Status of Breeding and Wintering Birds
in Petrified Forest National Park, Arizona
Alison Banks, Charles van Riper III, Steven S. Rosenstock(1)
     10 November 2002
     USGS Southwest Biological Science Center
     Colorado Plateau Field Station
     PO Box 5614
     Northern Arizona University, Flagstaff, AZ 86011

            1 Arizona Game and Fish Dept., 2221 W. Greenway Rd., Phoenix, AZ 85023

RESULTS
Study areas.
Vegetation measurements generally corroborated our habitat designations (Table 2). Multivariate ordination of ground cover types and numbers of trees (Fig. 2) indicated that the two grass transects (BLUE and RAIN) were very similar to one another (ANOSIM R=.1124), as were the two shrub transects (RT66 and BOPI; R=.0435). In addition, the dry wash (DRYW), grass-dune-shrub mosaic (FLAT), and dune habitats were not distinguishable from each other or from the grass and shrub transects. An ordination based on shrub species composition yielded similar results, with only the riparian and juniper woodlands as distinct habitat types. Shrub heights varied significantly among habitats (ANOVA F=18.177, p<.001), with shrubs significantly taller at the Puerco River site (Bonferroni post-hoc test p<.001); otherwise average shrub height did not vary among transects (Bonferroni post-hoc test; p>.10). A list of all species of woody vegetation found along transects is provided in Appendix VI.

At Pintado Point, we found 1 to 25 tall juniper or cliffrose shrubs per survey point; along Puerco River, the number of cottonwoods varied from 0 to 186 per point (Appendix VII). In riparian habitat at Puerco River, canopy cover varied from 0 to 27%, indicating a very open forest structure (Appendix VIII). Cottonwood was the dominant canopy type at four points, tamarisk at two points; willow was co-dominant with tamarisk or cottonwood at two points. Tamarisk and willow averaged 2-3 m in height even when dominant, while cottonwoods averaged 4-10 m tall.
Bird surveys.
Historical records. We found no published papers regarding birds in Petrified Forest. Several unpublished reports and checklists were found in the National Park Service bibliographic database (Appendix IX). Our museum search yielded responses from 31 museums indicating only two museums with specimens collected from Petrified Forest (Appendix X). Six specimens are held in the Museum of Southwestern Biology, University of New Mexico, and 131 specimens representing 68 species are contained within the Petrified Forest National Park Museum (Appendix XI). Natural History Field Observation Cards of bird sightings totaled 900 observations of 187 species (summarized in Appendix XII).

Bird checklist. Our revised park checklist includes 234 species of birds, 31 more than the 1984 list. One newly listed species, the Eastern Meadowlark (Sturnella vulgaris), was found breeding in the park. Of the newly identified avian species, 24 are year-round residents, 24 reside in the park only during breeding season, while 13 are only present during winter. Thirty-two species have been confirmed to breed in the park and 7 more are probable breeders. Forty-three species are rare in the park, having been sighted only once or twice within the past 40 years. Two species appear to have been extirpated within the park. The Yellow-billed Cuckoo bred for at least several years at Puerco River during the 1970's, and was last recorded in the park in 1984. The Canyon Wren was listed on the 1959, 1963, and 1984 lists as a resident, but there are no historical records the species in the park since 1966.

Field survey results. In 1998 and 1999 we counted a total of 2,812 birds, representing 51 species. Horned Larks were the most frequently detected bird (49% of all detections). The next most numerous birds were Black-throated Sparrows, Rock Wrens, and Meadowlarks, comprising 9%, 7%, and 5% of all detections, respectively. Relative abundances of species surveyed in the park are provided in Table 3. We detected additional bird species on all subsequent surveys in 1998 and 1999 (Fig. 3). Some species were detected in one year and not the other, so that the total number of species seen over the 2-years was substantially higher (Fig. 4).
Method differences. We present numbers of birds and species detected by our two survey types in Fig 5. If we consider the numbers of birds detected per hour of effort, more individuals (t=6.381, p<0.001) and more species (t=5.80, p=<0.001) were found using strip-transect surveys. However, if we consider only the time spent counting birds, (excluding time walking among points), more birds (t=3.766, p=.004) and more species (t=-4.806, p<.001) were counted per hour with point-counts

Seasonal differences. Abundance and species composition of birds varied greatly between winter and breeding seasons. Only 12% of all bird observations were made during winter, and abundances within species were reduced. For instance, Horned Larks were less than half as abundant in winter when compared to summer. In addition, only 18 species were detected in winter, in contrast to 43 during the breeding season. Lists of all birds seen in each season are provided in Appendices XIII and XIV.

The relative abundance of different bird species during breeding season (Table 4) are similar to those for both seasons combined (Table 3). During winter surveys, Horned Larks were also most abundant throughout the park, comprising 50% of detections (Table 5). However, the next most abundant species were Bushtits and Sage Sparrows, two of 7 species found only in winter. Some species found in both seasons were much less abundant in winter (e.g. Meadowlarks, Rock Wren), others were about equally abundant (e.g., Red-shafted Flicker), and only the Scaled Quail showed a slight trend for higher winter abundance.
Habitat differences. The numbers of birds counted varied significantly among habitat types during the breeding season with point counts ( F=3.64, p=.005) and strip-transects (F=3.03; p<.009). The numbers of species counted also varied significantly among habitat types during the breeding season with point counts (F=12.21, p<.001) and with strip-transects (F=13.05; p<.001). More species were found in riparian woodland than all other habitat types, except juniper woodland (Bonferroni post-hoc test p<.05). More birds were detected in grass habitats than in dune, wash or cliff (Bonferroni post-hoc test p<.05 During winter, habitats differed greatly in the number and species of birds using them. Of the 18 species detected during winter surveys, only one was detected in juniper woodland; 15 were present in riparian woodland.

Our ordination results suggest that bird communities are distinct among dune, cliff-side, juniper woodland, riparian woodland, and the other grouped habitats (Fig. 6; ANOSIM R>.268). Bird communities in the grass, shrub, mosaic, and wash habitats were not distinguishable (ANOSIM R<.167). In this grass-shrub habitat group, Horned Larks were most abundant, and number of species varied from 4 to 12 per transect (Table 6). Numbers of species tended to increase with increased numbers of shrubs. In the cliff-side and dune habitats, Rock Wrens were most abundant (Table 7). It should be noted that Rock Wrens rarely used dune vegetation, but rather were abundant in the rocky canyon to the north of the DUNE transect, which we sampled when walking along the crest of the dune. In juniper woodland, Black-throated Sparrows were the most abundant of 13 detected species (Table 8). In the riparian woodland along the Puerco River, we detected 37 species; Cliff Swallows were the most numerous, nesting colonially under the Puerco bridge (Table 9).

We found that the three most abundant species in the park had differing breeding habitat uses. Horned Lark abundance varied significantly among habitat types (F=21.618, p<.001), from no recorded birds in the Pintado Point juniper woodland, to an average of 79 individuals per km2 in the Rainbow Forest grassland. Horned Larks were positively correlated with percentage of grass cover (r=0.788, p=0.007) and negatively correlated with shrub/half-shrub cover (r=-0.747, p=0.013); grass cover and shrub/half-shrub cover were inversely correlated (r=-0.703, p= .023). Black-throated Sparrows also varied among habitats (F=6.181, p<0.001) and were positively correlated with shrub density (r=.860, p=.003). More Black-throated Sparrows were found in shrub and juniper woodland habitats than in grassland or riparian woodland, where they were essentially absent (t=s>1.46; p=s<.014). The abundance of Rock Wrens varied among habitats (F=27.025, p<0.001) and was positively correlated with the amount of open ground (r=0.663, p=0.037). More Rock Wrens were present in cliff-side habitat than in any other habitat (t=s>3.14; p=s<.001).

Mist-netting. We captured 19 individuals of 11 species (Appendix XV), including a MacGillivary’s Warbler, the only one detected in the park in 1998. We found evidence of breeding, in the form of brood patches or cloacal protuberances, for four species, with two (Ash-throated Flycatchers and Blue Grosbeaks) not previously known to have bred in the park.
Breeding evidence. We found nests of 17 species and other evidence of breeding for 7 additional species (Appendix XVI). Of the 24 species for which we detected breeding, 13 were not previously known to have bred in the park.


DISCUSSION
Study area differences.
We found that several habitats in the park differed distinctively in vegetative characteristics. Riparian woodland, as represented at our Puerco River study site, and juniper woodland, sampled near Pintado Point, were distinctively different from shrub and grass habitats in the rest of the park. The greater tree densities of the riparian and juniper woodlands, differentiated these transects. The other habitat types, which we initially identified as grassland, shrubland, dry wash, and dune habitats, were not statistically distinguishable in terms of ground cover types or shrub species composition. The density of shrubs varied continuously among transects, so that while some transects were dominated more by grass, and others by shrubs, there were no clear dividing points among grass or shrub habitats within the park.

Additions to our knowledge of the park’s avifauna.

We documented 13 additional bird species that breed in the park, one being the Eastern Meadowlark (Sturnella magna). In the general environs of Petrified Forest, Eastern Meadowlarks coexist with Western Meadowlarks (Sturnella neglecta), both similar looking species. These two species are best distinguished in the field by song, but identification can be difficult because males may sing songs that are typical of both species. In addition, the two species can form interspecific pairs (Lanyon 1994, 1995). The lack of previous Eastern Meadowlark records could indicate that the species was previously present but not separately identified, or that they recently expanded their range into the park. Data from the Arizona Breeding Bird Atlas suggest that the latter scenario may indeed be the case, and that the species has recently increased in abundance throughout the park. In 1993 few Eastern Meadowlarks were detected relative to Westerns on survey plots bordering the park, while in 1999 Eastern were far more common than Western Meadowlarks on the same plots. Our 1998-1999 formal surveys indicate that at present, nearly three times as many Eastern- as Western Meadowlarks occur in the park. It will be informative to track relative abundances of these two species over time. In addition, the park provides an ideal site for the study of habitat partitioning and hybridization among sibling species.

The Yellow-billed Cuckoo and the Canyon Wren have been previously recorded in the park, but did we record either species during our study, nor have sightings been recorded for the past 15 years. The historic status of the Canyon Wren in the park is not well documented, and historically observed birds may have been vagrants. However, the Yellow-billed Cuckoo bred for numerous years during the 1970's in the Puerco River riparian area, so the disappearance of this species represents a loss of a former resident bird. Yellow-billed Cuckoos have undergone a drastic reduction of range and numbers throughout the Southwest (Layman and Halterman 1987; Rosenberg et al. 1991) and are currently receiving attention for listing as a threatened or endangered species. This illustrates how avian monitoring data collected in the park may be useful in indicating larger-scale changes in bird populations.
Survey method differences.
The strip-transect (vs. variable circle plot) survey method was the most efficient for Petrified Forest habitats, as more birds and species were detected per hour of effort. Higher numbers of bird detections enables one to better estimate bird abundances, and thus facilitates comparisons among habitats or years. Thus, we recommend continuing surveys with the strip-transect method (see Monitoring Recommendations section).

Seasonal differences.
Avian abundance and diversity at Petrified Forest is substantially higher during the breeding season, than in winter. During the breeding season, Petrified Forest National Park hosts well over twice the number of species found during the winter. In addition, Petrified Forest is a popular stop-over site for spring and fall migrant birds. Indeed, the greatest diversity of species that we noted was during the spring and fall migrations, and this is also true for the documented historical records of birds in the park. Thus, in an attempt to detect the greatest number and diversity of birds, we believe that the park should focus its monitoring and management efforts primarily toward the breeding season, and secondarily during the spring and fall migration periods.
Habitat differences.
We found that several habitats in the park differed in the bird communities that they hosted. Birds found in riparian woodland, juniper woodland, cliff-side and dune habitats during breeding season were different from those found in the rest of the grass-shrub habitats. We will discuss each habitat type in turn:

Riparian woodland hosted the greatest diversity of breeding season species in the park (37 species), three times more than in any other park habitat. In winter, riparian woodland habitat was also the most diverse, hosting 83% of the species present in the park. The disproportionate use of riparian woodland by birds has been found by other studies in the Southwest (Carothers 1974; Szaro 1980). In addition, during migration, most birds are seen along the Puerco River, near the housing complexes, or on the sewage ponds. Thus, these watered habitats appear to attract migrating birds, and continued maintenance of this habitat type will significantly affect the diversity of avian species in the park. It should be noted that we sampled two types of wash or riparian habitat in the park: one with cottonwood over story (riparian woodland at Puerco River) and one without an over story (Dry Wash). The avifauna of these two habitat types differed significantly, with Puerco River hosting twice as many birds and three times as many species as Dry Wash. The birds of Dry Wash were not distinguishable from those along other shrub and grass transects.
The juniper woodland also hosted a high diversity of species during the breeding season, including Lark Sparrows, Ash-throated Flycatchers, and Northern Mockingbirds, species that were only otherwise detected at our Puerco River site. This suggests that presence of large shrubs, such as the cliffrose and junipers that are present in the Pintado Point area, account for much of the variation in bird use within the park (see also Rosenstock and van Riper 2001).

Cliff-side habitat was distinctive in regard to bird community composition during the breeding season. Species richness was relatively high (13 species), including nesting species such as American Kestrels, Cliff Swallows, Common Ravens, and Rock Wrens, all who built their nests in the cliff rock face. In addition, other species including Prairie Falcons and Golden Eagles are known to nest in rock faces in other locations within the park. Rock Wrens were the most abundant birds in this habitat type, and they showed a strong habitat association with bare ground and rock. Wiens and Rotenberry (1981) also found a strong association between the abundance of Rock Wrens and rocky outcrops.
The dune habitat was distinctive in bird species composition, mostly due to the large numbers of Rock Wrens. However, this result is somewhat misleading, as the Rock Wrens were detected in a steep arroyo to the north of the transect, rather than in the dune vegetation. This further suggests the distinctiveness of rock outcrops as important habitat for some avian species.

Within the grass-shrub habitats, we found that bird species richness was enhanced with increased shrub cover. In southern Utah grassland habitat Willey (1994) also found a positive association between bird species richness and vertical habitat structure, a combination of the density and number of vegetation layers. Other studies have also demonstrated higher avian species diversity associated with more structurally complex habitats (e.g., Rotenberry and Wiens 1980, Rosenstock and van Riper 2001). While species numbers that we detected was positively associated with shrub cover, increased shrub cover might not benefit all members of the grassland bird community. Black-throated Sparrows were not found in high numbers in areas of high shrub density and low grass cover, and Horned Larks were positively associated with grass cover and negatively associated with shrubs. Willey (1994) also found that Horned Larks and Black-throated Sparrows were positively associated vertical vegetation cover and little bare ground, but he did not distinguish between grass and shrubs. Wiens and Rotenberry (1981) found that Black-throated Sparrows were positively associated with shrub cover. Our findings are consistent with those of other workers who have studied grassland habitats in the western US, all whom have found a high degree of individuality among species in their responses to habitat characteristics. Thus, it may be difficult to generalize management objectives to the entire avian community at Petrified Forest NP, and management actions will more likely need to be made on an individual basis for each target species.
Despite the obvious importance of the woodland and cliff-providing habitats within the park, importance of the prairie habitat should not be overlooked. The grass-shrub habitats actually support higher densities of birds than do the other habitats, largely due to the numbers of Horned Larks. There is reason to believe that high quality grassland habitat of Petrified Forest NP is extremely valuable in terms of conservation, in that grassland birds have been identified on a continental scale as one of the most dramatically declining avian guilds (Peterjohn and Sauer 1999). Thus, for species reliant on grasslands, access to breeding areas that are free of excessive disturbance (such as the grasslands and shrublands found in the park), this habitat is important for the avian short-grass prairie guild.

MONITORING RECOMENDATIONS

1. Continue monitoring avian populations, focusing on the breeding season. To gain the most reliable information on breeding bird abundance in the park, strip-transect surveys should be conducted on a yearly basis. If yearly surveys are not possible, then surveys at increments of 2, 3, or 5 years could provide monitoring data, if accrued regularly over a long enough time period. Surveys should be conducted between mid-May and the end of June, with the best weeks being the last week of May through the third week of June. Each transect should be surveyed three times, allowing about a week between successive surveys. Several days flexibility should be allowed for ‘make-up surveys’ due to bad weather. One or two transects may be surveyed per morning, starting at sunrise. If two counts are conducted, each transect should be surveyed at least once first and once later in the morning, and observer(s) should move as rapidly as possible between surveys to decrease the effects of rising temperature. Observer(s) should be trained in bird survey techniques and familiar with species found in the park. One observer is preferable, but if two observers are utilized, observers should train together, then each person count birds at least once on each transect. One-to-two weeks prior to surveying, observer(s) should locate all points on transects, marking prominent shrubs near points to facilitate rapid location during the surveys. Although not necessary to the protocol, if observer(s) feel comfortable collecting distance information, they should practice with a range-finder prior to the formal count. Observer(s) should follow the survey protocol described in the Methods section, being sure not to count during days with rain, or when winds exceed an average of 8 mph or gust to 10 mph, because aural detections are seriously compromised during periods of high wind.
2. Vegetation along bird survey transects need to be measured in every year in which bird surveys are conducted. However, if this is not possible, other repeated monitoring increments, of 2, 5, or 10 years, would also provide worthwhile habitat monitoring data for the park. A minimum of two vegetation plots should be measured per bird survey point. It should be noted that the vegetation measurement protocol described in this report is quite simple, requires identification only of the shrub species, and is relatively time efficient. Most of the vegetation measurement could be completed by two or three observers within the several weeks of the bird survey season. Because of the high correlation between measures of shrub cover and shrubs counted in an area around the survey line, we recommend that shrubs need not be tallied, unless more specific information regarding shrub composition is desired. Likewise, we found measurements of shrub heights were fairly invariant among sites, and should be conducted only if there is specific interest in shrub heights. Because the riparian woodland at the Puerco River is so dynamic due to flooding, the presence of invasive species, and management efforts including planting of trees, continued measurements of average shrub heights, tree heights, and tree densities would be recommended for this habitat type.

3. Analysis of bird and habitat data should be reduced and analyzed each year following surveys, then collated for comparisons among years, after 5 years of data collection. An effective way to determine if species have changed in abundance is to compare relative abundances of species among years. This is accomplished by calculating the total number of birds counted for each species, then taking the number of the most abundant species (which will undoubtedly be the Horned Lark), and dividing that number into each of the total counts for every other species. For example, if a total count results in 156 birds (100 Horned Larks, 30Meadowlarks, 25 Black-throated Sparrows and one Loggerhead Shrike), the Horned Lark would have a relative index of 1.0 (100 divided by 100). Meadowlarks would have a relative index of .30 (30 divided by 100), Black-throated Sparrows .25 (25 divided by 100) and the Loggerhead Shrike a relative abundance index of .01 (1 divided by 100). Thus, by comparing among years, the park can determine if one species is either stable, in decline or increasing, relative to the general abundances of other species. In addition, if the relative abundances of birds among habitat types varies significantly among years, comparisons should be made back to vegetation measurements to determine if vegetation changes within particular habitats might explain changes in bird numbers.
4. Natural History Observation Cards should continue to be maintained as an effective part of the park’s avian monitoring program. The park should continue to encourage visitors and staff to report bird sightings utilizing Observation Cards, making sure that observers provide details explaining their identification of less common species (e.g., distance at which the bird was seen, duration of observation, and how the species was distinguished from other similar species). All Observation Cards should be input on a regular basis (not less than yearly), into the Avian Observation Database provided with this Technical Report. This database can be easily made available to all researchers and interested visitors through the Resource Management Office.
5. The Petrified Forest National Park bird species checklist should be reviewed every 10 years and updated as necessary. In 16 years since publication of the last PEFO bird list, we were able to contribute substantial revisions, including the addition of 31 species. An updated checklist ensures that visitors receive accurate interpretive information, while providing the park an opportunity to solicit information regarding species that may appear to be changing in numbers.