Are Nest Predation and Brood Parasitism Correlated in Yellow Warblers? A Test of the Cowbird Predation Hypothesis

Brown-headed Cowbirds (Molothrus ater) are generalist brood parasites that reduce the productivity of many of their passerine hosts through egg removal and competition between cowbird and host nestlings (Lowther 1993, Payne 1998, Lorenzana and Sealy 1999). Recently, it has been suggested that cowbirds exert a previously unappreciated influence on the demography of their hosts through predation (Arcese et al. 1992, 1996; Smith and Arcese 1994; Arcese and Smith 1999). The cowbird predation hypothesis (Arcese et al. 1992) was suggested to explain the frequently observed link between nest predation and interspecific brood parasitism in many passerines (e.g. Wolf 1987, Payne and Payne 1998). Specifically, the hypothesized link is a direct result of cowbird predation of host nests that are discovered too late in the host's nesting cycle for successful parasitism. This creates future opportunities for parasitism by causing hosts to renest (Arcese et al. 1996). Alternatively, a link between predation and parasitism may result if cowbird parasitism facilitates predation, or if both factors are correlated with a third variable such as nest vulnerability (Arcese and Smith 1999). Determining the independent effects of brood parasitism and nest predation on host populations is difficult because of this potential functional dependence (Pease and Grzybowski 1995, Arcese and Smith 1999). In a resident population of Song Sparrows (Melospiza melodia) on Mandarte Island, British Columbia, the frequency of nest failure was positively related to the frequency of cowbird parasitism (Arcese et al. 1996). In 5 of 19 years when cowbirds were rare, absent, or removed on the island, nest failure was lower than when one or more cowbirds were present on a consistent basis. To explain these trends, Arcese et al. (1992) proposed the cowbird predation hypothesis, which they subsequently tested directly (Arcese et al. 1996). A critical prediction of the hypothesis was supported when Arcese et al. (1996) found that parasitized nests failed less often than unparasitized nests when female cowbirds maintained exclusive territories, because cowbirds should regularly depredate unparasitized nests but not nests they have already parasitized. To assess whether results from this island population were unique, J. N. M. Smith and M. J. Taitt (unpubl. data) and Rogers et al. (1997) analyzed comparable data from Song Sparrows on Westham Island, near mainland British Columbia. In both studies, lower cowbird numbers were associated with higher host nesting success, as predicted. However, Rogers et al. (1997) found no difference in the frequency of nest failure in parasitized and unparasitized nests, contrary to the key prediction of the hypothesis. Published analyses of potential correlates of cowbird predation behavior are limited to these populations of Song Sparrows. Arcese et al. (1996) cited several anecdotal accounts of cowbirds destroying passerine nests and also called for additional independent tests of the cowbird predation hypothesis (see also Arcese and Smith 1999). We compared the frequency of failure of parasitized versus unparasitized nests in a population of Yellow Warblers (Dendroica petechia) in the Delta Marsh region of southern Manitoba, using data collected from 1974 to 1976. The cowbird predation hypothesis predicts that unparasitized nests should fail more often than parasitized nests when cowbird egg-laying ranges do not overlap. In contrast, parasitized nests may fail more often than unparasitized nests if cowbird egg-laying ranges overlap, or if cowbirds facilitate rather than cause predation. We also examined the evidence in support of the prediction that unparasitized hosts whose nests are depredated should have a high probability of their replacement nests being parasitized. We further tested the general hypothesis that nest predation is linked to brood parasitism, either directly or through facilitation, by examining the relationship between frequency of nest predation and parasitism in seven Yellow Warbler populations across North America (cf. Arcese and Smith 1999). Yellow Warblers are a common host of the cowbird at Delta Marsh (19.1% of 2,163 nests parasitized; Neudorf and Sealy 1994) and elsewhere. They typically accept cowbird eggs laid after the midpoint of the laying stage but respond to eggs laid earlier either by burial of the cowbird egg (and any host eggs) under a new nest, or by desertion followed by renesting at a different site (Clark and Robertson 1981, Sealy 1995). During 1974 to 1976, Yellow Warblers at Delta Marsh suffered high cowbird parasitism and high predation. Further details of the study site, species, and methods used to find and follow nests are given in Goossen (1978) and Sealy (1995). We examined data for 331 Yellow Warbler nesting attempts. Parasitism frequency (proportion of all nests that received one or more cowbird eggs) was 26%. We calculated the frequency of failure of parasitized and unparasitized nests in the subset of nests surviving at least to the second day of incubation to eliminate the bias of unequal exposure to parasitism. Nest-failure frequency is defined as the proportion of all nests that fledged no young. A parasitized nest in which the cowbird egg (and Yellow Warbler eggs, if present) was buried and a new clutch was laid was considered a single parasitized nesting attempt. Where possible, we identified cause of nest failure for each unsuccessful nesting attempt. For example, weather or predation could be implicated confidently in many cases. Only nests that clearly failed as a result of loss of most or all of the eggs or nestlings were considered to have been depredated. Nesting stage at which failure occurred was assigned on the basis of laying date of the first egg for nests found during building or laying or by backdating from the first day of incubation (i.e. the day after the last warbler egg is laid) or from the date of hatching for nests found later. For nests that were partially depredated over several days, the stage at the first day of egg or nestling loss was considered the stage of predation. Common nest predators at Delta Marsh include red squirrels (Tamiasciurus hudsonicus), Common Grackles (Quiscalus quiscala), Yellow-headed Blackbirds (Xanthocephalus xanthocephalus), Red-winged Blackbirds (Agelaius phoeniceus), Baltimore Orioles (Icterus galbula), and Gray Catbirds (Dumetella carolinensis; see Sealy 1994). We used corrected χ2 tests to compare the frequency of failure of parasitized and unparasitized nests, combining data for all years. We also performed an additional test confined to nests for which failure could be confidently attributed to predation. We examined the relationship between parasitism and predation in seven populations of Yellow Warblers from Ontario (S. Yezerinac unpubl. data), Michigan (Batts 1958, DellaSala 1985), Colorado (C. Ortega and J. Ortega unpubl. data, Howe and Knopf 2000), Montana (J. Tewksbury unpubl. data) and Manitoba (this study). We used Spearman rank correlation, weighted by sample size, to determine whether the percentage of nests depredated was related to the percentage of nests parasitized. The frequencies of failure of parasitized and unparasitized nests were opposite to predictions of the cowbird predation hypothesis: parasitized nests failed more often (52% of 64 nests) than unparasitized nests (37% of 214 nests; χ2 = 3.81, P = 0.05). To assess whether predation and parasitism are linked, we compared the frequency of failure of parasitized and unparasitized nests for failures that we could confidently attribute to predation. Again, the trend was opposite to predictions, although the difference was not statistically significant (40% of 52 parasitized nests vs. 30% of 192 unparasitized nests depredated; χ2 = 1.69, P = 0.19). These results did not support the cowbird predation hypothesis. In contrast to et al.'s (1996) results, unparasitized nests were less likely to fail than were parasitized nests. This result was not unexpected because Yellow Warblers frequently respond to parasitism by nest desertion (Clark and Robertson 1981, Burgham and Picman 1989, Sealy 1995). However, because we considered only nests that failed after the window for parasitism (after the second day of incubation), any nests deserted in response to parasitism within this window were not included. Similarly, Payne and Payne (1998) found that parasitized Indigo Bunting (Passerina cyanea) nests failed more often than unparasitized nests. This result may occur if cowbirds facilitate predation by their activities near host nests (Arcese and Smith 1999). However, because the frequency of failure was not significantly higher in parasitized nests, we have no evidence to support the cowbird facilitation hypothesis. Identification of nest predators is crucial to understanding the association between nest failure and brood parasitism. Field signs after nest predation only occasionally are useful in identifying nest predators (Arcese and Smith 1999, Thompson et al. 1999); hence, we have no clear record of whether cowbirds or some other predator were responsible for nesting failures. However, many nest predators exist in the study area, and of several observed predation events at Yellow Warbler nests, none involved cowbirds (Sealy 1994). In a similar habitat, Thompson et al. (1999) documented only one case out of 25 predation events recorded on video of nest predation by a cowbird. We conclude that cowbirds probably were not responsible for many of the predation events at the Yellow Warbler nests in our sample. The critical prediction of the cowbird predation hypothesis is that parasitized nests should fail less often than unparasitized nests, but this prediction would hold only when individual female cowbirds have distinct egg-laying ranges, because a female should destroy unparasitized nests but not nests she has parasitized (Arcese et al. 1996). When female cowbirds defend exclusive intraspecific nesting territories, any parasitized nest a female encounters is likely to contain her own egg. On the other hand, if cowbird egg-laying ranges overlap, different females would be likely to discover the same nests if some nests are especially conspicuous to cowbirds in general, and subsequent females may depredate a parasitized nest. Therefore, parasitized nests should fail more often than unparasitized nests (Arcese et al. 1996). One could argue that the trend we documented resulted from overlap of egg-laying ranges among female cowbirds. However, recent DNA work has shown that egg-laying ranges of female cowbirds do not overlap at Delta Marsh (Alderson et al. 1999), and we have no reason to believe that this was not the case 25 years ago. Indeed, egg-laying ranges likely overlapped less often then, because Yellow Warbler nesting density was about twice what it is at present, but cowbird densities were similar (S. Sealy unpubl. data). If host nests are abundant, female cowbirds should not need to search wide areas for host nests. The low frequency of multiple parasitism in the Yellow Warbler data set (8 of 85 parasitized nests received more than one cowbird egg) also supports the claim that egg-laying ranges overlapped little. Yellow Warblers frequently respond to parasitism by deserting (burying) the cowbird egg. In the present sample, burial occurred at 39% of parasitized nests. A nest at which burial has occurred and a new clutch is laid may suffer increased exposure to predation compared with a nest that proceeds directly to incubation. This difference would bias our results against the prediction of higher frequency of failure of unparasitized nests. When we excluded nests at which burial had occurred, our results were unchanged. In addition, the frequency of nesting failure did not differ between parasitized nests in which cowbird eggs were buried versus accepted (χ2 = 3.02, P = 0.08), and the trend was opposite to expectation because “burial” nests failed less often (36% of 25 nests) than “acceptance” nests (62% of 39 nests). Predation of a parasitized nest in which the cowbird egg is buried is predicted from the cowbird predation hypothesis only if cowbird egg-laying ranges overlap. Otherwise, a female probably remembers the nests she has parasitized (Sherry et al. 1993), and predation of these nests would be a form of the “Mafia” behavior described in parasitic cuckoos (Zahavi 1979, Soler et al. 1995). If Mafia-like behavior occurs, failures due to predation of parasitized Yellow Warbler nests should occur more often when egg rejection (i.e. burial) has taken place than when it has not. This was not the case in our study. Another prediction of the cowbird predation hypothesis is that unparasitized hosts whose nests are depredated should have a high probability of their replacement nest being parasitized. We could not test this prediction because the identities of Yellow Warblers that renested were not known. However, in five cases where a first nest was depredated (two during laying, three during incubation), the same nests were reused for the new nesting attempts, presumably by the same adults. None of these replacement clutches was parasitized. We present two explanations for our failure to support predictions of the cowbird predation hypothesis. First, host density is high at Delta Marsh, and opportunities for parasitism should not be limiting. In contrast, this may not be the case on Mandarte Island, where few alternate host species occurred. However, Smith and Taitt (unpubl. data) documented further support for the cowbird predation hypothesis in Song Sparrows on Westham Island, British Columbia, where alternate hosts were present. Despite this range of opportunity, alternate hosts are rarely used on Westham Island (J. N. M. Smith pers. comm.). Second, the timing of breeding of alternate host species is less synchronous at Delta Marsh than on Mandarte Island, and the presence of hosts late in the breeding season means that opportunities for parasitism are greater there. On Mandarte Island, cowbirds sometimes continue to lay until the end of the Song Sparrow laying season (Smith and Arcese 1994), but after this date, they are rarely seen on the island. At Delta Marsh, however, several other frequently parasitized hosts continue to breed later in the season, including Song Sparrows that lay second clutches. Both of the explanations noted above presuppose that nest predation confers a cost for cowbirds. If a cowbird incurs no cost by depredating a host nest, nests discovered too late for parasitism should always be destroyed regardless of the number of opportunities for parasitism. Surprisingly, we found that across seven populations of Yellow Warblers, predation and parastism were significantly negatively correlated (rs = −0.85, P = 0.008; Fig. 1), in contrast to what has typically been documented (e.g. Johnson and Temple 1990, Donovan et al 1997, Arcese and Smith 1999). We suggest that the frequently observed link between predation and parasitism cannot be explained by the cowbird predation hypothesis or the cowbird facilitation hypothesis, but instead represents a coincidental preference by nest predators and brood parasites for particular habitat features (Donovan et al. 1997, Tewksbury et al 1998). The only way to separate the independent influence of cowbirds as predators is to compare predation frequencies in populations where cowbirds are naturally or experimentally variable in abundance (Arcese and Smith 1999). In Song Sparrows, this test suggests that changes in the frequency of parasitism are directly linked to changes in the frequency of nest predation (Arcese and Smith 1999). Further experiments on other locally abundant hosts, such as Yellow Warblers, are needed. Percent of nests depredated versus percent parasitized in seven populations of Yellow Warblers, weighted by the number of nests studied. Size of open circles indicates sample sizes for each population: Ontario (ON, n = 74; S. Yezerinac unpubl. data); Manitoba (MB, n = 331; this study); northern Colorado (nCO, n = 73; Howe and Knopf 2000); southwestern Colorado (swCO, n = 93; C. Ortega and J. Ortega unpubl. data); southern Michigan (sMI, n = 20; Batts 1958); Montana (MO, n = 358; J. Tewksbury unpubl. data); southeastern Michigan (seMI, n = 25; DellaSala 1985) We thank J. P. Goossen for collecting the original data and the staff of the University of Manitoba Field Station at Delta Marsh for logistical support. The comments of P. Arcese, A. L. A. Middleton, S. I. Rothstein, J. N. M. Smith, J. J. Tewksbury, and an anonymous reviewer greatly improved the manuscript. We also thank C. P. Ortega, J. N. M. Smith, J. J. Tewksbury, and S. Yezerinac for kindly supplying unpublished data. Funds were provided by a research grant to SGS from the Natural Sciences and Engineering Research Council of Canada.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.