The Condor96~761-768 0 The CooperOrnithological Society 1994

PREDATION ON NORTHERN CARDINAL NESTS: DOES CHOICE OF NEST SITE MATTER? ’

TAMATHA S. FILLIATER,~ RANDALL BREITWISCH AND PAUL M. NEALEN~ Department of Biology, Universityof Dayton, Dayton, OH 45469-2320

Abstract. The fatesof 121 nestsof NorthernCardinals (Cardinalis cardinalis) in south- westernOhio weredetermined in 1991 and 1992.Success rate wasonly 1So/o, estimated by the Mayfield method.All failed nestswere known or suspectedto haveheen preyed upon. Severalhypotheses to explaindifferences between the locationof successfuland failednests weretested. None of thoseconsidered explained why the contentsof particularnests were taken. We proposethat a high incidenceof predationby a rich guild of nest predators precludesthe existenceof predictablysafe nest sites for cardinals.Instead, cardinals appear simply to be well-adaptedto renestrapidly in responseto the near randomnessof nest predation.A similar adaptationmay characterizemany passerines. Key words: Northern Cardinal: Cardinaliscardinalis; nesting success; nest site selection; nestpredation.

INTRODUCTION Bent 1968, Rinser 1973) and builds a new nest Selection of nest site may be of crucial impor- in a different location on the territory for each tance to the reproductive successof birds. Lack nesting attempt. Females also appear to be the (1954) estimated that 75% of all eggsand nest- sex that selectsnest sites (Rinser 1973; however, lings lost from open cup nestsare taken by pred- see Laskey 1944). ators. Ricklefs (1969) estimated predation per- Every passerine studied to date displays some centagesin six passerine speciesand concluded intrapopulational variation in nest site selection. that predation accounted for 55% of egg losses Some studies have revealed patterns in this vari- and 66% of nestling losses.Many aspectsof the ation associatedwith the probability of success nesting behavior of birds appear to be adapta- (see Martin and Roper 1988) while others have tions to avoid predation of the nest contents. not (e.g., Best 1978, Barnard and Markus 1990, Many birds hide their nests or build them in Morton et al. 1993; see also Gottfried and inaccessible sites (Collias and Collias 1984). In Thompson 1978 for an experimental study). We addition, the behavior of parent birds visiting attempted to test predictions of eight hypotheses nestsis notably stealthy (Skutch 1976, Breitwisch for the placement of successful versus unsuc- et al. 1989). Active defenseagainst predatorsalso cessfulnests by cardinals. These hypothesesad- occurs in some speciesof birds, but defense en- dress lossesto predators only; lossesdue to abi- tails a risk of injury to or death of the parent. otic environmental factors are not considered Clearly, birds may be advantaged by selecting here. In addition, our assumption is that pred- sites where nests will be unlikely to be found by ators find nests by actively searching for them predators. rather than by simply watching parent birds at- Northern Cardinals (Cardinalis cardinalis) are tend to the needs of eggsor nestlings (Collias and multibrooded, and females lay up to five clutches Collias 1984, Martin 1988a). in a season,of which as many as four have been The “nest concealment” hypothesis predicts recorded as successfully fledging (Shaver and that nests that are concealed regardless of the Roberts 1930, Laskey 1944). The female usually particular plant structure will be more successful builds the entire nest (Laskey 1944, Bums 1963, than those that are not so concealed. The “nest inaccessibility” hypothesispredicts that neststhat are lessaccessible (but not necessarilybetter con- ’ ’ Received27 December1993. Accepted 29 March cealed) will be more successfulthan those that 1994. are more accessible. Inaccessible nests in this 2 Presentaddress: U.S. ForestService, Klamath Na- tional Forest,37805 Highway 97, Macdoel,CA 96058. study are defined as those over water, in thorny 3 Presentaddress: Department of Biology,Univer- vegetation,among cane (Arundinaria sp., a woody sity of Pennsylvania,Philadelphia, PA 19104-60 18. herb), or at the end of thin branches or twigs (cf.

[7611 162 T. S. FILLIATER, R. BREITWISCH AND P. M. NEALEN

Collias and Collias 1984). The “nest height” hy- form of predation) increasesfrom forest interior pothesis predicts that nests placed high or low to forest edge (Brittingham and Temple 1983). relative to the frequency distribution of nest The “distance to human activity” hypothesis heights in the population will be more successful predicts that nests placed closer to areas of hu- than those of more typical height. The first pre- man activity will be more successfulthan those diction (high nests are more successful)assumes that are farther away (Collias and Collias 1984). that nests built high in plant crowns are con- Human activity may discouragepredators from cealed better than nests built lower, especially as searching near heavily used trails and in other the breeding season progressesand plants leaf areas frequented by people. We added this hy- out (Nolan 1978) and, in this sense,is a version pothesisand tested it in late 199 1 and 1992 after of the nest concealment hypothesis. Such nests finding that successfulnests tended to be closer may also be found less frequently by or be less to areasof human activity than unsuccessfulnests accessibleto terrestrial predatorsthan lower nests. earlier in 199 1. The second prediction (low nests are more suc- cessful)assumes that aerial predators are of ma- jor importance and are less successfulin either STUDY SITE AND METHODS finding or accessinglow nests.The “mid-height” This study was conducted at the Aullwood Au- hypothesis predicts that nests placed at mid- dubon Center and Farm, located about 15 km height in a tree or shrub will be more successful northwest of Dayton, Ohio (39”52’N and than those placed either higher or lower in the 84”16’W) from April to August 1991 and 1992. plant. This hypothesis assumes that nests built The Aullwood property is an 80 ha sanctuary higher are more easily seen by aerial predators, with habitats in the following proportions: 26% while those built closer to the ground are located croplands and orchards, 17% mature woodlands more easily by mammals and snakes(see Alonso (dominated by beech, Fagus spp.; oaks, Quercus et al. 199 1). The “needle in a haystack” hypoth- spp.; maple, Acer spp.; and buckeye, Aesculus esis predicts that nests placed in a common spe- glabra), 17% secondary growth (e.g., ash, Frax- cies of plant will be more successfulthan those inus spp.; and maple), 16% pasture, 8% meadow, that are placed in uncommon plant species(see 5% prairie, 5% residential area, 2% pine plant- Martin and Roper 1988). If a predator restricts ings, 2% wetlands and 2% wet woods (e.g., ash its search for nests to a few appropriate plant and red maple, Acer rubrum) (J. Ritzenthaler, species, then nests in the most common plant pers. comm.); see Filliater-Lee (1992) for further specieswill be more difficult to find becausethere details on vegetation. Cardinals frequent nearly are more individual plants to search. The “rare all habitats on the property and most densely site” hypothesispredicts that nests placed in un- populate secondarygrowth and edgesof wooded commonly used plant specieswill be more suc- areas. cessful than those placed in commonly used We discovered nests by intensively searching plants. If a predator searchesplant speciescom- at least every other day and by following the monly used for nesting, then it is best to be atyp- “chipping” vocalizations of adult cardinals ical and nest elsewhere. This hypothesis differs (Lemon 1968, Montgomerie and Weatherhead from the others in that successdepends on where 1988). Nearly all nests were observed daily to other members of the population are nesting (as record progress. Many nests were observed proposed among species for communities of through 10 x 40 binoculars at a distance of 5- nesting birds by Martin 1988b). The “edge dis- 15 m. We recorded nest height and tree or shrub tance” hypothesispredicts that nestsfarther from height with a meter stick to 0.1 m where possible a habitat edge (defined here as where a closed and estimated where necessary (nests in dense habitat changesobviously to another more open multiflora rose [Rosa multiflora] or higher than habitat, e.g., forest to field) will be more suc- several meters). In addition, in late 199 1 and in cessful than those placed closer to an edge. The 1992, we estimated distance to the closest area distance to habitat edge is important because ofhuman activity to 0.5 m. Human activity areas some mammalian and avian predators actively were defined as trails (most are heavily used) and search near these edges(Gates and Gysel 1978). other areasfrequented by people. Although some Furthermore, Brown-headed Cowbird (Molo- clearings were foci of human activity, distance thrus ater) parasitism (which can be viewed as a to the nearestclearing for any particular nest may CARDINAL NEST SITES AND NEST PREDATION 763 have been different from that to the closestarea central tendencies. Kolmogorov-Smirnov of human activity. 2-sample tests (D,,, value reported) were used We recorded nest visibility from six vantage to compare shapes of frequency distributions points (above, below, and from four horizontal (Siegel and Castellan 198 1). Spearman rank cor- directions [N, S, E, W]) at a distance of l-2 m relations (r, value reported) were used for tests using the following classification scheme: “visi- of monotonic relationship. Results are reported ble” (visible from five or six vantage points), as significant if they are associatedwith an alpha “ambiguous” (visible from three or four), and value ofP < 0.05. All testswere two-tailed. Means “not visible” (visible from none to two). Al- f SD are reported for descriptive statistics. though this division is somewhat subjective, two We also conducted a seriesof eight parametric observers agreed on the visibility of each nest. discriminant function analyses(SAS 1989) of nest The details of placement of each nest in veg- site variables to compare successfulwith failed etation were recorded in order to address the nests. This series included analyses with as few concealment and inaccessibility hypotheses. In as two independent variables and as many as addition to the direct measurements of visibility, seven. The discriminant function analysesdo not we recorded if the nest was built < 10 cm below directly addressthe predictions of the individual leaves, very effectively concealing it from above hypothesesbut rather search for a combination (a nest not so located could still be “not visible” of variables that is associated with successful from above at a distance of l-2 m). As men- nesting. tioned in the introduction, a nest was also clas- Data from the two years of the study were sified in terms of accessibility. pooled for analysis after testing for differencesin We recorded all measurements only after each nest variable distribution between years. Except nest had failed or succeeded,in order to avoid where noted, no differenceswere found between possible disturbance. We recorded the date of years. failure or successas the first day the nest was found to be inactive, i.e., when eggsor nestlings disappeared due to predation or when fledglings RESULTS were found in the vicinity of the nest. Nests dis- GENERAL covered inactive after a two day or rarely three or more day gap in observations were considered We found 121 active cardinal nests, 43 in 199 1 inactive at the midpoint of the hiatus. We con- and 78 in 1992. Cardinals nested in 22 species sidered any nest fledging at least one young as of plants. Seventy-nine of the 121 nests (65%) successful.All nest failure appeared to be due to were built in the dense shrubs multiflora rose predation rather than to any other cause, such and honeysuckle (Lonicera spp.). Eight of the as violent weather (one nest failed due to uncer- remaining nests (7% of total) were built in red tain cause). cedar (Juniperus virginiana). Only 30 of the 12 1 Nests were discovered at various stagesin the nests (25%) were successful. nesting cycle, and the majority were already at the egg or nestling stage. Such biased discovery can lead to an overestimate of the true nesting SURVIVAL RATE successin a population. Therefore, we calculated The mortality rate (Mayfield 1975) for the nest- total nest-days of observation for three stages: building/egg-laying period was 0.074 failures per (1) nest-building and egg-laying, (2) egg incuba- nest-day (14 failures/l88.5 nest-days), and the tion, and (3) nestling periods, and used May- survival rate for this 5-day period was 0.68 (=[ 1 field’s (1975) calculation for success. - 0.07415). The mortality rate for the egg incu- All univariate statistical analyses employed bation period (=14 days) was 0.065 failures per non-parametric tests.Tests of independence (Gadj nest-day (42 failures/644.5 nest-days), and the value reported) were used to ask if associations survival rate for this period was 0.39. The mor- existed between pairs of variables; all were cor- tality rate for the nestling period was 0.054 fail- rected due to small sample sizes by using Wil- ures per nest-day (17 failures/3 13 nest-days),and liams’ correction (Sokal and Rohlf 198 1). Mann- the survival rate for this period (= 10 days) was Whitney U Tests (z value reported) and Median 0.57. In this population, the overall successrate Tests (x2 value reported) were used to compare was 15%(0.68 x 0.39 x 0.57 x 100%). 764 T. S. FILLIATER, R. BREITWISCH AND P. M. NEALEN

TIME OF SEASON 1992(G, = 12.18, df = 1, P -c 0.05) and a Before testing hypothesesfor nest successversus higher proportion of nests with leaf cover were failure, we examined the relationship between successfulin 1991 than in 1992 (G,, = 4.82, df probability of successand date in the season.We = 1, P < 0.05). Yet, there was no difference found the first nest of each seasonat the end of betweenyears in the proportion of successfulnests April and the last nest each seasonin the middle for those without leaf cover (G, = 0.11, df = 1, of August. The temporal distributions of nests P > 0.05). Similarly, there was no difference in monitored in the two years had similar means the probability of successwhether nestswere vis- (z = 0.88, P > 0.05). Overall, we monitored six ible or not from above (G, = 0.017, df = 1, P nests in April (none of which fledged), 54 nests > 0.05) or visible or not from below (Gad,= 0.44, in May (eight fledged), 34 nests in June (12 df = 1, P > 0.05). fledged),24 nestsin July (eight fledged),and three There was no correlation between visibility and nests in August (two fledged). We compared date in the season(r, = 0.13, n = 121, P > 0.05) “early” and “late” nests,conducting Median tests and no association between leaf cover and date for seasonalchanges where correlations with date in the season (G,,, = 0.10, df = 1, P > 0.05). of season were inappropriate. Nests “early” in There was also no differencein successfulvs failed the season were built in April-May, and nests nests in the distance from the nest to the nearest “late” in the seasonin June-August. The prob- outer edge of the plant’s foliage (z = 1.39, P > ability of successwas greater late in the season 0.05). The mean distance for all nests was 0.47 (x2 = 7.21, df = 1, P < 0.05). Similarly, there f 0.48 m. However, cardinals built deeper in was a trend between month of the breeding sea- vegetation in 1992 than in 1991 (z = 3.11, P < son and probability of nest success(rs = 0.92, n 0.05). = 5, P -c 0.05) although the sample sizes for NEST HEIGHT HYPOTHESIS April and August were small. We tested for an association between nest height and successby both a test of central tendency NEST CONCEALMENT HYPOTHESIS and a distribution test. The mean nest height was We classified 55 nests (45%) as visible, 3 1 (26%) 2.1 + 1.6 m (range: 0.7-12 m). The mean nest as not visible, and 35 (29%) as ambiguous re- height for successfulnests was 2.7 + 2.6 m and garding visibility. We also classified 40 nests for failed nests I .9 f 1.O m (z = 0.47, P > 0.05). (33%) as located < 10 cm below leaves and thus There was also no difference in the shapesof the not visible from above. We ranked nest con- frequency distributions (D,,, = 0.156, P > 0.05). cealment in six categoriesfrom most to least vis- There was a correlation between the height of ible: (1) nest visible and not located below leaves, the nest and date in the season in both years (rs (2) nest visible but below leaves, (3) nest ambig- = 0.26, n = 12 1, P < 0.05). Cardinals built nests uous and not below leaves, (4) nest ambiguous higher as the seasonprogressed. and below leaves, (5) nest not visible but not below leaves, and (6) nest not visible and below MID-HEIGHT HYPOTHESIS leaves. The height of each nest in the plant divided by We found no correlation between nest con- the height of the plant gave the relative height cealment and the proportion of successfulnests of each nest in the tree or shrub. The mean height (rs = -0.14, n = 6, P > 0.05). Twelve of 45 of nests was 2.1 m (see above), and the mean (27%) category 1 nests, four of 10 (40%) category height of trees and shrubs in which nests were 2 nests, two of 18 (11%) category 3 nests, four built was 3.8 + 2.9 m (range: 1.1-20 m). The of 17 (24%) category 4 nests, four of 18 (22%) mean relative height of nests was 0.60 + 0.20 category 5 nests, and four of 13 (31%) category (range: 0.13-0.96). Successfulnests were located 6 nests were successful.Furthermore, there was at similar relative heights as failed nests (z = no association between “visibility” regardlessof 0.92, P > 0.05). Similarly, there was no differ- leaf cover and probability of success(G,, = 1.69, ence in the shapesof the frequency distributions df = 2, P > 0.05) or between leaf cover regardless for relative heights of successfulvs. failed nests of visibility and probability of success(G,, = (D,,, = 0.136, P > 0.05). 0.83, df = 1, P > 0.05). However, a higher pro- We tested for an association between date in portion of nests had leaf cover in 199 1 than in the season and the location of the nest in the CARDINAL NEST SITES AND NEST PREDATION 165 plant. Cardinals built nests relatively higher in mammals, and predatory birds (Kinser 1973), plants as the season progressed(rs = 0.42, n = but we do not know whether these predators re- 121, P < 0.05). We tested for a difference in the strict their searchto a few, commonly used plant proportionate height of successfulvs. failed nests species.Further, we did not map the vegetation for both early and late nests. There was no dif- of the Aullwood property and do not know rel- ference for either early nests(z = 1.48, P > 0.05) ative abundances of plant species. or late nests (z = 0.71, P > 0.05). However, the three most common specieson Second, we found a correlation between pro- the study site are multiflora rose, honeysuckle, portionate height and visibility for each half of and ash (J. Ritzenthaler, pers. comm.), and the the season and for the entire season (r, = 0.35, first two were frequently chosen as nest sites by 12= 12 1, P < 0.05). Relatively higher nestswere cardinals. In both 1991 and 1992, 65% of the less visible than lower nests. nests were built in these two species.We tested Third, we tested for an association between for an association between plant species and proportionate height and leaf cover for early and probability of nest successby dividing all nests late nests.Nests located high in a plant were more into two categories: those in the two common likely to be covered by leaves if they were built shrubs and those in all other species.The prob- early in the season(Gad, = 9.22, df = 1, P < 0.05), ability of successwas not associatedwith species but this was not so for nests built later in the category; 22% of nests in the common shrubs season(G,, = 2.43, df = 1, P > 0.05). were successfulvs. 3 1% of those in other species (G, = 1.25, df = 1, P > 0.05). EDGE DISTANCE HYPOTHESIS There was no difference in relative use of com- There was no difference in the distance of suc- mon speciesfrom early to late in the season(Gadj cessful vs failed nests from the nearest edge of = 0.006, df = 1, P > 0.05). Forty (51%) of 79 open habitat. Successfulnests were a mean 3.4 nestsin common speciesand 19 (5 1%) of 37 nests * 4.2 from habitat edge, and failed nestsa mean in other species were constructed early in the 4.6 -t 6.2 m (z = 0.82, P > 0.05). Nests in 1991 season. were a mean 2.0 + 2.2 m from habitat edge, and those in 1992 were 5.6 f 6.6 m (z = 4.28, P < DISTANCE TO HUMAN ACTIVITY HYPOTHESIS 0.05). In neither year was the probability of suc- cessrelated to the distance from the nest to hab- Human activity in the study area is restricted to itat edge. particular sites,in addition to heavily-used paths. The mean distance of successfulnests from hu- NEST INACCESSIBILITY HYPOTHESIS man activity areas was 6.2 + 7.3 m (range: O- Similar proportions of inaccessible nests (23%) 25 m, IZ = 23), and of failed nests was 8.7 -t 9.9 and accessiblenests (26%) were successful.There m (range: 0.4-50 m; IZ= 73) (z = 1.35, P > 0.05). was no association between accessibility of the There was no correlation between distance of a nest and the probability of successin either year nest to human activity and date in the season(r, or for the years pooled (Gad,= 0.10, df = 1, P > = -0.14, n = 96, P > 0.05). 0.05). However, a higher proportion of inacces- sible nests were successfulin 199 1 than in 1992 MULTIVARIATE ANALYSES (G, = 4.56, df = 1, P < 0.05). Accessible nests None of the discriminant function analyses were no more successfulin one year than the yielded a combination of variables associated other (Gad,= 0.78, df = 1, P > 0.05). with successfulnesting. In each of the eight anal- Furthermore, there was no relationship be- yses, at least 34% of the nests were incorrectly tween the proportion of accessiblenests and the assigned status of successful or failed (and as date in the season. Forty-nine percent of inac- many as 48% in one analysis). These results thus cessible nests and 50% of accessiblenests were agreed with those of the univariate analyses. built early in the season (G,, = 0.02, df = 1, P > 0.05). DISCUSSION NEEDLE IN A HAYSTACK AND RARE SITE None of the eight hypothesesfor explaining car- HYPOTHESES dinal nest successwas supported by the results These hypothesescannot be tested directly. Car- of this study. For this population, there seems dinal predators likely include snakes, small to be neither a single nor a combined predictor 166 T. S. FILLIATER, R. BREITWISCH AND P. M. NEALEN of nest successamong the nest variables mea- gument can be inverted: a rich guild of nest pred- sured. Cardinals nonetheless appear to follow a ators may-by its very diversity-eliminate pre- few simple behavioral rules in placing nests: (1) dictably safe nest sites. This does not preclude provide some concealment for the nest, (2) build the possibility that the simple behavioral rules within a few m of the ground, (3) build the nest suggestedabove for cardinals are an evolutionary higher as the seasonprogresses, and (4) build the responseto selection for safe sites. Our proposal nest relatively higher in plants as the seasonpro- does, however, predict that such rules will be of gresses.Although nests built later in the season only limited benefit in predator-rich communi- tended to fare better than earlier nests, none of ties. Predators search in different ways, and a these behavioral rules is specifically associated safe site with respectto one predator may make with an increased probability of nest success. the nest more vulnerable to a different predator. Conner et al. (1986) also failed to find a corre- For instance, depredations by terrestrial preda- lation between probability of cardinal nest suc- tors should favor nests placed higher in vegeta- cessand either nest height or concealment. Sim- tion, while predation by birds such as crows may ilarly, Best (1978) and Morton et al. (1993) found well favor lower-placed nests. Although heavy what might be simple behavioral rules followed predation pressure independent of predator di- by nesting Field Sparrows (Spizella p&la) and versity should also make avoidance of nest pre- White-crowned Sparrows (Zonotrichia leuco- dation difficult, we suggestthat the latter is the phrys), respectively, without effectson the prob- more important variable because diversity of ability of nesting success. predatory tactics forecloses options. As por- Variables related to nest site not measured in trayed in Figure 1, this population of cardinals this study include the density and homogeneity is then proposed to occupy a point in the fore- of vegetation within a several m radius of the ground of the “selection landscape.” nest. Martin and Roper (1988) found that suc- Cardinal nesting behavior in general appears cessfulHermit Thrush (Catharus guttatus) nests to fit this scenario. Parents are weak defenders tended to be surrounded by similar vegetation of eggsand nestlings against predators (Nealen as that in which the nest was placed. Kelly (1993) and Breitwisch, unpubl. manuscript). Rather than showed the same for Dusky Flycatchers (Empi- risk injury or death in defense of the nest, car- donax oberholseri). Conner et al. (1986) found dinals are instead well-adapted for rapid renest- that more successfulpairs of cardinals had ter- ing following predation. Nest intervals are as short ritories with higher foliage density and patchi- as four days from nest loss to initiation of the ness at 2 m height (mean nest height was 1.6 m) next clutch (Scott et al. 1987) and the cardinal than less successfulpairs. breeding seasonis long, from mid-April to mid- Cardinals in this population are subject to a August. This allows as many as six nesting at- high incidence of nest predation, similar to that tempts in a single season (TSF, RB, PMN, un- found by Kinser (1973) and Best and Stauffer publ. data). (1980). The conventional argument is that the There also appears to be a rich guild of pred- high incidence of nest predation characteristic ators on cardinal (and other passerine)eggs and/ for passe&es should select strongly for choice or nestlings in southwestern Ohio. These likely of less vulnerable nest sites (Best and Stauffer include Blue Racers (Coluber constrictor), Rat 1980, Martin and Roper 1988, Li and Martin Snakes(Elaphe obsoleta),Milk Snakes(Lampro- 199 1). For example, in a study of nesting success peltis doliata), Gray and Red Squirrels (Sciurus in riparian bird communities, Best and Stauffer carolinensis and Tamiasciurus hudsonicus), (1980) found that speciescharacterized by more Eastern Chipmunks (Tamias striatus), Blue Jays “general” choice of nesting sites suffered higher (Cyanocitta cristata), and American Crows (Cor- nest predation rates than specieswith more re- vus brachyrhynchos) (cf. Kinser 1973, Nolan stricted choice of sites. They interpreted the high 1978). A pattern of predation characterized by incidence of predation as the consequenceof gen- empty, undisturbed nests was common, and this erality in choice. We argue here for the opposite observation does not exclude any of the above interpretation of causality: a high incidence of predators (I. Lovette, pers. comm.). As a group, predation may be the causeof generality in choice snakes, small mammals, and birds also account- of nest sites. ed for the majority of failed cardinal nests in We propose that the usual strong-selectionar- another study (Best and Stauffer 1980). CARDINAL NEST SITES AND NEST PREDATION 767

Strength of selection for a safe nest site

FIGURE 1. Strengthof selectionfor a safe nest site is shown on the vertical axis as a function of both predator speciesrichness and overall intensity of predation.

It is worth noting that the general design of submitted for publication). Although the criti- nest predation studies probably favors finding cism of incorrect variables is an alternative in- some relationship between at least one variable terpretation of our own results, we contend that related to nest site and probability of success. it is useful to put forward a novel interpretation. Most such studies include an array of variables, We doubt that cardinals are unusual in being and as the number increases,it clearly becomes subject to a rich guild of nest predators. Further, more likely that one or more will display a re- their weak defense of offspring against predators lationship with probability of nest successby is typical of many passerines(Nealen and Brei- chance alone (see Knopf and Sedgwick 1992). twisch, unpubl. manuscript; P. M. Nealen, un- Consistency across years in the variables asso- publ. data). Finally, short relaying intervals fol- ciated with successwould be more persuasive lowing nest failure are also common in passerines than data gathered in a single breeding season. (Scott et al. 1987 and referencestherein). There- It would also be instructive to have more in- fore, we predict that future research will show formation on the studies of nest predation that that many passerines nesting in predator-rich have not revealed patterns in nest site selection habitats display the same suite of aspectsof nest- and success.When this occurs, the study is al- ing behavior as cardinals: simple behavioral rules ways subject to the criticism that the meaningful for nest placement, weak defense of offspring, variables were simply not examined. Indeed, no and rapid renesting following failure. We also revealed pattern may well increase the proba- predict that studies of nest sites and predation bility of the study not even being published (or will frequently either fail to reveal interpretable 768 T. S. FILLIATER, R. BREITWISCH AND P. M. NEALEN

patterns or display “statistically significant” but GO-ED, B. M., AND C. F. THOMPSON.1978. Ex- biologically insignificant patterns of nest site se- perimental analysis of nest predation in an old- field habitat. Auk 95:304-312. lection and the successor failure of nests. KELLY, J. P. 1993. The effect of nest predation on habitat selection by Dusky Flycatchers in limber ACKNOWLEDGMENTS pine-juniper woodland. Condor 95:83-93. We thank Charity Kruger and John Ritzenthaler of KINSER, G. W. 1973. Ecology and behavior of the Aullwood Audubon Center and Farm for permission cardinal, Richmondenacardinalis, in southernIn- to conduct this research on the Aullwood property. diana. Ph.D.diss., Indiana Univ., Bloomington, They and other staff members were helpful to us in IN. many ways. Robert Brua, Albert Burky, Kevin Eckerle, KNOPF,F. L., ANDJ. A. SEEGWICK. 1992. An exper- SusanLinville, Irby Lovette, and Kelly Williams pro- imental studyofnest-site selectionby Yellow War- vided useful comments and suggestionson previous blers. Condor 94~734-742. drafts of this manuscript. We also thank Glenn Wals- LACK, D. 1954. The natural regulation of animal berg and two anonymous reviewers for helpful com- numbers. Oxford Univ. Press, Oxford, England. ments. Michael E. Rayle of the Office of Computing L&KEY, A. R. 1944. A study of the cardinal in Ten- Activities, University of Dayton, provided critical aid nessee.Wilson Bull. 56:27-44. in the discriminant function analysis and its interpre- LEMON,R. E. 1968. The displays and call notes of tation. During the courseof this study, TSF and PMN cardinals. Can. J. Zool. 46: 14 l-l 5 1. were supported by Summer Graduate Fellowships at LI, P., AND T. E. MARTIN. 1991. Nest-site selection the University of Dayton. In the second season,RB and nesting successof cavity-nestingbirds in high was supported by a University of Dayton Research elevation forest drainages.Auk 108:4054 18. Council Faculty Grant. MARTIN, T. E. 1988a. Processesorganizing open- nestingbird assemblages:competition or nest pre- LITERATURE CITED dation? Evol. Ecol. 2~37-50. MARTIN, T. E. 1988b. On the advantagesof being ALONSO,J. A., R. MUNOZ-PULIDO,AND L. M. BAU- different: nest predation and the coexistence of TISTA. 1991. Nest-site selectionand nesting suc- bird species.Proc. Natl. Acad. Sci. USA 85:2196- cessin the Azure-winged Magpie in Central Spain. 2199. Bird Study 38:45-5 1. MARTIN,T. E., ANDJ. J. ROPER. 1988. Nest predation BARNARD,P., AND M. B. MARKUS. 1990. Reproduc- and nest-site selectionof a Western population of tive failure and nest site selection of two estrildid the Hermit Thrush. Condor 90:5 l-57. finches in Acacia woodland. Ostrich 6 1:177-l 24. MAYFIELD,H. 1975. Suggestionsfor calculatingnest BENT, A. C. 1968. Life histories of North American success.Wilson Bull. 87:4561166. cardinals, grosbeaks,buntings, towhees, finches, MORTON,M. L., K. W. SOCKMAN,ANDL.E. PETERSON. sparrows,and allies, Part 1. U.S. Nat1 Mus. Bull. 1993. Nest predation in the Mountain White- 237. crowned Sparrow. Condor 95:75-82. BEST,L. B. 1978. Field Sparrow reproductive success MONTGOMERIE,R. D., ANDP. J. WEATHERHEAD.1988. and nesting ecology. Auk 95:9-22. Risks and rewards of nest defenceby parent birds. BEST,L. B., AND D. F. STAUFFER.1980. Factors af- Q. Rev. Biol. 63:167-187. fecting nesting successin riparian bird commu- NOLAN, V., JR. 1978. The ecology and behavior of nities. Condor 82: 149-158. the Prairie Warbler Dendroica discolor.Omithol. Banrrwtscrr, R., N. Gornmn, AND J. ZAlAs 1989. Monogr. No. 26, American Ornithologists’ Union, Behtvioral differencesin nest visits between male Washington, DC. and female Northern Mockingbirds.Auk 106:659- RICKLEFS,R. E. 1969. An analysis of nesting mor- 665. tality in birds. Smithson. Contr. Zool. 9: 148. BIUT~INGHAM,M. C., ANDS. A. TEMPLE. 1983. Have SAS INSTITUTE. 1989. SAS/STAT user’s guide, ver- cowbirds caused forest songbirds to decline? sion 6, fourth ed., vol. 1. SAS Institute, Cat-y,NC. Bioscience 33:3 l-35. - Scorr, D., M., R. E. LEMON,AND J. A. DAWY. 1987. BURNS.R. D. 1963. Michiean coooerative cardinal Relaying interval after nest failure in Gray Cat- study nesting data. Jackkine Warbler 4 1:56-6 1. birds and Northern Cardinals. Wilson Bull. 99: COLLIAS,N. E., ANDE. C. C~LLU\S. 1984. Nest build- 708-7 12. ing and bird behavior. Princeton Univ. Press, SHAVER,J. M., AND M. B. ROBERTS. 1930. Some Princeton, NJ. nesting habits of the cardinal. J. Tenn. Acad. Sci. CONNER,R. N., M. E. ANDERSON,AND J. G. DICKSON. 5:157-170. 1986. Relationshipsamong territory size, habitat, SIEGEL,S., AND N. J. CASTELLAN.198 1. Nonpara- song, and nesting successof Northern Cardinals. metric statistics for the behavioral sciences,2nd Auk 103:23-3 1. ed. McGraw-Hill, New York. FILLIATER-LEE,T. S. 1992. Parental roles in feeding SKUTCH,A. F. 1976. Parent birds and their young. nestlings,and nest sitesand nest successin North- Univ. of Texas Press, Austin. em Cardinals (Cardinalis cardinalis). M.Sc.thesis, SOKAL,R. R., ANDF. J. ROHLF. 1981. Biometry, 2nd Univ. of Dayton, Dayton, OH. ed. W. H. Freeman, San Francisco. GATES,J. E., AND L. W. GYSEL. 1978. Avian nest dispersionand fledging successin field-forest eco- tones. Ecology 59:871-883.