Apr 2, 1984 - species, I examined the hypothesis that this geographic trend is the result of ... that geographic trends in fecundity in some cases may be largely ... soides, and the results differ onlâ¢f in detail from those ... Clutches for which incubation was reported to have ..... portant, it is unrealistic to expect any single.
GEOGRAPHIC
VARIATION
IN CLUTCH
SIZE IN THE
NORTHERN FLICKER (COLAPTES AURATUS): SUPPORT
FOR ASHMOLE'S
HYPOTHESIS
WALTER D. KOENIG
Hastings Reservation andMuseumof Vertebrate Zoology, University of California, StarRouteBox80, CarmelValley,California93924USA
ABSTRACT.--The NorthernFlicker(Colaptes auratus), a commonhole-nesting North Americanwoodpecker,exhibitsa highly significantlatitudinalincreasein clutchsize.Usingpreservedclutchesand informationon climaticconditionsandbreedingdensitiesof confamilial species,I examined the hypothesisthat this geographictrend is the result of seasonalfluctuations in resources("Ashmole'shypothesis";Ashmole 1961, 1963;Ricklefs 1980). Clutch
sizeis significantlycorrelatedbothwith the ratioof summerproductivityto estimatedbreeding densitiesof all woodpecker species andwith the ratioof summerto winterproductivity, supportingAshmole'shypothesis. In addition,a substantial portionof the latitudinalgradient can be directly explainedby the seasonalityof resources, primarily becauseof an inversecorrelationbetweenclutch size and winter productivity.Theseresultsprovide the firstcorroboration of Ashmole'shypothesisfrom an examinationof intraspecific clutch-size variationand suggestthat geographictrendsin fecundityin somecasesmay be largely attributableto trendsin seasonalfluctuations of resources. Received 5 December 1983,accepted 2 April 1984.
DESPITEchallenges and numerous modifica-
sive information,it is possibleto examineany particular hypothesisonly by comparing the 1983), there is general agreementthat clutch consistencyof geographictrendswith the spesize is ultimately adjusted to the food supply cificpredictionsthat stemfrom it. Here, I focus by natural selection to reflect the number that, on "Ashmole's hypothesis" (Ashmole 1961, when considered over the lifetime of the in1963; Ricklefs 1980), which suggeststhat geodividual, resultsin the most young surviving graphicvariation in clutchsize is primarily deto maturity and attaining reproductive status termined by absolute resourceabundance rel(Williams 1966). Consensus breaks down, howative to breeding population density. This ever, when the environmental and ecological hypothesishas recently been supportedby factors that influence clutch-size variation are Ricklefs(1980)throughhis analysisof the mean considered.Variables such as daylength (Lack clutch size of severalpasserinecommunities.I 1954); population stability (Slagsvoid 1981); examine this hypothesisby using data from a habitat stability (Cody 1966, MacArthur and single altricial species, the Northern Flicker Wilson 1967);risk of nest predation (Slagsvoid (Colaptes auratus).This hole-nestingwoodpeck1982a);allocation of time and energy among er was chosen because of its wide distribution competing activities, including antipredator and the large number of clutchesavailablefor tacticsand interspecificcompetition(Cody1966, it. Specifically,I (1) analyze geographictrends Ricklefs 1970); and seasonality of resources in clutchsize in this speciesand (2) investigate (Ashmole 1961, 1963;Boyce1979;Ricklefs 1980) factorscontrolling clutch size, primarily with have all been suggestedto be crucial in deter- respectto Ashmole'shypothesis,by comparing mining the patternsof clutch-sizevariation that these trends with climatic data and estimates correlatewith various geographicaland eco- of the breeding population densitiesof comlogical gradients. peting taxa. These hypothesesare not necessarilymuMETHOOS tually exclusive,but to test them simultaneously would require extensivedata gathered over Clutch-sizedatawere gatheredfrom variousNorth a wide geographicarea, not only on clutchsize American museum collections; all forms of the Cobut on demography,competitiveand predato- laptesauratuscomplexwere included except for the tions (see Stearns 1976, Winkler and Walters
ry community interactions, and climatic con-
"Gilded" Flicker (C. a. chrysoides), whose clutch size
ditions. In lieu of obtaining such comprehen- is sodifferentfrom that of other flickers(Koenig 1984) 698
The Auk 101: 698-706.
October
1984
October1984]
FlickerClutchSize
699
that it was excluded from the analyses. (All procedures were also run with the inclusion of C. a. chry-
were used as estimates of the breeding-population densities of woodpeckers. These surveys are done soides, andtheresultsdifferonl•fin detailfromthose throughout the United Statesand Canadaand consist presentedhere.) Original data cardswere consulted of censusingthe breeding birds for 3 min at a series to obtain the locality, date of collection, stage of in- of 50 stops0.8 km apart along a road transect (Byscubation, and the number of eggs of each clutch. track 1981).Data from thesesurveyswere condensed Clutches for which incubation was reported to have as follows. First, the number of individuals of each been "slight," "just begun," or "one-quarter or be- speciesof woodpecker recorded along each route was yond" were considered to be complete; those for averaged over all years that the survey was run (this which the incubation stage was either unknown or varied from 1 to 16 yr). Then, the latitude and lonstated to have been "fresh" or "none" were considgitude of each route was rounded to the nearest deered to be potentially incomplete. For the analysisof gree, and censusvalues for a route were averaged the influence of first-eggdate on clutch size, the first- with others within the same 1ø block of latitude and egg date was taken to be the date of collection, minus longitude ("latilong"). Finally, the mean censusvalthe number of eggs in the set, minus a correction for ues for all speciesof woodpeckers combined were the stage of incubation. The incubation period of added to each clutch-size record. In all, data from flickersis 11-14 days(Jackson1977),and the assumed over 18,000censusescoveringseveralthousandroutes clutch-initiationdate was correctedby subtractingthe were used. These estimates, although rough, are following number of days for the associatedstageof nonethelessobtained in a uniform way and are thus incubation: 2 ("slight," "just begun," "•A," or "2-5 likely to be comparable throughout the large geodays"), 4 ("begun," "commenced," or "started"), 7 graphic region of interest. (."smallembryo," "well begun," "•A," or "5-7 days"), Data analysis.--The combined clutch-size, geo10 ("advanced," "large embryo," or "well ad- graphic, climatic, and population-density data were vanced"). analyzed in two ways. First, each clutch-sizerecord For each clutch, information on geographic posi- was considered independently; in all, 1,164 clutches tion, climate, and population density were added as of Colaptesauratuswere obtained, of which 411 were follows. complete. Only these 411 complete clutches were Geographic position.--Thelatitude and longitude of used. Second,I combined all complete clutchesfrom eachlocality, to the nearestdegree,aswell aswheth- each latilong and used the meansfor each latilong as
er the locality was coastal(insular or within 1ø of longitude of the coast) or inland (not within 1ø of longitude of the coast), were added to each clutch record.
Climate.--Thestationnearesteachlocality for which climatic water-balance data are given by Thornthwaite Associates (1964) was determined; stations were usually within 50 km of collecting localities. From monthly data on mean precipitation (PPT) and actual evapotranspiration (AE), the following variables were recorded: (1) overall mean annual PPT, (2) overall mean annual AE, (3) combined PPT for
April, May, and June, (4) combined AE for April, May, and June, (5) the maximum AE for 3 consecutive months in the year, and (6) the minimum AE for 3 consecutivemonths in the year. AE, defined as PPT minus runoff and percolation (Sellers 1965), is qualitatively related to the amount of vascular plant activity (primary productivity) in terrestrial environments (Rosenzweig 1968) and is thus probably a consistent estimate of overall resourceavailability for virtually any animal, regardless of exact diet or trophic level (see Discussion). Following Ricklefs (1980), I added the three highest consecutivemonthly AE values for an estimate of "summer" for "winter"
AE and the 3 lowest
consecutive
a single datum. Thesevalueswere then recombined with the geographic, climatic, and censusdata. This modification
minimized
the statistical
difficulties
re-
suiting from repeated use of the same climatic and censusdata for all clutchesnear a particular locality (within the samelatilong block). Hence, theseresults are used
whenever
the
climatic
or census
data
are
involved. Logarithmic transformationwas usedwhen the ratio of two variables
was of interest,
as this
transformation, followed by a multiple regression, provides a convenient way to assessinterrelationships of variables. The distributions
of both clutch-size
data sets were
significantly non-normal by the Kolmogorov-Smirnov goodness-of-fittest (Sokaland Rohlf 1981);thus, Spearman rank testswere used to assesscorrelations. In lieu of comparablenonparametric tests,however, regressionsand analysesof variance were performed when the interrelationships of several variables were of interest. Cochran (1947) and Donaldson (1968) conclude that F-tests, used here, are robust to modest
violations of normality. Nonetheless, results from theseanalysesmustbe taken with caution.Two-tailed tests are employed, and P values less than 0.05 are considered significant.
values
AE.
RESULTS
Populationdensity.--Censusdata from the North American breeding-bird-survey program adminis-
tered by the United StatesFish and Wildlife Service
The mean size of the 411 complete clutches used in the analyseswas 6.5 eggs (SD = 1.4).
700
WALTER D. KOENIG =
clutch size increasesby slightly over 1 egg for every 10ø increasein latitude. Second,after adjusting for the stageof incubation (seeabove), I found that clutch size increasessignificantly with date (Spearman rank r• = 0.15; n = 405; P < 0.01). Finally, I found that there is a tendency for clutches collected near the coast to
I
• 20
ß-> •0
be smaller
2
3
4
5
6
?
Clutch
8
9
I0
II
than
those obtained
inland.
Inland
clutchesaverage 6.9 eggs (n = 197), compared with 6.1 eggsfor coastalclutches(n = 207); this difference is significant (Mann-Whitney U-test, z = 5.2; P < 0.001). In a multiple regressionof
r•
o
[Auk,Vol. 101
12
size
Fig. 1. The distribution of clutch sizes among complete clutchesof the Northern Flicker, exclusive of C. a. chrysoides. Mean ñ SD = 6.5 _+1.4 eggs.
clutch size on both date and latitude, however
date is not significant(F = 0.03;df = 1,404;ns). Similarly,the coastalinfluencedisappearswhen latitude is controlled for in an analysis of covariance (F-value for coastal effect controlled
The distribution ranged from 3 to 12 eggs,with the majority being between 4 and 9 (Fig. 1). Two geographicand one temporal trend were examined
for their
influence
on clutch
size.
First, using either all complete clutchesor the lumped latilong data, I found a highly significantpositivecorrelationbetweenclutchsize and latitude (Fig. 2, Table 1). In both cases,mean
for latitude = 2.8; df = 1,401; ns). Refining the coastal category to include only clutches collected either within 1/2ø or •o of longitude of the coastfails to alter the significanceof this difference.Thus, although thesevariablesmay be important in particular casesor within individual populations, neither date nor proximity to the coast has a detectable effect on clutch size independent of latitude on the macrogeographicscale under considerationhere.
I0-
•x
4
ß
N:410
Latitude
Fig. 2. Mean clutch size ñ SE by latitude for Colaptes auratus.Samplesizesfor each degreeof latitude vary from 1 to 77. Regression(using individual clutches) is: clutch size = 0.118 (latitude) + 1.97 (r2 = 0.21; F = 106.81; df = 1,408; P < 0.001).
October 1984]
FlickerClutchSize
701
TABLE1. Full and partial correlation coefficientsof mean clutch of Northern Flickers by latilongs with geographic,climatic, and breeding-density variables. Sample size refers to latilong blocks. Pearson
Spearman r Latitude
Controlling for summer AE Controlling for winter AE Controlling for breeding density of all picidae Climatic
or
partial r
P-value
n
0.48
0.49
< 0.001
121
----
0.49 0.30 0.40
< 0.001