Identifying western yellow-billed cuckoo breeding habitat with a dual modelling approach
Introduction
The western population of the yellow-billed cuckoo (Coccyzus americanus) (hereafter “cuckoo”) is a species of conservation concern and was recently listed as threatened under the federal Endangered Species Act (U.S. Fish and Wildlife Service (USFWS), 2014). The cuckoo is a Neotropical migrant that breeds in the western portions of the United States, Canada, and Mexico, where it nests in low to moderate elevation (328–6,902 ft; 100–2,104 m) riparian woodlands (Hughes, 1999). The riparian habitats upon which they depend are dynamic and can undergo repeated cycles of establishment, growth, and destruction (Paxton et al., 2007). Anthropogenic alterations of watersheds and riparian systems are degrading and destroying its habitat range-wide, while habitat fragmentation results in small isolated populations of cuckoos that contribute to its vulnerability. The cuckoo is included in several habitat conservation plans, including the Lower Colorado River Multi-Species Conservation Program’s (LCR MSCP, 2004), which aims to minimize and mitigate the effects of projects or actions on cuckoos.
Conservation of the cuckoo requires detailed information about its breeding habitat requirements at multiple spatial scales. Currently, most information indicates the importance of specific, finite, patch-level vegetation-community composition and structure, generally consisting of multi-structured or multi-layered riparian vegetation with substantial canopy cover provided by native riparian trees, particularly willows (Salix spp.) and cottonwoods (Populus spp.; Corman and Wise-Gervais, 2005, Girvetz and Greco, 2009, Johnson et al., 2010). There is evidence that consideration of patch-level habitat variables alone does not adequately describe cuckoo breeding habitat nor the potential effects of landscape-level habitat characteristics and processes (Girvetz and Greco, 2009, Greco, 2013, Johnson, 2009). Habitat features such as vegetation communities boundaries, patch size, patch shape, patch fragmentation and patch connectivity can have different influences when measured at different scales (Bissonnette, 1997; Givetz and Greco, 2009).
Understanding the spatial scales at which organisms respond to environmental factors and processes is especially essential for sustaining species of conservation concern and threatened habitats. Recognizing the need for assessing species’ habitat needs at multiple spatial scales, in conjunction with the availability of remotely sensed measures of vegetation and landscape composition, has prompted ecologists to model habitat associations with hierarchical approaches that incorporate predictor variables measured at multiple spatial scales (Girvetz and Greco, 2009, Meyer and Thuiller, 2006, Saab, 1999, Scott et al., 2002, Wiens, 1989). These approaches often combine variables of vegetation structure or composition that are measured at a relatively fine scale (e.g. <1 ha), and landscape variables (e.g. percent cover within a given area or patch configuration) measured at considerably broader spatial scales (e.g. entire watershed; Meyer and Thuiller, 2006). Comparing the relative strength of predictor variables measured at different spatial scales may be one tool for understanding the appropriate scale of measurement (Holland et al., 2004) and, ultimately, management actions. Through the identification of habitat features at multiple spatial scales, spatially explicit models increase our ecological understanding of a target species while providing site-specific habitat maps (Hatten and Paradzick, 2003, Paxton et al., 2007).
Previous efforts to spatially model cuckoo habitat used maps of vegetation communities obtained from orthorectified aerial photography (hereafter “aerial-photo models”). While aerial-photo models are comparatively easy to interpret, they are expensive to obtain and difficult to accurately reproduce due to classification errors, subjectivity, and the lack of standardized methods (Cherrill and McClean, 1995, Hearn et al., 2011). In contrast, vegetation maps created from Landsat imagery (hereafter “satellite models”) can characterize vegetation features across broad areas with the normalized difference vegetation index [NDVI] with near perfect repeatability (Hatten et al., 2010). However, satellite models usually lack the ability to identify specific plant communities and can be difficult to translate into target habitat conditions for conservation purposes (Paxton et al., 2007). Thus, the goal of the project is an important consideration when selecting imagery for classification purposes. When projecting habitat models across very large geographic areas, Landsat imagery is ideal (Hatten, 2016); when classifying vegetation at species or community levels, higher resolution imagery is preferred (Carter et al., 2009, Xun and Wang, 2015).
The goal of our study was to characterize and map cuckoo breeding habitat along the Lower Colorado River (LCR) with a dual modelling approach, at multiple spatial scales. We used aerial-photo and satellite models to accomplish five objectives; (1) identify factors associated with high-quality habitat, (2) develop species distribution models, (3) map habitat quality throughout the study area, (4) identify areas where habitat restoration will be most effective, and (5) increase our understanding of cuckoo breeding habitat by comparing and contrasting the two modelling techniques. To facilitate the modelling process, we developed a conceptual model of cuckoo breeding habitat that identified habitat features hypothesized to be important (Fig. 1). Specifically, we hypothesized that cuckoo habitat selection is influenced by riparian patch size and fragmentation, vegetation density and structure, and plant community and patch location in relation to land cover (e.g. agriculture, urban development, surface water).
Section snippets
Study area
The study area consists of riparian habitats within the LCR MSCP boundaries, which we refer to as the LCR. It includes areas up to and including the full-pool elevations of Lake Mead, Lake Mohave, Lake Havasu, and the historical floodplain of the Colorado River from Lake Mead to the United States–Mexico Southerly International Boundary, a distance of about 644 river km (400 mi; Fig. 2). Yellow-billed cuckoos were once considered abundant throughout the study area’s riparian floodplain (Grinnell
Yellow-billed cuckoo satellite model
We developed multiple satellite models of yellow-billed cuckoo breeding habitat that were comprised of unique combinations of explanatory variables (Table 1); we present the top two model performers in Table 3. Satellite model 1 contained three variables (all positive coefficients): (1) terrain ruggedness (TRI) broken into four classes of ruggedness (flat, low, moderate, high), (2) heterogeneity in vegetation density in a 480-m radius (ND_SD480), and (3) amount of dense vegetation (NDVI > 0.41)
Discussion
Our dual modelling approach leveraged the advantages of fine- and coarse-scale data extracted from digital ortho-photos or Landsat imagery, respectively. The satellite maps can be produced quickly and efficiently, applied across large areas (e.g. drainages, regions), and produce replicable results. The weakness of the satellite model is its inability to distinguish vegetation communities, something we did achieve with the aerial-photo model. Thus, combining them produced the most interpretable
Conclusion
Spatially explicit models are valuable tools when assessing the quantity and quality of avian breeding habitat. The dual modelling approach that we used provided us with information that individual modelling approaches could not. The aerial-photo model provided detailed vegetation maps that informed us about species-specific plant associations with breeding cuckoos, but acquiring vegetation cover data for such models is expensive, difficult to repeat, and poorly suited for tracking changes over
Acknowledgements
The authors wish to thank: the field crews that conducted yellow-billed cuckoo surveys, in particular Christopher Calvo who also supervised the field work; USGS Southwest Biological Science Center, Colorado Plateau Research Station, and Northern Arizona University for institutional support; Terry Arundel from USGS Southwest Biological Science Center for assisting with GIS data preparation; Teresa Olsen from Bureau of Reclamation for comments on an earlier draft. This work was funded by the LCR
References (62)
- et al.
Rivers, dams, and willow flycatchers: a summary of their science and policy connections
Geomorphology
(2002) - et al.
Modeling the dynamic habitat and breeding population of southwestern willow flycatcher
Ecol. Modell.
(2010) - et al.
The repeatability of vegetation classification and mapping
J. Environ. Manage.
(2011) - et al.
A simulation study of the number of events per variable in logistic regression analysis
J. Clin. Epidemiol.
(1996) - et al.
Vegetation type maps of the lower Colorado River from Davis Dam to the southerly international boundary
U.S. Department of the Interior
(1976) - et al.
Process-based principles for restoring river ecosystems
Bioscience
(2010) - Bio-west Inc., GEO/Graphics, Inc., 2006. 2004 Lower Colorado Region vegetation type mapping, backwaters delineation,...
Scale-sensitive ecological properties: historical context, current meaning
- et al.
Modeling relationships between birds and vegetation structure using airborne LiDAR data: a review with case studies from agricultural and woodland environments
IBIS
(2005) - et al.
Remote sensing and mapping of tamarisk along the Colorado River, USA: a comparative use of summer-acquired hyperion, thematic mapper and QuickBird data
Remote Sens.
(2009)
An investigation of uncertainty in field habitat mapping and the implications for detecting land cover change
Landsc. Ecol.
Arizona Breeding Bird Atlas
Cell-Based Modeling with GRID
Review of the status of the yellow-billed cuckoo in California: sacramento Valley populations
Condor
Multi-scale predictive habitat suitability modeling based on hierarchically delineated patches: an example for yellow-billed cuckoo nesting in riparian forests, California, USA
Landsc. Ecol.
Geographic modeling of temporal variability in habitat quality of the yellow-billed cuckoo on the Sacramento River, miles 196–219, California
Patch change and the shifting mosaic of an endangered bird’s habitat on a large meandering river
River Res. Appl.
The Distribution of the Birds of California
Distribution and habitat use of the yellow-billed cuckoo (Coccyzus americanus occidentalis) on the Sacramento River, California, 1987–90
Masters’ Thesis
Population Status, Site Tenacity, and Habitat Requirements of the Yellow-Billed Cuckoo at the Bill Williams River, Arizona: Summer 1998
A Natural history summary and survey protocol for the western distinct population segment of the yellow-billed cuckoo
U.S. Fish and Wildlife Techniques and Methods
Breeding characteristics of the yellow-billed cuckoos in Arizona. Proceedings of the Calif
Acad. Sci.
A multiscaled model of southwestern willow flycatcher breeding habitat
J. Wildl. Manage.
A satellite model of Southwestern Willow Flycatcher
U.S. Geological Survey Open-File Report 2016–1120
Determining the spatial scale of species’ response to habitat
Bioscience
Applied Logistic Regression
Yellow-billed cuckoo (Coccyzus americanus)
Yellow-billed cuckoo distribution, abundance and habitat use along the lower Colorado and Gila Rivers in La Paz and Yuma Counties, 2005
Report Submitted to Arizona Game and Fish Department Heritage Program
Yellow-billed cuckoo distribution, abundance, and habitat use along the lower Colorado and tributaries, 2006
Annual Report: U.S. Geological Survey Open-File Report 2007–1097
Yellow-billed cuckoo distribution, abundance, and habitat use along the lower Colorado River and its tributaries, 2007
Annual Report: U.S. Geological Survey Open-File Report 2008–1177
Yellow-billed cuckoo distribution and habitat associations in Arizona, 1998–1999
Cited by (9)
A river runs through it: Robust automated mapping of riparian woodlands and land surface phenology across dryland regions
2024, Remote Sensing of EnvironmentA spatial model of streaked horned lark breeding habitat in the Columbia River, USA
2019, Ecological ModellingCitation Excerpt :The utility of this tool would facilitate and enhance the ability of USACE to plan and manage deposition schedules by tracking and predicting habitat suitability while also meeting regulatory obligations. We used a combination of GIS and remote sensing techniques found useful for mapping and monitoring bird habitats across large landscapes (Hatten and Paradzick, 2003; Johnson et al., 2017). We streamlined the process and reduced costly labor intensive geoprocessing steps by utilizing free orthorectified satellite imagery and cloud-based mapping techniques.
Substrate and flow characteristics associated with White Sturgeon recruitment in the Columbia River Basin
2018, HeliyonCitation Excerpt :We used Froude number in our analysis because it can identify riffle/pool habitats (Jowett, 1993) and benthic invertebrate abundance (Jowett et al., 1991), predict fish traits (Lamouroux et al., 2002), and characterize anthropogenic impacts of hydropower operations (Hatten and Batt, 2010). We used a dual-modeling approach when analyzing White Sturgeon spawning habitat because it allowed us to examine our results with multiple lines of evidence (Burnham and Anderson, 2002; Johnson et al., 2017). We created a Mahalanobis distance model (Mahalanobis, 1936) since it has been used to characterize white and green sturgeon habitats in the Lower Columbia and Sacramento rivers, respectively (Hatten and Parsley, 2009; Mora et al., 2009), and logistic regression (Hosmer and Lemeshow, 2000) because of its proven flexibility to predict fish habitat at multiple spatial scales and flows in the Columbia River (Hatten et al., 2009; Tiffan et al., 2002).
Early successional riparian vegetation is important for western Yellow-billed Cuckoo nesting habitat
2021, Restoration Ecology
- 1
Present address: Colorado Plateau Research Station, Northern Arizona University, Box 5614, Flagstaff, AZ 86011, United States.