Andrew A. Millward B.Sc. (Env.), M.Sc., Ph.D. (Geography)
Dr. Andrew Millward is an Associate Professor of Geography and the principal investigator for the Urban Forest Research & Ecological Disturbance (UFRED) Group at Ryerson University. He joined the Department of Geography at Ryerson in 2006 and established UFRED in 2008. UFRED's primary focus is applied research that uses geospatial technology to document and explain urban forest presence and condition, with the end goal of furthering protection and enhancement of natural spaces within cities. Dr. Millward instructs undergraduate courses in physical geography, quantitative methods and environmental decision-making. At the graduate level, he teaches a course dedicated to the design and construction of interactive internet-based maps detailing Toronto's urban parks and treed landscapes. Dr. Millward has served as the Vice and Interim President of the Board of Directors for the Environmental not-for-profit Local Enhancement and Appreciation of Forests (LEAF), a Toronto-based organization dedicated to the protection and improvement of the urban forest. He is a graduate of Environmental Science and Geography from the University of Guelph, and holds a Ph.D. in Geography from University of Waterloo. After receiving his doctorate, Dr. Millward worked in the United States for three years where he held a Fulbright Fellowship in the Natural Resources Department at Cornell University and a visiting professorship at the George Washington University in Washington, DC.
Master's student creates first detailed inventory of university's 800 trees, 'an urban oasis' on campus
By Dana Yates
Aaron Pothier, left, with Andrew Millward in front of a little-leaf linden tree - an excellent species to provide shade in urban areas. Pothier, a graduate student, is creating a detailed inventory of the university's 800 trees as part of his master's thesis. Millward is his thesis advisor.
Read Full Article
In his popular children's book, The Giving Tree, the late author Shel Silverstein wrote of the many ways that trees provide for humans. Today at Ryerson, a graduate student is working to quantify those contributions.
As part of his master's thesis, spatial analysis student Aaron Pothier, geographic analysis '05, is meticulously studying the 800-plus trees on Ryerson's campus. It is the first detailed inventory of its kind to be conducted at the university.
Naturalization as a strategy for improving soil physical characteristics in a forested urban park
Millward, A.A., Paudel K., Briggs S.E. (2011) Naturalization as a strategy for improving soil physical characteristics in a forested urban park. Urban Ecosystems 14:261-278.
Ensuring good soil quality is essential to promoting plant growth in urban parks and building ecological resilience into a cityscape. Periodically used to restore a degraded urban ecosystem, parkland naturalization is a management approach designed to facilitate the return of an area to a natural state by largely discontinuing maintenance activities and restricting public access. This study investigates the potential for parkland naturalization to improve soil quality in a forested portion of Kew Gardens Park, Toronto, Canada by comparing soil physical properties in three six-year-old naturalization enclosures with those found in adjacent managed parkland. Soil texture, compaction, bulk density, and surface water infiltration rate were measured at 104 sample sites, while ordinary kriging was used to interpolate two-dimensional prediction surfaces representative of soil properties. Sand and loamy sand were the dominant soil texture classes found across the study site. Highly compacted soil (>2,000 kPa) and soil with a bulk density >1.8 Mg/m3 (values sufficiently elevated to restrict tree root growth and respiration and impair soil water infiltration) were spatially correlated with high pedestrian traffic areas and corridors used by festival and park maintenance vehicles. In contrast, compaction and bulk density measurements in the naturalized areas were at or below thresholds known to impair root function (10 and 30 cm depth: 849 and 1,311 kPa, 1.15 and 1.51 Mg/m3, respectively). Similarly, water infiltration rates were rapid (470 mm=hr) within the naturalization enclosures but retarded to the point of surface pooling in parkland subjected to regular public use. In the absence of quantitative baseline data, our use of spatial analysis demonstrates that parkland naturalization is a good management practice for restoration of soil physical characteristics. While our results show improvements to soil properties in a relatively short period of time, variability in the soil response rate to parkland naturalization will be dependent upon disturbance history as well as on soil and climate type.
Benefits of a forested urban park: what is the value of Allan Gardens to the city of Toronto, Canada?
Millward, A.A., Sabir, S. (2011) Benefits of a forested urban park: what is the value of Allan Gardens to the city of Toronto, Canada? Landscape and Urban Planning 100:177-188.
Treed urban parks provide numerous social, environmental, and economic services of measurable value to a city. To better understand the importance of a forested urban park we employ the Street Tree Resource Assessment Tool for Urban Forest Managers (STRATUM) to investigate the value of services provided by trees in Allan Gardens, a historic public park in downtown Toronto, Canada. A full inventory, conducted in 2008, found 309 trees representing 43 species. Park trees provided 26,326USD in annual benefits ($16,665 environmental; $9661 aesthetic) during 2008, and delivered a benefit-to-cost ratio of 3.4:1. Tree size and leaf area are larger in Allan Gardens when compared with trees growing in other Toronto parks and across the city as a whole. The flow of benefits from Allan Gardens’ urban forest is heavily dependent upon Norway Maple (Acer platanoides), a finding mirrored across much of Toronto’s urban forest. Norway Maple provides the greatest overall annual benefits ($4846 total; $113 per tree) and as a species contributes 17.5% of the environmental and 20% of aesthetic value provided by trees in the park. This work offers a model to urban planners, providing a straightforward methodology for quantifying the value of nature in public city spaces, in the form of treed parks.
Who is likely to plant a tree? The use of public socio-demographic data to characterize client participants in a private urban forestation program
Greene, C.S., Millward, A.A., Ceh, B. (2011) Who is likely to plant a tree? The use of public socio-demographic data to characterize client participants in a private urban forestation program. Urban Forestry & Urban Greening 10:29-38.
City trees, and the ecosystems of which they are a part, provide important benefits to urban residents. In many cities across North America, suitable locations for the planting of trees – expansion of the urban forest – are mostly confined to privately owned land. Our primary motivation for conducting this study was to investigate whether aggregate socio-demographic characteristics, represented geographically by census tract, have explanatory value concerning participation in a large urban forestation program. Specifically, we used 2006 Statistics Canada census data and known geographic locations of participants in a privately administered urban forestation program to conduct a two-stage multiple regression analysis for East York, Etobicoke, Markham, North York, Scarborough, Toronto, and York (all densely populated centres within the Greater Toronto Area of southern Ontario, Canada). A priori assumptions about program participants were evaluated first based on a review of the literature and through solicitation of expert opinion. The second step employed an exploratory data analysis approach to identify variables that may have differed from a priori assumptions. Results indicate that there are marked regional differences in both the a priori assumptions, as well as in the variables identified through the exploratory regression analysis. The explanatory ability of the baseline regression model is strongest for East York and weakest for Markham, whereas the ability to explain program participation using the exploratory regression model is strongest for Markham and weakest for North York. While participation of Toronto and York residents is largely explained by a dwelling-specific variable (the number of homes constructed pre-1946), the participation of Markham residents is typified by a gender-specific income variable (the number of females reporting a gross income range of $50 to <$60k). Beyond provision of location-specific client information, our study presents a methodological framework that is of value to the refinement of current forestation efforts and to future target marketing of similar initiatives.
Urbanisation viewed through a geostatistical lens applied to remote-sensing data
Millward, A.A. (2011) Urbanisation viewed through a geostatistical lens applied to remote-sensing data. Area 43:53-66.
The purpose of this study is to investigate the usefulness of variography for landscape change detection when applied to a time series of unclassified remote-sensing data. Specifically, the challenge was to identify and describe land-cover change, the result of rapid urbanisation, across a 12-year chronology of satellite images for which little temporally specific ground information was available. Using semivariograms, and the remote sensing technique of band-overlay for visual reference, the change in spatial extent of land-cover type, as well as feature richness (variance in reflectance values), was determined for Landsat and SPOT imagery obtained for the Sanya Region of Hainan, China in 1987, 1991, 1997 and
1999. Comparison of results with a traditional post-classification change trajectory confirms that timeseries semivariograms are instructive at identifying general changes to land cover resulting from urbanisation. They are complementary of traditional post-classification approaches where sufficient in-situ and time-specific data exist; where these data are absent, the semivariogram approach to change analysis is recommended as a precursory tool for monitoring land-cover change.
Structure of a forested urban park: implications for strategic management
Millward, A.A., Sabir, S. (2010) Structure of a forested urban park: implications for strategic management. Journal of Environmental Management 91:2215-2224.
Informed management of urban parks can provide optimal conditions for tree establishment and growth and thus maximize the ecological and aesthetic benefits that trees provide. This study assesses the structure, and its implications for function, of the urban forest in Allan Gardens, a 6.1 ha downtown park in the City of Toronto, Canada, using the Street Tree Resource Analysis Tool for Urban Forest Managers (STRATUM). Our goal is to present a framework for collection and analysis of baseline data that can inform a management strategy that would serve to protect and enhance this significant natural asset. We found that Allan Garden’s tree population, while species rich (43), is dominated by maple (Acer spp.) (48% of all park trees), making it reliant on very few species for the majority of its ecological and aesthetic benefits and raising disease and pest-related concerns. Age profiles (using size as a proxy) showed a dominance of older trees with an inadequate number of individuals in the young to early middle age cohort necessary for short- to medium-term replacement. Because leaf area represents the single-most important contributor to urban tree benefits modelling, we calculated it separately for every park tree, using hemispheric photography, to document current canopy condition. These empirical measurements were lower than estimates produced by STRATUM, especially when trees were in decline and lacked full canopies, highlighting the importance of individual tree condition in determining leaf area and hence overall forest benefits. Stewardship of natural spaces within cities demands access to accurate and timely resource-specific data. Our work provides an uncomplicated approach to the acquisition and interpretation of these data in the context of a forested urban park.
Ice storm disturbance greater along terrestrial-aquatic interface in forested landscapes
Millward, A.A., Kraft, C.E., Warren, D.R. (2010) Ice storm disturbance greater along terrestrial-aquatic interface in forested landscapes. Ecosystems 13:249–260.
Ice storms are an important and recurring ecological disturbance in many temperate forest ecosystems. In 1998, a severe ice storm damaged over ten million hectares of forest across northern New York State, eastern Canada, and New England impacting ecosystem processes across the landscape. This study investigated the spatial arrangement of forest damage at the terrestrial-aquatic interface, an ecological edge of importance to aquatic habitat and nutrient cycling. Vegetation indices, derived from satellite imagery and field-based data, were used to measure forest canopy damage across a 2045 km2 region in northern New York State affected by the 1998 storm. We investigated the forest damage gradient in the riparian zone of 13 stream segments of varying size (92.5 km total length) and 13 lakes (37.4 km of shoreline). Large streams (fourth and fifth order), occurring in forests that received modest ice damage (<15% disturbance coverage), exhibited significantly more damage in the riparian zone within 25 m of the water than in adjacent forest sections; F(3,12) = 7.3 P = 0.005. In similar moderately damaged forests, lake shorelines were significantly more damaged than interior forests; F(3,9) = 6.4 P = 0.013. Analysis of transitions in damage intensity revealed that canopy disturbance followed a decreasing trend (up to 3.5 times less) with movement inland from the terrestrial-aquatic interface. The observed predisposition of forest to disturbance along this ecosystem interface emphasizes the role of the physical landscape in concentrating the movement of wood from the forest canopy to locations proximate to water bodies, thus reinforcing findings that ice storms are drivers of ecological processes that are spatially concentrated.
Time-series analysis of medium-resolution, multisensory data for identifying landscape change
Millward, A.A., Piwowar, J.M., Howarth, P.J. (2006) Time-series analysis of medium-resolution, multisensory data for identifying landscape change. Photogrammetric Engineering and Remote Sensing 72:653-663.
The overall goal of this study is to use medium-resolution satellite imagery to determine recent changes in the landscape of the coastal zone near Sanya in the Province of Hainan, China. A search for suitable satellite imagery revealed that the
only way to identify the changes was to use data from three different sensors acquired over a 12-year time period: a 1987 Landsat 5 Thematic Mapper (TM) image, a 1999 Landsat 7 Enhanced Thematic Mapper Plus (ETM+) image, and two SPOT 2 High Resolution Visible (HRV) images acquired in 1991 and 1997. Given that the Landsat and SPOT images have different spatial resolutions and that the spectral bands cover somewhat different spectral ranges, the challenge was how to combine the images in digital format to be able to detect subtle changes in the landscape. Measures of brightness, greenness, and the normalized difference vegetation index
(NDVI) were explored using standardized principal components analysis (PCA). Approximately 38 percent of the scene was occupied by water, so tests were performed with the water included and also with the water masked out to remove these low-variance pixels. Factor loadings and input-band contributions were used to interpret component images. Results show that PCA of the visible bands, representing brightness, is the superior approach for identifying new urban features in the landscape. For identification of changes to vegetation, the near-infrared (NIR) bands outperformed NDVI. Selected standardized PCA images with visible and NIR bands are recommended for identifying general changes to an urban landscape using a time-series of imagery acquired by different satellite sensors. Benefits of using a mask are believed to be dependent upon study-site characteristics.
Physical influences of landscape on a large-extent ecological disturbance: the northeastern North American ice storm of 1998
Millward, A.A., Kraft, C.E. (2004) Physical influences of landscape on a large-extent ecological disturbance: the northeastern North American ice storm of 1998. Landscape Ecology 19:99-111.
The 1998 ice storm was a large-extent ecological disturbance that severely affected the eastern Adirondack forests of northern New York. Ice damage produced widespread breakage of limbs and trunks in susceptible trees. Although ice storms are common within northeastern North American forests, the magnitude and extent of the 1998 storm far exceeded damage caused by typical ice storms in the recent past. While plot and stand-scale ecological impacts of ice storms have received attention insofar as tree species vulnerability, stand age susceptibility, and microhabitat alterations, larger-extent damage patterns have not been previously evaluated. The normalized difference vegetation index (NDVI) was employed to assess forest vigor and canopy density in atmospherically corrected Landsat Thematic Mapper (TM) satellite imagery of the Adirondacks. Digital change analysis of the baseline forest condition (1990 NDVI data), and the condition encountered in a post-storm image (1998 NDVI data) was conducted. Forest damage was separated from natural variations in canopy reflectance by employing a generalized linear model that incorporated in situ measurements. A robust empirical variogram analysis revealed that locations of tree damage were significantly correlated for distances up to 300 meters. Intensely-damaged forest exhibited greater spatial dependence, but over a smaller distance. Canopy damage was not greater proximate to stream and forest boundaries, and did not follow our hypothesis of decreasing damage with distance from the boundary. Overall, we show that local topography (elevation and aspect), forest composition (deciduous or coniferous), and the meteorological characteristics of the disturbance event acted together to determine the spatial extent of ice storm damage.
Realizing the potential of GIS in community-based management of protected areas
Mersey, J.E., Millward, A.A., Martinez-R, L.M. (2002) Realizing the potential of GIS in community-based management of protected areas. International Journal of Sustainable Development and World Ecology 9:208-222.
This paper explores the roles GIS technology can play in support of land management, focusing in particular on its application in protected environments such as Biosphere Reserves. These ecologically significant areas present complex planning situations since, unlike areas set aside solely for conservation, they must continue to support human populations that depend on natural resources for their economic and social well being. In situations such as marginal tropical locations, land managers often face major economic and political disincentives to the conservation of precious natural resources. The necessity for community participation and local expertise in the planning process becomes increasingly critical as GIS evolves from a basic presentation aid for spatial data to a synthesizing problem-solving tool. Coupled with enhanced community participation is the opportunity for greater education of stakeholders; the potential for GIS to serve as a medium for information gathering and dissemination is discussed. A case study in the Sierra de Manantlan Biosphere Reserve (SMBR), Mexico, is presented.
A.A. Conservation strategies for effective land management of protected areas using an erosion prediction information system (EPIS)
Millward, A.A., Mersey, J.E. (2001) A.A. Conservation strategies for effective land management of protected areas using an erosion prediction information system (EPIS). Journal of Environmental Management 61:329-343.
This research demonstrates the predictive modeling capabilities of a geographic information system (GIS)-based soil erosion potential model to assess the effects of implementing land use change within a tropical watershed. The Revised Universal Soil Loss Equation (RUSLE) was integrated with a GIS to produce an Erosion Prediction Information System (EPIS) and modified to reflect conditions found in the mountainous tropics. Research was conducted in the Zenzontla subcatchment of the Rio Ayuquila, located within the Sierra de Manantlan Biosphere Reserve (SMBR), Mexico. Expanding agricultural activities within this area will accentuate the already high rate of soil erosion and resultant sediment loading occurring in the Rio Ayuquila. Two land-use change scenarios are modeled with the EPIS: (1) implementation of soil conservation practices in erosion prone locations; and (2) selection of sites for agricultural expansion which minimize potential soil loss. Confronted with limited financial resources and the necessity for expedient action, managers of the SMBR can draw upon the predictive capacity of the EPIS to facilitate rapid and informed land-use planning decisions.
Adapting the RUSLE to model soil erosion potential in a mountainous tropical watershed
Millward, A.A. Mersey, J.E. (1999) Adapting the RUSLE to model soil erosion potential in a mountainous tropical watershed. Catena 38:109-129.
This research integrates the Revised Universal Soil Loss Equation (RUSLE) with a Geographic Information System (GIS) to model erosion potential for soil conservation planning within the Sierra de Manantlan Biosphere Reserve (SMBR), Mexico. Mountainous topography and a tropical uni-modal precipitation regime characterize this region. These unique climatic and topographic characteristics required a modification of the standard RUSLE factors and their derivation. The resulting RUSLE–GIS model provides a robust soil conservation planning tool readily transferable and accessible to other land managers in similar environments. Future pressure to expand agriculture and grazing operations within the SMBR will unquestionably accentuate the already high rate of soil erosion and resultant sediment loading of watercourses occurring in this region. Until recently there did not exist a reliable or financially viable means to model and map soil erosion within large remote areas. An increase in the reliability and resolution of remote sensing techniques, modifications and advancements in watershed scale soil erosion modelling techniques, and advances in GIS, represent significantly improved tools that can be applied to both monitoring and modelling the effects of land use on soil erosion potential. Data used in this study to generate the RUSLE variables include a Landsat Thematic Mapper image land cover, digitized topographic and soil maps, and tabular precipitation data. Soil erosion potential was modelled within Zenzontla, a sub-catchment of the Rio Ayuquila, located in the SMBR, and the results are presented as geo-referenced maps for each of the wet and dry precipitation seasons. These maps confirm that high and extreme areas of soil loss occur within the Zenzontla sub-catchment, and that erosion potential differs significantly between wet and dry seasons.
A spatio-temporal index for heat vulnerability assessment
Kershaw, S.E., Millward, A.A. (In Press) A spatio-temporal index for heat vulnerability assessment. Environmental Monitoring and Assessment. DOI: 10.1007/s10661-011-2502-z
The public health consequences of extreme heat events are felt most intensely in metropolitan areas where population density is high and the presence of the urban heat island phenomenon exacerbates the potential for prolonged exposure. This research develops an approach to map potential heat stress on humans by combining temperature and relative humidity into an index of apparent temperature. We use ordinary kriging to generate hourly prediction maps describing apparent temperature across the Greater
Toronto Area, Canada. Meteorological data were obtained from 65 locations for 6 days in 2008 when extreme heat alerts were issued for the City of Toronto. Apparent temperature and exposure duration were integrated in a single metric, humidex degree hours (HDH), and mapped. The results show a significant difference in apparent temperature between built and natural locations from 3PM to 7AM; this discrepancy was greatest at 12AM where built locations had a mean of 2.8 index values larger, t(71)05.379, p <0.001. Spatial trends in exposure to heat stress (apparent temperature, ≥30°C) show the downtown core of the City of Toronto and much of Mississauga (west of Toronto) as likely to experience hazardous levels of prolonged heat and humidity (HDH ≥72) during a heat alert. We recommend that public health officials use apparent temperature and exposure duration to develop spatially explicit heat vulnerability assessment tools; HDH is one approach that unites these risk factors into a single metric.