On the morning of April 9, 2023, Tommy Sheets stopped walking. He had been contracted to remove Burmese pythons for the Florida Fish and Wildlife Conservation Commission since 2019, working mostly in Big Cypress National Preserve and the East Everglades Wildlife Management Areas west of Miami. By the spring of 2023, he had logged more than 800 python removals. He knew what a dead python looked like in every form the Everglades produced: one he had caught himself, one killed by an alligator with its distinctive jaw bruising, a roadkill, or a snake that had died in a hard cold snap and rotted into the grass before anyone reached it.
He had never seen anything like what he found that morning in a pine forest section along a firebreak in the Fakahatchee Strand. The animal was 14 feet long. An adult female in breeding condition, her body mass estimated at nearly 40 kilograms when he lifted her torso to search for a tracking implant. The body was perfectly intact from the base of the neck to the tail.
The muscle tissue was still fresh. The dorsal scales showed no injuries, no puncture wounds, no pressure damage consistent with an alligator attack, no claw marks, and no sign of dragging in the pine needles around the carcass. The pine needles themselves were undisturbed in every direction around the snake. No struggle had taken place at that site.
What the snake did not have was its head. The neck ended cleanly about eight inches from where the skull should have been. The cut was not surgical in any sense of the word. The separation was the result of a bite.
A single bite from an animal whose jaw geometry Sheets could not immediately identify, leaving the rest of a 14-foot predator lying undisturbed in the pine needles, as if the forest had removed one specific part of it and left everything else behind. He stood over it for a long time. He photographed it from every angle before touching it, kneeling to capture the site where the neck ended from three different distances, then stepping back to fit the animal’s full length into a single shot. He filed his recovery report with the commission that evening and added a single line at the bottom of the digital form, the line he would repeat more than a dozen times over the next eighteen months.
Cause unknown? Head missing. To understand why the headless carcasses mattered, you first had to understand what they were cutting. Burmese pythons reached the Florida Everglades through the exotic pet trade in the 1980s, and the situation worsened in 1992 when Hurricane Andrew scattered animals from commercial breeding facilities along the Tamiami Trail.
What those first snakes found was a subtropical ecosystem that provided everything their biology required. Warm temperatures year-round. An abundance of prey. Dense vegetative cover.
And the complete absence of any native predator capable of targeting an adult constrictor of the size it reaches in Florida. The ecosystem had no evolutionary reference for them. The prey species they encountered had no defensive adaptations against them. The native predators that shared their habitat had no behavioral model for dealing with them.
A 14-foot snake in the Florida Everglades was, for most of its adult life, functionally immune to everything the ecosystem could field against it. Raccoon populations in the python’s core range collapsed by more than 99 percent. Opossum populations by 98 percent. Bobcat populations fell by nearly 90 percent in affected areas.
The Everglades filtration system provides clean water to more than 9 million people and depends on a food web that the python had spent four decades dismantling. Florida launched the Python Elimination Program in 2017. More than 14,000 animals have been removed since then, and researchers will tell you plainly that the snake’s reproductive capacity calculations absorb every removal within a single breeding season and continue expanding. The python population has been defeating human effort for four decades.
What the headless carcasses represented was different, and the difference began with where they appeared and what they did not show. The FWC Wildlife Research Laboratory in Gainesville maintains a forensic database of unusual python carcass recoveries. From 2017 through 2020, the database contained zero entries indicating a missing head. Beginning in 2021, the entries began to accumulate.
Three in 2021, seven in 2022, 14 in 2023. A clear trend, and all cases shared the same set of characteristics. Adult animals, mostly female, bodies largely intact with no drag marks, and a cut in the neck at a fixed point, eight to 14 inches from the skull. No jaw-pressure bruising indicating an alligator attack.
No clusters of puncture wounds indicating a panther attack. Dr. Laura Heim, a wildlife forensic specialist at the FWC Wildlife Research Laboratory in Gainesville, requested skull fragments from field teams. Most recoveries yielded nothing usable because the head had been removed entirely.
But in 11 of the recorded cases, skull fragments were found under the body or within one meter of the neck-severing point. Skull bone fragments, parts of the jaw, and tooth marks embedded in the bone where the bite force had been greatest. Eleven partial sets to work with. The forensic analysis of those 11 samples covered jaw-angle geometry, bite force estimation from the depth of bone compression, tooth-spacing measurement from the embedded impressions, and the geometry of the wound edges at the sever point.
The methodology was identical to standard animal attack investigations. The 11 samples yielded the same result from every analytical direction at once. The jaw angle was consistent with a medium-sized canid, ruling out alligator and panther on geometry alone. Bite force ranged between 80 and 160 pounds per square inch, far below an alligator’s crushing capacity.
Tooth spacing was consistent across the samples in a range matching a predator weighing between 11 and 18 kilograms. And the wound-edge geometry: every sample showed the separation occurred in a single bite, rather than the progressive multiple-bite pattern a larger predator follows. One bite, the right angle, and complete separation in a single moment of contact. Heim requested field data on every predator species documented in the recovery areas.
She was looking for a candidate in the 11 to 18 kilogram range, present in sufficient numbers in the specific management units where the headless carcasses were concentrated. The field data pointed back to a single species whose range, body mass, jaw anatomy, and documented presence in the Everglades matched every criterion at once. The coyote. Not a species appearing in any of the snake-management literature as a relevant predator.
The eastern coyote has been documented in Florida since the 1970s, and the South Florida population, while not forming the dense urban packs familiar in Texas or California, is stable and expanding. GPS collar studies in Miami-Dade and Collier counties show coyote home ranges overlapping directly with the high-density python zones in the East Everglades and Big Cypress management units. Adult coyote body mass in Florida ranges from 9 to 18 kilograms. Jaw anatomy produces a bite force within the documented range.
Skull geometry matches the forensic analysis. Tooth spacing matches the measured impressions. Every measurement aligned in every direction. What the measurements could not explain on their own was the mechanism.
How could an animal of at most 18 kilograms kill an adult Burmese python that could weigh 40 kilograms or more, whose constriction force produced enough pressure to immobilize a white-tailed deer? The answer required a different kind of investigation, focused not on coyote anatomy but on snake biology. Specifically, on what the snake is doing in the hours and days when a coyote, in the right position and timing, can approach and strike the skull before the snake’s defensive reaction begins. The Burmese python’s vulnerability profile is something the scientific literature has documented intermittently without assembling it into a coherent pattern, until the forensic data from the headless carcasses forced the question.
There are three distinct time windows in the snake’s behavioral and physiological cycle during which its ability to respond defensively is not merely low, but completely disabled. The first window is post-feeding digestion. When a large snake consumes a large prey item, the metabolic demand of digesting that meal triggers a massive physiological response. Blood flow is redirected from muscle tissue to the digestive organs at a rate unparalleled in mammalian physiology.
The heart physically enlarges by up to 40 percent during peak digestion to meet the circulatory demand needed to process a large meal. Muscle oxygenation drops significantly during this period. A snake that has consumed prey equal to 20 percent or more of its own body mass enters a physiological state in which its coiling reflex is completely absent for several hours. The animal is physically incapable of producing the constriction force that makes it dangerous.
It is a 40-kilogram mass of muscle temporarily stopped from functioning as muscle, functioning instead as a digestive organ. The only defensive response available during this period is evasive movement, and escape requires muscles that are currently out of service. The second window is egg incubation. A female Burmese python incubating her eggs coils entirely around them, maintaining that position with almost complete behavioral focus for an incubation period of two to three months.
She does not feed. She generates heat for the eggs through fine muscle contractions, a behavior unique among non-avian reptiles. Her attention, insofar as the concept of sensory attention can be projected onto a reptile, is on the eggs beneath her. She does not move to investigate stimuli at the edges of her site.
She is still. Still in a way that makes a predator approaching from behind her unnoticeable until it is close enough to strike. Her field of vision, determined by the resting position of her head, covers the arc in front and to both sides. It does not cover the arc behind.
The third window is the shedding cycle. Snakes shed their skin at intervals determined by growth rate and health. In the days immediately preceding a shed, a specific physiological change occurs in the fluid layer between the old skin and the new. This fluid produces a milky opacity that spreads to the eye membrane, the fixed transparent scales covering the eyes.
The snake’s eyes cloud over completely. Visual detection of approaching objects drops to a fraction of its normal capacity. The animal can detect broad light changes and moving shadows at close range, but cannot distinguish the approach of a predator-sized object unless that object is within two to three feet of physical contact. A snake in pre-shed condition is essentially blind to any threat approaching from beyond striking range, and it does not know what it cannot see.
All three windows share one structural feature. They are periods when the snake’s defensive response to an approaching threat is compromised at the point where the threat begins, not merely at the point where the response would be executed. The snake either cannot see the coyote clearly, or cannot move toward or away from it effectively, or cannot generate the constriction force that would neutralize a contact event. A coyote that exploits any of these windows is not facing a 40-kilogram apex predator.
It is facing a 40-kilogram animal that cannot fully defend itself. What the blind-spot analysis added to this picture was the specific approach geometry. The lateral positioning of a snake’s eyes produces a wide field of view covering nearly 320 degrees. But that same geometry produces a true, fixed blind spot in the arc directly behind the skull, from about 30 degrees to 45 degrees off both sides of the centerline, where the visual fields of the two eyes do not overlap.
A threat approaching from that specific arc is invisible to the snake regardless of how alert the animal is or how efficient its visual system is at that moment. The arc where the coyote’s jaws connected in every documented headless carcass case is that arc. The entry angle of the neck-terminating wound, measured across 11 forensic samples, clusters between 28 degrees and 43 degrees off the snake’s centerline on a rear-approach vector. Every documented kill was executed from within the blind spot.
The forensic data does not tell you whether the individual coyotes that carried out those killings consciously knew they were approaching a blind spot. What the data does tell you is that the attack angle converges with extreme precision across many independent events, so precisely that random approach geometry cannot explain it. Something in the coyote’s behavioral repertoire leads to an approach from that specific arc, whether through trial and error with pythons that succeeded or failed based on approach angle, through the systematic avoidance of the snake’s reactive striking range that any intelligent predator would develop after its first encounter with a constrictor, or through some mechanism of behavioral learning the data cannot distinguish. The documented approach pattern is consistent across killings separated by months and geographic distances that rule out repetition by any single individual animal.
Heim’s forensic report compared approach angles against trail camera footage from the carcass discovery areas. In three cases, footage from nearby stations showed coyote activity in the same areas and during the same time windows as the dated carcass recoveries, with the animals moving in directions consistent with the documented blind-spot approach arc. The footage was not conclusive. But added to the forensic measurements, the body-mass match, the jaw-force data, and the documented vulnerability windows, the evidentiary picture had passed the threshold of practical explanation.
The increase in documented headless carcasses, from three in 2021 to seven in 2022 then to 14 in 2023, followed a pattern documented in the behavioral ecology literature for learning that spreads through a population via observation and repetition. Just as crow flocks developed the behavior of placing hard-shelled mollusks on road surfaces at intersections and retrieving them after vehicles crushed them, just as orca pods developed beach-hunting of seals by watching pod members who had discovered the technique, this was not instinct. It was knowledge transmitted through a community capable of observational learning. A coyote that discovers, through an encounter with a digesting, incubating, or pre-shed python, that approaching from the correct angle makes a skull strike low-risk and high-yield will repeat that behavior.
A coyote that watches another coyote complete that behavior will add it to its own behavioral repertoire. The brain of an adult Burmese python is the most calorie-dense food item in the coyote’s operational environment in the Florida Everglades marsh. The risk of a properly executed approach from the documented arc during one of the three documented vulnerability windows is clearly low. The forensic record shows no evidence of coyote injury at any of the 14 recovery sites.
The reward is large. The risk approaches zero when timing and angle are right. This is not behavior confined to a single individual. This is behavior spreading.
What this accumulation means for the snake population in the affected areas requires context. Fourteen documented headless carcasses in 2023 is not an elimination number that changes the growth trajectory of an Everglades python population on its own. Fourteen animals against a population that may number 300,000 individuals is a statistical rounding error. But the forensic database only records carcasses that contracted hunters actually found and reported in actively monitored areas.
The Big Cypress prairies and the dense sawgrass of Shark River Slough contain carcasses no human reaches and no database records. The documented headless kills are a sample of a behavior, not a complete census of it. What the trend in that sample shows is a behavior absent from observation records before 2021, appearing at low frequency in 2021, and increasing at a rate of nearly 100 percent annually over the documented period. A behavior that spreads through observation will not increase linearly.
It will increase until it reaches a saturation point in the population. That saturation point has not yet been reached. Tommy Sheets extracted his second headless python from the same Fakahatchee Strand area in October 2023, six months after the April carcass that began his file on the pattern. This one was 11 feet long.
A juvenile female, not yet of breeding age, meaning not one of the high reproductive-value animals the removal program prioritizes. He photographed it, filed the report, and added the same line at the bottom. He knew by then what the Gainesville forensic lab had concluded about the April animal and those that came before it. He sat at the edge of the pine forest for a while before getting in his truck.
He later told a field coordinator with the Fish and Wildlife Conservation Commission that he had been running surveys along that exact corridor for four years and had never once seen a coyote in all his years working from dawn to dusk in that forest. He heard them at night on the road leading to the trailhead. He knew they were there from the track surveys the commission ran periodically in the area. But he had simply never seen one.
Something had been working in that forest all along, hidden, and learning something in that forest that everyone had missed. A coyote walked into a pine forest in the Fakahatchee Strand sometime around 2020 or 2021, approached from the one angle a 40-kilogram reptile could not track, and discovered something no wildlife manager had put into any management plan. The behavior it displayed in that moment is now documented in forensic records across multiple management units, increasing in frequency every year, transmitted by animals that no one managing the Florida python crisis has included in any model, budget, or strategic document. No one planned it.
No one funded it. No one designed the approach angle. The forest already knew it.
