Relationships Between the Density of Mourning Doves and Long-term Habitat Change Along Call-count Routes in the Western Management Unit

Mourning Dove (Zenaida macrura) populations in the Western Management Unit (WMU) have declined a significant (P<0.01) 2.2% annually during the last 30 years (1966-1995) as determined with annual call-count data from WMU states (Dolton 1995). California (3.5% annually) and Nevada (3.9% annually ) have experienced the greatest declines (P<0.01 and P<0.05, respectively), whereas lesser (P>0.05) declines have occurred in all other WMU states (Dolton 1995). These long-term estimates are conservative, because annual population declines between 1966 and 1987 were more severe (4.9% in California, 6% in Washington and Oregon, and 5.2% in Nevada; all P<0.01) (Reeves et al. 1993), and essentially no change occurred between 1987 and 1995, except a significant (P<0.05) continued decline in Nevada (Dolton 1995).

Harvest and number of hunters have declined along with population indices in the WMU. Changes in hunters and harvest in Nevada and Oregon have been less marked. Some of the decrease in harvest was due to regulation changes, such as shortened seasons and smaller bag limits. However, declines in hunter participation has been important. For example, in California, the number of hunters dropped from nearly 285,000 in 1969 to 135,000 in 1987 and the harvest declined from near 5 million to 1.9 million. In Washington, hunters declined from 30,000 to 8,000 and harvest from 400,000 to 66,000 over the same time period (Reeves et al. 1993).

Several factors have been presented to explain mourning dove population losses in the WMU. Loss of nesting habitat, agricultural changes that degrade habitats and reduce foods important to doves, application of pesticides in orchards and rangelands, eruptive or chronic disease problems (especially trichomoniasis), increased predation, and over-harvest (Subcommittee on Mourning Doves 1992) have been examined, but results have been inconclusive. Based on analyses of banding programs from the 1960s and 1970s, hunting was thought not to be a causative factor in declines of dove populations, at least not with the recruitment rates derived at the time of the banding program (Tomlinson et al. 1988). Preliminary recruitment rates (young fledged per pair) estimated from a 4-year field study of mourning doves nesting in California (M.R. Miller and C.L. Stemler, unpubl. data) were probably adequate to offset mortality if the sites studied were representative of the WMU, and if survival rate estimates derived from the 1960-70s banding data are still applicable. At one site with historical records from three time periods (Gray Lodge Wildlife Area), dove recruitment was highest in 1949-1950 (nest success = 44-56%), declined to 1979-1980 (nest success = 23-25%), and increased to an intermediate level in 1992-1993 (nest success = 30-35%) (all differences among periods significant, P<0.01) (M.R. Miller and C.L. Stemler, unpubl. data). Only in 1980, did the number of fledglings produced per active pair, 2.4, fall below the estimated 2.5 needed to maintain populations in California, again, based on the analyses of banding done in the late 1960s and early 1970s (Tomlinson et al. 1988).

If reduced recruitment rate has not been a determinant of population size of WMU mourning doves, then large scale detrimental changes in habitat and farming practices could have produced the observed population declines. Although mourning doves nest throughout the U.S., certain regions and habitat complexes support more doves than others: "The most favorable areas contain an intermixture of wooded edge habitat used by doves for nesting and farmland that produces grain and weed seeds used for food" (Tomlinson and Dunks 1993).

In the WMU, comprehensive data on changes in dove habitat are not available, but Reeves et al. (1993) solicited observations from several local areas within the Unit. Observers indicated that changes in agricultural practices correlated to declines in local dove densities. Often mentioned practices in California included elimination of tree groves, intensive farming, changes in crop types (not specified), removal of odd areas of brush, and replacement of dirt-lined with concrete-lined irrigation ditches. In southern Oregon, introduction of turbine pesticide sprayers in pear orchards, and in eastern Washington, introduction of dwarf fruit trees and overhead sprinklers in apple orchards were mentioned as potential causative agents for dove declines. In Washington, for example, Zeigler (1977) and Knight et al. (1984) found few dove nests in apple trees less than 15 years old, and the greatest nest success occurred in trees greater than 25 years old that were not overhead irrigated. Tomlinson and Dolton (1987) and Tomlinson et al. (1987) found that declines in dove density on call-count routes in the WMU could not be consistently attributed to changes in cropping patterns; however, strong correlations between dove abundance and sorghum, corn, and cotton acreages showed that changes in agricultural practices could have impacted doves. Some doves in the Great Basin nest under sage brush (Howe and Flake 1989), but large areas of this habitat have been converted to grasslands for cattle grazing and urban development. In California, blue oak woodlands, a major nesting habitat (M.R. Miller and C.L. Stemler, unpubl. data), have been clear-cut for fuel wood, and diversified crops in the San Joaquin Valley, have been converted to monocultures of cotton, reducing weed and agricultural seeds. Tobin and Hothem (1983) reported seeing no doves using large tracts of vineyards in a diversified landscape of grasslands, small orchards, and pastures, in the San Joaquin Valley, doves were plentiful (M.R. Miller and C.L. Stemler, unpubl. data). Citrus trees are heavily used by doves for nesting (Reeves et al. 1993). The above information is useful to formulate hypotheses concerning changes in agricultural practices and their effects on dove populations. However, this does not conclusively explain WMU dove population trends because the observational, as opposed to experimental, nature of the studies precluded cause and effect interpretations. Importantly, habitats that may control dove numbers at one site or within one region, may not have any impact at another site because of the relative value of other habitats (Grue et al. 1983).

The relationship between dove abundance, as measured with call-counts, and crop status was investigated quantitatively in the Eastern Management Unit (EMU) states. Overall, in a state-by-state comparison, researchers found a highly significant positive correlation between dove population size and the proportion of land area in crops (Martin and Sauer 1993). However, most correlations between land area in specific crops and doves heard on call-count routes were not significant (P>0.05). The authors felt that latitude may have been more important in determining abundance of doves along call-routes than were specific crops. Likewise, relationships between: 1) weighted trends in dove abundance and crops acreages; and 2) annual indices of dove abundance and yearly changes were not consistent. The authors emphasized that the analyses relied on the assumption that call-count routes adequately sampled all crop types, but, given the small sample of routes and some crop types, analyses of dove trends and crops at the state level probably had little power to detect correlations, especially considering that weather and latitude may have masked effects of crops (Martin and Sauer 1993).

In Iowa farm groves, density of dove nests declined with removal of a large percentage of elm trees from the groves and complete elimination of other groves (LaPerriere and Haugen 1972). In Missouri, during 1966-1983, increased soybean production in concert with decreased corn production correlated with declining abundance of doves (U.S. Fish and Wildlife Service, Albuquerque, MN, unpubl. data). George et al. (1987) examined the relationship between dove trends and trends in crop acreages, farm size, and number of farms from 1966-85 in Minnesota, Iowa, Missouri, Arkansas, and Texas. They verified the correlation between corn/soybean production for Missouri, but not for other states. Instead, the number of farms correlated positively and the size of farms correlated negatively to dove abundance over time. Thus, they concluded that the trend toward fewer and larger farms, and the loss of fence rows, shelter belts and diversified habitats consistent with that trend, caused reductions in dove populations. In contrast to a chronological examination of habitat changes over time, Grue et al. (1983) examined general habitat characteristics on 133 call-count routes in Texas and correlated dove numbers with the distribution, abundance and structural features of habitats that met the requirements for dove nesting and survival. Their analysis suggests that the habitats provided by most agricultural crops studied were beneficial to doves, especially habitat variable associated with edge, clearings, and the absence of understory in continuous forests (pine).

Similar quantitative work needs to be done in the WMU. The California Department of Fish and Game (D. Yparraguirre, unpubl. data) has completed a pilot effort comparing habitat change within 8km (5miles) either side of call-count route 120 in northern Sacramento Valley. Route 120 was chosen because raw survey data seemed to indicate that doves had declined since 1961. This area consisted of a mix of blue oak woodland, grasslands, riparian habitat, small orchards, ranches, and housing. This analysis compared habitats from 1966, 1979, and 1988 using existing black/white aerial photography. A partial analysis of the route showed a nearly 90% decline in agriculture (open areas showing signs of plowing) (2.43 to 0.32 sq km), a 7% decline in woodlands (blue oak wooded areas of >10% canopy coverage) (73.92 to 68.87 sq km), an 11% increase in grasslands (open areas without signs of plowing) (60.19 to 66.74 sq km), a 13% increase in riparian areas (1.22 to 1.37 sq km), a 56% increase in ponds (0.23 to 0.36 sq km), and a 4,000% increase in orchards (0.01 to 0.34 sq km). Thus, the technique was able to detect small to large changes in a variety of habitats covering wide or narrow portions of the area studied. However, the limited knowledge of complete habitat requirements of nesting mourning doves in California limits interpretation of these results. Also, the use of raw data to show the downward trend in doves heard on route 120 declined in magnitude each time there was an observer change (3 different observers), and the period prior to 1966 contained one extreme outlier that probably biased the trend computed from the raw data. Thus, the declining trend may simply be an artifact of the raw data. Use of the route regression technique, which takes into account changes in observers (Geissler and Sauer 1990), would yield more consistent and reliable results in these types of analyses. Also, the roughly 500 sq km area (16 km wide by 32 km long) may have introduced irrelevant habitat changes into the analyses, because as the distance away from the routes increases, the likelihood of obtaining spurious correlations increases because the probability increases of including habitat changes with no impact on dove numbers along the route.

The difficulty in relating call-count results with actual nesting populations on small areas has long been known (see review by Baskett 1993). This lack of relationship between density of calling doves and nesting effort probably results from the greater calling rate of unmated doves and differences in calling rates of mated males according to stage and nesting cycle (Baskett 1993). Also, of critical importance to previous studies, related to the mobility of doves, and not mentioned previously, is the small size of the area around the listening points within which investigators have searched for nests (Grue et al. 1983, and review by Baskett 1993). Thus, doves nesting outside the searched area probably called from perches within the searched area, and doves nesting within the searched area may have had calling perches outside the searched area but within hearing range of observers, thus, confounding results. A similar problem could exist when trying to determine the relationship between habitat change and the resulting effect on the density of doves along the call-count routes. Intuitively, if habitat controls the nesting population, habitat change should affect the density of doves heard calling on transects bisecting the habitats. This is provided that the extent of area selected for study is consistent with the movements of nesting doves of unmated males during the nesting season. The presence of unmated doves, with their higher calling rate, could mask the loss of nesting habitat.

The distance away from call-count routes selected for inclusion in habitat analysis must encompass the effective distance that doves can be heard, the "radius of audibility." This is about 0.8km (Davey 1953), and was used by Grue et al. (1983) in their analysis. This distance must also enclose the average territory or home range size of nesting doves (Sayre et. 1980). During 2-hour early morning observation periods, Sayre et al. (1980) found that 5 radio-tagged unmated males used cooing perches a maximum of 3.2km apart, with an average range of 0.4 to 1.4 km. Unmated males called and displayed over areas ranging from 54-352 ha, whereas, mated nesting males ranged over areas ranging from 0.4 to 1.3 ha. About 70-80% of all on-territory activity occurred during the 2-hour morning period. This time is consistent with the nation-wide call-count protocol (Dolton 1993). Thus, considering the area used by mated and unmated doves, maximum distance between cooing perches for unmated doves, and the radius of audibility, a reasonable area to include in habitat analyses would be within 1.6 km of call-count routes.

We intend to build on work done by the California Department of Fish and Game by expanding the analysis of habitat change along call-count routes to correlate these changes with changes in dove density in the WMU. We propose to use a GIS analysis of black/white aerial photography and call-count data adjusted with the route-regression procedure. Tomlinson et al. (1994) in, Migratory Shore and Upland Game Bird Management in North America, ranked as the first priority nationwide, the need for this kind of GIS-driven approach to habitat-dove population relationships.

Determine if there is a relationship between long-term changes in dove density and habitat changes along call-count routes in the Western Management Unit.

We will obtain route-regression trends for each mourning dove call-count route in the WMU from the Office of Migratory Bird Management (MBMO), U.S. Fish and Wildlife Service. Using route trend estimates, we will sort routes in order of decreasing slope, and categorize routes into those in which doves have increased, decreased, and not changed between 1966 and 1987, when populations began to stabilize (Dolton 1995). Within the WMU, we will select for continuing analysis the 5 routes in which doves have decreased the most (greatest negative slope), the 5 routes in which doves have increased the most (greatest positive slope), and the 5 routes in which doves have changed the least (slope nearest zero) since 1966 when counts began. We will emphasize selection of California and Pacific coastal state routes, but we will select from all WMU states as necessary. We will attempt to select routes such that we analyze 1 increasing, 1 decreasing, and 1 no-change route in each of the 5 physiographic regions. This will provide nested routes within similar habitats. Successfully meeting this criteria may be difficult because of the constraints imposed by all other criteria. Selected routes must have means of at least 8-10 doves (MBMO data) to increase the likelihood of accounting for biological real, as well as statistically real, information. Additionally, we will examine raw data and eliminate from consideration routes in which zeros (no doves heard) make up more than 20% of all years unless they terminate a long-term decline or precede a long-term increase. We will also reject for analysis routes in which large numbers of doves were heard erratically interspersed among years of zeros (no doves heard), an indication of inclusion of migrant doves or some factor such as annual changes in crop types. In addition, we will select only those routes which have not been relocated during the study periods. Routes with the fewest observer changes over time will be preferred, but we will rely on the route regression data to provide reliable trends by route.

Analysis will consist of identifying and quantifying (hectares) habitat types within 1 mile (1.6 km) either side, and from each end, of each call-count route studied (Grue at al. 1983) for the mid 1960s, mid 1970s, and mid 1980s.The three periods selected for study encompass the time of most severe population decline in the WMU. The area selected along call-count routes includes the radius of audibility and typical home range sizes of nesting doves and unmated males, but will not be so wide as to include habitat changes not relevant to changes in dove density along the route.

We will obtain available 1:20,000, 1:40,000, or 1:60,000 scale black/white aerial photography as appropriate from the Natural Resource Conservation Service (formerly the Agricultural Stabilization and Conservation Service), U.S. Department of Agriculture, Aerial Photography Field Office in Salt Lake City, Utah, for the three study periods. Routes that meet previously described criteria, but which do not have aerial coverage for each study period, will be excluded from our analyses. Other sources of aerial photos will be explored if not available from the Aerial photography Field Office, but costs will likely be markedly greater for photography secured from other sources. Obtaining appropriate aerial photography is expected to consume considerable time and will be subject to the shipping schedules of the source agencies. Photos used will be 10" x 10" (25.4 cm x 25.4 cm) and have a scale of (1" = 1667' (2.54cm = 508m). Sectional enlargements will be obtained, if possible, for smaller scaled photos (>1:20,000). Each call-count route will require 20-40 prints for complete coverage per study period depending upon photo scale. Biologists will identify cover types on each photo and a technician will enclose them in hand-drawn polygons. The technician will the scan the photos, and register them to points derived from features on existing U.S. Geological Survey quadrangles. From these registered aerial photographs, we will digitize and classify each habitat polygon for inclusion in the GIS system at the Dixon Field Station, California Science Center. The GIS will include elevation adjustments as needed, using DEM (digital elevation model) data, to hectares of each habitat type. ARC/INFO will be used to create the GIS data set, and ArcView will be used to analyze these data spatially. In particular, spatial analysis will be used to determine if certain sections of call count routes have experienced more habitat changes than have others.

Habitats to be digitized will include small grain agriculture (wheat/barley, corn, rice, sorghum, etc.), irrigated pasture, other agriculture (cotton, beans, beets, etc.) native grasslands/pastures, oak savannah (oak coverage <20%), oak woodlands (oak coverage >20%), weedy odd areas, riparian, urban (houses, roads, other), orchards (fruits and nuts), vineyards, stockponds and reservoirs, etc. as required for each route. In addition, two indices to farm size and the number of farms will be recorded: 1) the size and number of fields; and 2) the number of farm buildings (homes, barns, equipment sheds, etc.).

Hectares of each digitized habitat type, mean field sizes, and number of farm buildings will be compiled for each of the three study periods for each route. These data will be compared with the dove density trend in each route with linear correlation analysis. Based on the available literature reviewed here, we predict that routes, or portions of routes, with declining dove densities will show, as appropriate: 1) increased field sizes and fewer farm buildings (Grue et al. 1983, George et al. 1987); 2) decreased weedy areas and small grain crops (Tomlinson and Dolton 1987, Tomlinson et al. 1987, Reeves et al. 1993); 3) decreased riparian habitat, tree groves, and shelterbelts (Olson et al. 1991); 4) increased orchard/vinyard sizes (Tobin and Hothem 1983, M.R. Miller and C.L. Stemler, unpubl. data); 5) decreased orchard tree canopy diameter (Zeigler 1977, Knight et al. 1984) (if photography scale allows); 6) decreased area of woodlands, sage, and other brush (Howe and Flake 1989); and 7), if routes have not been relocated, increased urban expansion (Reeves et al. 1993). We predict that routes experiencing increases in doves will show opposite habitat trends, and routes with no change in dove density will show a mixture of habitat changes or none at all. Analysis of the results will provide information on the types of habitat changes that influence dove numbers in the WMU.

Baskett, T.S. 1993. Biological evaluation of the call-count survey. Pages 253-268 in T.S. Baskett, M.W. Sayre, R.E. Tomlinson, and R.E. Mirachi (eds.). Ecology and Management of the Mourning Dove. Wildl. Management Institute, Stackpole Books, Harrisburg, Pennsylvania.

Davey, S.P. 1953. A study of the mourning dove in southern Michigan. M.S. Thesis, Univ. Michigan, Ann Arbor.

Dolton, D.D. 1993. The call-count survey: historic development and current procedures. Pages 233-252 in T.S. Baskett, M.W. Sayre, R.E. Tomlinson, and R.E. Mirachi (eds.). Ecology and Management of the Mourning Dove. Wildl. Manage. Institute, Stackpole Books, Harrisburg, Pennsylvania.

Dolton, D.D. 1995. Mourning dove breeding population status, 1995. U.S. Fish and Wildl. Serv. Laurel, Maryland.

Geissler, P.H., and J.R. Sauer. 1990. Topics in route-regression analysis. Pages 54-57 in J.R. Sauer and S. Droege, (eds.). Survey designs and statistical methods for the estimation of avian population trends. U.S. Fish Wildl. Serv., Biol. Rep. 90(1).

George, R.R., R.E. Tomlinson, D.L. Hallett, S.C. Yaich, W.J. Suchy, and A.H. Berner. 1987. Effects of agricultural changes on mourning dove call-count data in the eastern tier states of the Central Management Unit. Unpubl. Final rept. To the Central Migratory Shore and Upland Game Bird Tech. Comm. Texas Parks and Wildl. Dept., Austin.

Grue, C.E., R.R. Reid, and N.J. Silvy. 1983. Correlation of habitat variables with mourning dove call counts in Texas. J. Wildl. Manage. 47:186-195.

Howe, F.P., and L.D. Flake. 1989. Nesting ecology of moruning doves in a cold desert ecosystem. Wilson Bull. 101:467-472.

Knight, R.L., D.G. Smith, D.M. Gaudet, and A.W. Erickson. 1984. Nesting ecology of mourning doves in fruit orchards in north central Washington. Northwest Sci. 58:230-236.

LaPerriere, A.J., and A.O. Haugen. 1972. Trends of mourning dove populations in Boone County, Iowa farm groves. Iowa State J. Sci. 46:417-424.

Martin, F.W., and J.R. Sauer. 1993. Population characteristics and trends in the Eastern Management Unit. Pages 281-304 in T.S. Baskett, M.W. Sayre, R.E. Tomlinson, and R.E. Mirachi (eds.). Ecology and Management of the Mourning Dove. Wildl. Manage. Institute, Stackpole Books, Harrisburg, Pennsylvania.

Olson, T.E., C.E. Braun, and R.A. Ryder. 1991. Agricultural land use and mourning doves in eastern Colorado: Implications for nesting and production in the Great Plains. Prairie Naturalist 23:1-10.

Reeves, H.M., R.E. Tomlinson, and J.C. Bartonek. 1993. Population characteristics and trends in the Western Management Unit. Pages 341-376 in T.S. Baskett, M.W. Sayre, R.E. Tomlinson, and R.E. Mirachi (eds.). Ecology and Management of the Mourning Dove. Wildl. Manage. Institute, Stackpole Books, Harrisburg, Pennsylvania.

Sayre, M.W., T.S. Baskett, and K.C. Sadler. 1980. Radio-telemetry studies of the mourning dove in Missouri. Terrestrial Ser. No. 9, Missouri Dep. Conserv., Jefferson City.

Subcommittee on Mourning Doves. 1992. Pacific Flyway management plan for the Western Management Unit mourning dove. Western Migratory Upland Game Bird Tech. Comm. [c/o USFWS, MBMO] unpubl. Report. Portland Oregon.

Tobin, M.E., and R.L. Hothem. 1983. Diurnal and seasonal abundance patterns of birds in vineyards. Proc. Bird Control Sem. 9:143-149.

Tomlinson, R.E., and D.D. Dolton. 1987. Current status of the mourning dove in the Western Management Unit. Proc. Ann. Conf. West. Assoc. Fish and Wildl. Agencies 67:119-133.

Tomlinson, R.E., and J.H. Dunks. 1993. Population characteristics and trends in the Central Management Unit. Pages 305-340 in T.S. Baskett, M.W. Sayre, R.E. Tomlinson, and R.E. Mirachi (eds.). Ecology and Management of the Mourning Dove. Wildl. Manage. Institute, Stackpole Books, Harrisburg, Pennsylvania.

Tomlinson, R.E., P. Smith, and L.A. McKibben. 1987. Agricultural practices and their possible effects on mourning dove populations in the Western Management Unit. Unpubl. Rep. To the Western Migratory Upland Game Bird Comm.

Tomlinson, R.E., D.D. Dolton, H.M. Reeves, J.D. Nichols, and L.A. McKibben. 1988. Migration, harvest, and population characteristics of mourning doves banded in the Western Management Unit, 1964-1977. Fish and Wildl. Tech. Rep. 13. U.S. Dept. Interior, Fish and Wildlife Service. Wash., D.C.

Tomlinson, R.E., D.D. Dolton, R.R. George, and R.E. Mirarchi. 1994. Mourning Dove. Pages 1-26 in T.C. Tacha, and C.E. Braun (eds.). Migratory shore and upland game bird management in North America. Int. Assoc. Fish and Wildl. Agencies, Washington, D.C.

Wells, D., B.W. Anderson, and R.D. Ohmart. 1979. Comparative avian use of southwestern citrus orchards and riparian communities. J. Ariz.-Nev. Acad. Sci. 14:52-58.

Zeigler, D.L. 1977. Evaluation of the breeding habitat of the mourning dove in eastern Washington. Unpubl. rep., Washington Game Dept, Olympia.

Final Report September 30, 1998

Publication in refereed science journal, "The Relationship Between Habitat Change and Mourning Dove Density Change Along Call-count Routes in the Western Management Unit."