian or Ordovician periods. The model also places aggregate-form sperm in the ancestor of the hexapods, the group that includes insects and their wingless relatives.
Those ages come from a composite evolutionary tree calibrated with 14 fossil nodes and two estimates from earlier phylogenetic studies. They are model-based reconstructions, not fossilized sperm bundles.
So “dating back roughly 500 million years” does not mean one cellular team appeared once and survived unchanged. The study’s central finding is almost the opposite. Arthropod lineages repeatedly gained, remodeled and discarded several ways of binding sperm together. Across the full tree and the outgroups, the modeled branches spent about two-thirds of their combined evolutionary time in a conjugated state, largely because the trait evolved again so often.
Cooperation does not mean tiny agents choosing to help
The word cooperation needs care here. The cells perform collectively, but that does not show that individual sperm control the arrangement or sacrifice their own prospects. A 2011 review by Dawn Higginson and Scott Pitnick argued that sperm conjugation should usually be understood as a male-level reproductive trait, like ejaculate size or seminal proteins, rather than altruism selected through the haploid genes of individual sperm.
One proposed advantage is motility. Aligned tails might generate greater force without a matching increase in drag, especially if their beating synchronizes. But the 2026 authors are explicit that direct measurements across conjugated species remain sparse and have produced mixed results.
Speed cannot explain every form anyway. Some sheathed conjugates do not move until they break apart. Some rods have long bare stretches with no sperm attached. And a 2012 experiment on paired diving-beetle sperm found no significant difference in viability between paired and mechanically separated cells under the conditions tested.
The new paper develops another possibility: SAM may be cargo as well as scaffolding. By remaining attached to sperm, it could carry seminal-fluid proteins farther into the female reproductive tract and help them persist near sperm-storage organs. The authors present this as a hypothesis, not a demonstrated universal function.
Separate lineages reused part of the same protein toolkit
The molecular comparison focused on spermatostyles, the rods to which sperm attach. Ancestral-state modeling indicated that true bugs and beetles evolved these structures independently from simpler aggregates. Their last common ancestor, living roughly 400 million years ago, was assigned aggregate sperm rather than a spermatostyle.
Yet the two lineages built their rods with some of the same molecular parts. Across the sampled true bugs and whirligig beetles, an average of 21.7% of identified protein families occurred in both sets of structures, more overlap than the team’s random simulations predicted. Sperm leucyl aminopeptidases, or S-LAPs, were among the most abundant shared proteins.
That is evidence of molecular convergence, but it does not establish that every shared protein has the same role. The paper notes that proteins could be recruited together because they share patterns of expression or function. The result suggests an available toolkit that evolution drew on twice, not one ancient rod passed intact down both family trees.
I recently wrote about trees as a form that unrelated plant lineages evolved repeatedly. Sperm teams offer a cellular version of the same principle: similar designs can reappear because evolution keeps encountering related physical and reproductive problems.
The evolutionary map is better than the functional map
The study is unusually broad, but it inherits the limits of the literature it assembled. Species entered the dataset only when researchers could find an adequate sperm description and enough molecular information to place the species on the tree. Forty-two recognized arthropod orders, mostly small and poorly studied, were not sampled.
The phylogeny is also a custom supertree made by grafting relationships from many earlier studies. Gomez and colleagues say this provides a practical, biologically informed framework for 621 species, while acknowledging that some internal relationships may be inconsistent with the true evolutionary history. The source matrix and tree files are available in the team’s Dryad archive.
What the work establishes most clearly is recurrence: sperm grouping is old, widespread and evolutionarily easy to lose and reinvent. What it cannot yet settle is why a pair helps one species, why a thousand-cell rouleau helps another, or why some lineages abandon grouping altogether.
Answering that will require direct comparisons among close relatives with and without conjugation, measuring swimming, storage, protein transport and fertilization success inside the reproductive tract. The tree now shows where those experiments would be most informative.
