522 
THE WrLSON JOURNAL OF ORNITHOLOGY • Vol 124. No. 3. September 2012 
treatments. We used n. < 0.05 for all statistical 
analyses, and conducted all analyses in R 2.10.1 
(R Development Core Team 2010). 
RESULTS 
One hundred and four (58%) of the 180 eggs 
were depredated. Sixty-seven of 120 (56%) 
imitation eggs were depredated, compared to 37 
of 60 (62%) real eggs. Predation rates were not 
different between real and imitation eggs (W = 
4,000; n = 180. P ~ 0.2). Each block contained 
60 eggs of which 20 (4 of each treatment group) 
were real and 40 (8 ol each treatment group) were 
imitation eggs. Fifty eggs (83%) in block one, 31 
(52%) eggs in block two, and 23 (38%) eggs in 
block three were depredated. 
Predator type could only be ascertained for 
imitation eggs. Forty (59%) of the 67 depredated 
imitation eggs had rat tooth-marks, II (16%) had 
avian peck-marks, four (6%) were dragged from the 
nest but left unmarked, and 12 (18%) were broken 
from the end of their strings and lost from the nest, 
preventing predator identification. One egg had 
both rat and mouse tooth-marks, and this was the 
only egg with non-rat mammalian tooth marks. 
Adhesive paprika-treated (z = 2.458, n - 180. 
P = 0.013) and untreated (z - 2.586. n = 180, P = 
0.009) nests had higher hazard rates than adhesive 
capsaicin-treated nests in the all-comparisons test. 
Hazard rales were higher for the first block than the 
second (z = 4.255, n = 180. P < 0.001) or third 
(z - 5.907. n = 180. P < 0.001) block. Hazard 
rates were higher in the second block than the third, 
but this contrast was not in itself significant (z = 
1.652. u ~ 180. P — 0.099). There were no other 
significant contrasts in this test. 
The hazard rates for eggs treated with adhesive 
capsaicin were significantly lower than for eggs 
pooled across all other treatment groups (z = 
2.409. n = 180, P = 0.016. Fig. 2) after 
considering the effect of block and egg type. 
Adhesive-capsaicin treated nests had greater mean 
survival times than other nest treatment types 
using the robust, but conservative, Wilcoxon 
signed-rank test (W = 201. n = 60, P = 0.05). 
DISCUSSION 
Adhesively-applied chili powder significantly 
reduced the rate of egg predation in imitation 
nests constructed to resemble the open-cup nests 
oi introduced thrushes in New Zealand, relative to 
untreated eggs, eggs treated with non-adhesively 
applied chili and paprika, or eggs treated with 
adhesively-applied paprika. However, all artificial 
nests had markedly-reduced predation when these 
treatments were repealed several times in a small 
area. It seems unlikely that predation likelihood 
was independent for all eggs in a nest, but it . 
not possible to correct for nest effects wiihm ih: 
Kaplan-Meier analysis, as they were confounded 
by treatmem effects. The statistical pattern' 
observed in Kaplan-Meier analysis were aho 
observed in the highly robust, although conser.a- 
tive, Wilcoxon signed-rank test. The interpreta¬ 
tion of the Kaplan-Meier analysis support' our 
main conclusions, although the significance 
values associated with specific treatment contrasts 
are probably slightly under-estimated. 
Rats had been considered the most likely 
predator of eggs in suburban New Zealand pnor 
to this study (Brown 1997. Flux and Bradfieid 
2006) and tooth marks on imitation eggs in this 
study matched tooth marks obtained from museum- 
specimen rat jaws in size and shape. Rats have 
extensive home ranges, which often exceed Id 0 *m 
linear dimensions and cover a variety of habitats 
(Dowding and Murphy 1994, Hooker and lanes 
1995, Pryde et a), 2005). Thus, it is possible that one 
or more chili-treated nests were within a randomly 
selected rat home range, especially as the path 
through ARBG doubled back on itself in several 
locations. One consequence is that rats in the study 
area would have had a good opportunity to learn the 
majority of eggs in imitation nests were typically 
indigestible and/or at times highly unpleasant to 
bite into during the first block treatment. 
The diet of rats in New Zealand also vanes 
seasonally (Miller and Miller 1995), and our 
experimental blocks are confounded by season./ 
effects. Eggs in imitation nests were depredated at 
a much lower rate in later blocks regardless or 
treatment type; thus, it would be valuable to 
investigate whether rats learn to avoid nests 
treated with capsaicin and. if so, how they learn 
to recognize nest types, to assess whether thb 
behavior can be modified to benefit native an. 
endemic bird species. 
The physiological side effect of applying m' 
adhesive based nest-predation deterrent to die 
surface ot live incubated, developing eggs was no 
investigated in this study. Avian eggshells arc 
known to be selectively permeable, allowing the 
developing embryo to respire (Burton and Tullett 
1983). Thus, adhesive paralleling of complementing 
the egg's permeability characteristics would need to 
he found (or developed) or the adhesive would need 
