Honey, I ate the kids: and maybe it wasn't a bad idea.

The giant thing swam closer to the sergeant damselfish's nest on the coral reef, but the paterfamilias guarding the eggs didn't back down one bit. He darted and nipped threateningly at the huge, intruding shape, that is, to the extent that a 13-centimeter fish can threaten something that's more than 10 times its length.

"He bit my finger," remembers the looming menace, aka Andrea Manica of the University of Cambridge in England. "It was not even a particularly painful pinch. He didn't have any chance of hurting me in anyway except for becoming rather a nuisance. It's quite difficult when trying to take accurate measurements when the [tape measure] gets continually grabbed and pulled."

Swimming in the Seychelles, an archipelago in the Indian Ocean, Manica was just measuring fish nests. But the fierce little defender couldn't have known that. Even though the fish was far overmatched, he stayed by his eggs. His steadfast efforts against overwhelming odds would make him a hero in human terms. Yet, Manica saw this model father eat some of the eggs that he defended so fiercely.

Among fish, childcare is typically a male's job. In many species, a male claims a nesting site, and a female shops around the male territories. She eventually deposits a layer of eggs in a lucky guy's nest, and the male then releases sperm over them.

"This is where things get interesting," says Manica. "Basically, the female just goes away, and Daddy is left with all these eggs to look after."

Scientists have long observed fish and other animals eating their young. But researchers tended to dismiss the practice as confusion or a weird behavior caused by captivity. Since the 1970s, however, researchers have been rethinking the practice, looking for a positive side to cannibalizing the kids.

Eating all the kids in one meal, sometimes in one gulp, may be a salvage operation to boost the odds for future generations, some researchers suggest. They see different explanations for parents nibbling a small proportion of the eggs at a time.

Perhaps a single parent reaches a point where it's impossible to go on without a snack. Or perhaps the little ones will breathe more easily overall if a few of them just disappear. These options raise an additional question: Which kids to eat?

BIG GULP Creatures as different as newts and storks occasionally eat their own young, but fish are the most studied of the filial cannibals, as the animals that eat their own offspring are called. Species with cannibalistic parents show up in at least 17 different fish families.

Because fertilization is external, both fish parents can immediately swim away from their offspring, and plenty of morns and dads do just that. The typical cannibals aren't the swim-away parents but those that offer some care, at least for the early days.

Parenting for most fish means tending the eggs. Once they hatch, currents sweep away the minute larvae, "and good luck to them--it's a harsh world out there," says Manica. Out of 100,000 eggs, maybe 2 or 3 will reach adulthood.

Eating the kids in one gulp is "conceptually different" from consuming just a portion of them in a series of snacks, says Manica. Oddly enough, total filial cannibalism is the easier observation for biologists to explain.

Gulping the whole brood is all about the future. Among nesting fish, "the amazing thing that tends to happen is that you can end up with a male looking after something like 70-, 80-, 100-thousand eggs," says Manica. "He could easily be looking after two times his body mass in eggs." To achieve such a brood, a male typically advertises during the first several days of babysitting, for females to keep adding eggs to his stash.

Scientists propose that a male ends up with more offspring over the course of his lifetime if he doesn't run himself ragged for a small brood. He's better off starting over and trying to get more eggs. But when he abandons a small brood, some creature will eat it, so the father might as well get the benefit.

"Males sometimes just give up, eat all the eggs, and swim away" says Manica.

Egg eating also occurs among egg-guarding fish that don't create nests. Instead, a parent gathers the fertilized eggs into its mouth. The adult then swims around for days with a mouthful of eggs until the young fish hatch and ride off on the currents.

Mouth brooders tend to be female. However, the rare male mouth brooder often swallows the eggs if a female that's more attractive--that is, bigger--than the original brood's mother shows up with eggs to offer. Manica says that this happens when the father fish has invested only a day or two in the first brood's care.

Total filial cannibalism occurs less often among mouth-brooding mothers. It wastes a lot of energy, Manica says. The mother puts more into making the eggs than the father does, and she won't recover all of it should she eat them.

SNACK ATTACK In contrast to gulping down all the eggs, nibbling the occasional 100 or so could benefit the current brood. The first paper suggesting a bright side to cannibalizing the young, by Sievert Rohwer in 1978, proposed that the youngsters serve as an alternative food source for overworked caregivers.

Manica says, "Put yourself in the shoes of this poor father." A luscious pile of eggs attracts great interest in the crowded, competitive world of a coral reef. Manica has clocked a male fish fending off an intruder every minute and a half during a day of egg guarding.

"Within half an hour, you were absolutely certain that the whole place had been cleaned spotless of any sign that eggs had ever been there," he says.

The father fish can't take even a coffee break until the eggs hatch. "He's running a marathon for 10 days, and he's got no food" says Manica. "Now, all of a sudden, those babies are starting to look attractive."

"You can see the male as a cold-hearted individual eating his own offspring" Manica says. "Or, on the other hand, you can see it as: Basically some offspring get sacrificed because that's the only way to look after the rest."

Biologists have tested whether feeding a father fish cuts down on egg snacking, but results varied. In 2000, Manica staged a large field test with sergeant damselfish (Abudefduf sex-fasciatus) on a patch of reef in Malaysia.

Three times a day, he swam around with a meat baster to feed half the male fish in his study population. He offered them either sergeant damselfish eggs that he'd picked from another nest or a paste of crabmeat and bread. The males that had meals delivered cannibalized only about one-half to two-thirds as many of their eggs as the unfed neighbors did, Manica reported. Hunger and nutrition play a role in cannibalism, he concludes.

Dinner by delivery didn't eliminate the cannibalism, however. Manica proposes that even well-fed fathers weed their eggs, eating ones with infections or developmental glitches. The lowered rate of cannibalism among fed fathers was similar, says Manica, to the mortality rate that he found when he covered nests with a fine mesh so that no creature, not even Dad, could get to them.

Manica has also tested for nutritional influence in an insect cannibal, the assassin bug Rhinocoris tristis. Lurking in shrubbery in Africa, fathers protect their eggs. Like fish, male assassin bugs keep soliciting eggs during the early days of guarding a clutch.

When Manica and Lisa Thomas, also of Cambridge, checked on male assassin bugs as they stood guard over their broods, the researchers never found one out foraging for food. Yet the fathers didn't lose weight during their vigils.

In the debris left over from hatching, Manica and Thomas found egg cases that had been mysteriously emptied of their contents. These didn't have popped-open lids, which would have signaled that the young bugs had climbed out. Nor did the eases show the signature holes of attacks by a parasitic wasp.

The empty egg eases revealed the work of a subtler predator with narrow, stabbing mouthparts--like the fathers', the researchers argue. The male assassin bug sacrifices some of his offspring during his vigil, says Manica. "It if didn't have six legs and four wings, it would be a perfect fish, he notes."

BREATHING ROOM In 2002, a research group at Queen Mary, University of London proposed another explanation of filial cannibalism. Maybe Dad's doing it to avoid a shortage of oxygen for overcrowded eggs.

Adam G. Payne, now a fisheries consultant in London, and his colleagues have looked at beaugregory damselfish (Stegastes leucostictus) guarding clutches of eggs in Discovery Bay, Jamaica.

The fish eagerly ate freeze-dried nutrient tablets provided by the researchers. Males that each received a tablet daily grew faster than did males left to their own devices. As for cannibalism, though, feeding "had absolutely no effect," says Payne.

Thinking through the wreckage of that hypothesis, he noticed that the males picked eggs to eat that were scattered throughout their broods instead of just chewing big patches here and there. He also discovered that the eggs around the edges of a bald spot clearly developed faster. Payne decided to look for some benefit of lowered egg density.

Diminished oxygen availability in lab experiments did increase the parental egg eating. Payne and his colleagues also found that densely packed eggs developed more slowly than eggs that had been picked over.

The idea that the dads are providing the eggs with breathing room intrigued Hope Klug of the University of Florida in Gainesville. "There seem to be a lot of systems where either personal condition or food availability doesn't affect the amount of cannibalism," she says.

To test the importance of crowding, Klug looked to the sand gobies (Pomatoschistus minutus) common in the Baltic Sea. Males pile sand around shells or rock crevices to make nesting enclosures with openings only a centimeter or two across. If a female favors a particular male, she slips into his shelter and spends several hours upside down depositing a patch of eggs in a single layer on the ceiling. After the male fertilizes the eggs, the mother leaves them to his care.

In the University of Helsinki's zoological station in Finland, Klug placed split flower pots of two sizes into the fish tanks. The fish spread their eggs loosely on the large pottery pieces or crammed the eggs densely into the small-pot halves. Then, Klug moved egg batches, which were roughly equal in number, into a medium-size structure to standardize the conditions.

Males with tightly packed eggs cannibalized more eggs than did the males with loosely packed eggs.

In a separate experiment, Klug found that low egg densities improved the chances of embryo survival. Oddly enough, though, the oxygen concentrations had no effect on survival. Perhaps some other quirk of density, such as low concentrations of waste products building up around the eggs, improves embryo survival.

Whatever the mechanism that boosts survival, Klug calculates that the males could eat 40 percent of their eggs, on average, without reducing the outcome of their reproductive efforts. In the absence of those eggs, the remaining eggs survive at a rate high enough to compensate for the drop from the original number, Klug and her colleagues reported in the October 2006 Evolution.

WHO'S NEXT? Eggs aren't equal in their chances of getting eaten by a parent. Theorists have predicted that a hungry parent guarding a brood would choose the younger eggs because they hatch later and delay re-nesting.

Tests of this prediction during the 1980s and 1990s by Paul Sikkel of Murray (Ky.) State University found a twist. During the first days of their parenting ordeal, male garibaldi fish (Hypsypops rubicundus) eat more of the older eggs. At that time, the males are still flirting with females and accepting contributions to the egg pile. Tests showed that females looked more favorably on a male with young eggs in the pile. Once the flirting period was over, however, the males tended to snack on the younger eggs.

Besides egg age, certainty of paternity can influence a father's choice of edible offspring. Earlier tests had suggested that when another male tries to fertilize the eggs during spawning, the brood-guarding male is more likely to turn cannibal.

New work shows that an intruding male has this effect even on a fish that doesn't take care of his eggs. The colorful males of Telmatherina sarasinorum compete fiercely to spawn with the drab females of Indonesia's Lake Matano. Even after a couple has paired up, another male may dash over to it and release sperm that might reach at least some of the eggs. The original male then might understandably make the best of a bad job and get lunch.

For three breeding seasons, Suzanne M. Gray of Simon Fraser University in Burnaby, British Columbia, watched several hundred encounters of courting T. sarasinorum. If a second male showed up, the chances tripled that the original male would turn around right after spawning and gobble some of the brood off the lake bottom. If a third male showed up, the chances of egg cannibalism rose almost sixfold, Gray and her colleagues report in the February American Naturalist.

This is the first time that anyone has shown cannibalism increasing as evidence of cuckoldry increases, says Gray.

In calculating the difference between the effects of one bounder and of two, scientists may be making a fine distinction in working out the nuances of filial cannibalism. But that research demonstrates how dramatically ideas have changed since the days when eating the kids was considered just a crazy, stupid mistake.