Return to METRIA 12

Urban Plant Ecology: A Horticultural Perspective.
Initial Results From a Long Term Ecological Research (LTER) Site

Chris A. Martin, Linda B. Stabler, Sarah B. Celestian, and Jean C. Stutz

Mailing Address:
Urban Horticultural Ecology Research Lab
Department of Plant Biology
Arizona State University
PO Box 871601, Tempe, AZ 85287-1601

A paper from the Proceedings of the 12th Metropolitan Tree Improvement Alliance Conference (METRIA 12), Landscape Plant Symposium: Plant Development And Utilization, held in Asheville, NC, May 23-25, 2002, co-sponsored by the North Carolina State University, North Carolina Division of Forest Resources, USDA Forest Service Southern Region, North Carolina Landscape Association, North Carolina Association of Nurserymen, The Landscape Plant Development Center, The North American Branch of The Maple Society, and The International Ornamental Crabapple Society.


Growing interest in urban ecology has given horticultural scientists a unique opportunity to study plant processes and interactions in an environment that is foreign to most ecologists. Classic ecological pedagogy depicts exotic ornamental plant species as invaders, and human manipulations of the geosurface as disturbance events. Since 1998, we have been studying net primary production (NPP) in an arid, urban ecosystem from a horticultural perspective, one that considers landscape plant communities as part of a system that has been carefully designed and is intensively managed to deliver ecological services. In conjunction with the Central Arizona Phoenix (CAP) LTER, we are engaged in multi-scalar, long-term research on PP, which presupposes that urban systems are an interacting triad of plants, people, and the physical environment. Our research projects have ranged from studies of landscape plant function and plant/microbe interactions to use of remote sensing to relate urban vegetation density to land use and microclimate. We have found that the principal driver of urban NPP at this LTER site is irrigation of landscape vegetation, and that carbon acquisition potential of irrigated landscape plants, per unit of canopy cover, was 2.8 times higher than that of surrounding Sonoran Desert vegetation. Irrigation of landscape vegetation also has a demonstrable effect on city microclimates which are most pronounced in Phoenix during early morning hours in summer in areas having the highest vegetation density. We have also found that human landscape preferences in Phoenix are shaped by a desire for the color green. Also, regionally common landscape maintenance practices such as the frequent pruning of desert adapted landscape trees and shrubs, which are intended to restrain plant size and promote plant canopy density, reduce plant water use efficiency. Though air in Phoenix is carbon enriched, built surfaces have a significant negative impact on landscape tree size (as much as an 87% reduction in tree canopy volume), apparently largely due to increased rhizosphere temperatures. Managed landscape vegetation also changes the composition of arbuscular mycorrhizal fungal communities in a manner that is spatially and temporally specific. An increased understanding of ecological processes and interactions of urban plants might lead to design and management strategies that maximize benefits associated with plants in cities and improve human well being and quality of life.


According to the United States Census, Maricopa County added more residents than any other county in the USA between 1990 and 1996 (U.S. Census) with land development occurring at a rate of over 1 acre per hour (Morrison, 1995). Prior to 1970, urban residential and commercial development in Phoenix occurred mostly on agricultural lands. More recently, urban sprawl of the Phoenix metropolitan area has encroached onto large tracts of Sonoran Desert lands previously undisturbed by agricultural activities. The abundance, diversity, and distribution of managed vegetation in Phoenix differ greatly from that of the surrounding desert because human preferences and anthropogenic activities dictate the community structure of urban and suburban landscape plantings (Peterson & Martin, 2000). Historically, the character of urban vegetation in Phoenix has been that of an oasis in the desert, and has been typified by landscape greenery. Prior to the advent of air conditioning in the early 1960s, amenity landscapes exemplified by large shade trees, lush vegetation and grassy lawns were commonly planted. Since the late 1980s, public policy has shifted towards advocating water conservation in urban landscapes because of concerns about the distribution, abundance, and quality of fresh water resources.

Since 1998, we have been studying net primary production (NPP) in the greater Phoenix metropolitan area from a horticultural perspective, one that considers urban landscape plants as part of a ecosystem that has been carefully designed and is intensively managed to serve human interests. In conjunction with the Central Arizona Phoenix (CAP) LTER (Hope, 2002), we are engaged in long-term study of plant ecology in an arid environment through use of remote sensing, extensive monitoring of NPP at uncontrolled sites across the city, and controlled experimentation at research plots. This paper presents an overview of studies undertaken since 1998. Data shown here gives an overview of ways that that humans have modified a former desert and agrarian habitat to create an urban forest in an arid environment.

Research Overview

Phoenix, AZ, is located in the northeast quadrant of the Sonoran Desert at an approximate elevation of 1000 feet. Mean annual precipitation in this region is 180 mm with approximately 50% occurring as late summer thunderstorms, the remainder normally associated with winter frontal systems originating in the Pacific Ocean. Our research activities are in response to compelling questions about human-plant interactions in an urban setting.

What are people's preferences for and perceptions of vegetation in a desert city?

During 1999, mail surveys were distributed to 1800 homes in residential communities throughout the Phoenix metropolitan area to gather information from homeowners about their landscape design and plant preferences and landscape maintenance practices. The survey also gathered background information about socioeconomic and other factors that influence plant-people interactions (Peterson and Martin, 2000). The survey response rate was 52%. We found that 70% of respondents preferred landscapes designs with turf while only 30% of respondants preferred a native desert or xeric landscape design without turf. Contrary to our expectactions, the average number of years respondents had lived in Arizona was least for those homeowners who preferred xeric landscape designs. In regards to landscape maintenance, 60% of homeowners found plants growing in their natural, unpruned form to be more appealing than those sheared into geometric shapes. Ninety percent of respondents reported using fertilizers, while only 57% reported use of pesticides. The most important characteristics of landscape plants cited by respondents were ease of maintenance (64%), flowering (38%) and low water use (26%). Most respondents maintained their own landscapes; only 29% used professional landscape maintenance contractors. Less than 1% of respondents reported use of no supplemental irrigation; the most common (49%) scenario was use of a combination of drip and overhead spray irrigation. Ninety per cent of respondents felt that water conservation was an important issue in Phoenix despite a majority preference for green landscapes designs.

Does human land use affect the spatial distribution of vegetation in a desert city?

Several studies have been undertaken to examine the spatial distribution of vegetation in the CAP LTER urban ecosystem. One quantitative measure of vegetation cover is use of a normalized differential vegetation index (NDVI), a value calculated from satellite spectral data. We found that land use affected NDVI along five transects in the Phoenix metropolitan area, with highest values in agricultural land uses and lowest in industrial areas (Stabler and Martin, 2000). Residential landscapes had intermediate NDVI values. We noted a clear difference in the vegetation densities of urban Phoenix and the surrounding desert. In a survey of 200 randomly selected sites across the CAP LTER study region, tree and shrub canopy area was highest in single family residential plots, averaging 131.8 m2, and lowest in transportation, vacant, and construction areas, with an average of 24.0, 21.2 and 0m2, respectively (Hope, et al. 2000). For sites within the Phoenix urban area, 55% of land cover was pervious surfaces such as turf (16%), soil (25%), and decomposing granite mulch (14%). The remainder of the urban area was impervious, comprised of asphalt (20%), concrete (10%), buildings (12%) and other surfaces (3%).

Is NPP of normally irrigated urban Phoenix vegetation seasonally different from that of vegetation in surrounding irrigated agricultural and non-irrigated desert lands?

During 1998-99, we monitored plant gas exchange and water status at eighteen uncontrolled sites in the Phoenix metropolitan area. Measurements of monthly mid day and seasonal diurnal patterns of photosynthesis were used to calculate the carbon acquisition potential of plants in agricultural, residential, and remnant desert sites (Martin and Stabler, in press). We found that maximum carbon acquisition potential to be 423.6, 1191.6, and 1907.0 Kg C/m2leaf surface area/year for desert, residential, and agricultural land uses, respectively. This equates to 2.8 times greater uptake of atmospheric carbon for the suburban residential lands than for desert lands on a leaf surface area basis. On a monthly basis, differences were most pronounced during June, a very dry month in Phoenix, and least pronounced during the August monsoon season. Water was the factor that most limited NPP.

Is water use efficiency (WUE) of urban vegetation affected by human management practices?

During May 1999, we established an experiemntal reseach station consisting of 14-10 x 10 m controlled simulated landscape plots, each planted with 12 shrubs, two trees and one prostrate ground cover, based on typical plant distributions in local residential landscapes in the Phoenix area. At these plots we test how human landscaping practices such as irrigation and pruning affect plant productivity and water use efficiency (WUE). We also installed an automated micrometeorlogical station that continuously monitors environmental factors such as air and soil temperature, relative humidity, and soil moisture content at this experimental site. A graph of soil moisture content as a function of irrigation volume and precipitation shows how important irrigation is to plant water availability in the CAP ecosystem. We have found that landscaping practices affect plant productivity and water use efficiency, defined as biomass produced per unit of water applied. Frequent shearing of two common landscape shrubs reduced plant WUE by as much as 59% relative to unpruned controls. In a glasshouse study, we also found that increased irrigation frequency reduced water use efficiency of containerized ornamental landscape shrubs and trees (Stabler and Martin, 1999). One goal of this ongoing study is to redefine WUE as applicable to urban amenity landscape plants. We intend to incorporate factors such as plant aesthetic value, maintenance requirements, as well as more traditional measures such as biomass accumulation and CO2 uptake into our definition of urban plant water use efficiency.

To what extent is NPP negatively impacted by proximity to built surfaces such as asphalt and concrete?

Landscape trees in commercial parking lots provide shade as well as enhance aesthetic value. However, planting location might compromise tree growth and function. In 2001, we studied the effect of commercial parking lots on the size of established landscape trees (Brachychiton populneus, Fraxinus velutina, Pinus canariensis, Pinus halapensis, Prosopis alba, and Ulmus parvifolia) in Phoenix, AZ, USA (Celestian and Martin, 2002). Data was collected for trees in medians of 15 commercial parking lots and of trees growing in adjacent perimeter planter beds. For all tree species, mean canopy volume, height, and trunk diameter at breast height were 73, 36 and 41%, respectively, compared with trees in adjacent perimeter planter beds. Size of Pinus halapensis and Ulmus parvifolia was most negatively affected by median planting location, while size of Brachychiton populneus and Prosopis alba was least affected by median planting location. For all trees, an infrared thermometer was used to measure summer mid-day surface temperatures of common surface types found in commercial parking lots (asphalt, concrete, turf, mulched surface). Temperatures of asphalt surfaces around medians were up to 63°F higher than surface temperatures of vegetated and non-vegetated surfaces in adjacent perimeter planter beds.

Viewing urban landscape installation as a disturbance analog, how are patterns of arbuscular mycorrhizal fungal diversity and carbon uptake potential of urban trees affected by this disturbance?

Reseach was undertaken beginning in 1997 to determine the effect of time lapsed since landscape installation on arbuscular mycorrhizal (AM) fungal diversity (Stutz and Martin, 1998). Soil samples were collected under the canopies of common landscape Fraxinus uhdei (Shamel ash) or Fraxinus velutina (Arizona ash) trees along a temporal transect (time since landscape establishment) ranging from 3 to 45 since the time of landscape installation. These sampling sites were arrayed along a transect starting in the urban core of Tempe, AZ and extending south to the greater Phoenix metropolitan urban fringe in Chandler, AZ. Trap cultures were established for each soil sample to detect non-sporulating species. Spores were extracted from soils samples and trap cultures to determine AM fungal species richness and composition. We found a gradual increase in the number of AM fungal species occurred over time in urban landscapes with species richness matching or surpassing that previously reported for sampling sites in the surrounding Sonoran Desert. Four species were detected in recently established landscapes in comparison to 12 species detected in landscapes established over 45 years ago. All AM fungal species, with the exception Glomus macrocarpum, were small (#150 Fm) and many were hyaline or white. Most of the species detected, including an undescribed species, have been previously detected in the Sonoran Desert surrounding the greater Phoenix area. In another study of carbon dynamcis and AM fungal function we found that landscape trees in residential landscape plots were not as highly colonized by AM fungi as were trees in adjacent remnant desert sites (Stabler et al, 2000). Finally, results from another glass house study suggest that tree carbon storage might be increased nearly ten fold my AM fungal colonization relative to uncolonized controls.


Based on these and other data, we have concluded that NPP in the CAP ecosystem is closely linked, and likely principally regulated by human activities. Human land use largely determines the distribution of vegetation in Phoenix, and human preferences dictate species composition. Environmental factors that severely limit NPP in the surrounding desert, such as low soil moisture and nutrient content, are overcome by human irrigation and fertilization practices in the city. Other human activities, when viewed from an ecological perspective, represent disturbance events and might limit productivity via changes in soil microbial populations. Management practices such as pruning decrease water use efficiency, but may have a role in enhancing plant appearance by increasing plant visual density and controlling plant shape, important aspects of the ecosystem services provided by landscape vegetation to urban inhabitants.

Literature Cited

Return to METRIA 12

Web Crafter: Anne S. Napier ~ Email:

Format updated January 19, 2010