A so-called conservation biologist at St. Lawrence University in Canton, New York, named Susan Willson has five cats, two that live outside and three indoors. One of the outdoor cats, named “the Gorilla,” is an admitted “serial killer” cat, one that is “especially adept at killing birds.” She tried but decided not to keep this cat indoors because it gets into “a peeing war” with her three indoor cats, so she is willing to accept her cat killing a lot of birds because she does not want to euthanize her cat. All things considered, she would fall under what many conservation biologists call a “cat lady.”
But she evidently feels at least some guilt over the needless loss of wildlife at her hands because she bought a Birdsbesafe Collar Cover that is claimed to reduce bird kills, and then conducted a study to at least partly justify the collar cover as adequate mitigation for those wildlife kills. Her study is reported in Birdsbesafe: Can a novel cat collar reduce avian mortality by domestic cats (Felis catus)? S.K. Willson, I.A. Okunlola, and J.A. Novak. Global Ecology and Conservation 3(2015) 359-366.
Birdsbesafe’s product consists of a tube of brightly colored cloth enclosing a break-away nylon collar and having a reflective edge trim to make it easier to see by motorists at night. Collar covers are brightly colored so that birds and some other prey species can more easily spot cats stalking them. Reds and oranges are used extensively along with yellows; however, they noted that reds, oranges, and yellows can match autumn leaf colors and thus might camouflage cats somewhat and be responsible for the increase observed in autumn bird predation. Many mammals, including nearly all nocturnal mammals, cannot see red colors, rendering Birdsbesafe’s product of little value to them. Conversely, reds and yellows possibly come across as camouflage to color-blind mammals under certain conditions. Also, collar covers reflecting ultraviolet could provide additional visual warning to some wildlife, but this was not addressed.
Willson et al. found that 71% of tested cats got used to collar covers within 5 days. That means the collar covers will be problematic for 29% of the kind of cats they tested. Since all tested cats were fully domesticated human-owned animals, it is probable that feral cats would exhibit significantly lower collar cover acceptance rates. In combination with the difficulty in trapping and then re-trapping feral cats to install and replace lost break-away collars, effective use rates are likely to be quite low and thus ineffective mitigation. This point was not reported by authors or noted by Birdsbesafe on its website.
Birdsbesafe collar covers are made of an unidentified “fabric” and designed to enclose a nylon collar, both of which will absorb and hold water and other liquids. If the product is never washed, such as with feral cats, cat hoarders, and other low-quality cat owners, collar covers can become stiff with embedded algae and dirt (and age). This could result in open sores on the cats from abrasion and could snag on brush and fencing when running from predators. This may not be an issue in a fenced-in suburban back yard, but both issues could constitute cruelty when applied to feral cats whose collars cannot be replaced due to trap-shyness. Willson et al. and Birdsbesafe do not address this issue even though authors suggest that collar covers be used on feral cats.
Willson et al. tested the successfulness of collar covers only on serial killer cats like “the Gorilla.” They placed collar covers on 54 cats in the fall and 19 cats in the spring for twelve weeks each, and claimed that collar covers reduced bird kills by a factor of 19 times in the spring and 3.4 times in the fall and decreased small mammal predation by about half in the fall but had no effect in the spring. In comparison, belled collars were shown by a second research group to reduce bird and mammal kills by 50%, although a third study revealed that cats can compensate for the bells and develop different hunting strategies. A fourth study showed that the CatAlert product, which produces a beeping sound every seven seconds, resulted in cat kill reductions that were not significantly different from bells. The CatBib is a large piece of colored cloth attached to a collar that not only visually warns birds and color-vision enabled mammals but also places a physical barrier between cat paws and wildlife victims. A fifth study found that 81% of CatBibbed cats stopped catching birds and 45% stopped catching mammals, but noted that owner acceptance would likely be small because the CatBib is bulky and appears awkward.
The serial killer cats that were employed in the Bridsbesafe collar cover study habitually brought home killed but allegedly “intact” prey. Obviously, this protocol does not account for kills that were simply left in the field, stolen by other predators, or eaten. This could be a significant flaw in the data because large numbers of uncounted kills could go unreported:
Assume that an uncollared cat over a week’s time kills 10 birds, eats 2 of them, and brings 8 home. Assume that a Birdsbesafe collar cover is then put on that same cat and it then kills only 2 birds but eats both of them over the next week. Willson et al.’s conclusions would then be that collar covers reduced kills from 8 to 0, or by 100%, when the real rate would be 80%.
Alternatively, assume that the uncollared cat over a week again killed 10 birds, ate 2 of them, and brought 8 home, but a collar cover used over the following week resulted in only 4 bird kills, 2 of which were eaten and the other 2 brought home. This would result in a calculated 75% reduction in kills when the actual number would be only 60%.
Birdsbesafe and Willson et al. can play with the numbers and come up with any success rate they want, but any way that I look at the numbers, authors’ results are optimistic and may actually be gross overestimates of success because they do not contain a correction factor to account for kills not brought home. Has any research been done on quantifying such a correction factor? I did a lot of literature reviewing on this subject a few years ago and do not remember any such correction factor being published in the scientific literature, but if a correction factor does exist, I am sure that a scientist like Willson at an institute of higher learning like St. Lawrence University would have applied it. If a correction factor does not exist and if it is potentially important, then why do Willson’s and other research groups not ascertain it? It would clearly be a significant gap in our knowledge of cat kills and cat prey mitigation.
Willson et al.’s data collection occurred during two seasons, spring and fall. There were no reasons cited for those seasons nor for not sampling during other seasons. I suspect it had to do with summer college vacations and winter cold, but authors should have addressed whether these two seasons are representative of a full year or else should have pointed out that their results may not be indicative of an annual period. Birdsbesafe’s 87% reduction claim implies an annual basis, which I think is misleading.
Willson et al.’s tested cats brought home 19 species of birds from 12 families and 11 species of mammals from 8 families, totaling 50 birds and 136 mammals. Unfortunately, authors did not include a raw-numbers table of before-and-after kills, so we cannot scrutinize the actual data ourselves. The only raw data provided by authors was a list of kill numbers by species, and the only reduced data provided is a bar chart of average kill numbers per cat per week. They then report their results in terms of reduction multiples, stating that uncollared cats killed 19 times more birds in the spring and 3.4 times more in the fall. Similarly, uncollared mammal kills were twice as high as collar-cover kills in the fall but were equal to collar-cover kills in the spring. In other words, if I have done my math correctly in the absence of raw data, collar covers reduced bird kills by 77-95% and mammal kills by 0-50%. However, considering the uncounted-kills flaw, those numbers may be overestimates by 10%, 20%, or even more.
Willson et al. reported that only one herpetofaunal (herp) prey item was brought home, a spring peeper frog. They claimed that the reason for this low number is that the study was done in northern New York, implying that this locale has low herp populations. I am not sure if that is true, but even if it is, the implication should have been clearly stated as a fact and should have been referenced.
They quote a sixth report that free-ranging cats wearing kitty-cams caught more herps than any other type of animal. Additional reports have found that cat owners are more likely to bag and freeze certain animals than others. Owners (1) readily collect small birds and intact (non-bloody) mammals and lizards, (2) are less likely to handle shredded carcasses, (3) are unlikely to even see, much less collect small herps half-eaten and left by the cat in the lawn, and (4) snakes of any kind or size. Therefore, it is likely that Willson et al.’s tested cats did bring home more herps than owners were willing to collect and report, so authors were correct to eliminate herps from their study.
The generally accepted and most often way in the biological literature to convey before-and-after changes is to report percent changes, whereas Willson et al. reported only reduction multiples. The latter are indeed fine to impart, but only if percent changes are also reported. Authors’ peers cannot otherwise confirm the original findings, which is important if herps or other prey items are left out of the equation.
Willson et al.’s data is based on kills that were collected by cat owners and saved for seasonal pickup, thus relying on the cat owner claims that they saved all the brought-home kills. It is known from other studies that many prey items brought home are in varying states of intactness and that owners will leave a particularly stinky mess on the lawn rather than bag and freeze it. This was not mentioned by the authors even though it is a known flaw of owner-based data collection and will underestimate predation rates.
Willson et al. initially used collars that easily broke away and were lost approximately 50% of the time. They replaced that model with one that broke away less than 10% of the time. Obviously, different break-away rates will produce different efficacy rates, but authors neither gave different efficacy rates for the two collar models nor reported whether the two data sets were determined separately or lumped together. This should have been clarified. Regardless, authors did not name the brands/models of the tested collars, contrary to standard methods. This failure means that their experiment cannot be independently repeated by others.
Willson et al. did not address whether any of the tested cats learned how to open a break-away collar or if the broken collars were accidentally separated. Although not integral to their study, cats may find it easier to learn how to disengage some collar models than others, and this might be why one of their two models broke away so often. This is yet another reason why authors should have listed the collar models tested.
Only 26% of Willson et al.’s participating cat owners said they would buy and use cover collars in the future. Their reasons were break-away losses, cat comfort, and owner dislike of the product. It is good to note that Birdsbesafe developed a new break-away collar from stretchy nylon fabric in order to lessen the break-away issue.
Willson et al. focus on bird kill reductions and discuss mammal kill reductions but little. This is perplexing, since mammals are as ecologically valuable as birds and cats evidently kill more mammals than birds. It is obvious to conservation biologists that it is at least as important to protect mammals as birds. True, Birdsbesafe collar covers are only claimed to reduce bird predation, but the correct goal is to protect all wildlife from free-roaming cats, so a birds-only mitigation device is woefully inadequate. It begs the question of why authors expended resources on a product with such limited value.
In summary, Willson et al.’s report is flawed in terms of (1) its reason for being, which was to assuage guilt, (2) non-standard and misleading methods of reporting kill reductions, (3) failure to admit to probably low feral cat use rates, (4) failure to recognize potential cruelty to feral cats from dirty, stiff collars, (5) failure to compensate for uncounted kills in efficacy rate estimates, (6) failure to justify projecting results from two seasons out over the entire year, (7) inadequate reporting of raw data, (8) failure to justify using a product designed for bird protection even though it is evidently unsuitable for other taxonomic groups such as herps and mammals, (9) failure to list the tested collar models despite very different loss rates, and (10) failure to admit that low owner acceptance means that collar covers can mitigate only for relatively small numbers of prey items. The cat kill reduction values reported by Willson et al. and highlighted by Birdsbesafe are therefore unwarranted at best and deliberately misleading at worst.