WP)"; Date llll“HIIIIHZIIUIIIHHIHUI!!!llllllmlllllIIIIHUUIHII 1293 10485 9750 uuuuu ‘i'ua- RA .12 Y ,7 EEiIChigan St; if. E 3:; wLUnivcrsity (4; ‘I‘ This is to certify that the thesis entitled SITE SELECTION AND NESTING SUCCESS OF OLDFIELD SONGBIRDS presented by Bruce Wi lber Becker has been accepted towards fulfillment of the requirements for a Ph.D. degreein Fisheries and Wildlife Major professor April 21, 1981 0-7639 wow» M131: 4 ‘j‘éuli; sm‘ (“9% JUN? 18? 2002 m. W: 25¢ per day per item RETUMIM LIBRARY MATERIALS. ' Place in book return to remove charge from circulation records SITE SELECTION AND NESTING SUCCESS OF OLDFIELD SONGBIRDS By Bruce Wilber Becker A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Fisheries and Wildlife 1981 1-.) hr", a I -" "| / a» ' I) c r ABSTRACT SITE SELECTION AND NESTING SUCCESS 0F OLDFIELD SONGBIRDS By Bruce Wilber Becker Site selection and nesting success were examined for an avian community inhabiting shrubby oldfields adjacent to woodlots. Contrary to edge-effect theory, nest distribution at the community level was random throughout the patchy oldfield vegetation. Species-Specific differences in nest-site selection were demonstrated, however, by discriminant function analysis. Total vegetation cover and tree cover were the most important discriminating variables, with shrub cover and nest height of lesser importance. Examination of the nesting substrates selected by the various birds showed that all species were basically shrub nesters, although early field sparrow and rufous-sided towhee nests were located on the ground. The size of the nesting shrub was strongly correlated with bird length. Certain plants were shown to be preferred nesting sites, and these were utilized by all the bird spe- cies. Preferred nesting plants were shown to be much less available in transition zone and forest habitats, which may partially explain the insignificant use of these areas by Oldfield birds. Success rates for the various species were shown to be equal by analysis of variance. A multiple regression approach was used to relate nesting success to habitat parameters. Success was not related to edge, or to vegetation Bruce Wilber Becker cover, but was positively correlated with nest height and negatively correlated with nesting date. These correlations explained only about 8% of the variation in nesting success, so the fate of an individual nest seemed to be largely the result of randomly operating factors. T-tests showed that ground nests were half as successful as above-ground nests, which appeared to explain the general preference for nesting in shrubs. The decline in success over time was shown to result from smal- ler clutch sizes, rather than a change in predation rate. It was hypoth- esized that an edge effect will occur when two seral stages which nor— mally occur far apart along a successional gradient become Sharply juxtaposed. Birds characteristic of the intervening seral stages res- pond with a supernormal settling response, because of the accentuated proximate habitat features. The edge effect lasts until the younger seral stage develops into the intermediate stage, thus allowing the intermediate species of birds to disperse. ACKNOWLEDGEMENTS I wish to express my appreciation to a number of individuals who were especially helpful to me during the course of this study: Dr. Leslie W. Gysel gave me the opportunity for advanced study by accepting me as his student. He provided a research assistantship through the Michigan Agricultural Experiment Station, and was very gen- erous with his time and experience. I am indebted to Dr. Rollin H. Baker who assumed the chairmanship of my committee upon Dr. Gysel's retirement. Dr. Baker's counsel and leadership were invaluable. Drs. Richard W. Hill, Peter G. Murphy, and Harold H. Prince served as my committee members. They made many valuable comments during the planning stage of this study, and were able editors of the draft. Drs. Donald L. Beaver, Jonathan B. Haufler, and J. Edward Gates also reviewed the draft and made many helpful comments. Dr. Stanley J. Zarnoch pro- vided advice on the statistical analyses. I am grateful to the Michigan Department of Natural Resources for allowing me to use the Rose Lake Wildlife Research Area as a study site. Dr. Glenn Y. Belyea of the DNR provided access to maps of the area which proved very useful. Jim Ruhl and Tom Slager assisted me with vegetation measurements. Their help was appreciated. I would especially like to acknowledge the support of my family ii and my friends: Scott Shalaway, Keith Williams, and Bonnie Jacobs dur- ing this period of my life. They were there when I needed them. iii LIST OF TABLES ... LIST OF FIGURES INTRODUCTION ......OOOOOO... TABLE OF CONTENTS OO.........OOOOOOOOOOOOOOOO...... STUDY AREA .................................................. METHODS ...................... ... ................. . . ....... RESULTS .. .................... . .............................. Study Area Vegetation ...................... ..... ....... Distribution of Vegetation Parameters ............ ...... The Nesting Community ......... Nest Distribution NeSt'Site SElECtiOI‘l oooooooooooooooooooooooao oooooooooo 0 Relationship to Cross Habitat Features .... ....... . Vertical Stratification .................... ..... .. Temporal Segregation ..... Substrate Selection .... Use of Transition Zones Nesting Success The Avian Community in Relation to Plant Succession .... Relationship to Habitat Variables ........ ......... Nesting Success by Substrate Category ....... ...... Comparison of Species-specific Success Rates ...... Decline of Nesting Success Over Time .............. Causes of Nest Failure DISCUSSION ....... Nest-site Selection .......... Nesting Success ......OOOOOOOOOOOOOOOO0.0... The Mechanism of Edge Effect ........................... LITERATURE CITED iv 10 10 19 21 23 25 25 29 31 31 37 39 39 39 41 42 42 44 46 46 49 52 55 Table LIST OF TABLES Page Percent coverages of shrub and tree strata plants for old- field homogeneous vegetation types.................. ..... .... 11 Distribution of nests by species and study areas... ......... . 22 Discriminant function analysis of community nest-site seleCtioDOOOIOOOO ....... O ..... ... 0000000000 00...... 0000000000 26 Predicted group membership based on the discriminant function classification procedure............................ 28 Distribution of community nests by substrate types. .......... 32 Comparison of mean bird length with mean height of the nesting shrub ............................... ...... ........... 36 Percent occurrence of preferred nesting plants by habitat type ......................................................... 38 £192 LIST OF FIGURES Locations of study areas in the Rose Lake Wildlife Research Area............ 00.00.000.000...OOOOOOOOOOOOOOIOOOO . Homogeneous vegetation types (HVT) of Area 1.. ......... ..... Homogeneous vegetation Homogeneous vegetation Homogeneous vegetation Homogeneous vegetation types types types types (HVT) (HVT) (HVT) (HVT) of Area 2 ..... . ......... . of Area 3................ of Area 4 ..... ..... ...... of Area 5 ................ Distribution of nests by height. ............................ Distribution of nests by species in hawthorn (Crataegus spp. ) by 0.5 meter height classes. vi l3 14 15 17 18 30 35 INTRODUCTION Good evidence exists to Show that avian habitat selection is in- fluenced by the structural features of plant communities (Balda 1975). Several studies have shown that bird species diversity increases as herbaceous, shrub, and tree layers are added, with species composition of the vegetation being relatively unimportant (MacArthur and MacArthur 1961, MacArthur et a1. 1962, Karr 1968, Recher 1969). Willson (1974) found, however, that the addition of trees in a vegetational series had a disproportionate effect on the addition of bird species, primarily through the addition of new avian guilds. Roth (1976) concluded that new avian species were accommodated in preforest habitats by the addi- tion of new patches of vegetation, with patches being differentiated by the layering of the vegetation. In contrast, species packing in late shrub or forest stages occurred primarily through vertical segregation. Holmes et al. (1979) found that the importance of vegetation height to bird species diversity was related to foraging opportunities. The ad- dition of shrubs and trees along a vegetational gradient increases fo- liage layering and complexity, and provides a new foraging region, e.g., the tree holes and branches. They also considered plant species com- position to be a factor influencing bird species diversity because dif- ferent plant morphologies require different foraging techniques, and because differences in plant chemistry affect the quantity of the in- sect fauna found on various plant Species. 2 From this review, it is evident that the diversity of bird species is related to the structural complexity of the vegetation in both vertical and horizontal dimensions. A woodlot, for example, with her- baceous, shrub, and tree Strata should support a more diverse avian community than an herbaceous Oldfield. With areas being equal, a hab- itat containing the junction of the above two vegetation types would be diverse in the horizontal dimension and would be expected to Show higher bird species diversity, as a result. When vegetation types with different structural features are jux- taposed (as above), an edge is created. A second theory of avian hab- itat selection follows from this relationship. Leopold (1933:131) in- troduced the concept of edge effect by noting that many game animal species required several different vegetation types within their re- spective habitats. Leopold considered game to be a phenomenon of edges. Odum (1971:157) reflects current thought by defining edge effect as a tendency for variety and density of organisms to increase at the junctions between communities. Several studies involving avian species along woodlot margins support Odum's definition (Lay 1938, Johnston 1947, Gates and Gysel 1978, Strelke and Dickson 1980). A similar edge effect has been noted with fence rows in agricultural fields (Shalaway 1979). Habitat manipulation designed to benefit animal populations by creating edge is a widely accepted tenet of wildlife management (Bur- ger 1973:19, Giles 1978:135, Thomas 1979:48), and interest continues in trying to relate edge to management goals (Schuerholz 1974, Patten 1975, Ranney 1977, Taylor 1977, Thomas 1979248, Gates and Mosher 1981). Recent studies, however, have shown that edge effect may have negative 3 consequences. Gates and Gysel (1978) found that an avian community ex- hibited high nesting density but low nesting success along a woodlot- oldfield junction. Robertson and Flood (1980) showed that edge areas created by shoreline land development had significantly more birds than natural areas, but lower species diversity. These findings indicate the need for further study of edge effect at the avian community level, including examination of nesting success. The present study examines the nest-site selection ecology and success of an avian community in oldfield vegetation adjacent to wood- lots. The study was designed to determine if Gates and Gysel's (1978) findings of high nest density and low success continue when the oldfield vegetation includes prominent shrub and tree strata, in addition to herbaceous cover. It is hypothesized that the junction between the old- field and woodlot would be less sharply delineated under such conditions, and that this would inhibit the edge effect. A second objective, related to the first, was to examine the nest-site selection ecology of the oldfield avian community from a comparative viewpoint. Species-specific preferences are expected from the competitive exclusion principle (Whittaker 1975:78), and are sug- gested by Nolan (1963) and Harrison (1975). A review of the literature failed to produce a quantitative study of this nature for an oldfield avian community. A study of nest-site selection would thus be a con- tribution to the literature, and should provide insight into the edge- effect phenomenon. STUDY AREA To insure the best comparison with the findings of Gates and Gysel (1978), an upland site providing mature forest adjacent to ad- vanced oldfield was required for the study. Multiple study areas proved necessary to provide an adequate sample of nests. Four sites were located in the Rose Lake Wildlife Research Area which lies 19 km northeast of Lansing, Michigan in Clinton and Shiawassee Counties. Two of the sites (Figures 4 and 5) adjoined the same woodlots studied by Gates and Gysel. The study began in the spring of 1977. In 1978, a fifth site was added to increase the sample size of nests. Observa- tions and measurements were concluded in the fall of 1978, with the exception of Site 5 which was searched for nests during the fall of 1979 to provide two years of data. The Rose Lake Area comprises 1,349 ha of moderately rolling farm land, abandoned fields, upland and lowland woods, and marsh (Figure 1). The area was purchased by the Michigan Department of Natural Resources in 1939, and is administered primarily for wildlife research, and secondarily for wildland recreation. The area features many habitat development measures such as floodings, tree and shrub plantings, and food patches. Pregitzer (1978) provides the following description of climatic features. The inland location of Clinton County minimizes the effect of the Great Lakes on the climate, the most noticeable influence being the increased cloudiness which moderates the minimum temperature in 4 OATH 1' 0'08 J uramm———q— , V . I) It‘ r‘ u I 7 4 a . u, ~ ‘ v .... . . «2:1, TC‘. , ‘ ' S 4“,.7 ‘d‘. * ‘A'<~"* q‘ n ‘1')», . a. I .,.‘,5 "_;t. . l 4‘ . . \.' i Locations of study areas in the Rose Lake Wildlife Research Area. Figure 1. 6 fall and early winter. The annunal average of possible sunshine is 53 percent. Cloudy days are most common in the fall and early winter, and least common in late spring and summer. The growing season averages 143 days (May 13 to Oct. 3). Precip- itation is well distributed throughout the year, with the crop season receiving 45.3 cm of the average annual total of 76.3 cm. Evaporation from the class "A" pan during the crop season averages 88.9 cm. Soil moisture replenishment during fall and winter is important to agricul- tural success because potential evaporation exceeds the average annual precipitation during the growing season by 92 percent. Snowfall aver- ages 95 cm per year, with an average of 67 days per season with 2.5 cm or more of snow on the ground. Elevation is about 244 m above sea level. The major soils of the study areas are Spinks loamy sand (Psam- mentic Hapludalfs), Boyer loamy sand, and Boyer sandy loam (Typic Hapludalfs) (Pregitzer 1978, Threlkeld and Feenstra 1974). These soils are moderately suited to agriculture, with droughtiness, slope, and erosian being the limitations. METHODS Nests were located by repeatedly searching the study areas from May to September in 1977 and 1978. Nests were revisited until fates were known. Care was taken to minimize disturbance of the nests and surrounding vegetation to deter desertion or observer-induced predation. A mirror mounted on a pole was used to view nests from a distance. Data gathered at each nest site included: species of nest plant, height of nest plant, notes on surrounding vegetation, height of nest, and information relating to fate of the nest. An unbiased evaluation of the influence of edge effect on avian community nest-site selection and success required that the edge effect be factorable from other considerations. Preliminary examinations of the study areas showed that the oldfield vegetation was not structur- ally homogeneous. Differences in soils, topography, and encroachment of plant species, as well as plantings of grasses, shrubs, and trees resulted in patches of vegetation ranging in aspect from sparse shrub through medium and dense shrub-tree stages. The junctions of these patches formed edges, just as the oldfield-woodlot junctions did, and it was hypothesized that the presence of these patches might also af- fect nest-site selection and success. To enable Statistical analyses of the above factors, study area vegetation was classified into distinct, homogeneous vegetation types (HVT) according to three criteria: (1) An obvious difference had to exist between adjacent HVT in the crown coverages of the shrub (1-5 m) 7 8 and/or tree (5+ m) strata. Herbaceous cover was essentially complete in all HVT except for a few minor coniferous plantings, and was not used to differentiate HVT. (2) The size of the HVT had to be large enough to be ecologically relevant. Minimum size was determined from the ter- ritory or home range size of the various species. Because all the bird species in the study nest within their foraging areas (territor- ies), territory size is an indication of relevant patch size. Review of the literature revealed the following territory sizes for common species on the study areas: field sparrow (Spizella pusilla) 0.36-0.43 ha (Evans 1978), 0.31-1.62 (av. 0.76) ha (Best 1977), 0.4-0.8 ha (Wal- kinshaw 1945), and 0.81-2.43 ha (Walkinshaw 1968); cardinal (Cardinalis cardinalis) 0.51-2.32 ha (Dow 1969); and catbird (Dumetella carolinen- gig) 0.95-161 ha (Zimmerman 1963). A compromise of 0.4 ha was selected as the minimum HVT size because it was near to the lower end of the territory sizes reported. A minimum figure was chosen because it was likely that in at least some cases, a species' territory might encompass several distinct vegetation types (Evans 1964, Dow 1969, Best 1977). This would be expected also from the literature review on edge effect. (3) Plant species composition was also used to differentiate HVT. In many cases differences in species composition were reflected by obvious differences in shrub or tree strata crown coverages, but this was not always the case. In the latter instances, HVT could easily be lumped for statistical analyses where appropriate. Oldfield HVT and HVT adjacent to the study areas were identified from 35 mm aerial photos and ground observations, and were mapped at a scale of 1 cm = 6 m using compass and rangefinder. Locations of all nests discovered during the study were plotted on the maps. 9 Measurements of HVT vegetation were adapted from those employed by Karr (1968) and Carothers et a1. (1974). Crown coverages of the shrub and tree strata were determined by the line intercept method (Can- field 1941, Andresen and McCormick 1962). Eight transects were measured in each HVT following a restricted random pattern (Goldsmith and Har- rison 1976:104). Nominal transect length was 30.5 m, although this was altered if necessary to fit the shape of the HVT. All transects within a HVT were of approximately the same length. Percent coverages for individual shrub and tree strata species were calculated by measuring their horizontal interception on the transects. Where two species overlapped the transect within a height interval, only the one with the greatest area was measured. If areas were equal, the one on top was measured. Heights of the tallest trees in each HVT were measured with a Haga altimiter. RESULTS Study Area Vegetation Crown coverages of important shrub and tree Species present in each of the 16 oldfield HVT are listed in Table l by height strata. Shrub-strata coverages ranged from 2.0 to 91.2%, with a weighted average of 46.4% for the study area as a whole. The small overstory trees: Prunus, Ulmus, and Quercus dominated the shrub strata in most HVT, although 3222 occurred in very high densities in a few HVT. Tree stra- ta coverages ranged from 0 to 66.1%, with a weighted average of 26.2%. Prunus, Quercus, and HlEBE dominated the overstory. Because of the presence of old field—border trees, and the fact that succession had occurred undisturbed (except for some shrub and tree plantings) for 38 years, trees ranging in height from 12 to 24 m occurred in most HVT (Figures 2-6). Bromegrass (Bromus inermus) and bluegrass (Poa compressa) were the predominant grasses in the oldfield HVT, reflecting plantings made when the area was retired from agriculture in 1939. Little bluestem (An- drogogon scoparius) occurred in sporadic patches, indicating a slow return of native species. Common forbs included: goldenrod (Solidago spp.), panicled tick-trefoil (Desmodium paniculatum), cow vetch (Vigia cracca), round-headed bush clover (Lespedeza capita), hairy bush-clover (L, flirts), wild carrot (Daucus carota), yarrow (Achillea millefolium), flowering spurge (Euphorbia corollata), wild bergamont (Monarda fistu— losa), hoary alyssum (Berteroa incana), and heath aster (Aster 10 .eseosn 4m novsaocuo .mu0:u_dd nonnaucHn .audu vocunaou mo N0.o A seasons movsaucao 11 ~.0~ 0.0N q.ha o.ma 5.0 0.0 0.o~ n.~¢ 0.0 0.00 ~.00 H.NH m.o~ 0.00 n.¢~ 0.00 0.~n manuoa o.~ 0.m 0.0 m.H 0.0 n u 0.0 n ~.a a.H 0.0 a n 0.0 «.0 H.H I. uosuo 0.0 H.N a 0.0 u n u u u u n u u 0.0 a u 0.N Esvanan onumnnanm 0.0 0.0 a n i u u 0.0 n u a u u n n 0.0 0.N anocamou macaw ~.H 0.0 n N.m u u u u a u o.H u u N.m u 0.~ n novuoHDEouu anaammm 0.H u u 0.0 n u u a . o.H 0.00 a u u c u n oumucopaccwuw unaam0m o.N 5.0 0.0 0.0 n a 0.H o.~ u c.~ u 0.0 u 0.0 n 0.0 ~.H acnoaumEm mafia: «.0 0.0 0.m 0.0 n u u 0.m~ u 0.5 a 0.0 m.~ 0.0 a 0.0 0.0H nocuuaam> mauuomm c.0H H.m 0.N o.- H.n 0.0 0.0 ~.oH u n.~n 0.0 H.0 «.00 H.mu 0.0H o.HH 0.0H mcwuouoo mucosa auouum mesh «.00 0.05 0.mm 0.00 n.~m n.H~ ~.0m 0.m0 o.~ ~.Hm n.nm H.- H.00 p.00 0.5N 0.00 0.50 «Hauoa 0.0 m.HH m.oH m.HH 0.0 u o.~ «.0 «.0 5.0 H.~ 0.0 ~.N 0.0 0.~ 0.0 N.o uosuo To in - To To m6 «6 mi» - - - - u u - 0.~ 0; mac—Soon was: 0.0 0.0 o.~ «.0 u u n u . q.N 0.0 m.~ o.~ 0.0 0.H 0.0 0.0 wouuuuwu muouucog 0.0 u H.~ 0.0 I u s 0.0 u u I n N.0 0.o u u s moans MflMNM 0.H q.~ 0.0 0.0 u u u 5.0 u 0.0 . m.~ u 0.0 0.0 H.o 0.~ unmoeoomn macaoo o.a - - a.a N.o - - m.o - - 0.5 s.o - o.n H.o a.o 0.0 macaoH=Emnu aaamuoa 5H - - TN ~.o 04 0.3 05 04 u u - - TN To . . Scions: macwmmaym ~.a m.~ - ~.H - - - m.~ - «.5 H.5H a.o - m.n - ~.H 0.0 saunas “~04 a; - - is m.m - pm as - as a; M: «A «.0 «6 NS to ..E... g “.0 m.- 0.0 0.0 u m.~ . o.q~ . 0.0 0.0 0.o ~.~ 0.0 0.0 o.~ 0.0 nacwusao> maouoao 0.0 0.0m ~.0 u 0.0 «.0 m.~ H.~ n 0.~ 0.0 0.o u 0.0 0.0 0.5 0.0 ncnuuuwen mafia: Na .03 - - - - - - - ado ma - 9% 1H a.~ 98 «.2 a 21E o.NH 0.0 ~.n m.H~ «.0H H.0H 0.00 m.m u 0.m ~.0 H.m 0.0 0.0 0.0a 0.0 m.¢~ acauouom ancsum numuum nannm uwanu>< ~-m H-m a-s H-m ¢.~ n-~ ~-N H-~ m-a ~-a o-a m-~ ¢.H n-a ~-H H-H coauoam sauswams nonasz om>9 coauuuowo> a mou< zvnum mammxu angumuowo> msoocoono; uaofiupao new munmam cannon menu nan sauna mo nowmuo>oo ucouuom .H wanna 12 ericoides). Common ground-layer woody plants included: blackberry (33- bus allegheniensis), black raspberry (R; occidentalis), red raspberry (R; idaeus), strawberry (Fragaria virginiana), and poison ivy (Rhgg'ga- dicans). Characteristics of individual study areas and their surround- ings follow. Area 1 measured 6.03 ha and was divided into 8 oldfield HVT (Fig- ure 2). Slopes were 6-12 and 12-18%. The area was bordered on the nor- th by a woodlot (HVT l3) dominated in the tree stratum by red oak (Que;- gg§_£gbrg) 49.1% crown coverage, red maple (Age; rubrum) 20.5%, and big- tooth aspen (Populus grandidentata) 11.9%; on the east by a mixed wood (HVT ll) dominated by Scotch pine (Pinus sylvestris) 23.9%, bigtooth as- pen 17.5%, black cherry (Prunus serotina) 16.4%, and black oak (Quercus velutina) 9.6%; and on the west by a mature red pine (Pinus resinosa) 64.8% - white pine (§;_strobus) 21.4% planting (HVT 9). Oldfields ex- tended to the south. The transition zone (HVT 12) which comprised the north boundary of the study area was the vegetation between the 1939 fence line and the present woodlot margin. The tree stratum was domin- ated by red oak 54.5%, bigtooth aspen 15.8%, and black cherry 12.1%. Area 2 measured 1.66 ha and was divided into 4 oldfield HVT (Fig- ure 3). Slopes were 0-2 and 6-12%. The area was bordered on the north (HVT 7) by a black oak 50.6% - bigtooth aspen 28.4% woodlot, a highway to the east, and oldfields to the south and west. HVT 3 contained an autumn olive (Elaeagnus umbellata) planting. The transition zone (HVT 6) was dominated by black oak 27.5% and bigtooth aspen 17.8%. Area 3 measured 1.53 ha and comprised l oldfield HVT (Figure 4). Slopes were 12-18 and 18-25%. The area was bordered on the west by a pignut hickory (Carya glabra) 67.3%, ‘white oak (Quercus alba) 14.4%, 13 Woods 2‘— H.HN H.~H a an A00 a.co o.mm e «a Aoav meow .9 .H mwu< mo AH>mV mocha coHumum0m> msowcmono: uo>oo nausm N um>ou mosh N unwfiom maocmo amassz H>m ANHV mcooz u m.n~ - m.sH - s as AmV .N munwwm 0 0 C 0 O Q ‘. a.mm m.¢m a ma 3 Amy w.~ N.om a ma Aav muooz s.~w . 0.~m a s a ma m m.mo o.am 2 mm Amav l4 (6) 12 m (7) 58.6 20 m 67 7 95.9 Woods ° 57.. 1, T.Zone E (4) - HVT Number : 10 m - Canopy Height ; 5.0 - % Tree Cover : 21.7 - % Shrub Cover w ; o (5) . 21 11 m 5 18 0 - - « <3) (2) 64.3 :Tm\./\ 16m 5 10.9 42.7 : 58.1 65.6 (1) ; N 0 m I 0.0 2.0 L________4 0 30.5 m Homogeneous vegetation types (HVT) Figure 3. of Area 2. 15 (5) 25 m Woods 93‘8 70.1 I T. Zone (4) : 13 m .' 60.4 : 76.0 5 (1) - HVT Number ; 9 m - Canopy Height ; 3.7 - % Tree Cover : 32 2 - % Shrub Cover (2) ...... 6 m "- ~-‘____ 6.4 “~- 21.3 ....FJ y—————————-I 0 30.5 m Figure 4. Homogeneous vegetation types (HVT) of Area 3. Woods (3) 18 m 91.3 44.3 16 black oak 9.1% woodlot (HVT 5); on the north by a mature mixed wood (HVT 3) dominated by red pine 38.5%, black oak 19.5%, and black cherry 14.3%, and to the east by an oldfield (HVT 2) dominated by 6 m red pine. The transition zone (HVT 4) was dominated by pignut hickory 17.4%, black oak 16.7%, and black cherry 10.9%. Area 4 measured 1.27 ha and comprised l oldfield HVT (Figure 5). Slopes were 6-12%. The area was bordered on the south by a woodlot (HVT 3) dominated by pignut hickory 56.6%, white oak 19.3%, and black oak 18.4%, to the east by a wooded area (HVT 8) composed of black locust (Robinia pseudo-acacia) 65.6% and black cherry 23.5%; to the northeast by a mixed wood (HVT 7) dominated by red pine 43.6%, black cherry 25.4%, and honey locust (Gleditsia triacanthos) 10.3%; to the northwest by a mixed wood (HVT 6) dominated by black oak 49.5%, pignut hickory 17.8%, and red pine 10.5%; to the east by an oldfield dominated by 5 m Scotch pine 25.8% and autumn olive 10.3%; and to the southwest by the mixed wood (HVT 4) described for Area 3. The transition zone (HVT 2) was dom- inated by pignut hickory 23.8%, bigtooth aspen 19.6%, and black oak 13.5%. Area 5 measured 1.79 ha and was divided into 2 oldfield HVT (Fig- ure 6). Slopes were 6-12, 12-18, and 18-25%. The area was bordered on the west by woods (HVT 3) dominated by American elm (Ulmus americana) 28.7%, black cherry 11.4%, and red oak 9.0%; on the northwest by dense shrub; and on the east by herbaceous oldfield. Two small woodlots (HVT 4) dominated by black locust 66.0% and black cherry 14.4% were located in the oldfield. These were the result of plantings made in the 1950's. 17 Woods (6) ‘ Woods 18 m ---------- 90.9 (7) ‘ 78.6 15 m “ 91.9 49.5 5 12)m (8) 17 m 16.3 94.9 59.3 49 8 (1) - HVT Number ' 14 m - Canopy Height 43.0 - % Tree Cover W d 55.8 - % Shrub Cover 00 S (4) 18 m (2) 91.3 14 m T. Z 44.3 one 66.6 H TM 76.2 ( 27 m d VJ 95.6 W00 5 T 78.1 |——————-—i O 30.5 m Figure 5. Homogeneous vegetation types (HVT) of Area 4. Figure 6. 18 -¢--------_--_——---— ....... Woods (4) 21 m 94.0 48.1 (1) HVT Number 11 m Canopy Height 17.4 - % Tree Cover 35.8 % Shrub Cover 0 30.5 m Homogeneous vegetation types (HVT) of Area 5. 19 Distribution of Vegetation Parameters It is evident from Figures 2-6 that the oldfield vegetation was patchy. Additional habitat complexity resulted from juxtaposition with woodlot vegetation. A series of simple linear regressions (Nie et a1. 1975) was used to determine if vegetation parameters were distributed in a homogeneous manner. Heterogeneous distributions might influence the distribution of nests. The analyses were based on a total of 222 points which were shown to be randomly located on the study area maps (see the section on nest distribution). DIST, the straight-line distance from a random point in the oldfield to the nearest woodlot edge was the independent (X) variable. Dependent (Y) variables were: shrub crown cover (SHCC), tree crown cover (TRCC), total vegetation cover (VC), foliage height diversity (FHD), and homogeneous vegetation type diver- sity (HVTD). Data from all study areas were combined for this analysis. Values for the dependent variables were calculated from the study area maps using a scale 0.4 ha circle around each random point as a sampling area. Choice of a 0.4 ha circle followed from the discussion of HVT classification. A sampling unit smaller than the territory was desirable because territories are not necessarily circular, and nest placement is not always in the center of the territory (Nolan 1978:134). For each point, SHCC and TRCC were averaged values for the vegetation encompassed by the 0.4 ha circle. VC was the sum of SHCC and TRCC. FHD was the ShannonJWiener value calculated from SHCC and TRCC (Brower and Zar 1977:138). HVTD was the Shannon4Wiener value for the percent occurrence of different HVT in the 0.4 ha circle. HVT were considered "the same" if 95% confidence intervals overlapped for both the shrub and tree-strata crown coverages. This procedure was necessary to avoid 20 unrealistically high HVTD values in cases where the difference between HVT was taxonomic rather than structural. Results of the regressions were as follows. Slopes (m) signifi- cantly different from zero (p (.05) indicate non-uniform oldfield vege- tation. SHCC declined with DIST (m=-.3l4, F=4.146, p=.O43, r2=.019). TRCC declined with DIST (m=-.l7l, F=84.0SO, p= <.001, r2=.276). vc De- clined with DIST (m=-.202, F=42.849, p 4.001, r2=.163). FHD was unre- lated to DIST (m=-.686, F=3.444, p=.065, r2 =.015). HVTD declined with DIST (m=-.459, F=5.1l7, p=.025, r2=.023). These results require careful interpretation because large sample size may result in the rejection of the null hypothesis (m=0) even though the fitted linear relationship has little explanatory value (Wesolowsky 1976:61). Examination of the r2 values indicates that dis- tance from the woodlot edge accounts for very little of the variation in shrub cover, foliage height diversity, homogeneous vegetation type diversity, and total vegetation cover; and only 27.6 percent of the variation in tree cover. I conclude that a weakly defined gradient of decreasing tree cover existed from the 01df181d-W00d10t junction to the oldfield interior, but that the variations in shrub cover and foliage height diversity are essentially unrelated to distance from the edge. The relatively high r2 value for total vegetation cover reflects the pattern of tree cover. The low r2 for homogeneous vegetation type di- versity indicates that contrasting vegetation types occur throughout the oldfield, not just at the woodlot edge. The TRCC and SHCC regres- sions indicate that the contrast is mostly a function of difference in tree cover 0 Examination of plotted residuals indicated that in all cases the 21 variables were linearly related, and that the variances of the residuals were homogeneous. The large sample size rendered the assumption of normally distributed residuals unimportant. The Nesting Community Distribution of nests by species and study area is shown in Table 2. Of the total of 222 nests located, 46 were from years preceeding the study and 14 were from Area 5 in 1979. Nests were identified to species from observations of the nesting bird, with the exception of field sparrow and indigo bunting nests which were distinctive. Seventy-seven nests were built on 10.49 ha in 1977 and 85 on 12.28 ha in 1978, giving nest densities of 7.34 and 6.92 nests per ha, respec- tively. These data are thought to be nearly complete for the open- nesting altricial species to which my study was limited. Several tow- hee nests, however, were probably not found in Area 1 each year. A scarlet tanager (Piranga olivacea) nested in Area 1 in 1978, but was not included in the analyses because its nest height (8 m) was beyond the vertical range of the other species. One precocial species, a mal- lard (Anas platyrhynchos) was found on Area 3 in 1978, but was excluded from the analyses. No effort was made to include cavity nesters in the study because they differ significantly in clutch size, number of clut- ches, incubation period, nestling period, and success rate from open nesting species (Nice 1957, Welty 1975). Few snags were available on the study areas because of the seral nature of the vegetation, which limited populations of hole nesters. Black-capped chickadees (P3523 articapillus) were commonly observed, and probably nested on the areas. A great creasted flycatcher (Myiarchus crinitus) may have nested in Area 1 in 1977. 22 .Hmu0u Eouw ucouomev ouoaa mumw: %u0umH£nc3ocx mo umpEDZm Amavmmm am am mm as ea masses um 0H 0H m n HN wuwmz mHnwamsucmuHcp Auoum moufiuoHOZV H u u u u H qu0300 wovwmnnc3oum AmocmoHuoEm msumouoov H - - - - H ooxuso meHap-onHmw AmuwumHuo nuuwoocmmuv m H I u u N %mn msHm Aaowsu mEOumoxOHv A00m n N H N u nosmnufik :3oum AmDEHmnumounumwo manhoooov m m H H H m ooxooo voHHHnnxumHm AmDEHmcunmounuhuo onmwmv m H m H H 0 005309 vwvanmsoHsm wachawwwo wanaswwwov HNHVmH - m N s a Hawawwwo AmHmcocHHoumo mHHoumannv AmquH m u u N n qunumo AmoGMHU mcHummmmmv Hmva a H m H m wnHucnm ovacH HwHwasa wHHmwwamV manMVNm q 0 HN mm mm souumam 0Hon mawuoe Hma.av H-.Hv Hmm.av Hos.av Amo.ov mmwumam m 0 m N H AmwumUoomv mou< kvnum .mmmum hmoum 0cm mmHommm >0 mums: Ho cOHunanumHQ .N mHan 23‘ Table 2 indicates that most species were present on each of the study areas, although relative densities differed. A best estimate of community composition was obtained by combining data from 1977 and 1978, and apportioning the unknown-species nests according to the relative abundances of known species to which they could belong. The cowbird was not included in this analysis. Community composition was found to be: field Sparrow 42.2%, cardinal 11.8%, catbird 11.8%, indigo bunting 10.6%, towhee 8.1%, black-billed cuckoo 8.1%, brown thrasher 4.3%, blue jay 2.5%, and yellow-billed cuckoo 0.6%. Nest Distribution The null hypothesis that nest distribution was uninfluenced by edge was tested with Holgate's (1965) test of randomness (Brower and Zar 1977:123). Tests were based on sixty random points for Area 1, and 30 points each for the Smaller areas, giving a total of 180 random points. Holgate‘s A was calculated for each area and for all areas combined. The distribution of nests would be random if A = 0.5, contagious if AI>0.5, and uniform if A<10.5. The significance of the departure of A from 0.5 was determined from [A -o.5| «ll/12 n t: where n is the number of random points (Brower and Zar's 1977:123 for- mula was corrected). The computed t was compared to the t-table value for infinite degrees of freedom. The A-values and computed t's for Areas 1-5 and combined data were as follows: (1) A = 0.4864, t 0.6888; (3) 0.3649; (2) A = 0.5363, t A = 0.5205, t = 0.3890; (4) A 0.5056, t = 0.1063; (5) A 0.4976, 24 t = 0.0455; and combined data A = 0.5055, t = 0.2558. All the t's are smaller than the two-tailed d.= 0.05 value, 1.960, so there is no sta- tistical evidence to indicate that the distribution of the 222 community nests was not random. Nest distribution was also examined with linear regression (Gill 1978:86). The number of nests per hectare per year was calculated for each of the 16 oldfield HVT, and then regressed with HVT tree cover (TRCC). A plot of the points showed the relationship to be linear. The null hypothesis of equal nest densities was not rejected (m=-.052, t=1.480, p2>.05, r2=.l35). This test supports the previous finding of random nest distribution, and indicates that the avian community was able to utilize all of the oldfield vegetation types for nesting. The hypothesis that bird species diversity should be greater near the oldfield-woodlot junction was tested by listing the 101 nests ob- served while active by order of increasing distance from the nearest woodlot edge. The nests were then divided sequentially into 6 groups. The first 5 had 17 nests, the last 16. Bird species diversity was the Shannon-Wiener value for each group (Brower and Zar 1977:138). Bird species diversities were then linearly regressed with the midpoints for each distance interval. A plot of the points Showed the relationship to be linear. The null hypothesis was not rejected (m=-.0004, t=.218, p>>.05, r2=.015). The conclusion is that bird species diversity did not show an edge effect in this study. 25 Nest-site Selection Relationship to Cross Habitat Features Considering the random distribution of nests and the relatively uniform distribution of the oldfield vegetation parameters, the degree of nest-site segregation of the various bird species was of interest. Stepwise discriminant function analysis (Nie et a1. 1975) was used to test for species-specific differences in nest-site selection. Sample size was 155 nests distributed as follows: field sparrow 92, indigo bunting 18, catbird 14, cardinal 13, towhee 9, and black-billed cuckoo 9. Brown thrashers, blue jays, the yellow-billed cuckoo, and the cowbird were excluded from analysis because of small sample sizes. Seven hab- itat parameters were used as discriminating variables. DIST, SHCC, TRCC, VC, FHD, and HVTD were defined previously. Nest height (NHT) was the vertical distance from the ground to the bottom of the nest. Results of the analysis are presented in Table 3. DIST and HVTD were not useful as discriminating variables, and were eliminated by the stepwise selection feature. Apparently, none of the species showed a strong preference for nest sites near HVT boundaries. Five discriminant functions accounted for a total of 88.8% of the variance in nest-site selection among the six bird species. Function 1 with an eigenvalue of 1.25 was the most important discriminant function, accounting for 77% of the variance existing in the 5 functions, and 56% of the variance in nest-site selection. The standardized coefficients for discriminant function 1 show that total vegetation cover and tree cover were 3 and 2 times, respec- tively, more important than nest height or shrub cover for separating the bird species. Foliage height diversity was relatively unimportant. 26 Table 3. Discriminant function analysis of community nest-site selection. Discriminant Eigen- % of Cannonical Correl- Function value Variance ation Squared 1 1.2469 76.82 .555 2 .1604 9.88 .138 3 .1257 7.74 .112 4 .0901 5.55 .083 5 .0001 .01 .000 100.00 .888 Standardized Discriminant Function Coefficients Function Variable 1 2 3 4 5 NHT 1.00 -.12 .01 .01 .13 SHCC 1.24 8.30 -3.30 -8.25 -2.64 TRCC 2.01 12.58 -3.77 -16.59 -3.26 VC -3.02 -18.54 7.26 22.46 4.72 FHD -.11 —1.52 -.53 1.27 -.35 27 Total vegetation cover, tree cover, and shrub cover remained the most important discriminating variables in the other four discriminant func- tions. Difference in sign indicates whether the variable was making a positive or negative contribution. F-tests were preformed between pairs of species to determine if significant differences (P<:.05) existed in nest-site selection. Of the 15 pair-wise comparisons, all were significantly different (pi(.02) except towhee and cardinal (p=.33), which indicates: (1) that the set of discriminating variables was relevant, (2) that sample sizes were adequate, and (3) that niche segregation among the 6 species did have a nest-site-selection component, in addition to probable differences in food habits and foraging behavior. The extent of habitat segregation was examined further by using the original 155 nests in a classification procedure. Results are shown in Table 4. Sixty-three percent of the nests were correctly classified. Examination of individual species shows that considerable overlap in nest-site habitat occurred. If the values for percent correctly clas- sified are taken as measures of faithfulness to a preferred nesting habitat, the field sparrow (72%) and catbird (71%) appear to be habitat "specialists"; and the black-billed cuckoo (33%) and towhee (33%)are "generalists". According to Nie et a1. (1975:435), the statistical theory of dis- criminant analysis assumes that the variables have a multivariate nor- mal distribution, and that they have equal variance-covariance matrices within each group. They further stated that the technique is very ro- bust and that these assumptions need not be strongly adhered to. Har- ris (1975:232), however, cautioned that unequal sample sizes may lead 28 Table 4. Predicted group membership based on the discriminant function classification procedure. Predicted Group Membership (%) Species No. FS IB C RT CB BC F. Sparrow 92 71.7 13.0 1.1 8.7 3.3 2.2 I. Bunting 18 38.9 50.0 0.0 5.6 5.6 0.0 Cardinal 13 0.0 7.7 46.2 15.4 15.4 15.4 R. Towhee 9 33.3 0.0 22.2 33.3 11.1 0.0 Catbird 14 0.0 0.0 0.0 0.0 71.4 28.6 B. Cuckoo 9 0.0 11.1 0.0 22.2 33.3 33.3 29 to discrepancies between true and nominal significance levels. The equality of group covariances was tested with Box's M (Nie et a1. 1975). Covariances were not equal (p<(.001). Several transformations were tried: m , log10(X+l) , and (Z/WMO?) where K=32/5(- (Neter and Wasserman 1974:509), but none produced equal covariances. The smal- lest F-value, largest cannonical correlations, and highest percent of species classified correctly resulted from the‘filtg- transformation, and this was used for the analysis. The foregoing analysis of nest-site selection dealt primarily with gross habitat features. Although nest-site segregation occurred at this level of analysis, considerable overlap existed among the bird species. Nest-site segregation may also occur, however, within a veg- etation type through vertical stratification, different nesting periods, or choice of different nesting substrates. The following analyses ex- amined these aspects of nest-site segregation. Vertical Stratification It is evident from Figure 7 that the nests of the oldfield song- bird community occupied a narrow vertical dimension. Over half occur- red below 1 m, and 89 percent below 2 m. Differences in mean nesting heights among the six major species were examined by analysis of var- iance (Nie et a1. 1975). Mean nest height (meters), standard error, and sample size for the six species were as follows: field sparrow .33 105 (92), indigo bunting .48 12.04 (18), catbird 1.88:.25 (14), cardinal 1.40109 (13), towhee .92 1.22 (9), and black-billed cuckoo 1.79:1.29 (9). These data were transformed (Z/JE-)(J§-) to improve heterogeneous variances. The null hypothesis of equal nesting height was rejected (F=36.609, p= (.001). LSD, SNK, and Duncan's a posteriori Number of Nests 100 90 80 70 60 50 40 30 20 10 30 Cumulative Percent fi 17 f V v r v v *1 V W W 18% 56% 89% 97% 99% 99% 92 4 . 39 32 35 ll 4 6 3 1 2 3 4 5 Meters Nest Height Figure 7. Distribution of nests by height. 31 contrasts all gave the following groupings of species (p=‘<.05): (1) field Sparrow, (2) indigo bunting = towhee, and (3) cardinal = catbird = black-billed cuckoo. It appears that the oldfield avian community did show degregation in nesting height, although overlap existed to an ever increasing extent as the preferred nesting height increased. It was noted in the discriminant function analysis of nest-site selection that nest height had considerably less power as a discriminating variable than total vegetation cover. Temporal Segregation Mean nesting dates for the six species were tested by analysis of variance (Nie et a1. 1975). Days of the nesting season were numbered consecutively beginning with April 1. Date of first egg was used in the analysis. Data were not transformed because variances were homogen- eous. Sample sizes were as follows: field sparrow 37, catbird 13, car- dinal 12, towhee 9, indigo bunting 9, and black-billed cuckoo 9. The null hypothesis of same mean nesting date was not rejected (F=.529, p=.754). Therefore, it was concluded that the 6 species nested concur- rently and that nest-site segregation did not have an important temporal component. Substrate Selection The nesting substrates selected by the avian community are summar- ized in Table 5. Herbaceous vegetation Supported 18.92% of the nests, with the field sparrow the only species to utilize this cover consis- tantly. Nests in herbaceous vegetation were located on the ground ex- cept for 2 found in goldenrod, the only herb apparently strong enough to support a nest. Ground nests occurred in a great variety of 32 .mucmamunmmoe COHDMuowo> aw vwvuooou you was mcchHwHH>.de .AmNmHv mocHuuom umumw .Aaaumuum sousm cH moHooam ucmHa mo mucouunuoo m>HumHou NV \ Amanda 00 N00 .mcHucom ochcH can zouummm 0H0Hh cmsu umnuo moHoonm .AqunsoU .ooxooo voHHHnn3oHHw» .>m0 oon .umnmmuny a3oumv umsuouo .ooxooo n vaHHnuxomHmuom aqunumonmo .mosaoh vmvauwDOHQMHHm .Hmchumouo .mcHucom ochcHumH .Souummm 0H0Hmnmmm mmoHumam qum u 00.00H NNN mm 0H 0 0H 0 0H 0H N0 meuoa n 00.0 «H N N H u u n m 0 umsuo Ho. mm.H m - - H H - H - - ochdso once 00.0 mm.H m u u u a i u H N mHumcouoo mnumm 00.0 mN.N m N n u u a H H H mamoHHme monoum 0N. 0N.N m m n n n u n n u mcmoHumEm mDEHD 0m.NN mH.m N u u u u H u u 0 mH:=EEou monoansh 00.H 00.0 0 u u H u H N i q mumHHone: macwmomdm . Hm.0 0H H u u u i n m u .amm mmmmm occh Ho.m NH o - H - H N - N oancHwMH> oscoanon 0N.0 Hm.0 «H q a u H i H H N muonHane omom NN.N 00.0 0H NH u u m H a N H mUHHmuumu whoUH:0H oo.n Hw.oH om o - N o m H H a .odo nacnoo n N0.0H N0 H H u u H u i am msomomnuo: oo.o m~.m~ on cu a m m H m - NH .ooo oowononwo on N .oz a c on mo em 0 mH ma ooHooam chHm .moahu oumuumnom >0 mummc zuHcosaoo mo cOHuonHuumHm .m mHan 33 situations, ranging from dense stands of bromegrass or cow vetch to sparse grass-forb cover associated with a small shrub. Large spreading forbs such as panicled tick-trefoil and Lespedeza spp. were common nest sites. Eighty-one percent of the nests occurred in shrubby substrates. Of these, Egbgg spp. (mostly blackberry) was unique in that it seldom grew beyond the ground cover strata (<,1 m). Rgbgg supported 4.51% of the nests, with the indigo bunting the only bird to use this substrate significantly. The indigo bunting was the only member of the avian com- munity to have a pensile nest (Pettingill 1967:257). The ability to securely fasten the nest to the supporting plant facilitated use of Egbgs stems, which lack the forked branches necessary to support the nests of the other species. Hawthorn (Crataegus spp.) was the single most important nesting substrate, accounting for 25.23% of all nests. Eight of the 10 bird species utilized this plant, with the indigo bunting and cowbird the only exceptions. Dogwood (Cornus racemosa and Q; amomum), tartarian honeysuckle (Lonicera tartarica), and multiflora rose (Rosa multiflora) accounted for an additional 25.68% of the nests. Individual shrub spe- cies were used by most bird Species, with no species-specific preferen- ces obvious, except for the indigo bunting and 33233. A preference index (Petrides 1975) was calculated for the shrub layer species to determine if use of a given plant species was propor- tional to its occurrence in the shrub strata (PI=1). Table 5 shows that all species except elm (Ulmus americana) and sumac (Rhus gyphina) were used to a greater extent than expected (PI) 1). Black cherry, which was the dominant shrub layer species(Table 1). was not an important 34 nesting substrate. Evans (1978) found nest-site segregation between the field sparrow and chipping sparrow (Spizella passerina) based on shrub height, with the field sparrow preferring junipers (Juniperus communis) shorter than 1.5 m. The extent of shrub-size segregation in my study area was ex- amined by two methods. Figure 8 shows the species nesting in hawthorn by 0.5 m size class intervals. Size segregation was evident, with the field sparrow the only species preferring shrubs shorter than 2.5 m. Taller shrubs were avoided by the field sparrow, but preferred by the other species. In the second analysis, heights of the nest shrub, irrespective of plant species, were averaged for the 7 most prevalent bird species (Ta- ble 6). comparison of 95% confidence intervals showed three groupings of species: (1) indigo hunting, (2) field sparrow - towhee, and (3) towhee = cardinal = catbird = black-billed cuckoo - brown thrasher. It appeared that members of the oldfield avian community may be grouped as small or large shrub selectors, with the towhee utilizing both classes. The small-shrub selectors evidenced the greatest flexibility in nest- site selection, with the field sparrow also utilizing herbaceous cover, and the indigo bunting utilizing R3233. Regression of mean shrub height (Y) with bird length (X) (Table 6) gave an r2=.86, which strongly suggests that size of the bird is the factor determining the size shrub selected. The regression equation was logloY = .99 + .73 loglox. The loglo transformation was used be- cause a plot of the data points indicated a curvilinear relationship, with shrub size becoming aSymptotic when bird length reaches 23 cm. The null hypothesis that slope of the regression line was zero was .00000H0 u£0H0£ u0u0E 0.0 >0 H.000 0300000000 :uonusms :H 00H00am >0 0000: Ho :OHunnHuumHQ .0 00:0Hm mu0u0z u 0:0H0: 030:0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.N 0.N m.H 0.H m.0 35 00 D 00 90 D 00 0 mm 0m mm mm mm 4 00 0 00 0% 0 0 mm mm mm mm a D D a D 90 0 mm mm mm a D D 0 H0 0 00 mm o a a pm an a mm a 00 0 mm : D D mm .w 00HHHuc0cha u 0 son osHm - am a am ; 00500009 c3000 n Hm Hmchumo u o 0 4 ooxono 00HHHnuonH0> u ow ooxono 00HHHnuxomHm u 00 D + 005308 000Hmumoowam n 90 Socwdom oHon - mm a : saroads Kq snseN go IaqmnN 36 Table 6. Comparison of mean bird length with mean height of the nesting shrub. Shrub Ht. (m) Species Number (cm)8 (X i.std. err.) Indigo Bunting 18 11.43 1.15 i..08 Field Sparrow 53 12.70 1.64 i..12 Rufous-sided Towhee 8 18.42 2.25 i..25 Cardinal 13 19.69 2.88 I..11 Catbird 14 19.69 3.37 i..34 Brown Thrasher 5 25.40 3.30 i..41 Black-billed Cuckoo 9 27.94 3.36 i..36 aFrom Robbins et a1. (1966). 37 rejected (p < .05, t=5.637). Use of Transition Zones Edge-effect theory suggests that avian abundance and diversity should be high in transition zone areas because of the overlap of for- est and oldfield communities, in addition to species unique to the transition zone. To test this hypothesis, transition zone areas (Fig- ures 2-5) bounded by the 1939 fence lines at the old woodlot margins and the present oldfield borders were searched for nests. The four areas comprised a total of 1.82 ha, and produced only 3 nests: 1 field sparrow, 1 wood thrush (Hylocichla mustelina), and 1 unknown; compared to the 222 nests in 12.26 ha of oldfield. A chi-square test (Test of Binomial Proportion, Gill 1978:76) was used to determine if nest den- sities were the same in transition zone and oldfield vegetation. The null hypothesis was rejected (p= (.05, chi-square=25.85). The conclusion is that the transition zone was not suitable nesting habitat for mem- bers of the oldfield avian community. The presence of only one wood thrush nest indicates a lack of suitability for forest nesters, also. Additional insight into the complexities of avian-vegetation re- lationships is evident from Table 7. Occurrence of preferred (P12 1) nesting shrubs is presented for oldfield, transition zone, and woodlot vegetation types. Favored species declined in total cover from 7.4% in the oldfield to about 3% in both the transition zone and woodlot habitats. The relative contribution to total shrub cover declined from 16% to 5%. While it must certainly be simplistic to argue that reduc- tion in preferred nest sites is the major reason for the decline of the oldfield avian community in the wooded vegetation type, it may be a contributing factor. 38 Table 7. Percent occurrence of preferred nesting plants by habitat type. Transition Species Oldfield Zone Woodlot Crataegus spp. 1.94 .18 1.05 Elaeagnus umbellata 1.66 .80 - Cornus racemosaa 1.62 .45 1.92 Lonicera tatarica 1.43 - - Rosa multiflora .32 - .15 Prunus Spp. .24 - - gyigg'spp. .13 1.59 - Juniperus communis .06 - - Juniperus virginiana - - - Total % Cover 7.40 3.02 3.12 % of Total Shrub Cover 15.97 4.94 4.71 Total Shrub Cover 46.35 61.08 66.20 Total Tree Cover 26.17 77.28 97.02 8Includes 9; amomum. 39 The Avian Community in Relation to Plant Succession A final comment on avian-vegetation relationships involves the re- lationship between avian community composition and stage of plant suc- cession. The oldfields comprising my study area were retired from agriculture in 1939 and have been untended since that time. Succession started from plantings of bromegrass and bluegrass, and has been aug- mented by shrub and tree plantings for habitat development. In 1978, the mean tree cover for the oldfield areas was 26%, which was still considerably less than the 77% recorded for transition zones and 97% for woodlots (Table 7). Differences in shrub cover for the three habi- tat types were less pronounced, with oldfield at 46%, transition zone at 61%, and woodlot at 66%. In a typical grass-shrub-tree succession, we would expect to find corresponding changes in the avifauna (Odum 1950, Johnston and Odum 1956, Karr 1968). The birds on my areas were charac- teristic of early and late shrub stages (Karr 1968). Many of them con- tinue to occur in forest habitat, but they are not stenotopic for ei- ther forest or herbaceous habitats. Nesting Success Factors affecting nesting success were tested by the following analyses. Relationship to Habitat Variables Multiple regression (Nie et a1. 1975) was used to relate nesting success (SUCC) with a series of habitat variables: DIST, NHT, SHCC, TRCC, HVTD, FHD, VC, and DATE. SUCC was the number of young fledged divided by the mean clutch size for the respective species. Division by the mean clutch size was necessary to standardize the success values 40 because clutch sizes differed among the various species. Mean clutch sizes, standard errors, and sample sizes for the various species were as follows: field sparrow 3.79 i_.13 (29), indigo bunting 3.71 i..l8 (7), towhee 3.33 i .24 (9), cardinal 3.20 i .13 (10), catbird 3.45 i .16 (11), black-billed cuckoo 2.75 i .25 (4), brown thrasher 3.00 i .41 (4), blue jay 4.67 i,.33 (3), yellow-billed cuckoo 3.00‘i -- (1), and cowbird 2.00 i,-- (1). Only nests with completed clutches were used in calcu- lating means. DATE was the date the first egg was laid in the nest (see analysis of mean nesting dates). Other variables were described previously. A stepwise, forward inclusion regression procedure was preformed on the 98 nests whose success was known. DIST (p=.343), SHCC (p=.844), TRCC (.132), HVTD (p=.664), FHD (p=.206), and VC (p=.288) did not enter the equation (slopes were not different from zero). The conclusion from DIST and HVTD is that nesting success was not influenced by prox- imity to the oldfield-woodlot junction, or in the more general case, by proximity to the junction of any contrasting vegetation types. Nesting success did not show an edge effect. Variability in the gross habitat features: shrub cover, tree cover, total vegetation cover, and foliage height diversity was also unrelated to nesting success. NHT (p=.043, r2=.042) and DATE (p=.044, r2=.040) did enter the re- gression equation, giving an overall significance level of p=.017 and r2-.082. Nest height and date of nest establishment each accounted for about 4% of the total variation in nesting success, leaving 91.8% of the variation unexplained. It appeared that the success of an individ- ual nest was very close to being a function of randomly operating fac- tors o 41 Examination of residuals plotted against DATE showed that the as- sumptions of constant error variance, independence of error terms, and linearity of the regression function were met. The assumption of a normally distributed error term was relaxed because of the large sample size (Wesolowsky 1976:125). A X+.5 transformation was applied to the data because it resulted in slightly higher r2 values. Mean clutch size was used instead of simply the number of eggs in each individual clutch in formulating SUCC for the following reasons: (1) For 23 of 98 nests, the actual clutch size was not known because the nest either failed before the clutch was completed, or the nest was found just before fledging. (2) Using mean clutch size, a field spar- row, for example, that lays 5 eggs and fledges all 5 would have a high- er success rate than one that lays and fledges 3, which seems relevant ecologically. The analysis was repeated using the 75 nests for which clutch size was known, with SUCC defined as the number of young fledged divided by the clutch size for that nest. The multiple regression ap- proached significance (F=2.975, p=.057, r2=.076), with FHD (p=.063, r2=.047) and NHT (p=.133, :2 =.029) entering the equation. This analy- sis supports the previous conclusion that the success of an individual nest is close to being a function of randomly operating factors. SUCC calculated with mean clutch size values was used in all the following analyses except where indicated differently. NestingfiSuccess by Substrate Category The discussion of nest-site segregation showed that nest sites may be meaningfully grouped according to herbaceous (ground), low shrub ((2.5 m), and high shrub ()2.5 m) categories. Nesting success was calculated for each category, and mean values were compared by t-test 42 (Gill 1978:67). The assumption of homogeneous variances was met. No significant difference was found between small shrub (.5513 :,.0765, n=40) and large shrub (.5305 i .0719, n=41) categories (p) .05, t=.l747), but the difference between herb (.2488 i .1144, n=17) and shrub categor- ies was significant (p1(.05, t=2.1248). Apparently, it was advantage- ous for a bird to nest above ground level, but the actual height chosen seemed to make little difference. Comparison of Species-specific Success Rates Various studies have concluded that open-nesting altricial species have similar success rates (Nice 1957, Nolan 1963, Ricklefs 1969, Fret- well 1972:133, Gates and Gysel 1978). This hypothesis was tested for the nesting community using the SUCC variable by analysis of variance (Nie et a1. 1975). Mean nesting success, standard errors, and sample sizes for the 6 major species were as follows: field sparrow .328 i..073 (37), indigo bunting .570 i..l82 (9), cardinal .443 i..151 (12), towhee .600 I, .141 (9), catbird .714 I. .130 (13), and black-billed cuckoo .486 i..l60 (9). Variances were homogeneous. The null hypothesis of identical nesting success was not rejected (F=l.583, p=.l74). Decline of Nesting Success Over Time The decline of nesting success over time could be the result of several factors, acting singly or jointly: (1) reduction in clutch size, (2) increase in predation rate, (3) larger numbers of nests lost to natural disasters, (4) increased desertion rate, or (5) increased cow- bird parasitism. It will be shown later that (3), (4), and (5) were of little importance in this study. The relative contribution of (1) and (2) to the decline of nesting success was examined by regression 43 analysis (Nie et a1. 1975). A clutch size variable, CLCH, was formed by dividing the nunber of eggs in a clutch by the mean clutch size for the species. Only nests with completed clutches (n=75) were used in this analysis. A predation variable, PRED, was formed by subtracting SUCC from CLCH. Thus if a nest fledged the full clutch, PRED would be zero. If a nest failed (SUCC=0), PRED would be equal to CLCH. Partial failure of a nest would result in a PRED proportionally smaller than CLCH. The PRED variable actually reflects (3), (4), and (5) above in addition to predation, but their influence was negligible. SUCC (Y) was regressed with DATE (X) to see if the previously re- ported relationship held with a subsample consisting only of nests with completed clutcles. The null hypothesis of uniform success rate was rejected (m=-.O47, F=5.988, p=.017, r2 =.076). As before, nesting suc- cess did decline as the season progressed. The r2=.076 indicates that the effect was not very pronounced. CLCH (Y) was then regressed with DATE (X). The null hypothesis of constant clutch size was rejected (m=-.O32, F=30.988, p=i<.001, r2=.298), indicating that clutch size declined over the nesting season. About 30% of the variation in clutch size was attributable to DATE. PRED (Y) was then regressed with DATE (X). The null hypothesis of constant predation rate could not be rejected (m=.015, F=.626, p=.431, r2=.009). These regressions indicate that predation rate was constant throughout the nesting period. The observed decline in nest- ing success was, therefore, primarily a function of decreased clutch size. 44 Causes of Nest Failure Of the 98 known-history nests, 54 fledged at least one nestling, yielding a nest success rate of 55.10%. Causes of nest failure were as follows: large mammals 4 (9%), birds-snakes-small mammals 37 (84%), cow- bird parasitism 1 (2%), natural disasters 0 (0%), and desertion 2 (5%). Predation by large mammals was assumed if the nest was dislodged orthe surrounding vegetation disturbed. Empty nests showing no sign of dis- turbance were considered to have been lost to birds, snakes, or small mammals (Best and Stauffer 1980). It is obvious that large mammals, cowbird parasitism, natural disasters, and desertion were minor factors affecting community reproduction. The overwhelming cause of nest fail- ure was predation from the bird-snake-small mammal group of predators, which accounted for 84% of failed nests. Of this group, eggshells were found in or under the nest in 10 (27%) instances, suggesting predation by chipmunks (Tamias striatus) (Pettingill 1977, Nolan 1978:415). The most commonly observed potential predators on the study areas were the blue jay, chipmunk, red squirrel (Tamiasciurus hudsonicus), white-footed mouse (Peromyscus leucopus), blue racer (Coluber constric- 333), and garter snake (Thamnophis sirtalis). Pettingill (1977) noted, however, that nest predation seemed to be most common at night, so my diurnal observations may well be biased. Other potential predators known or thought to be present included: (birds) red—tailed hawk (33333_ jgmaicensis), sparrow hawk (Falco sparverius), crow (Corvus brachyrhyn- 3333), marsh hawk (Circus cyaneus), red-shouldered hawk (Buteo linea- 333), broad-winged hawk (Buteo platypterus), great horned owl (3333 virginianus), C00per's hawk (Acgipiter cogperii), long-eared owl (3333 otus), screech owl (Otus asio), (mammals) shorttail shrew (Blarina 45 brevicauda), woodland deer mouse (Peromyscus maniculatus), thirteen- lined ground squirrel (Spermophilus tridecemlineatus), fox squirrel (Sciurus 33g33), flying squirrel (Glaucomys volans), longtail weasel (Mustela frenata), striped skunk (Mephitis mephitis), opossum (23333: phis virginiana), raccoon (Procyon lotor), cat (Felis catus), red fox (Vulpes vulpes), dog (Canis domesticus), and (snakes) milk snake (33m- propeltis doliata) (Linduska 1950, Gates and Gysel l978)_ DISCUSSION Nest-site Selection Edge-effect theory has two parts: (1) that variety and density of organisms tend to increase at the junction of plant communities (Leo- pold 1933:131, Odum 1971:157), and (2) that reproductive success is low- er at the junction through the effect of density-dependent mortality factors (Gates and Gysel 1978). These hypotheses were tested in the present study by observing nest-site selection and relative success for a community of open-nesting altricial birds in shrubby oldfields adja- cent to woodlots. Contrary to expectation, distribution of nests at the community level was found to be random throughout the oldfield, and nesting success was found to be unrelated to distance from the woodlot edge. Thus, the edge effect demonstrated by Gates and Gysel (1978) along an herbaceous oldfield-woodlot edge was absent when the oldfield vegetation was dominated by shrubs. Examination of species-specific nest-site selections provided insight into the mechanisms leading to edge nesting in one case, and a random pattern in the other. In the present study, mean crown coverages for oldfield shrub and tree strata were 46% and 26%, respectively. The avian community was comprised of species characteristic of early and late shrub stages, with no grassland species remaining and no intrusion of forest species (Karr 1968). Discriminant function analysis demonstrated that species- specific nest-site selection occurred within the oldfield avian commun- ity based on coarse habitat features, although overlap among species was considerable. Total vegetation cover and tree cover were the most 46 47 important discriminating variables, with shrub cover and nesting height of lesser importance. Nest-site selection within the community showed two strategies, small and large shrub nesters. The large-shrub nesters (cardinal, cat- bird, black-billed cuckoo, yellow-billed cuckoo, brown thrasher, and blue jay) preferred shrubs and small trees higher than 2.5 m. No spe- cies-specific preferences were evidenced by the birds within the group of preferred shrubs. The small-shrub nesters (field sparrow and indigo bunting) preferred shrubs lower than 2.5 0. Both species utilized most of the same shrubs, but some nest-site segregation was evident, in con- trast to the large-shrub nesters. The field sparrow tended to place early nests in herbaceous cover (ground nests), while later nests were almost entirely in small shrubs (see Nolan 1963, Walkinshaw 1968, Best 1978). The indigo bunting never nested on the ground, but frequently nested in 33333_where the field sparrow was not found. The rufous-sided towhee nested both on the ground and in shrubs, a pattern characteristic for the species (Davis 1960, Nolan 1963, Dickinson 1964), and appeared to be intermediate between the two groups. The size of shrub selected was shown to be strongly correlated with the size of the nesting bird, as was also evident from Rich (1980). Presumably, the large-shrub selectors were unable to utilize the small shrub resource. The small-shrub selectors were capable of nesting in larger shrubs, as was demonstrated by a few field sparrows, but seldom did so. Remarkably little use was made of herbaceous nesting cover, indicating a change in strategy from the grassland situation. Converse- 1y, analysis of nesting heights showed that 89% of all nests were below 2 m, which suggests that the birds were not inclined to nest any higher 48 than was necessary to secure a sturdy substrate. Preston (1946) and Preston and Norris (1947) found most nests were located on or near the ground in both wooded and more open areas. Circumstances leading to an edge effect may be explained by a simple model based upon Hilden's (1965) proximate and ultimate habitat selection factors. Shrubs for nest sites and foraging, "openness of habitat", and high singing perches are important habitat selection cri- teria for oldfield birds (Nolan 1963, Harrison 1975, Hickey 1975:99, Welty 1975:213). In my study, shrub cover was shown to be uniformly distributed throughout the oldfield, tree cover was open, and singing perches were available everywhere. Nest distribution at the community level was shown to be random throughout the oldfield. The edge effect noted at field-woodlot junctions by Lay (1938), Johnston (1947), Gates and Gysel (1978), and Strelke and Dickson (1980), can be explained by noting that the junction was the only place where the habitat selection factors were simultaneously available, as was hy- pothesized by Gates and Gysel (1978). Under such conditions, the edge effect existed for only a short distance (10-20 m Gates and Gysel 1978, 25 m Strelke and Dickson 1980), and it was observed by Gates and Gysel (1978) that while oldfield birds nested in the woodlot, the reverse was not true. A similar edge effect was associated with fence rows in agri- cultural fields (Shalaway 1979). Neither the grassland-woodlot junction nor the fence row is visual- ly similar to the mosaic of herbaceous shrub-tree cover types found in advanced oldfields, but apparently, the combination of habitat selection factors is enough to exceed Hilden's (1965) "threshold for settling" through the mechanism of stimulus summation. Furthermore, the extremely 49 high nesting densities observed by Gates and Gysel (1978) and Shalaway (1979) seem to indicate that the stark contrast of the arrangement of the factors acts as a super normal stimulus (Manning 1967), which over- rides the negative stimuli of already present territorial birds until much higher than normal nesting densities are achieved. The low density of nests observed in my transition zone areas is further evidence supporting the habitat selection model. Transition zones (forest edges) may actually be marginal habitat for oldfield avian species, considering the striking decrease of preferred nesting shrubs observed both there and in the woodlot interior. Common woodland shrubs such as witch-hazel (Hamamelis virginiana) and flowering dogwood (Cornus florida) have a much less compact crown than preferred nesting species. The edge effect in Gates and Gysel's (1978) study may well have been much less pronounced were it not for the plantings of multi- flora rose and tartarian honeysuckle along their woodlot edges. Nesting Success Nests fledging at least one young totaled 55.1% for the avian com- munity, which compares closely with Nice's (1957) average of 49.3% for open nesting altricial species. Predation was the cause of failure for 93.2% of the unsuccessful nests. This was high compared to other studies: 45% Young (1949), 88% Nolan (1963), 75% Shalaway (1979), 79% Best and Stauffer (1980), but not surprising in view of the small losses to cowbird parasitism, desertion, and natural disasters. As was the case with nest distribution, nesting success was found to be unrelated to both distance from the woodlot edge and vegetation type diversity. Community reproduction was thus independent of the edge-related mortality demonstrated earlier in succession (Gates and 50 Gysel 1978). Furthermore, no relationship was observed between nest- ing success and the vegetation variables: shrub cover, tree cover, to- tal vegetation cover, and foliage height diversity, which suggests that the predator community functioned equally well throughout the diverse oldfield vegetation. Nest height and nesting date each accounted for about 4% of the total variation in community nesting success. Reduction in clutch size, rather than increased predation, was responsible for the decline in nesting success over time. Consequently, the only factor affecting predation rate was nest height. A community-wide preference for eleva- ted nests was noted, and this appeared to be a selective response to the low success rate of ground nests, which was only half that observed for the small and large-shrub categories. Nolan (1963), Best and Stauffer (1980), and this study showed that predation rate varied lit- tle with height for elevated nests. The field sparrow and rufous-sided towhee were paradoxical in their behavior of using both ground and elevated nest sites, in view of the low success rate of the former. The best explanation seems to be that these species begin nesting earlier than other species, and utilize re- sidual cover until deciduous plants have leafed out (Nolan 1963, Walkin- shaw 1968). Best (1978) noted, however, that the transition was not entirely synchronous. The field sparrow and towhee were among the ear- liest to nest on my areas, but comparison of mean nesting dates showed no significant differences at the community level. Larger sample Sizes might show a staggered pattern of community nesting (Young 1949), or perhaps, the early ground nests play a more important part in the pOpu- lation biology of the species nearer the centers of their ranges. My 51 data showed that most of the May nests, about half of the June nests, and almost none of the July and August nests were on the ground, which also indicated a lag in moving to shrub sites. Considering the selective advantage for elevated nests, why small- shrub nesters failed to utilize large-shrub sites is intriguing. An interesting model involves the relative availability of the respective resources, and the effect on predator-prey relations. Bowman and Harris (1980) showed that foraging efficiency of raccoons on artificial ground nests decreased as spatial heterogeneity increased. From personal ob- servation, the number of suitable nesting shrubs on my areas decreased as size class increased. Thus, environmental heterogeneity may be con- strued to decrease with height. Under this model, a selective advantage would accrue from nesting in the most heterogeneous environment, i.e., the small shrubs. If predation functions in a density-dependent manner (Fretwell l972:128,l32, Gates and Gysel 1978, Best 1978), a decrease in nest density following the decreasing availability of nest sites and de- creasing environmental heterogeneity would be expected with height; and nesting success should be independent of height. Figure 7 of this study and Preston and Norris (1947) Show that nest density does de- crease with height, and as hypothesized, the success rates for small and large-shrub nests were similar. Both my study and Rich (1980) show that community nest distribu- tion is random in the horizontal dimension in "natural habitats". From the above, it would seem that distribution approaches randomness in the vertical dimension, as well. Random nest placement would seem the best method to combat density-dependent predation. The near-random 52 nesting success in my study supports the hypothesis of density-depen- dent nest predation. The advantage in success rate shown for elevated nests over ground nests follows from Best and Stauffer's (1980) obser- vation that predation by large mammals decreases with nest height, while predation by the bird-snake-small mammal group is constant. Friday's (1978) study of mammalian activity along an oldfield- transition zone-forest continuum took place in habitat structurally similar to mine, thus providing an insightful comparison. He found that sciurid activity (chipmunks, red and fox squirrels) was highest in the transition zone, as was activity of the white-footed mouse (Peromygcus leucopus). Sciurids are commonly implicated as nest pred- ators (Nolan 1963, Thompson and Nolan 1973, Nolan 1978:415, Best and Stauffer 1980). Considering all species together (opossum, raccoon, striped skunk, red fox, woodchuck (Marmota monax), cottontail (Sylvi- 1agus floridanus), whitetail deer (Odocoileus virginianus) and the above, activity was highest in the transition zone, and generally low- est in the oldfield. It appears that the oldfield would be a safer place to nest than the transition zone under conditions similar to my study. This correlated well with my finding that nest density was significantly lower in the transition zone than in the oldfield. The Mechanism of Edge Effect It was evident from my study that shrub-sere birds were closely tied to shrubby nesting substrates. It was shown that when these re- sources were universally available, nest distribution and nesting suc- cess were random at the community level. The edge effect observed by Gates and Gysel (1978) resulted from the superimposition of shrub-sere birds (75% of all species recorded) upon herbaceous-sere and 53 forest-sere avian communities along the woodlot edge, which was the only place shrubby nest sites were available. From these observations, along with demonstrations that ordinations of avian species occur along gradients of plant succession (Johnston and Odum 1956, Karr 1968, Rick- lefs 1973:596), an hypothesis of the mechanism of edge effect may be formulated. It is hypothesized that an edge effect will occur when two seral stages, normally far apart along a successional gradient, become sharply juxtaposed. Birds characteristic of the intervening seral stages respond with a supernormal settling response, because of the ac- centuated proximate habitat features. The edge effect lasts until the younger seral stage develops into the intermediate stage, thus allowing the intermediate species of birds to disperse. The mechanism(s) resulting in decreased nesting success in edge habitat seems to be more complex. Several factors, acting singly or jointly, may be operant: (l) Edges such as the oldfield-forest junction (Bider 1968) and fence rows (Wegner and Merriam 1979) may function as travel corridors, thus concentrating predator activity. (2) Edge areas may be superior habitat for some potential predators (Friday 1978), which would also concentrate predator activity. (3) Edge areas may attract cowbirds, thus increasing losses due to parasitism (Best 1978, Gates and Gysel 1978). (4) Increased mortality at the edge may be a simple density-dependent function. Formation and retention of a "search image" by predators could result in higher nest losses, even if predator densities were not higher than usual (Fretwell 1972:128,132). (5) Clutch sizes may be smaller near the edge because of a density-de- pendent physiological effect (Gates and Gysel 1978). Decreased nesting success is not a universal characteristic of 54 edge habitat, as was shown by Shalaway's (1979) study of fence rows. The high nesting success rate he observed was attributed to the absence of chipmunks, ground squirrels, and arboreal snakes. This implies that the size of adjoining habitats determines species utilization to some degree (also see Moore and Hooper 1975, Galli et a1. 1976, Got- tfried 1979). It appears that avian nesting species may be more flex- ible in this regard than their potential predators. LITERATURE CITED LITERATURE CITED Andresen, J.W. and J. McCormick. 1962. An evaluation of devices for estimation of tree cover. Broteria 31:15-30. Balda, R.P. 1975. Vegetation structure and breeding bird diversity. Pages 59-80 33_Proceedings: A symposium on management of forest and range habitats for nongame birds. U.S. For. Serv. Gen. Tech. Rep. WO-l. Best, L.B. 1977. Territory quality and mating success in the field sparrow (Spizella pusilla). Condor 79:192-204. . 1978. Field sparrow reproductive success and nesting ecology. Auk 95:9-22. , and D.F. Stauffer. 1980. Factors affecting nesting success in riparian bird communities. Condor 82:149-158. Bider, J.R. 1968. Animal activity in uncontrolled terrestrial com- munities as determined by a sand transect technique. Ecol. Monogr. 38:269-308. Bowman, B.G. and L.D. Harris. 1980. Effect of Spatial heterogeneity on ground-nest depredation. J. Wildl. Manage. 44:806-813. Brower, J.E. and J.H. Zar. 1977. Field and laboratory methods for general ecology. Wm. C. Brown Co. Publ., Dubuque. 194 pp. Burger, G.V. 1973. Practical wildlife management. Winchester Press, New York. 218 pp. Canfield, R.H. 1941. Application of the line interception method in sampling range vegetation. J. For. 39:388-394. Carothers, S.W., R.R. Johnson, and S.W. Aitchison. 1974. Population structure and social organization of southwestern riparian birds. Amer. Zool. 14:97-108. Davis, J. 1960. Nesting behavior of the rufous-sided towhee in coastal California. Condor 62:434-456. Dickinson, J.C., Jr. 1964. Rufous-Sided towhee: eastern U.S. Sub- species. Pages 562-577 33 Austin, O.L., Jr., ed. 1964. Life histories of North American cardinals, grosbeaks, buntings, towhees, finches, sparrows, and allies. Part 1. Dover Publ., 55 56 Inc., New York. 602 pp. Dow, D.D. 1969. Home range and habitat of the cardinal in peripheral and central populations. Can. J. 2001. 47:103-114. Evans, E.W. 1978. Nesting responses of field sparrows (Spizella pusilla) to plant succession on a Michigan old field. Condor 80:34-40. Evans, F.C. 1964. The food of vesper, field, and chipping sparrows nesting in an abandoned field in southeastern Michigan. Am. Midl. Nat. 72:57-75. Fretwell, S.W. 1972. Populations in a seasonal environment. Prince- ton Univ. Press, Princeton. 217 pp. Friday, G.P. 1978. Vegetative structure and mammalian utilization of a forest-field transition and its adjacent habitats. Doctoral dissertation. ‘Mich. State Univ., E. Lansing. 63 pp. Galli, A.E., C.F. Leck, and R.T.T. Forman. 1976. Avian distribution patterns in forest islands of different sizes in central New Jersey. Auk 93:356-364. Gates, J.E. and L.W. Gysel. 1978. Avian nest dispersion and fledging success in field-forest ecotones. Ecology 59:871-883. , and J.A. Mosher. 1981. A functional approach to estimating habitat edge width for birds. Am. Midl. Nat. 105:189-192. Giles, R.H., Jr. 1978. Wildlife management. W.H. Freeman and Co., San Francisco. 416 pp. Gill, J.L. 1978. Design and analysis of experiments in the animal and medical sciences, Vol. 1. Iowa State Univ. Press, Ames. 409 pp. Goldsmith, F.B. and C.M. Harrison. 1976. Description and analysis of vegetation. Pages 85-155 33_Chapman, S.B., ed. 1976. Methods in plant ecology. John Wiley & Sons, New York. 536 pp. Gottfried, B.M. 1979. Small mammal populations in woodlot islands. Am. Midl. Nat. 102:105-112. ' Harris, R.J. 1975. A primer of multivariate statistics. Academic Press, New York. 332 pp. Harrison, H.H. 1975. A field guide to birds' nests. Houghton Mifflin Co., Boston. 257 pp. Hickey, J.J. 1975. A guide to bird watching. Dover Publications, Inc., New York. 252 pp. Hilden, 0. 1965. Habitat selection in birds. Ann. 2001. Fenrl- 2153’75- 57 Holgate, P. 1965. Some new tests of randomness. J. Ecol. 53:261-266. Holmes, R.T., R.E. Bonney, Jr., and S.W. Pacala. 1979. Guild structure of the Hubbard Brook bird community: a multivariate approach. Ecology 60:512-520. Johnston, D.W. and E.P. Odum. 1956. Breeding bird populations in re- lation to plant succession on the Piedmont of Georgia. Ecology 37 2 50-620 Johnston, V.R. 1947. Breeding birds of the forest edge in Illinois. Condor 49:45-53. Karr, J.R. 1968. Habitat and avian diversity on strip-mined land in east-central Illinois. Condor 70:348-357. Lay, D.W. 1938. How valuable are woodland clearings to birdlife? Wilson Bull. 50:254-256. Leopold, A. 1933. Game management. Charles Scribner's Sons, New York. 481 pp. Linduska, J.P. 1950. Ecology and land-use relationships of small mam- mals on a Michigan farm. Lansing, Game Division, Dept. Conserv. 144 pp. MacArthur, R.H. and J.W. MacArthur. 1961. On bird species diversity. Ecology 42:594-598. , ,.and J. Preer. 1962. On bird species diversity. II. Pre- diction of bird census from measurements. Am. Nat. 96:167-174. Manning, A. 1967. An introduction to animal behavior. Addison-Wesley Publ. Co., Reading. 208 pp. Moore, N.W. and M.D. Hooper. 1975. On the number of bird species in British woods. Biol. Conserv. 8:239-250. Neter, J. and W. Wasserman. 1974. Applied linear statistical models. Richard D. Irwin, Inc., Homewood. 842 pp. Nice, M.M. 1957. Nesting success in altricial birds. Auk 74:305-321. Nie, N.H., 0.H. Hull, J.G. Jenkins, K. Steinbrenner, and D.H. Bent. 1975. Statistical package for the social sciences. McGraw-Hill Book Co., New York. 675 pp. Nolan, V., Jr. 1963. Reproductive success of birds in a deciduous scrub habitat. Ecology 44:305-313. . 1978. Ecology and behavior of the prairie warbler (Dendroica discolor). Ornithol. Monogr. 26:1-595. Odum, E.P. 1950. Bird populations of the Highlands (North Carolina) 58 Plateau in relation to plant succession and avian invasion. Ecol- ogy 31:587-605. . 1971. Fundamentals of ecology. W.B. Saunders Co., Philadelpha. 574 pp. Patten, D.R. 1975. A diversity index for quantifying habitat "edge". Wildl. Soc. Bull. 3:171-173. Petrides, G.A. 1975. Principal foods versus preferred foods and their relation to stocking rates and range condition. Biol. Conserv. 7:161-169. Pettingill, O.S., Jr. 1967. A laboratory and field manual of ornithol- ogy. Burgess Publ. Co., Minneapolis. 381 pp. . 1977. Observed acts of predation on birds in northern lower ‘Michigan. Living Bird 16:33-41. Pregitzer, R.E. 1978. Soil survey of Clinton County, Michigan. U.S. D.A., S.C.S. in coop. with Mich. Agr. Exp. Sta. 89 pp. Preston, F.W. 1946. Nest heights of birds building in shrubs. Ecol- ogy 27:87-91. , and R.T. Norris. 1947. Nesting heights of breeding birds. Ecology 28:241-273. Ranney, J.W. 1977. Forest island edges - Their structure, development, and importance to regional forest ecosystem dynamics. EDFB/IBP- 77/1. Oak Ridge Natl. Lab., Oak Ridge, Tenn. 56 pp. Recher, H.F. 1969. Bird species diversity and habitat diversity in Australia and North America. Am. Nat. 103:75-80. Rich, T. 1980. Nest placement in sage thrashers, sage sparrows, and Brewer's sparrows. Wilson Bull. 92:362-368. Ricklefs, R.E. 1969. An analysis of nesting mortality in birds. Smithsonian Contributions to Zoology 9. '---. 1973. Ecology. Chiron Press, Portland. 861 pp. Robbins, 0.8., B. Bruun, and H.S. Zim. 1966. Birds of North America. Golden Press, New York. 340 pp. Robertson, R.J. and N.J. Flood. 1980. Effects of recreational use of shorelines on breeding bird populations. Canadian Field-Naturalist 94:131-138. Roth, R.R. 1976. Spatial heterogeneity and bird species diversity. Ecology 57:773-782. Schuerholz, G. 1974. Quantitative evaluation of edge from aerial 59 photographs. J. Wildl. Manage. 38:913-920. Shalaway, S.D. 1979. Breeding bird abundance and distribution in fence row habitat. Doctoral dissertation. Mich. State Univ., E. Lansing. 63 pp. Strelke, W.K. and J.G. Dickson. 1980. Effect of forest clearcut edge on breeding birds in east Texas. J. Wildl. Manage. 44:559-567. Taylor, M.W. 1977. A comparison of three edge indexes. Wildl. Soc. Bull. 5:192-193. Thomas, J.W., ed. 1979. Wildlife habitats in managed forests. USDA For. Serv. Agric. Handbook No. 553, Washington, D.C. 512 pp. Thompson, C.F. and V. Nolan, Jr. 1973. Population biology of the yel- low-breasted chat (Icteria virens L.) in southern Indiana. Ecol. Monogr. 43:145-171. Threlkeld, G.W. and J.E. Feenstra. 1974. Soil survey of Shiawassee County, Michigan. USDA, SCS in coop. with the Mich. Agr. Exp. Sta. 113 pp. Walkinshaw, L.H. 1945. Field sparrow, 39-54015. Bird-Banding 16:1-14. 1968. (Spizella pusilla pusilla, Wilson) eastern field sparrow. Pages 1217-1235 33DAustin, O.L., Jr., ed. 1968. Life histories of North American cardinals, grosbeaks, buntings, towhees, finches, sparrows, and allies. Part 2. Dover Publ., Inc., New York. 647 pp. Wegner, J.F. and G. Merriam. 1979. Movements by birds and small mam- mals between a wood and adjoining farmland habitats. J. Appl. Ecol. 16:349-357. Welty, G.W. 1975. The life of birds. W.B. Saunders Co., Philadelphia. 623 pp. Wesolowsky, 6.0. 1976. Multiple regression and analysis of variance. John Wiley & Sons, New York. 292 pp. Whittaker, R.H. 1975. Communities and ecosystems. MacMillan Publ. Co., Inc., New York. 387 pp. Willson, M.F. 1974. Avian community organization and habitat structure. Ecology 55:1017-1029. Young, H. 1949. A comparative study of nesting birds in a five-acre park. Wilson Bull. 61:36-47. Zimmerman, J.L. 1963. A nesting study of the catbird in southern Mich- igan. Jack-Pine Warbler 41:142-160.