Site and Soil Characteristics Driving White Oak Decline in the Chicago Region

  • Arboriculture & Urban Forestry (AUF)
  • November 2025,
  • 51
  • (6)
  • 629-656;
  • DOI: https://doi.org/10.48044/jauf.2025.028

Abstract

Background A widespread trend of declining and dying mature oaks has been observed in the Chicago region since 2019. While there have been studies on the stress-physiology of white oaks and the epidemiology of decline diseases, there have not been any studies focusing on the relationship between predisposing abiotic site and soil characteristics and white oak decline in the Chicago metro area. The objective of this study was to determine if site and soil characteristics are significantly related to white oak decline in the Chicago region.

Methods Soil and site characteristics were compared between 66 healthy and declining trees of the white oak group on 32 sites in the summer of 2023. During site visits, deep pedon samples, composite soil samples, fine roots for oomycete detection, site characteristics, tree health assessments, and site management information were collected. DAS-ELISA Phyt tests and soil lab analyses were conducted.

Results Low micronutrient levels, compaction, elevated soil sodium, and poor drainage appear to be the most relevant predisposing abiotic site characteristics for white oak decline through ANOVA and GLMMs. The ELISA Phyt tests were 88% positive for both healthy and declining trees.

Conclusions Results from this study can help arborists and urban foresters improve management and treatment design and implementation for white oaks, improve our understanding of the predisposing factors influencing the health of managed trees, and improve future planting guidelines.

Keywords

Introduction

Oak decline is a problem found throughout the world where oaks are common (Starkey and Oak 1989). Oak decline is not a single disease but instead is a complex of abiotic physiological stressors that make the trees vulnerable to attack by opportunistic pests and pathogens. Diseases such as oak decline are explained with a complimentary disease model: Manion’s (1981) spiral of decline. This decline spiral is triggered by predisposing factors such as unfavorable site and soil characteristics that make trees susceptible to being stressed by inciting factors, such as drought and flooding. Once stressed, individuals are then vulnerable and attractive to biotic stressors such as secondary pests and pathogens that may push individuals toward decline and death. For example, a secondary pest for the white oak group in the upper Midwest is the native two-lined chestnut borer, Agrilus bilineatus, which targets stressed oak trees (Kessler 1989). Opportunistic pathogens include root rots caused by genera of the phylum Oomycota, including Phytophthora, Pythium, and Phytopythium, hereafter collectively referred to as the oomycetes (Brasier 1996; Jung et al. 2000; Romero et al. 2007; De Vita et al. 2013; Jankowiak et al. 2015).

The genus Phytophthora includes species proven to be pathogens in historic oak decline trends in Europe (Brasier 1996; Jung et al. 2000), as well as other tree declines, such as Jarrah dieback in Australia with Phytophthora cinnamomi (Weste and Taylor 1971) and Port-Orford-Cedar dieback with Phytophthora lateralis (Hansen et al. 2000). Due to root disease and vascular damage in the trees by this group and other opportunistic pathogens and pests, symptoms of oak decline include tip dieback, or “stag horning,” and chlorotic, dwarfed, sparse foliage concentrated on the main stem and larger branches (Starkey and Oak 1989). The predisposing factors found to favor Phytophthora spp. in decline spirals include poor drainage (Weste and Taylor 1971), fine soil textures, and soil compaction (Bernardo et al. 1992; Cabral et al. 1992; Moreira et al. 1999). For these studies on mediterranean evergreen oaks, their sensitivity to poor drainage may have further predisposed them to these pathogens. These 3 environmental factors increase water saturation and duration (Saxton et al. 1986; Day and Bassuk 1994), thus reducing oxygen availability to roots and allowing for zoospore chemotaxis in the soil-water matrix (Zentemeyer 1980). This compounding effect of stressors creates a feedback loop that leads to visible decline symptoms and often tree mortality (Manion 1981).

Predisposing factors are growing conditions that are not ideal for the species and thus leave them more susceptible to inciting factors. What constitutes as predisposing factors for a particular tree species depends on its growing preferences, which may vary geographically (Manion 1981; Starkey and Oak 1989). Based on research on similar patterns of white oak decline reported in Europe, potentially important predisposing site and soil factors may include poor drainage, compaction, and fine soil textures.

Recently, property owners and managers in the Chicago region have noticed an increase in decline and mortality in the white oak group, including Quercus alba, Q. macrocarpa, and Q. bicolor. Many of these declining trees are mature, which are highly valued and provide more services than smaller trees; thus, their decline and loss directly impacts Chicago region residents (Nowak et al. 2008). Investigating the relationship between site and soil factors and white oak decline will improve the understanding of the regional pattern. Understanding the factors driving the decline and loss of these valuable oaks can help urban foresters and arborists recommend soil amendments and improve management and restoration strategies.

The objectives of this study are (1) to determine which site and soil characteristics are most strongly associated with declining Q. alba, Q. macrocarpa, and Q. bicolor in the Chicago region, and (2) to conduct a preliminary investigation on the role of oomycetes in the oak decline spiral. Based on similar patterns in Europe, we hypothesized that (1) a combination of predisposing abiotic factors, fine soil textures, compaction, and poor drainage will have a significant relationship to white oak decline, and that (2) the oomycetes are a contributing factor associated with the decline spiral.

Materials and Methods

Study Design

Thirty-two sites were selected throughout the seven county Chicago metropolitan area. They were identified in 2023 through collaborators at The Morton Arboretum, Davey Tree Expert Co., and regional county forest preserve districts. Sites were selected only if both healthy and declining groves of the same species were present but no closer than 2-times their dripline (Figure 1). At each site, at least one healthy and one symptomatic mature tree were selected, leading to a total of 66 trees studied across the region. Trees were categorized as healthy or declining by looking for symptoms of oak decline, which include not only tip dieback but also signs or symptoms of other secondary pests or pathogens, such as D-shaped exit holes, bleeding cankers, and leaf herbivory or necrosis. The “declining” tree condition was assigned if at least 20% of the canopy had tip dieback or heavy epicormic sprouts off larger branches and one other sign or symptom was observed. Otherwise, it was assigned the “healthy” tree condition. This assessment occurred in May and June to ensure that these symptoms were due to at least one year of decline and were not drought stress symptoms of that same summer.

Study site map with each of the counties of the 7 county Chicago metropolitan area outlined.
Figure 1.

Study site map with each of the counties of the 7 county Chicago metropolitan area outlined.

The Chicago region (41.8818, −87.6231) has a temperate and continental climate with 9.5 °C average annual temperatures, 92-cm annual rainfall, and a median growing season of 172 days (Illinois State Climatologist 2007). Soil can differ significantly between trees on the same site, especially due to the heterogeneity of soils in urban areas (Effland and Pouyat 1997; Pickett and Cadenasso 2009). The parent materials of Chicagoland soils are primarily of glacial origin, with the most common being Wisconsinian drift, Wisconsinian outwash, and loess over Wisconsinian drift (Fehrenbacher et al. 1984). Anthropogenic parent materials are also common within the more urbanized areas of these counties. Most of this land was originally prairies, oak woodlands or savannas, or wetlands, before being converted into agricultural or urban land. This leads to some of the most common soil orders being Alfisols and Mollisols in this region (Fehrenbacher et al. 1984). The climate should be relatively similar throughout the region, although the more urban areas will be affected by urban heat islands and therefore have more temperature extremes and less soil moisture, a topic reviewed by Arnfield (2003). Sites were relatively flat, except those with glacial kame landforms, human-altered topography, or ravines.

Field Methods: First Visits

All the following methods involving tree tissues during both site visits were completed following aseptic methods by disinfecting all equipment with 70% alcohol to prevent any potential pathogen spread (Brundrett et al. 1996). The first site visits were from 2023 May 25 to June 30.

These first visits focused on completing a soil profile description underneath each selected tree. A representative deep pedon core was extracted and assessed under each tree with a 7.62-cm diameter bucket auger, following the USDA-NRCS Field book for describing and sampling soils (Schoeneberger et al. 2012). The location for deep soil sampling was selected for being representative of the plot and at a parallel position in the slope profile with the tree. A plot consisted of twice the diameter of the dripline of the tree. Depth to redoximorphic features (RMF), structure, roots, and coarse fragments were assessed in the field. A core sampler was used to collect a bulk density core at the location within the plot with the most observed signs of compaction for later analysis (Blake and Hartge 1986). Signs of compaction and human activity were recorded.

Site conditions were recorded, including slope, slope profile position, aspect, and distances from infrastructure and water bodies. Aspect and slope were measured following the USDA-NRCS procedures (Schoeneberger et al. 2012). A 100-foot tape was used to measure distances from infrastructure and water bodies when possible. Otherwise, distances were measured using GPS. Diameter at breast height (DBH) for each tree was measured with a diameter tape. Observations of signs and symptoms of pests and diseases were recorded to account for seasonal variation between the first and second visit.

Field Methods: Second Visits

The second round of site visits, conducted from 2023 July 17 to August 19, focused on tree health assessments and collecting time-sensitive soil and root samples. We collected 4 composite surface soil samples in each of the 4 cardinal directions, 0 cm to 15 cm deep, excluding turf caps, 2 paces from the base of the tree, using a 6.35-cm diameter push probe to preserve soil structure. Vegetative samples of fine roots were collected from a combination of the composite samples as well as from the base of each tree, at depths of 0 cm to 15 cm. All samples were put in a chilled cooler after completing each tree’s assessment and were then kept continuously refrigerated until temperature-sensitive lab analyses were completed.

Exposure was visually assessed as the percentage of the canopy faces that were exposed to full sun. The percentage of each surface cover type was recorded. Crown base radius was measured with a 100-foot tape from the base of the tree in each cardinal direction and calculated following Moser et al. (2015). Total tree height was measured with a Suunto PM-5/360 PC Clinometer (Suunto, Vantaa, Finland). Any observed signs and symptoms of pests and diseases were recorded, along with the percentage of the trees that each affected.

Two different tree health indices, the Tree Condition Index (TCI) and Urban Tree Health (UTH) index, were used to quantify the quality and health of the trees. The TCI is a modified version of Webster’s (1978) Tree Condition assessment method following Scharenbroch and Catania (2012). Each metric was assigned a score of 0 to 3, following Table 1, and then an average was taken of the 3 scores for each tree. This score was later normalized by dividing it by 3 to give it a consistent scale with the UTH results. This index was included because it includes the overall condition of the tree.

View this table:
Table 1.

Scoring parameters for TCI adapted from Webster (1978) and Scharenbroch and Catania (2012). TCI (Tree Condition Index).

The UTH index was developed by Bond (2012). We used a modified version to only include the parameters applicable and practical to assess for tall, mature oaks that have likely been heavily pruned, especially if declining in urban environments. Following this logic, we removed Ratio (the percentage of the total tree height occupied by live crown) and Growth (the length of average twig extension over the past 3 years based on annual internodes). Our study included Opacity (the percentage of light intercepted by live crown), Vitality (the percentage of crown free from dieback), and Quality (the percentage of crown free of signs and symptoms of pests, diseases, and other stressors). The USFS Crown Density Scale was used to assess Opacity. Actual percentages were recorded to not lose detail. Each of these 3 parameters was scored 0 to 5 following Table 2. The average of the 3 scores was then taken and normalized by dividing by 5 to ensure a consistent scale with TCI results.

View this table:
Table 2.

Parameters for the UTH, adapted from Bond (2012). UTH (Urban Tree Health index).

Lab Analyses

The fine root samples were used for the qualitative detection of defined Phytophthora, Pythium, and Phytopythium spp. using Agdia double antibody sandwich enzyme-linked immunosorbent assay (DAS-ELISA) Phytophthora tests (Agdia Inc., Elkhart, IN, USA). Tests were performed within 3 weeks of samples’ original collection following the Agdia kit guidelines (Agdia, Inc. 2023).

Soil profile descriptions and the following lab analyses were assessed to correlate each tree’s representative soil profile to a soil series (Soil Survey Staff 1999). These soil series were used to provide supplemental information for the soil from the USDA-NRCS Soil Survey, such as confirming drainage class (Web Soil Survey 2019).

Subsamples of the composite surface soil samples were sieved to < 6 mm and air-dried before being tested for respiration immediately after the field season. These samples’ tests were conducted in chronological order of their original sampling. Respiration was measured following Stott (2019) using the Vaisala CO2 Probe GMP252 (Vaisala, Vantaa, Finland). Soil organic matter (SOM) and gravimetric soil moisture (GSM) content were also measured at the same time as the respiration tests with the loss on ignition method (Nelson and Sommers 1982). Bulk density was measured with the core samples following Blake and Hartge (1986) and corrected for coarse fragments and roots by sieving to < 2 mm and taking the mass of both components.

Subsamples of the composite surface soil samples were then prepared for the following methods. pH and electrical conductivity were analyzed following Doran and Jones (1996) using an Oakton PC 700 bench meter (Environmental Express, Charleston, SC, USA). Subsamples were then analyzed for wet aggregate stability (WAS)(Kemper and Rosenau 1986; Angers et al. 2007). Labile organic carbon was then estimated by permanganate oxidizable carbon (POXC), with triplicates to account for variability and only half of the typical soil sample mass to keep results within the standard curve’s range (Weil et al. 2003; Gruver 2015). Refrigerated and undisturbed subsamples of the composite surface samples were sent to Waypoint labs (Waypoint Analytical, Memphis, TN, USA) for nutrient analysis, conducted with Mehlich 3 extractable elements (Ziadi and Sen Tran 2007).

Surface texture was measured with the hydrometer method for particle size analysis from composite samples, or samples from a depth of 0 cm to 15 cm (Gee and Bauder 1986; Kroetsch and Wang 2007). These data were then used to help confirm the correlation to a soil series and then to select the appropriate control section following Soil Taxonomy (Soil Survey Staff 1999). Particle size analysis with the hydrometer method was then performed on each pedon’s control section to evaluate family particle size class (Soil Survey Staff 1999). The control section sample for this analysis was comprised of a weighted average of each horizon within the control section based on their thickness in the profile.

Management information for each tree was attained from property managers and owners without collecting identifying information. This included lawn and tree care programs, mulching history, pesticide application history, irrigation, deicing salt applications, drainage structures, and other relevant events on the site. This information was used to account for differences among sites, or possible causes for outliers.

Data Analyses

Before running data analyses, numerical data were prepared using R (R Core Team, Vienna, Austria). Preliminary screening indicated that there were too many parameters in our models for the sample size, so amending surface cover types (mulch and natural), Ca and Mg, Fe and Mn, and micronutrients (Cu, Zn, and B) were lumped together based on functional groups. Structure was assigned a score based on the weighted average of the sum of grades and the structure type scores in each profile. Granular structure type was assigned 3, subangular blocky and platy (in E horizons) were assigned 2, angular blocky and platy (in surface horizons) were assigned 1, and single grain and massive were assigned 0. Due to missing values in profiles’ depths to RMF, including depletions, which occurred when none were observed in the field, depths were pulled from the USDA-NRCS official series descriptions of the associated soil series. Depletions are a type of RMF that is associated with anaerobic, saturated conditions. If no depletions or other RMF were in the series description either, then a value of 201 was assigned following NRCS conventions (Schoeneberger et al. 2012). This was done to account for the full spectrum of drainage classes in our profiles.

The dataset was then transformed, standardized, and checked for outliers. All the numerical variables were then put through Box-Cox transformations to fix issues with non-normality following the Maaz et al. (2023) study’s data preparation for structural equation modeling. This was our study’s original intended model selection method but was deemed unsuitable for our sample size. These variables were then mean centered by subtracting the mean from each observation and standardized by dividing each by the standard deviation using the scale R function. The data were then examined with boxplots for any potential outliers. The sites W1 and L9 were excluded from analysis because they did not have a corresponding healthy tree present. The healthy tree of site DP6′s depth to RMF and depletions were excluded because they were deemed relic due to it being on a manmade berm made with the nearby lake’s materials. We also excluded this same tree’s soil sodium levels, because it was the only tree in a parkway-like landscape position. The declining tree of site DP5’s POXC value was excluded due to it being skewed by the long-term additions of woodchips on the surrounding trail, which extended over 6 in (15.24 cm) deep and influenced the composite sample. This reduced the sample size from 66 to 64, and the healthy tree of DP6 and declining tree of DP5 were also excluded from analyses that involved their outlier variables.

For the following analyses, α = 0.1 was chosen due to the limited sample size and inherent noise in the data. Two-way ANOVA, Welch’s, and Wilcoxon tests were run on the transformed, standardized, and lumped variables to assess their relationship with tree condition. Those that had a significant relationship with the Wilcoxon or Welch’s tests were considered significantly related to tree condition. Linear regressions were then performed for each parameter against UTH.

An exploratory factor analysis was run on JMP® Pro 17 (SAS Institute Inc., Cary, NC, USA) to understand which of the parameters were most important to include in the model selection processes. This included a principal component analysis (PCA) and variable clustering analysis on all site and soil parameters. Those in the clusters or principal components that explained most of the variance in the data were noted as potential parameters for models. A correlation matrix was also run on the site and soil characteristics and tree health indices using the corrplot R package (Wei et al. 2024). Those with the highest correlations with UTH were also noted as potential variables for model selection. Variables that had high correlations with each other were noted for potential redundancies for model selection. With low variances and correlations across the exploratory factor analysis, confirmatory factor analysis was not an ideal path for our study. The unsuitability for factor analysis was confirmed quantitatively with low results of the Kaiser-Meyer-Olkin (KMO) test.

After data preparation, generalized linear mixed models (GLMMs) were created and compared. We started with the same parameters that were chosen for the originally intended modeling method, structural equation modeling. These initial variables were those with the highest correlations with UTH in the correlation matrix, those with significant ANOVA or linear regression results, and those in the first two clusters that explained the most variance of the data set. Only the two strongest variables from each category were selected: physical, biological, chemical, and site properties. We continued to simplify the model by sequentially removing variables whose P-values were least significant. The generalized χ2/df was compared after each variable’s removal to see if it improved or weakened the model. A variety of models were selected with different levels of complexity and combinations of variables. The species variable was chosen as a random effect for GLMMs to account for differing preferences among species. Adding any other random effects made the model too complex to run for this sample size.

This GLMM selection process was also started with parameters not influenced by the original structural equation modeling method. For the original model, we selected all parameters that were significant in their ANOVAs, linear regressions (α = 0.1), or which had relatively higher correlations with UTH (> 0.1) in the correlation matrix. Parameters that were correlated with each other in the correlation matrix were removed to reduce redundancy.

All models created were included in an Akaike information criterion (AIC) table to compare the fit of the regression models. The best overall fit model was selected based on the ∆AIC scores using the AIC-cmodavg package in R. Models with ∆AIC scores < 2 were considered competitive.

For those parameters that were significant in ANOVAs or linear regressions, or were in the best performing GLMMs, their linear relationships with the other parameters were examined with linear regressions on JMP® (α = 0.05). A mosaic plot was also tested comparing tree condition and drainage classes using a Chi-squared test.

Results

Including both healthy and declining trees, 88% of the trees’ roots tested positive for antigens from a member of Phytophthora, Pythium, or Phytopythium genera. Of the healthy trees, 77% tested positive, whereas 97% of the declining trees were positive (Figure 2). The health indices accurately represented the healthy looking and declining tree labels, as they differed significantly between the two groups (P = < 0.0001).

Stacked bar chart comparing the number of declining and healthy trees that tested positive or negative with ELISA tests.
Figure 2.

Stacked bar chart comparing the number of declining and healthy trees that tested positive or negative with ELISA tests.

ANOVA and Bivariate Fit Results

A comparison of 7 variables’ distributions before and after transformations, scaling, and exclusions are compared between Figures 3 and 4. Soil zinc concentration (ppm) had a significant positive relationship with the healthy tree condition (P = 0.0123) with ANOVA, as well as with Wilcoxon (P = 0.0179) and Welch’s (P = 0.0133). Zinc also had a significant positive relationship in the linear regression with UTH (P = 0.0044) and an R2 of 0.1235. After lumping the data, the ANOVA (P = 0.0104), Welch’s (P = 0.0113), Wilcoxon (P = 0.0192), and linear regression with UTH (P = 0.00073, R2 = 0.1103) all found a significant positive relationship between the lumped micronutrients (Zn, Cu, and B) and tree condition.

Stacked histograms of the distributions of 7 variables’ results before scaling, transformations, and exclusions. RMF = redoximorphic features; POXC = for permanganate oxidizable carbon. Bars’ colors are based on whether their condition was healthy looking (H) or declining (D).
Figure 3.

Stacked histograms of the distributions of 7 variables’ results before scaling, transformations, and exclusions. RMF = redoximorphic features; POXC = for permanganate oxidizable carbon. Bars’ colors are based on whether their condition was healthy looking (H) or declining (D).

Stacked histograms of the distributions of 7 variables’ results after scaling, transformations, and exclusions. RMF = redoximorphic features; POXC = permanganate oxidizable carbon. Bars’ colors are based on whether their condition was healthy looking (H) or declining (D).
Figure 4.

Stacked histograms of the distributions of 7 variables’ results after scaling, transformations, and exclusions. RMF = redoximorphic features; POXC = permanganate oxidizable carbon. Bars’ colors are based on whether their condition was healthy looking (H) or declining (D).

Depths to depletions (cm) only had a significant positive relationship with the healthy tree condition in the Wilcoxon test (P = 0.0255). Depths to RMF also had one test which was significant, Welch’s (P = 0.0994). Without Box-Cox transformations, depth to depletions (not RMF) was still significant and only with the Wilcoxon test (P = 0.0302).

Soil sodium concentration (ppm) was significantly negatively related to UTH (P = 0.0416) and in ANOVA (P = 0.0923), Wilcoxon (P = 0.0968), and Welch’s (P = 0.0889) with tree condition. Before the Box-Cox transformation, soil sodium was only significant with the Wilcoxon test, with the same P-value above.

Exposure also had significant negative relationships with tree condition, confirmed through ANOVA (P = 0.0027), Welch’s (P = 0.0026), Wilcoxon (P = 0.0029), and the linear regression with UTH (P = 0.0125). No other soil or site parameters had significant relationships with either condition or UTH (Figure 5 and Table 3).

Linear regressions against UTH (x-axis) for each of the scaled and transformed variables that were either significant in ANOVAs, Welch’s, and/or Wilcoxon tests, linear regressions with UTH (α = 0.1), or were in the best GLMM models. Points’ color and shape are based on whether their condition was healthy looking (H) or declining (D). RMF = redoximorphic features; POXC = permanganate oxidizable carbon.
Figure 5.

Linear regressions against UTH (x-axis) for each of the scaled and transformed variables that were either significant in ANOVAs, Welch’s, and/or Wilcoxon tests, linear regressions with UTH (α = 0.1), or were in the best GLMM models. Points’ color and shape are based on whether their condition was healthy looking (H) or declining (D). RMF = redoximorphic features; POXC = permanganate oxidizable carbon.

View this table:
Table 3.

Minimums, maximums, means (SE), and medians of tree, site, and soil properties of declining and healthy oaks after exclusions and before transformation and standardization. ANOVA, Welch’s, and Wilcoxon P-values for each of the tree, site, and soil properties of declining and healthy oaks after transformations, scaling, and exclusions of outliers. SE (standard error); DBH (diameter at breast height); RFM (redoximorphic features); POXC (permanganate oxidizable carbon).

Along with the ANOVA and linear regression results, the following information was also used to help select variables for models. The correlation matrix of these parameters is displayed in Figure 6. This displays the generally weak relationships between variables and the UTH and TCI indices. The strongest correlation coefficients were between UTH and soil micronutrients (0.3321), soil sodium (−0.2575), soil potassium (0.1964), pH (0.1533), soil phosphorous (0.1425), magnesium and calcium (0.1294), depth to RMF (0.1285), control section clay percent (−0.1196), control section sand percent (0.1142), bulk density (−0.1095), and amending surfaces (0.1081). The clustering analysis results for the 2 clusters that explain the highest amount of the total proportion of the dataset’s variation are shown in Table 4. Clustering analysis only accounted for a proportion of 0.615 of the total variation in the dataset. The relevant principal components and their eigenvalues and eigenvectors are compared in Table 5, and the first and second principal components are plotted in Figure 7. Each principal component accounted for a low amount of variation, with principal component 1 only accounting for 24.4%.

Correlation matrix of each soil and site parameter compared to each other as well as 2 tree condition indices. Tr_s = the data being standardized and transformed; FPS = control section’s proportion of Sand (S), Silt (Si), and Clay (C); STR = structure; DB = bulk density; Micros = micronutrients; Ammend_Surf = amending surfaces; SURF = surface textures; SOM = soil organic matter; GSM = gravimetric soil moisture; EC = electrical conductivity; WAS = wet aggregate stability; POXC = permanganate oxidizable carbon; RES = soil respiration; DEP = depletions; RMF = redoximorphic features; DIST_INF = tree’s distance to infrastructure; TCI = Tree Condition Index score; UTH = Urban Tree Health index score; K = soil potassium; P = soil phosphorous; Na = soil sodium.
Figure 6.

Correlation matrix of each soil and site parameter compared to each other as well as 2 tree condition indices. Tr_s = the data being standardized and transformed; FPS = control section’s proportion of Sand (S), Silt (Si), and Clay (C); STR = structure; DB = bulk density; Micros = micronutrients; Ammend_Surf = amending surfaces; SURF = surface textures; SOM = soil organic matter; GSM = gravimetric soil moisture; EC = electrical conductivity; WAS = wet aggregate stability; POXC = permanganate oxidizable carbon; RES = soil respiration; DEP = depletions; RMF = redoximorphic features; DIST_INF = tree’s distance to infrastructure; TCI = Tree Condition Index score; UTH = Urban Tree Health index score; K = soil potassium; P = soil phosphorous; Na = soil sodium.

View this table:
Table 4.

Clustering analysis results of the 2 clusters with the highest proportions of variation. WAS (wet aggregate stability); SOM (soil organic matter); GSM (gravimetric soil moisture); POXC (permanganate oxidizable carbon).

View this table:
Table 5.

PCA results table of each variable’s eigenvector for PC and each PC’s eigenvalue. The PCs considered were limited to PC8 due to an eigenvalue cut off at 1. PCA (principal component analysis); PC (principal component); RMF (redoximorphic features); POXC (permanganate oxidizable carbon); WAS (wet aggregate stability).

Ordination plot of the first 2 principal components (PC) of the data, along with each PC’s variation. Tr_s = the data being standardized and transformed; FPS = control section’s proportion of Sand (S), Silt (Si), and Clay (C); STR = structure; DB = bulk density; Micros = micronutrients; Ammend_Surf = amending surfaces; SURF = surface textures; SOM = soil organic matter; GSM = gravimetric soil moisture; EC = electrical conductivity; WAS = wet aggregate stability; POXC = permanganate oxidizable carbon; RES = soil respiration; DEP = depletions; RMF = redoximorphic features; DIST_INF = tree’s distance to infrastructure; TCI = Tree Condition Index score; UTH = Urban Tree Health index score; K = soil potassium; P = soil phosphorous; Na = soil sodium.
Figure 7.

Ordination plot of the first 2 principal components (PC) of the data, along with each PC’s variation. Tr_s = the data being standardized and transformed; FPS = control section’s proportion of Sand (S), Silt (Si), and Clay (C); STR = structure; DB = bulk density; Micros = micronutrients; Ammend_Surf = amending surfaces; SURF = surface textures; SOM = soil organic matter; GSM = gravimetric soil moisture; EC = electrical conductivity; WAS = wet aggregate stability; POXC = permanganate oxidizable carbon; RES = soil respiration; DEP = depletions; RMF = redoximorphic features; DIST_INF = tree’s distance to infrastructure; TCI = Tree Condition Index score; UTH = Urban Tree Health index score; K = soil potassium; P = soil phosphorous; Na = soil sodium.

GLMM Results and Model Selection

Six GLMM binomial models with log functions and species as a random effect were selected for comparison (Table 6). Any further simplification from the Simplest model reduced performance. The variance for the random effect was zero and removing it did not affect results.

View this table:
Table 6.

AIC table comparing performance metrics of each of the GLMM models. AIC (Akaike information criterion); GLMM (generalized linear mixed models); SEM (structural equation modeling); POXC (permanganate oxidizable carbon); SOM (soil organic matter); RMF (redoximorphic features).

Based on the AIC table (Table 6), although the Simplest model had the best ∆AIC score of 0, the Best Significant model’s ∆AIC of 1.71 was not significantly worse. A seventh model, Second Best Significant, was then created that added back in control section clay percent. This model performed significantly worse and was not chosen. Based on the P-values (Table 7), only soil micronutrients and sodium had significant relationships in the 2 best performing models. Micronutrients and control section sand percent had a positive effect on tree condition, whereas POXC, bulk density, and sodium had negative effects on tree condition. The actual vs. predicted plots of each of these models (Figure 8) display an imperfect fit of the models.

Actual vs. predicted plots for each of the binomial GLMMs. An actual value of 0 indicates the tree was observed as declining, and 1 indicates healthy looking tree condition.
Figure 8.

Actual vs. predicted plots for each of the binomial GLMMs. An actual value of 0 indicates the tree was observed as declining, and 1 indicates healthy looking tree condition.

View this table:
Table 7.

Comparison of the P-values of the predictor variables in the 2 best performing GLMMs based on the AIC table (Table 6). GLMM (generalized linear mixed models); AIC (Akaike information criterion); POXC (permanganate oxidizable carbon).

The boxplots comparing the declining and healthy trees for each of the parameters that were significant in any testing method or were in the best GLMMs are compared in Figure 9. The linear regressions between each of these parameters and UTH are also shown in Figure 5, and their performance metrics are compared in Table 8.

Boxplots comparing the declining (D) and healthy (H) trees for each of the variables that were either significant in ANOVAs, Welch’s, and/or Wilcoxon tests, linear regressions with UTH (α = 0.1), or were in the best GLMM models.
Figure 9.

Boxplots comparing the declining (D) and healthy (H) trees for each of the variables that were either significant in ANOVAs, Welch’s, and/or Wilcoxon tests, linear regressions with UTH (α = 0.1), or were in the best GLMM models.

View this table:
Table 8.

Performance metric comparison of each of the linear regressions between the transformed and standardized UTH and each parameter that was either significant in ANOVAs, Welch’s, and/or Wilcoxon tests, linear regressions with UTH (α = 0.1), or were in the best GLMM models. UTH (Urban Tree Health index); RMSE (root mean square error); GLMM (generalized linear mixed models); RMF (redoximorphic features); POXC (permanganate oxidizable carbon).

Other Observed Relationships

Other relationships were examined and visualized to better understand our results. Each of the parameters that were significant in analyses also had significant relationships with other parameters (Table 9). Figure 10 shows the distribution of surface and control section textures amongst all the healthy and declining trees. Figure 11 shows the distribution of declining and healthy trees between each of the drainage classes, which had a nonsignificant (Prob > ChiSq = 0.2961) result with a mosaic plot’s Chi-Square test.

Soil texture triangle with the distribution of surface and control section textures for healthy and declining trees.
Figure 10.

Soil texture triangle with the distribution of surface and control section textures for healthy and declining trees.

Bar chart of the distribution of declining and healthy trees between the observed drainage classes.
Figure 11.

Bar chart of the distribution of declining and healthy trees between the observed drainage classes.

View this table:
Table 9.

Significant linear relationships (α = 0.05) with other parameters for each parameter that was either significant in ANOVAs, Welch’s, and/or Wilcoxon tests, linear regressions with UTH (α = 0.1), or were in the best GLMM models. UTH (Urban Tree Health index); GLMM (generalized linear mixed models); RMF (redoximorphic features); POXC (permanganate oxidizable carbon); SOM (soil organic matter); GSM (gravimetric soil moisture); EC (electrical conductivity); WAS (wet aggregate stability).

Discussion

This research has revealed associations of several parameters in white oak decline of the Chicago region. ELISA Phyt tests detected the presence Phytophthora spp. in root tissues but may also have cross-reacted with other oomycete genera, including Pythium and Phytopyhtium. Results were considered positive if the sample’s optical density was more than twice the negative control, which was an average of 0.22 optical density unit for our analyses (Agdia, Inc. 2023). More extensive sampling of the root system would be necessary to accurately detect different levels of antigens throughout the root system.

Although almost all the DAS-ELISA negative trees were apparently healthy, with such a high positivity rate, these results warrant further research to understand why oomycetes may be so consistently present in the white oak group in this region. Other ongoing research projects by collaborators have found a similar positivity rate of oak trees in this region: 86% in another Chicagoland study in 2021 (Watson and Adams 2025) and > 80% in a 2023 statewide Illinois survey (Adams and Miller 2023, unpublished data). DAS-ELISA tests are a limited qualitative detection method of antigens and do not confirm infection. Our DAS-ELISA results are intended as a preliminary screening for future pathological research by collaborators. Collaborators are using root samples from this study and the other two Illinois-based studies to investigate these potential pathogens with more advanced detection methods, DNA analysis, and Koch’s postulates. Phytophthora spp. have been found to be an important pathogen in other oak decline studies in Europe (Brasier 1996; Jung et al. 2000; Balcì and Halmschlager 2003). A review of European oak decline studies shows positivity rates of 0% to 96% for declining Q. suber and Q. ilex, although their detection methods across those studies varied (de Sampaio e Paiva Camilo-Alves et al. 2013).

General observations have given us insights on the tree health data. Based on UTH’s consistently similar relationships with the parameters as the assigned condition, it appeared that UTH more effectively represented healthy-looking and declining trees than TCI. This is why linear regressions with UTH instead of TCI were reported. Only 2 swamp white oak trees were observed in the study. This may be due to mature members of this species not being as common in the Chicago landscape or that they are not as frequently impacted by decline. Future studies designed to compare species performance may help confirm this observation. Although exposure showed a significant negative relationship with tree condition in both ANOVA and linear regressions with UTH, it was not included in the exploratory factor analysis nor model selections because it was skewed by trees in declining groves, which tend to be more exposed because of the surrounding trees’ decline, death, or removal around them.

Due to nonsignificant differences between performance metrics, both the Best Significant and Simplest models are considered competitive. With a variance of zero for the random effect of species, it appears that differences in species preferences are not significant in the study. Further research and data collection could help improve model selection and performance.

Significant Predisposing Factors

Soil micronutrients had a strong positive relationship with tree health across all analysis methods. Because only zinc of the 3 lumped micronutrients was significant on its own in these same analyses, zinc is driving the statistical significance of the micronutrient parameter. Zinc is known to be an important nutrient for plant immune systems, plant-pest interactions, and responses to climate stressors, reviewed by Cabot et al. (2019). Either a deficiency or excess in this nutrient can increase plant susceptibility to pests, pathogens, and abiotic stressors. Zinc has also been found to be related to infections of Phytophthora spp. in other plants. For example, Singh Grewal (2001) found that zinc fertilizations reduced Phytophthora spp. disease severity in alfalfa and improved their crop production.

The range of soil zinc observed in this study was 2.3 mg kg−1 to 36.1 mg kg−1. None of the samples were near the 100 mg kg−1 threshold for phytotoxicity for crop plants (Kabata-Pendias 2000). A threshold of 0.78 mg kg−1 of soil zinc was found when rice responded positively to zinc fertilizations (Sakal et al. 1982). With our range of soil zinc appearing to be within the established ideal levels, these trees may not be suffering with deficiencies or toxicities.

Zinc’s relationships with other parameters indicate that this metric may be indicating overall soil health and nutrient cycling. This is supported by its strong relationships with dynamic soil health and fertility properties including POXC, SOM, and magnesium and calcium. Rhizosphere microbial communities are also known to be correlated with concentration of micronutrients like zinc (Peng et al. 2022). It may be that zinc concentrations are associated with these soil health parameters because of zinc’s influence on microbial activity and therefore soil health. Further research would be required to investigate if there is a direct or indirect effect of soil zinc on oak condition and what mechanisms may be driving this relationship.

Bulk density was consistently in the best performing GLMMs. However, its negative relationships with tree condition in each testing method were not significant, and the means between both conditions were only 0.05 g/cm3 apart. It may be that this study, which avoided parkways that experience the most root-limiting and compacting situations (Day and Bassuk 1994), did not have a wide enough range of bulk density to capture a direct relationship between compaction tree condition. This means that this variable may not have a direct effect on tree condition and instead may be in these models due to its interactive relationships with other factors. This includes its negative relationships with properties that influence fertility and aeration: micronutrients, magnesium and calcium, SOM, GSM, POXC, and respiration (Table 9).

Bulk density’s negative relationship with tree condition in the best performing models, and its relationships with other important soil health factors, supports that it may be an important predisposing factor for white oak decline. None of our samples were past the root-limiting thresholds based on their textures (Daddow and Warrington 1983). However, in the field, visible signs of compaction, such as bare soil and penetration resistance, were frequently observed, and many of the bulk density core samples remained moist and retained core shape weeks after sampling and transportation. Elevated levels of bulk density, not necessarily past root-liming thresholds, will still negatively impact soil and root health. Compaction impairs site drainage and urban woody plant growth (Alberty et al. 1984), and it is associated with urban trees’ susceptibility to droughts and other inciting factors due to the reduced rooting area (Manion 1981; Rosolem et al. 1994). The reduced macropore space and aeration from higher bulk densities is associated with less SOM and microbial activity (Celik et al. 2010), and therefore reduced fertility, all of which are supported by the study’s data (Table 9). The associated decreased water availability, aeration (Kozlowski 1999), fertility and microbial activity (Nawaz et al. 2013), and limited root penetration (Halverson and Zisa 1982; Nambiar and Sands 1992; Müller et al. 2001) may all be contributing to urban white oak’s susceptibility to inciting factors, such as extreme weather events. This aligns with past research on decline events in German oaks, in which Gaertig et al. (2002) found reduced soil aeration was an important predisposing factor. Nejad et al. (2021) also found a similar relationship with bulk density and the severity and frequency of mortality in Iranian oak decline.

Soil sodium appears to be another important variable for predicting tree condition. The soil sodium concentration (mg kg−1) had a significant negative relationship with UTH and tree condition, meaning the healthier trees had lower concentrations of sodium in their soil. It was also consistently one of the significant explanatory factors in the GLMMs, including the best performing models. This negative relationship could be due to the sodium levels being tied to continuous deicing salt applications, the most common deicing agent being sodium chloride (NaCl).

The negative relationship of soil sodium to urban tree condition and its association with deicing salts stress agrees with a wide body of research. For example, Czerniawska-Kusza et al. (2004) found that soil sodium concentrations were higher in areas where NaCl deicing salts were applied. In these areas, tree symptoms such as leaf chlorosis and necrosis were present at 132 mg kg−1 Na soil dry mass, and the severity of these salt injury symptoms increased with its concentration. Soil sodium had toxic effects on soil protozoa starting at 260 mg kg−1 Na soil dry mass. The only tree with soil sodium levels above the 260 mg kg−1 Na threshold suggested by Hootman et al. (1994) was excluded and considered an outlier because of its parkway-like position. It appears that the declining trees in our study were not suffering from salt injury. Based on its relationships with other parameters, elevated sodium levels in declining trees may indicate disrupted soil conditions, reduced drainage, and urban stressors.

Rozas and Sampedro (2013) found that dead trees’ soil had lower soil sodium concentrations than the soil of alive trees in Q. robur decline. The difference in our results may be due to the different study settings. This study took place in natural European forests and investigated if there were limited nutrient availabilities due to recent excessive precipitation and how they influenced tree condition. Their range of soil sodium ranges from 4.6 mg kg−1 to 6.2 mg kg−1, which is a lower and narrower range than in our study. In their study’s context, sodium may be viewed as a limited and necessary macronutrient, not a sign of continuous deicing salt applications.

Permanganate-oxidizable carbon (POXC) is believed to measure the labile portion of SOM (Culman et al. 2012; Hurisso et al. 2016). Although POXC had nonsignificant negative relationships with UTH and tree condition in initial tests, this metric was consistently one of the parameters in the best GLMMs. Like bulk density, its P-values in the GLMMs were nonsignificant, indicating it may not have a direct effect on tree condition. This metric may be in the best performing models because of its complex and interactive relationships with other important soil health factors. This is supported by POXC’s positive relationships with GSM, SOM, magnesium and calcium, electrical conductivity, and micronutrients.

Instead of a direct predisposing factor, POXC’s negative relationship with tree condition may be indirectly indicating soil drainage. This property may have been higher in declining trees’ sites due to them being more poorly drained, which slows decomposition and allows for buildup of organic matter and labile organic carbon (Schuur 2001). Although POXC did not have a strong relationship with depths to depletions or other RMF, it did have a strong positive relationship with GSM. Therefore, soil of declining trees may have characteristics that allow the soil to retain relatively more moisture during the summer, such as more SOM, in the surface soil (Stevenson 1994).

García-Angulo et al. (2020) found that soil labile organic carbon decreased on sites where oak decline was present. Our urban trees may have altered nutrient cycling dynamics compared to the natural forests of study. Also, the García-Angulo et al. (2020) study sites were in the Mediterranean, an area where decline events had been occurring for decades (de Sampaio e Paiva Camilo-Alves et al. 2013). There, the continued decline of oaks may have led to a net decrease of labile organic carbon after the initial defoliation and increased usage of stored starches for immune system responses (Högberg et al. 2001). Meanwhile, this outbreak of oak decline started relatively recently in the Chicago region, where most of the affected trees had only started showing symptoms in 2020. The Mediterranean study also involved evergreen oaks, which have different site and soil preferences than the white oak group in North America. Therefore, our sites may not yet be experiencing this long-term decrease in labile organic carbon or may have different dynamics between soil and tree condition.

The control section sand percentage, or the average percentage of sand in the subsoil of the plot, was in one of the two best performing GLMMs. However, it had a nonsignificant positive relationship with tree condition in ANOVAs and linear regressions, and its P-value was not significant in the GLMM. Instead of having a direct effect on tree condition, it may have improved model performance through its strong relationships with other parameters. Subsoil sand had a negative relationship with GSM, respiration, and SOM, illustrating that this factor may be indicating soil drainage indirectly by influencing infiltration and water holding capacity (Saxton et al. 1986).

Other studies have also found similar relationships between soil sand percentage and oak decline and Phytophthora spp. infections. In Iran, Nejad et al. (2021) found a similar inverse relationship between percent sand and the frequency of mortality and severity of decline symptoms. Corcobado et al. (2013) also found that fine surface textures favored P. cinnamomi infections in declining Q. ilex. A similar relationship was found with Q. alba stands in Southern Ohio (Nagle et al. 2010).

Redoximorphic features, including depletions, are signs of the depths to water table, anaerobic saturated conditions, or areas of seasonal saturation. Because the healthy trees in this study had on average deeper depths to depletions or any RMF than declining trees, healthier-looking trees were more likely to be found in better drained soils than declining trees. Their significance in the Wilcoxon test for depletions and Welch’s for RMF indicate that soil drainage may be an important predisposing factor for white oak decline. Their weak relationship in the linear regressions with UTH are likely due to their non-normal distributions, which was caused by substitutions of 201 cm for missing values where no depletions or other RMF were present in the profile nor its official series description. However, drainage class distribution between healthy and declining trees was not significant when tested with a mosaic plot’s Chi-Square test. Depths to RMF allow us to quantify drainage of the soils, instead of broader categories, and may be why we detected a relationship with these metrics instead.

Our results support that our hypothesized predisposing factors—poor drainage, fine textures, and compaction—may be important in white oak decline in this region. Depths to RMF and depletions were not consistently significant, but when using methods that accounted for test assumption violations, significant positive relationships with tree condition were found. Fine-textured soils appear to favor decline, because the control section sand percentage in models was positively related to tree condition. This relationship also indicates sandier, and therefore better drained, soils tended to support healthier trees. Permanganate-oxidizable carbon (POXC) and soil Na had negative relationships with tree condition, each of which could be related to soil drainage. Poorly drained soils allow the accumulation of organic matter, and therefore labile organic carbon, and sodium from deicing salts. Lastly, bulk density had a negative relationship with tree condition in models, showing that elevated levels of bulk density, although below root-limiting thresholds, may be a predisposing factor in this decline spiral.

As explained previously, a potential contributing factor to oak decline is the abundance of oomycetes like Phytophthora spp., which are favored in poorly drained soils (Zentemeyer 1980). Poor soil drainage was found to be one of the primary predisposing factors for Jarrah dieback in Australia, a decline spiral involving P. cinnamomi (Weste and Taylor 1971). Our other proposed predisposing factors, fine textures and compaction, are also tied to site favorability for oomycetes by increasing soil moisture after rainfall events and exacerbating drought stress (Saxton et al. 1986). These predisposing factors were also found to be related to past European evergreen oak decline patterns in which P. cinnamomi was an associated contributing factor (Cabral et al. 1992; Brasier 1996; Moreira et al. 1999; Corcobado et al. 2013). At this stage, we are unable to confirm if an oomycete is playing a significant role in white oak decline, but the potential association warrants future research.

Our study also revealed 2 other potential predisposing factors for oak decline: lower levels of zinc and elevated levels of soil sodium. Lower levels of zinc may be weakening trees’ immune system responses to climate stressors and pathogens in the water mold group. It appears these lower levels are tied to altered nutrient cycling. Elevated soil sodium and its associated deicing salts may increase drought stress and may indirectly measure intensity of the surrounding urban environment. Based on our data and analyses, we propose that the strongest predisposing factors for the white oak group in the Chicago region are poorly drained, compacted, and fine-textured soils that are exposed to deicing salts and disrupted nutrient cycling. Therefore, we have updated our proposed oak decline spiral (Figure 12).

Spiral of decline for the white oak group in the Chicago region based on the results of this study.
Figure 12.

Spiral of decline for the white oak group in the Chicago region based on the results of this study.

Seven factors stood out in either linear regressions, ANOVA, or modeling. However, many other factors were nonsignificant across testing, despite some of them being significant in other oak decline studies. These differences may be due to our measurement techniques, data distributions, different site types and histories, narrow ranges of values, or sample size.

Study Limitations and Future Research

This study had weaknesses that may be remedied by future research and collaboration. The sample size limited the kinds of analyses possible and the performance of our models. To fully capture the variability and allow for more complex models, further data collection in the Chicago region, or in other cities in the upper Midwest with similar climate, geology, and management, should continue to take place. With enough data, structural equation modeling and GLMMs may have improved performance and allow more complexity.

These results warrant future research to help understand white oak decline in the Chicago region and the upper Midwest. A study focusing on more natural areas in Illinois may help us understand if predisposing factors differ between urban and rural landscapes. This study could also compare the severity and frequency of decline in different ecosystem types. Comparing increment core samples of both healthy and declining trees may also help improve our understanding of when the decline pattern started and of the associated inciting factors. Comparisons of branch internode length may also help reinforce such a study. Greenhouse studies comparing effects of zinc levels and fertilizations could also help confirm if zinc alone is influencing decline, or if it is an indicator of overall soil health or urban environments.

Ongoing research projects are addressing contributing factors of oak decline in the Chicago area. One includes collaborators who are isolating the potential oomycete pathogens from root samples, including those collected in this study. Trials for treatments of declining oaks based on positive ELISA tests are also ongoing.

Conclusions

Site and soil characteristics appear to be important for the white oak decline spiral in the Chicago region. Our results support that poor drainage, fine textures, and compaction may be important predisposing soil characteristics for white oak decline in the Chicago region. Lower micronutrient levels, especially zinc, and elevated sodium levels are also associated with declining oaks. It appears that these 2 factors may be associated with urban stressors like deicing salts and altered nutrient cycling. Further research is necessary to investigate the mechanisms behind these relationships.

Knowledge of predisposing factors can inform planting and management strategies for the white oak group in the Chicago region’s residences, parks, and forest preserves. These applications include planting white and bur oaks on sites that have less intense urban pressures, loamy textures, and better drainage, where predisposing factors can be minimized. Soil quality for these vulnerable species needs to be monitored, maintained, and improved following BMPs. Appropriate irrigation and drainage management are essential to maintain mature white oak health in the face of continued inciting climate factors.

Poor drainage, compaction, fine textures, low soil zinc, and elevated soil sodium appear to be playing a significant role in the decline spiral of the white oak group in the Chicago region. Future research on contributing and predisposing factors will further our understanding of white oak decline in this region. We believe this study’s findings can help provide guidance for management and future research during this period of widespread loss and decline of mature white oak trees.

Conflicts of Interest

The authors reported no conflicts of interest.

Acknowledgements

Thank you to the arborists and park and forest preserve managers that helped with site selection and visits. Thanks to our collaborators, including Gary Watson, at The Morton Arboretum for their help with site selection, study design, and shared lab space for ELISA testing. Funding for the field season was provided by The Davey Tree Expert Company and Garden Club of America. Soil nutrient analyses were funded by Bartlett Tree Experts. Lab analysis support was provided by UWSP Pedology Lab members Maya Desai, Gregor Willms, and Logan Brice.

Literature Cited

  1. ↵
    Agdia, Inc. 2023. ELISA test kits for Phy (Phytophthora). Elkhart (IN, USA): Agdia, Inc. https://orders.agdia.com/agdia-set-phyt-alkphos-sra-92601
    Google Scholar
  2. ↵
    Alberty CA, Pellett HM, Taylor DH. 1984. Characterization of soil compaction at construction sites and woody plant response. Journal of Environmental Horticulture. 2(2):48-53. https://doi.org/10.24266/0738-2898-2.2.48
    Google Scholar
  3. ↵
    Angers DA, Bullock MS, Mehuys GR. 2007. Aggregate stability to water. In: Carter MR, Gregorich EG, editors. Soil sampling and methods of analysis. 2nd Ed. Boca Raton (FL, USA): CRC Press. p. 811-819. https://doi.org/10.1201/9781420005271
    Google Scholar
  4. ↵
    Arnfield AJ. 2003. Two decades of urban climate research: A review of turbulence, exchanges of energy and water, and the urban heat island. International Journal of Climatology. 23(1):1-26. https://doi.org/10.1002/joc.859
    Google ScholarCrossRefWeb of Science
  5. ↵
    Balcì Y, Halmschlager E. 2003. Phytophthora species in oak ecosystems in Turkey and their association with declining oak trees. Plant Pathology. 52(6):694-702. https://doi.org/10.1111/j.1365-3059.2003.00919.x
    Google Scholar
  6. ↵
    Bernardo A, Gomes AA, Evaristo I, Tinoco I, Saraiva A. 1992. Um exemplo de mortalidade de montado de sobro na região do Cercal-Herdade da Cordeira. In: 2° Encontro Sobre os Montados de Sobro e Azinho. 1992 June 4–5. Évora (PT): Universidade de Évora. p. 238-247.
    Google Scholar
  7. ↵
    Blake GR, Hartge KH. 1986. Bulk density. In: Klute A, editor. Methods of soil analysis: Part 1 Physical and mineralogical methods. 2nd Ed. Madison (WI, USA): American Society of Agronomy, Inc. p. 363-375. https://doi.org/10.2136/sssabookser5.1.2ed
    Google Scholar
  8. ↵
    Bond J. 2012. Urban tree health: A practical and precise estimation method. Geneva (NY, USA): Urban Forest Analytics, LLC. 130 p.
    Google Scholar
  9. ↵
    Brasier CM. 1996. Phytophthora cinnamomi and oak decline in southern Europe. Environmental constraints including climate change. Annals of Forest Science. 53(2-3):347-358. https://doi.org/10.1051/forest:19960217
    Google Scholar
  10. ↵
    Brundrett MC, Bougher N, Dell B, Grove T, Malajczuk N. 1996. Working with mycorrhizas in forestry and agriculture. Canberra (Australia): Australian Centre for International Agricultural Research. 347 p. http://doi.org/10.13140/2.1.4880.5444
    Google Scholar
  11. ↵
    Cabot C, Martos S, Llugany M, Gallego B, Tolrà R, Poschenrieder C. 2019. A role for zinc in plant defense against pathogens and herbivores. Frontiers of Plant Science. 10:1171. https://doi.org/10.3389/fpls.2019.01171
    Google Scholar
  12. ↵
    Cabral MT, Ferreira MC, Moreira T, de Carvalho EC, Diniz AC. 1992. Diagnóstico das causas da anormal mortalidade dos sobreiros a sul do Tejo. Scientia gerundensis. 18:205-214. https://dugi-doc.udg.edu/bitstream/handle/10256/5341/45489.pdf
    Google Scholar
  13. ↵
    Celik I, Gunal H, Budak M, Akpinar C. 2010. Effects of long-term organic and mineral fertilizers on bulk density and penetration resistance in semi-arid Mediterranean soil conditions. Geoderma. 160(2):236-243. https://doi.org/10.1016/j.geoderma.2010.09.028
    Google Scholar
  14. ↵
    Corcobado T, Solla A, Madeira MA, Moreno G. 2013. Combined effects of soil properties and Phytophthora cinnamomi infections on Quercus ilex decline. Plant and Soil. 373:403-413. https://doi.org/10.1007/s11104-013-1804-z
    Google ScholarCrossRefWeb of Science
  15. ↵
    Culman SW, Snapp SS, Freeman MA, Schipanski ME, Beniston J, Lal R, Drinkwater LE, Franzluebbers AJ, Glover JD, Grandy AS, Lee J, Six J, Maul JE, Mirksy SB, Spargo JT, Wander MM. 2012. Permanganate oxidizable carbon reflects a processed soil fraction that is sensitive to management. Soil Science Society of America Journal. 76(2):494-504. https://doi.org/10.2136/sssaj2011.0286
    Google ScholarCrossRef
  16. ↵
    Czerniawska-Kusza I, Kusza G, Dużyński M. 2004. Effect of deicing salts on urban soils and health status of roadside trees in the Opole region. Environmental Toxicology. 19(4): 296-301. https://doi.org/10.1002/tox.20037
    Google ScholarCrossRefPubMed
  17. ↵
    Daddow RL, Warrington GE. 1983. Growth-limiting soil bulk densities as influenced by soil texture. Fort Collins (CO, USA): USDA Forest Service, Watershed Systems Development Group. Report No. WSDG-TN-00005. 17 p. https://forest.eecs.wsu.edu/smp/solo/documents/RPs/daddow_warrington_root_limiting_bulk_density.pdf
    Google Scholar
  18. ↵
    Day SD, Bassuk NL. 1994. A review of the effects of soil compaction and amelioration treatments on landscape trees. Journal of Arboriculture. 20(1):9-17. https://doi.org/10.48044/jauf.1994.003
    Google Scholar
  19. ↵
    de Sampaio e Paiva Camilo-Alves C, da Clara MIE, de Almeida Ribeiro NMC. 2013. Decline of Mediterranean oak trees and its association with Phytophthora cinnamomi: A review. European Journal of Forest Research. 132:411-432. https://doi.org/10.1007/s10342-013-0688-z
    Google Scholar
  20. ↵
    De Vita P, Serrano MS, Ramo C, Aponte C, García LV, Belbahri L, Sánchez ME. 2013. First report of root rot caused by Pythium spiculum affecting cork oaks at Doñana Biological Reserve in Spain. Plant Disease. 97(7):991. https://doi.org/10.1094/PDIS-10-12-0952-PDN
    Google Scholar
  21. ↵
    Doran JW, Jones AJ, editors. 1996. Methods for assessing soil quality. Madison (WI, USA): Soil Science Society of America, Inc. 410 p. https://acsess.onlinelibrary.wiley.com/doi/pdf/10.2136/sssaspecpub49.frontmatter
    Google Scholar
  22. ↵
    Effland WR, Pouyat RV. 1997. The genesis, classification, and mapping of soils in urban areas. Urban Ecosystems. 1:217-228. https://doi.org/10.1023/A:1018535813797
    Google ScholarCrossRef
  23. ↵
    Fehrenbacher JB, Alexander JD, Jansen IJ, Darmody RG, Pope RA, Flock MA, Voss EE, Scott JW, Andrews WF, Bushue LJ. 1984. Soils of Illinois. Urbana-Champaign (IL, USA): University of Illinois at Urbana-Champaign College of Agriculture. Report No. 778. 85 p. https://archive.org/details/soilsofillinois00unse/page/n1/mode/2up
    Google Scholar
  24. ↵
    Gaertig T, Schack-Kirchner H, Hildebrand EE, Wilpert KV. 2002. The impact of soil aeration on oak decline in southwestern Germany. Forest Ecology and Management. 159(1-2):15-25. https://doi.org/10.1016/S0378-1127(01)00706-X
    Google Scholar
  25. ↵
    García-Angulo D, Hereş AM, Fernández-López M, Flores O, Sanz MJ, Rey A, Valladares F, Curiel Yuste J. 2020. Holm oak decline and mortality exacerbates drought effects on soil biogeochemical cycling and soil microbial communities across a climatic gradient. Soil Biology and Biochemistry. 149:107921. https://doi.org/10.1016/j.soilbio.2020.107921
    Google Scholar
  26. ↵
    Gee GW, Bauder JW. 1986. Particle-size analysis. In: Klute A, editor. Methods of soil analysis: Part 1 Physical and mineralogical methods. 2nd Ed. Madison (WI, USA): American Society of Agronomy, Inc. p. 383-411. https://doi.org/10.2136/sssabookser5.1.2ed
    Google Scholar
  27. ↵
    Gruver J. 2015. Evaluating the sensitivity and linearity of a permanganate-oxidizable carbon method. Communications in Soil Science and Plant Analysis. 46(4):490-510. https://doi.org/10.1080/00103624.2014.997387
    Google Scholar
  28. ↵
    Halverson HG, Zisa RP. 1982. Measuring the response of conifer seedlings to soil compaction stress. Broomall (PA, USA): USDA Forest Service, Northeastern Forest Experiment Station. Res. Pap. NE-509. 6 p. https://doi.org/10.2737/NE-RP-509
    Google Scholar
  29. ↵
    Hansen EM, Goheen DJ, Jules ES, Ullian B. 2000. Managing Port-Orford-cedar and the introduced pathogen Phytophthora lateralis. Plant Disease. 84(1):4-14. https://doi.org/10.1094/PDIS.2000.84.1.4
    Google ScholarPubMed
  30. ↵
    Högberg P, Nordgren A, Buchmann N, Taylor AFS, Ekblad A, Högberg MN, Nyberg G, Ottosson-Löfvenius M, Read DJ. 2001. Large-scale forest girdling shows that current photosynthesis drives soil respiration. Nature. 411:789-792. https://doi.org/10.1038/35081058
    Google ScholarCrossRefPubMedWeb of Science
  31. ↵
    Hootman RG, Kelsey PD, Reid R, von der Heide-Spravka K. 1994. Factors affecting accumulation of deicing salts in soils around trees. Journal of Arboriculture. 20(3):196-201. https://doi.org/10.48044/jauf.1994.035
    Google Scholar
  32. ↵
    Hurisso TT, Culman SW, Horwath WR, Wade J, Cass D, Beniston JW, Bowles TM, Grandy AS, Franzluebbers AJ, Schipanski ME, Lucas ST, Ugarte CM. 2016. Comparison of permanganate-oxidizable carbon and mineralizable carbon for assessment of organic matter stabilization and mineralization. Soil Science Society of America Journal. 80(5):1352-1364. https://doi.org/10.2136/sssaj2016.04.0106
    Google Scholar
  33. ↵
    Illinois State Climatologist. 2007. Chicago O’Hare International Airport Station 111549: Precipitation summary. Champaign (IL, USA): Illinois State Climatologist. [Accessed 2022 November 14; Updated 2007 September 11]. https://stateclimatologist.web.illinois.edu/data/illinois-climate-summaries/chicago-ohare-international-airport-station-111549
    Google Scholar
  34. ↵
    Jankowiak R, Stępniewska H, Bilański P. 2015. Notes on some Phytopythium and Pythium species occurring in oak forests in southern Poland. Acta Mycologica. 50(1):1-10. https://doi.org/10.5586/am.1052
    Google Scholar
  35. ↵
    Jung T, Blaschke H, Oßwald W. 2000. Involvement of soilborne Phytophthora species in Central European oak decline and the effect of site factors on the disease. Plant Pathology. 49(6):706-718. https://doi.org/10.1046/j.1365-3059.2000.00521.x
    Google Scholar
  36. ↵
    Kabata-Pendias A. 2000. Trace elements in soils and plants. 3rd Ed. Boca Raton (FL, USA): CRC Press. 432 p. https://doi.org/10.1201/9781420039900
    Google Scholar
  37. ↵
    Kemper WD, Rosenau RC. 1986. Aggregate stability and size distribution. In: Klute A, editor. Methods of soil analysis: Part 1 Physical and mineralogical methods. 2nd Ed. Madison (WI, USA): American Society of Agronomy, Inc. p. 425-442. https://doi.org/10.2136/sssabookser5.1.2ed.c17
    Google Scholar
  38. ↵
    Kessler KJ Jr. 1989. Some perspectives on oak decline in the 80’s. In: Rink G, Budelsky CA, editors. Proceedings of a meeting held at Southern Illinois University at Carbondale, Carbondale, Illinois, March 5–8, 1989. Seventh Central Hardwood Forest Conference; 1989 March 5–8; Carbondale, Illinois, USA. Carbondale (IL, USA): USDA Forest Service, North Central Forest Experiment Station. p. 25-29. https://books.google.com/books?id=tbFHAQAAMAAJ
    Google Scholar
  39. ↵
    Kozlowski TT. 1999. Soil compaction and growth of woody plants. Scandinavian Journal of Forest Research. 14(6):596-619. https://doi.org/10.1080/02827589908540825
    Google ScholarCrossRefWeb of Science
  40. ↵
    Kroetsch D, Wang C. 2007. Particle size distribution. In: Carter MR, Gregorich EG, editors. Soil sampling and methods of analysis. 2nd Ed. Boca Raton (FL, USA): CRC Press. p. 713-726. https://doi.org/10.1201/9781420005271
    Google Scholar
  41. ↵
    Maaz TM, Heck RH, Glazer CT, Loo MK, Zayas JR, Krenz A, Beckstrom T, Crow SE, Deenik JL. 2023. Measuring the immeasurable: A structural equation modeling approach to assessing soil health. Science of the Total Environment. 870:161900. https://doi.org/10.1016/j.scitotenv.2023.161900
    Google ScholarPubMed
  42. ↵
    Manion PD. 1981. Tree disease concepts. Englewood Cliffs (NJ, USA): Prentice-Hall. 399 p. https://archive.org/details/treediseaseconce0000mani
    Google Scholar
  43. ↵
    Moreira AC, Ferraz JFP, Clegg J. 1999. The involvement of Phytophthora cinnamomi in cork and holm oak decline in Portugal. In: Hansen EM, Sutton W, editors. Phytophthora diseases of forest trees: IUFRO Working Party 7.02.09, proceedings from the First International Meeting on Phytophthoras in Forest and Wildlife Ecosystems. IUFRO Working Party; 1999 August 30–September 3; Grants Pass, Oregon, USA. Corvallis (OR, USA): Forest Research Laboratory, Oregon State University. p. 132-135. https://www.researchgate.net/publication/235154527_the_involvement_of_phytophthora_cinnamomi_in_cork_and_holm_oakdecline_in_portugal
    Google Scholar
  44. ↵
    Moser A, Rötzer T, Pauleit S, Pretzsch H. 2015. Structure and ecosystem services of small-leaved lime (Tilia cordata Mill.) and black locust (Robinia pseudoacacia L.) in urban environments. Urban Forestry & Urban Greening. 14(4):1110-1121. https://doi.org/10.1016/j.ufug.2015.10.005
    Google Scholar
  45. ↵
    Müller MML, Ceccon G, Rosolem CA. 2001. Influência da compactação do solo em subsuperfície sobre o crescimento aéreo e radicular de plantas de adubação verde de inverno. Revista Brasileira de Ciência do Solo. 25(3):531-538. https://doi.org/10.1590/S0100-06832001000300002
    Google Scholar
  46. ↵
    Nagle AM, Long RP, Madden LV, Bonello P. 2010. Association of Phytophthora cinnamomi with white oak decline in southern Ohio. Plant Disease. 94(8):1026-1034. https://doi.org/10.1094/PDIS-94-8-1026
    Google ScholarPubMed
  47. ↵
    Nambiar EKS, Sands R. 1992. Effects of compaction and simulated root channels in the subsoil on root development, water uptake and growth of radiata pine. Tree Physiology. 10(3):297-306. https://doi.org/10.1093/treephys/10.3.297
    Google ScholarCrossRefPubMedWeb of Science
  48. ↵
    Nawaz MF, Bourrié G, Trolard F. 2013. Soil compaction impact and modelling. A review. Agronomy for Sustainable Development. 33:291-309. https://doi.org/10.1007/s13593-011-0071-8
    Google Scholar
  49. ↵
    Nejad AZ, Badehian Z, Nejad R. 2021. The relationship between Iranian oak decline (Quercus brantii Lindl.) and some properties of soil and determining the ecophysiological responses of this. Iranian Journal of Forest. 13(23):221-236. https://doi.org/10.22034/ijf.2021.141074
    Google Scholar
  50. ↵
    Nelson DW, Sommers LE. 1982. Total carbon, organic carbon, and organic matter. In: Page AL, editor. Methods of soil analysis: Part 2 Chemical and microbiological properties. Madison (WI, USA): American Society of Agronomy, Inc. p. 539-579. http://doi.wiley.com/10.2134/agronmonogr9.2.2ed.c29
    Google Scholar
  51. ↵
    Nowak DJ, Crane DE, Stevens JC, Hoehn RE, Walton JT, Bond J. 2008. A ground-based method of assessing urban forest structure and ecosystem services. Arboriculture & Urban Forestry. 34(6):347-358. https://doi.org/10.48044/jauf.2008.048
    Google Scholar
  52. ↵
    Peng Z, Liang C, Gao M, Qiu Y, Pan Y, Gao H, Liu Y, Li X, Wei G, Jiao S. 2022. The neglected role of micronutrients in predicting soil microbial structure. npj Biofilms Microbiomes. 8:103. https://doi.org/10.1038/s41522-022-00363-3
    Google ScholarPubMed
  53. ↵
    Pickett STA, Cadenasso ML. 2009. Altered resources, disturbance, and heterogeneity: A framework for comparing urban and non-urban soils. Urban Ecosystems. 12:23-44. https://doi.org/10.1007/s11252-008-0047-x
    Google Scholar
  54. ↵
    Romero MA, Sánchez JE, Jiménez JJ, Belbahri L, Trapero A, Lefort F, Sánchez ME. 2007. New Pythium taxa causing root rot on mediterranean Quercus species in southwest Spain and Portugal. Journal of Phytopathology. 155(5):289-295. https://doi.org/10.1111/j.1439-0434.2007.01230.x
    Google Scholar
  55. ↵
    Rosolem CA, Vale LSR, Grassi HF, De Moraes MH. 1994. Sistema radicular e nutrição do milho em função da calagem e da compactação do solo. Revista brasileira de ciência do solo. 18(3):491-497.
    Google Scholar
  56. ↵
    Rozas V, Sampedro L. 2013. Soil chemical properties and dieback of Quercus robur in Atlantic wet forests after a weather extreme. Plant and Soil. 373:673-685. https://doi.org/10.1007/s11104-013-1835-5
    Google ScholarCrossRef
  57. ↵
    Sakal R, Singh BP, Singh AP. 1982. Determination of critical limit of zinc in soil and plant for predicting response of rice to zinc application in calcareous soils. Plant and Soil. 66:129-132. https://doi.org/10.1007/BF02203412
    Google Scholar
  58. ↵
    Saxton KE, Rawls WJ, Romberger JS, Papendick RI. 1986. Estimating generalized soil-water characteristics from texture. Soil Science Society of America Journal. 50(4):1031-1036. https://doi.org/10.2136/sssaj1986.03615995005000040039x
    Google ScholarCrossRefWeb of Science
  59. ↵
    Scharenbroch BC, Catania M. 2012. Soil quality attributes as indicators of urban tree performance. Arboriculture & Urban Forestry. 38(5):214-228. https://doi.org/10.48044/jauf.2012.030
    Google Scholar
  60. ↵
    Schoeneberger PJ, Wysocki DA, Benham EC, Soil Survey Staff. 2012. Field book for describing and sampling soils. Version 3.0. Lincoln (NE, USA): USDA Natural Resource Conservation Service. 298 p.
    Google Scholar
  61. ↵
    Schuur EAG. 2001. The effect of water on decomposition dynamics in mesic to wet Hawaiian montane forests. Ecosystems. 4:259-273. https://doi.org/10.1007/s10021-001-0008-1
    Google Scholar
  62. ↵
    Singh Grewal H. 2001. Zinc influences nodulation, disease severity, leaf drop and herbage yield of alfalfa cultivars. Plant and Soil. 234:47-59. https://doi.org/10.1023/A:1010544601267
    Google Scholar
  63. ↵
    Soil Survey Staff. 1999. Soil taxonomy: A basic system of soil classification for making and interpreting soil surveys. 2nd Ed. Agricultural Handbook No. 436. Washington (DC, USA): USDA Natural Resource Conservation Service. 886 p. https://www.nrcs.usda.gov/sites/default/files/2022-06/Soil%20Taxonomy.pdf
    Google Scholar
  64. ↵
    Starkey DA, Oak SW. 1989. Site factors and stand conditions associated with oak decline in southern upland hardwood forests. In: Rink G, Budelsky CA, editors. Proceedings of the Seventh Central Hardwood Conference. Seventh Central Hardwood Forest Conference; 1989 March 5–8; Southern Illinois University, Carbondale, Illinois, USA. St. Paul (MN, USA): USDA Forest Service, North Central Forest Experiment Station. General Technical Report NC-132. p. 95-102. https://www.nrs.fs.usda.gov/pubs/gtr/gtr_nc132.pdf
    Google Scholar
  65. ↵
    Stevenson FJ. 1994. Humus chemistry: Genesis, composition, reactions. 2nd Ed. New York (NY, USA): J. Wiley and Sons. 512 p.
    Google Scholar
  66. ↵
    Stott DE. 2019. Recommended soil health indicators and associated laboratory procedures. Washington (DC, USA): USDA Natural Resources Conservation Service. Soil health technical note No. 450-03. 76 p. https://www.naptprogram.org/files/napt/publications/method-papers/2019-nrcs-technote-450-03.pdf
    Google Scholar
  67. ↵
    Watson G, Adams S. 2025. Involvement of soilborne Phytophthora species in northeast Illinois rapid oak decline and the effect of site factors on the disease. Arboriculture & Urban Forestry. 51(1):65-73. https://doi.org/10.48044/jauf.2024.015
    Google Scholar
  68. ↵
    Web Soil Survey. 2019. Web soil survey. Washington (DC, USA): USDA Natural Resource Conservation Service. [Updated 2019 July 31]. https://websoilsurvey.nrcs.usda.gov/app
    Google Scholar
  69. ↵
    Webster BL. 1978. Guide to judging the condition of a shade tree. Journal of Arboriculture. 4(11):247-249. https://doi.org/10.48044/jauf.1978.061
    Google Scholar
  70. ↵
    Wei T, Simko V, Levy M, Xie Y, Jin Y, Zemla J, Moritz F, Cai J, Protivinsky T. 2024. corrplot: Visualization of a correlation matrix. Version 0.95. https://doi.org/10.32614/CRAN.package.corrplot
    Google Scholar
  71. ↵
    Weil RR, Islam KR, Stine MA, Gruver JB, Samson-Liebig SE. 2003. Estimating active carbon for soil quality assessment: A simplified method for laboratory and field use. American Journal of Alternative Agriculture. 18(1):3-17. https://doi.org/10.1079/AJAA200228
    Google Scholar
  72. ↵
    Weste GM, Taylor P. 1971. The invasion of native forest by Phytophthora cinnamomi. I. Brisbane Ranges, Victoria. Australian Journal of Botany. 19(3):281-294. https://doi.org/10.1071/BT9710281
    Google Scholar
  73. ↵
    Zentemeyer GA. 1980. Phytophthora cinnamomi and the diseases it causes. St. Paul (MN, USA): American Phytopathological Society. 96 p.
    Google Scholar
  74. ↵
    Ziadi N, Sen Tran T. 2007. Mehlich 3-extractable elements. In: Carter MR, Gregorich EG, editors. Soil sampling and methods of analysis. 2nd Ed. Boca Raton (FL, USA): CRC Press. p. 107-114. https://doi.org/10.1201/9781420005271
    Google Scholar
Loading
Loading
Loading
PDF
Back to top
In this Issue
Arboriculture & Urban Forestry (AUF)
Arboriculture & Urban Forestry (AUF)

Vol. 51, Issue 6

November 2025

Table of Contents Table of Contents (PDF) Index by author
Jump to section

Keywords

Quercus alba, Quercus macrocarpa, Site Quality, Tree Health, Urban Soil