A Fine Line between Phytotoxicity and Blue When Producing Hydrangea macrophylla in a Nursery at a Low Substrate pH

Abstract

Hydrangea macrophylla exhibiting blue sepals (versus purple or pink) have improved marketability; however, little research has been conducted to evaluate aluminum (Al), the element responsible for bluing, on crop growth, effectiveness of bluing sepals, and characteristics of flower clusters in an outdoor nursery. This study compared substrate Al availability, crop growth, flower color, number, and size over a 56-week period in two locations. A polymer coated (90-day release) or ground aluminum sulfate [Al₂(SO₄)₃; water soluble] was either incorporated into a non-limed pine bark substrate, applied to the surface of the substrate as a top dress, or as a routinely applied Al₂(SO₄)₃ drench (low concn.) or applied once (high concn.). In general, application of Al increased plant foliar Al concentration, but also decreased substrate pore-water pH and increased electrical conductivity (EC) with varying effects based on the applied product’s solubility and subsequent longevity. Aluminum sulfate increased the potential of Al phytotoxicity negatively affecting root morphology and creating an undesirable rhizosphere electrochemistry due to the pH being continually acidic, <4, and the EC being temporarily increased to >1.5 mS·cm⁻¹. These suboptimal rhizosphere conditions resulted in a lower quality or smaller plant. No plants exhibited clear, deep blue flower cluster sought by consumers. Neither the effect of pore water pH or EC could, alone or in combination, account for the lack of plant vigor or blue flower clusters when substrate and foliar Al concentrations were adequate in flowering H. macrophylla. More research is needed to investigate the effect of pore-water electrochemical properties, possible mineral nutrient co-factors that provide Al synergisms or toxicity protections, and holistic plant health on ensuring blue coloration of a vigorous H. macrophylla.

1. Introduction

Bigleaf hydrangea (Hydrangea macrophylla Thunb. Ex J.A. Murr.) sepals can be manipulated during cultivation to range from pink to blue. The presence of aluminum (Al) is considered essential for blue sepal color in H. macrophylla [1,2]. Hydrangeas with blue sepals are in high demand; being preferred ten to one by customers when compared to hydrangeas with pink sepals, resulting in rejection of plants that show lavender, “blurple”, or pink flowers [3,4]. For most cultivars, the showy sepals are pink in basic or neutral soilless substrates, and shades of purple to blue in acidic substrates where free aluminum ions (Al³⁺) are increased and available for crop uptake. In sepals, Al³⁺ form a complex with delphinidin 3-glucoside, an anthocyanin, and 3-p-courmaroylquinic and 5-caffeoylquinic acids, co-pigments, resulting in blue color [5,6,7]. Sepal Al concentration and subsequent content is strongly related to sepal color, with pink having the lowest Al content, purple sepals having moderate Al content, and blue having the greatest Al content [8]. The minimal Al concentration associated with blue sepals is 40 µg·g⁻¹ of sepal fresh weight [8], except in some of the deepest colored cultivars. However, Kodama et al. [9] did not find Al to be statistically higher in either stable blue cultivars or variable-color cultivars when grown in acid soils compared to alkaline soils. They attributed this response to reduced availability of “free” Al³⁺ in the soil solution due to precipitation or sorption to phosphoric acid when grown in alkaline conditions in which Al was not “free” and thus unavailable for chelation with anthocyanin.

Soilless substrates contain minimal amounts of lasting mineral nutrients and metals, including Al, and thus require a continuous supplement of macro- and micro-nutrients to ensure crop vigor and health and Al to aid in hydrangea sepal bluing. One solution has been to incorporate different Al containing clay aggregates such as kaolin, zeolite, bentonite, or pozzolan into the container substrate [10,11,12,13,14,15]. These amendments vary in Al content and subsequent labile Al based on structure and solubility of existing Al complexes or minerals. High levels of clay amendments have also been shown to reduce plant growth, thus reducing marketability of plants [11,12].

Researchers have investigated various rates and timing of aluminum sulfate [Al₂(SO₄)₃] applications to container substrate when forcing hydrangeas (i.e., pushing growth for sale under ideal temperature conditions) in the greenhouse [16,17,18]. Blom and Piott [16] received plants in 15-cm pots that were either treated with three applications of Al₂(SO₄)₃ at 15 g·liter⁻¹ the previous year, or no applications. During greenhouse forcing, plants that received no treatment the previous year were subjected to either early or late applications of one of 4 rates of Al₂(SO₄)₃, ranging from 4–16 g per container, through subirrigation. Controls received no Al₂(SO₄)₃. The bluest sepals were achieved with early applications of approximately 12 g of Al₂(SO₄)₃ during greenhouse forcing. Edziak (2015) forced hydrangeas in 15-cm azalea pots in a greenhouse by applying two rates of slow release Al₂(SO₄)₃ either as a topdress application (10 or 15 g per pot), a substrate incorporation (3.6 or 5.4 kg·yard⁻³), or as drenches. Drenches were applied at 28 g per pot or 56 g per pot every two weeks or at 56 g per pot or 112 g per pot monthly. A rate of 15 g per container topdressed slow release Al₂(SO₄)₃ or drenches at either rate of Al₂(SO₄)₃ every two weeks yielded the bluest sepals. Landis [19] applied seven treatments ranging from 0 to 15 g total of Al₂(SO₄)₃ per pot as a drench and then measured Al levels in the leaves. Higher rates were applied over two to three applications delivered every two weeks. They found that plants with foliar Al concentrations between 1300 to 2000 µg·g⁻¹ dry weight at weeks 6 to 8, such as provided by 10 to 15 g Al₂(SO₄)₃, produced the bluest sepal color. However, plants treated with 15 g per plant were stunted and had leaf scorch. These controlled environment experiments were described as using peat-based substrates, which have a higher cation exchange capacity (CEC) and greater pH buffering capacity compared to the bark-based substrates used in outdoor nurseries, which could greatly influence response to fertilizer. Therefore, none of these experiments identified safe, effective rates for Al₂(SO₄)₃ use in an open-air or outdoor nursery setting.

Producing H. macrophylla with blue flowers successfully and consistently remains a challenge for nursery producers across the U.S. and Canada. Using the same protocol does not yield the same results season after season at a given nursery. This may be because outdoor nurseries are subjected to a range of meteorological conditions and extremes that are not experienced in the greenhouse such as unpredictable rainfall and fluctuating humidity and temperature, the latter being a result of increased solar radiation/irradiance that can result in substrate temperatures exceeding ambient air temperature. Additionally, different cultivars are typically grown in outdoor production compared to greenhouse production. Yet, few studies that have been conducted in an outdoor nursery setting have been reported in the literature. In one, Midcap [20] grew three remontant varieties of hydrangea in #3 (~13 L) containers. Al₂(SO₄)₃ and flowable lime were applied alone or in combination for a total of three treatments. Al₂(SO₄)₃ at a rate of 1.5 g per container alone produced the bluest sepals at a pH of 3.5–3.6, while the addition of 80 g per container flowable lime with Al₂(SO₄)₃ produced purple sepals at a pH 5.2–5.3. Identifying the optimal supplemental amount and type of Al, optimal pH, and timing of application(s) that can be safely applied during hydrangea production in nurseries is critical to consistently producing hydrangea crops that are highly desirable to consumers.

Many nursery producers incorporate 7.4 kg·m⁻³, up to 8.9 kg·m⁻³ of Al₂(SO₄)₃ (personal communications with industry) into a pine bark-based substrate; however, irrigation water alkalinity as well as acid injection containing phosphorus to treat alkalinity, substrate incorporation of lime and phosphorus, rainfall, temperature, and root system health can impact the effectiveness of Al applications [4,21]. Additionally, because this is a specific fertilizer amendment for a single crop among nurseries that often grow hundreds of different species [22], it’s application can be overlooked until sepals begin to show color. In such close proximity to the sales window, a “rescue” application of Al may be used that typically involves surface applying or drenching the crop with high rates of Al₂(SO₄)₃ [4]. Nursery producers rely on trial and error when this critical, high-risk treatment is required.

Research using economically significant cultivars grown in an outdoor nursery production system is necessary to establish standard Al application, both product and rate. Therefore, the objective of this research is twofold: (1) identify supplemental Al fertilization applications that create blue flowers in H. macrophylla Endless Summer^® The Original for the year sales are scheduled, and (2) identify rescue treatments to use when the entire crop is not sold in the intended year and a portion is carried over for sale the following season. To see the effect of Al alone, lime was omitted and, particular attention was paid to crop safety.

2. Materials and Methods

2.1. Bluing the Initial Year of Production: Year 1

2.1.1. Location, Plants and Substrate

Experiments were conducted concurrently at the University Tennessee in Knoxville, TN [Tennessee Appalachian Highlands (TAH); 35.954° N, 83.929° W] and Virginia Tech Hampton Roads Agricultural Research and Extension Center in Virginia Beach, VA, USA [Coastal Virginia Flats (CVF); 36.892° N, 76.179° W] in 2018 and 2019. On 26 April 2018 [0 weeks after initiation (WAI)], 5 × 5 cm H. macrophylla ‘Bailmer’ PP15,298 Endless Summer^® Original liners were transplanted into #1 (3.8 L) containers (C400; Nursery Supplies, Inc., Chambersburg, PA, USA) filled with a non-limed pine bark screened to 12.6 mm (Grower Pine Fines; Garick, Cumming, GA, USA) with the following properties: bulk density 0.19, air space 41.3%, container capacity 44.1%, total porosity 85.5%.

2.1.2. Treatments and Other Fertilization

Plants received one of nine bluing treatments either soil-incorporated before transplanting, topdressed as a rescue treatment, or drenched. Incorporated treatments included controlled release Al (Florikan Sapphire 90-day at 26.7 °C; 15% Al by wt. derived from aluminum sulfate [Al₂(SO₄)₃]; Florikan E.S.A, LLC, Sarasota, FL, USA) incorporated at either the labeled low 4.15 (FSL) or high 5.93 (FSH) kg·m⁻³ rate; or ground Al₂(SO₄)₃ (17.0–17.5% Al by wt.; USALCO^®, Baltimore, MD, USA) incorporated at the industry low 5.93 (ASL) or high 8.90 (ASH) kg·m⁻³ rate. There were approximately 264 containers filled with one cubic meter of amended substrate, with each receiving between approximately 2.35 to 3.37 or 3.90 to 5.91 g of Al per container when controlled release or ground Al₂(SO₄)₃ was incorporated at the low or high rate, respectfully. Two rescue treatments were applied as topdress; 90-day controlled release Al₂(SO₄)₃ topdressed (Florikan Sapphire; FST) at 15 g per container in CVF and 9 g per container in TAH; or ground Al₂(SO₄)₃ top dressed (AST) at 32.66 g per container in CVF and 26.17 g per container in TAH. Drench treatments were applied every 2 weeks beginning 1 week after experiment initiation (WAI) at a rate of 250 mL per pot of Al₂(SO₄)₃ at 17.97 g·L⁻¹ (CD), or as a rescue treatment at a one-time rate of 500 mL per pot of Al₂(SO₄)₃ at 23.97 g·L⁻¹ (RD). A summary of treatments is provided in

Table 1. The abbreviation and description of aluminum (Al) application (appl.) treatments to containerized H. macrophylla ‘Bailmer’ PP15,298 Endless Summer^® Original occurring in the Coastal Virginia Flats (CVF, Virginia Beach, VA, USA) or Tennessee Appalachian Highlands (TAH, Knoxville, TN, USA) ecophysiographic regions for 56 weeks after initiation (WAI).

Application Method	Treatment/Product	Product Appl. Rate	Frequency of Appl.	Aluminum (Al) Added	Abbreviation	Location
Year 1 (2018) planted liner; base substrate or “control” (CO) contained no Al sulfate (0.00 kg Al·m−3)
Pre-plant incorporated	90-day controlled release Al	4.15 kg·m−3	1 appl.	0.62 kg·m−3	FSL	CVF, TAH
(provide Al throughout production)		5.93 kg·m−3		0.89 kg·m−3	FSH
	Ground Al sulfate	5.93 kg·m−3	1 appl.	1.03 kg·m−3	ASL	CVF, TAH
		8.90 kg·m−3		1.56 kg·m−3	ASH
Topdressed	90-day controlled release Al	9.00 g·cont.−1	1 appl.	1.35 g·cont.−1	FST	TAH
(provide Al at time of flowering, rescue)		15.00 g·cont.−1		2.25 g·cont.−1		CVF
	Ground Al sulfate	26.17 g·cont.−1	1 appl.	4.58 g·cont.−1	AST	TAH
		32.66 g·cont.−1		5.72 g·cont.−1		CVF
Drench	Dissolved Al sulfate
(continuously every 2 wks.	250 mL of 17.97 g·L−1	4.49 g·cont.−1	7 appl.	5.50 g·cont.−1	CD	CVF, TAH
One time rescue appl.)	500 mL of 23.97 g·L−1	11.99 g·cont.−1	1 appl.	2.10 g·cont.−1	RD
Year 2 (2019) 1-year old plant; base substrate or “control” (UC) contained 5.93 kg·m−3 aluminum sulfate (1.03 kg Al·m−3)
Topdressed	90-day controlled release Al	9.00 g·cont.−1	1 appl.	1.35 g·cont.−1	TDL	CVF, TAH
(provide Al at time of flowering, rescue)		15.00 g·cont.−1		2.25 g·cont.−1	TDH
Drench	Dissolved Al sulfate	8.98 g·cont.−1		1.57 g·cont.−1		CVF
(one time rescue appl.)	500 mL of 17.97 g·L−1		1 appl.		ASD
	500 mL of 23.97 g·L−1	11.99 g·cont.−1		2.10 g·cont.−1		TAH

. A control group of plants received no bluing treatment (CO). Rescue treatments were applied the week of 25 June 2018 (9 WAI) in CVF and the week of 2 July 2018 (10 WAI) in TAH.

Plants were hand watered after transplanting and 250 mL per pot of surfactant (Aquagro^®L, Aquatrols, Paulsboro, NJ, USA) was applied as a 600 mg·L⁻¹ drench. Plants were placed in a bowhouse with either 60% (CVF) or 50% (TAH) shade cloth. All plants were fertilized on 30 April 2018 (1 WAI) at the medium rate of 11 g per container with a 5–6 month (26.7 °C) 19N:1.7P:6.6K (19-4-8, Harrell’s, LLC, Lakeland, FL, USA) controlled release fertilizer containing micronutrients. Five fallow (i.e., unplanted) pots were placed in the bowhouse in TAH. pH and electrical conductivity (EC) were measured 3 May 2018 and extractant was collected for analysis of Al in the substrate. Irrigation water was sampled in September 2018. In TAH, pH was 8.1 and total alkalinity was 54.9 mg·L⁻¹; in CVF, pH was 5.8 and total alkalinity was 18.8 mg·L⁻¹. At both locations, Al concentration in irrigated water was below detectable limits of 0.20 mg·L⁻¹.

2.1.3. Substrate Pore-Water Chemical Properties

Beginning the week of 14 May 2018 (3 WAI) the Virginia Tech Extraction Method [23], using 120 mL of deionized (CVF) or irrigation (TAH) water to displace pore-water, was used to measure EC and pH of extractant on half of the replications bi-weekly, one day prior to the CD. pH and EC of each extract were measured using a Hanna Instruments 9814 handheld meter (Hanna Industries, Woonsocket, RI, USA) in CVF, and an Agri-meter 6 (Myron L Co., Carlsbad, CA, USA) or a Hanna Instruments 9811-5 handheld meter in TAH, until and including the final harvest in August (17 WAI). Extractant samples were collected with the final pH and EC measurements, filtered to 0.45 um yielding 15 mL for analysis, and analyzed for Al at an academic service laboratory (Virginia Tech Soil Testing Laboratory, Blacksburg, VA, USA) using ARCOS II Multi-View ICP model FHM22 with CETAC ASX560 Autosampler (Spectro Analytical Instruments, Inc., Mahwah, NJ, USA).

2.1.4. Flower Color Measurements

Flower buds were removed through the week of 11 June 2018 (7 WAI) at both locations, at which point roots were visible at the container sidewall interface. Beginning the week of 9 July 2018 (11 WAI), the first inflorescence on each plant was removed and photographed from above using a light-emitting diode (LED) 50.8 × 50.8 cm studio-in-a-box with the standard grey background (FotodioX, Inc., Gurnee, IL, USA). The diameter of each inflorescence was measured, and area calculated using the equation for a sphere (1)

Flower index = (4/3) πr³

(1)

Subsequent to taking photographs, images were imported into ImageJ, a java-based open-source software developed by the US National Institutes of Health [24]. A color card was photographed at the beginning of each session to calibrate the software. In short, the images were imported as a batch and cropped to a rectangular area that included only sepals from the inflorescences. Images were then split into their red, green, and blue (RGB) components. Software color calibration was done by measuring individual swatches on the color card from the RGB components, typing in the actual color values corresponding to the swatches, and plotting the difference on a linear grid. Measurements were then taken for each image for the area of the rectangle, average value, minimum value, and maximum value for each color component. Files were saved in an excel spreadsheet for statistical analysis. Percent blue was defined by Equation (2) and was calculated for each treatment and analyzed

Percent Blue = [Blue/(Red + Blue + Green)] × 100

(2)

2.1.5. Plant Growth, Quality and Flower Index Measurements

During the week of 20 August 2018 (17 WAI) total flower number, plant growth index, and plant quality were assessed on all plants. Photos of representative samples from each of the treatments and quality ratings were taken after measurements were finished. Recently expanded, full mature leaves (approximately 5 g fresh weight per treatment) were collected for foliar analysis, and four replications per treatment were harvested for root and shoot dry weight. Growth index was calculated using the Equation (3)

Growth Index = [(height from the substrate surface + widest width + width transversing the widest width)/3].

(3)

Plant quality was rated on a scale of 0 to 5 with 0 = the plant is entirely dead with little or no green visible; 1 = smallest plant with necrosis and/or off color (chlorotic or purple) leaves; 2 = medium size plant with short internodal distance, may have necrosis or off color (chlorotic or purple) leaves; 3 = medium size plant with longer internodal distance, may have minimal necrosis, some chlorosis, not uniform (asymmetric); 4 = larger size plant with new growth and some vigor, no necrosis, some chlorosis, more symmetrical canopy than 3; 5 = largest size plant showing signs of vigor or active growth, no necrosis, some chlorosis possible, symmetrical canopy (Figure S1). For root and shoot dry weight, shoots were removed at the substrate surface and roots were washed free of substrate. Each biomass was oven dried at 55 °C until no further weight change, and then the dry weight was recorded. Leaf tissue samples were oven dried at 55 °C and sent to a commercial lab (Brookside Laboratories Inc., New Bremen, OH, USA) for standard tissue analysis.

2.1.6. Saturated Media Extract

In the substrate incorporated treatments and the drench treatments, root development was stunted and “pancaked” (i.e., shallow and flat) in the top 2.5 to 3.8 cm of the pot, while roots were spread throughout the container profile in the control plants. Rescue treatments initially had well developed roots, but after treatments were applied, many of the roots were killed and new root growth was shallow, being limited to the upper portion of the pot. Thus, three replications of each treatment were placed in a cooler at 4.4 °C and watered as needed until the week of 10 September 2018, at which time shallow root systems were removed, and the container was evenly separated into upper, middle, and lower sections of substrate. Fertilizer granules were also removed. It was not possible to remove all substrate from the root system; therefore, the upper layer is not an exact representation of the root zone although it is the layer in closest proximity to the roots. Substrate pH and EC of each section were measured using saturated media extract (SME) [25]. In brief, deionized water was added to approximately 500 mL of substrate in a 600 mL glass beaker until water glistened at the surface. Each sample was stirred with a glass rod and then subjected to a 30-min equilibration period prior to measuring pH and EC. Following pH and EC measurements, two coffee filters placed together were used to filter the supernatant into 50 mL glass beakers. After a 2- to 3-h period to allow solids to settle, a 17 mL sample was extracted from the surface using a 20 mL syringe and filtered with 25- or 30- mm diameter 0.45 micron nylon filter. Samples were placed in 50 mL centrifuge tubes and analyzed for Al.

2.2. Bluing a Crop Held Over: Year 2

2.2.1. Treatments and Other Fertilization

Four treatments were potted with Al₂(SO₄)₃ incorporated at 5.93 kg·m⁻³, grown with 2018 experimental plants, then overwintered to simulate grower conditions when saleable plants do not reach market and must be held until the following spring. All plants were pruned to stem height of 12.7 cm and fertilized on 19 February 2019 (43 WAI) in both locations. Plants were topdressed with 11 g of the 19N:1.7P:6.6K controlled release fertilizer previously described. On the same date, two Al rescue treatments were applied with 90-day controlled release Al₂(SO₄)₃ (Florikan Sapphire) top dressed at 15 g per container (TDH) and 9 g per container (TDL). On 11 March 2019 (46 WAI) plants in both locations reached vegetative bud break prompting the Al₂(SO₄)₃ drench rescue treatment, which was applied at rate of 500 mL per pot at 23.97 g·L⁻¹ in TAH and 500 mL per pot at 17.97 g·L⁻¹ in CVF (ASD). The fourth treatment was an untreated control (UC), which received no additional bluing treatment.

2.2.2. Substrate Pore-Water Chemical Properties

During the growing season the Virginia Tech Extraction Method [23] was used to measure pH and EC on half of the replications bi-weekly, beginning 18 March 2019 (47 WAI). pH and EC were measured until final harvest 56 WAI (20 May 2019) using a HI9814 portable meter in CVF, or a HI9811-5 portable meter in TAH. Both locations collected 15 mL aliquots of extractant in conjunction with pH and EC measurements, filtered to 0.45 um, and analyzed for phosphorus and aluminum at the Virginia Tech Soil Testing Laboratory, Blacksburg, VA, USA (ARCOS II Multi-View ICP model FHM22 with CETAC ASX560 Autosampler).

2.2.3. Flower Color Measurements

Beginning 22 April 2019 (52 WAI) in TAH, and 29 April 2019 (53 WAI) in CVF, and continuing to final harvest on 56 WAI, the first inflorescence of each plant was photographed, the color analyzed using ImageJ as previously described, and flower diameter measured to calculate area using the equation for a sphere.

2.2.4. Plant Growth, Quality and Flower Index Measurements

During the week of 20 May 2019 (56 WAI) both locations completed final data collections and harvest which included total flower number, plant growth index, and quality ratings. Growth index and plant quality were measured as previously described. Photos of plants by treatment and quality rating were taken. Leaf tissue samples were collected for each treatment and four replications were harvested for root and shoot dry weight. Leaf tissue samples were oven dried at 55 °C, then sent to a commercial lab (Brookside Laboratories Inc.) for standard tissue analysis.

2.3. Statistical Analysis

In 2018, there were 10 replications of each treatment for both locations, CVF and TAH, and in 2019, there were 10 replications for CVF and 9 replications for TAH of each treatment. The experiments were arranged in a completely randomized block design, with each location and year being independently analyzed to determine the effect of treatment (α = 0.05) by analysis of variance (ANOVA) (JMP Pro 16 SAS Institute Inc., Cary, NC, USA). When a comparing dependent effect (e.g., pH versus dry weight), data was pooled across time (WAI) to compare grand means for each year by treatment. Means separation were conducted using Tukey’s HSD where appropriate (α = 0.05). For the Cluster analysis, RGB color data was averaged for each treatment at each location. Averaged data was imported into JMP (JMP Pro 15.1.0). Cluster analysis was performed using Hierarchical clustering with the Average method. Dendrograms were created for yearly data.

3. Results

3.1. Bluing the Initial Year of Production: Year 1 Results

3.1.1. Substrate Pore-Water Chemical Properties

The average pore-water pH was 4.6 at time of transplanting hydrangea cuttings. No application of Al (CO) resulted in the average pore-water pH ranging from 4.0 to 5.3 over the course of the experiment (Figure 1). The pore-water pH in rescue treatments (FST, AST, RD), which were applied 9 (CVF) or 10 (TAH) WAI, were not different from the control (CO) before treatments were applied. After application, pore-water pH decreased to an average of 3.1 to 3.5. The continuous drench (CD) application decreased average pH to 3.5 by four weeks after initiating drenches, and pH ranged from an average of 3.2 to 3.5 during year 1. Incorporation of Al in the substrate (ASL, ASH, FSL, FSH) decreased pore-water pH (3.2–3.5, p < 0.0001) when compared to all other treatments (pH 3.9–4.4) in both locations within the first three weeks. Soil incorporation buffered the Al, allowing for slight recovery over time with average pH ranging from 3.8 to 4.1 by the end of year 1 (17 WAI), though pH was still lower than the CO, ranging from 4.4 to 4.7 in either location (p < 0.0001).

Substrate incorporation of Al initially increased average pore-water EC, ranging from 0.5 to 1.1 mS·cm⁻¹, approximately two to three-fold when compared to the control receiving no Al (Figure 1). Pore-water EC was higher when receiving a bi-weekly Al₂(SO₄)₃ drench (CD) than any other treatment for the duration of year 1, reaching a maximum of 1.16 mS·cm⁻¹ in Virginia (CVF) at 7 WAI and 0.92 mS·cm⁻¹ in Tennessee (TAH) at 5 WAI. For rescue treatments, which were applied at 9 (CVF) or 10 (TAH) WAI, pore-water EC reached a maximum at 11 WAI, but sharply declined at 13 WAI. Pore-water EC maximums were 1.67, 3.21, and 2.26 mS·cm⁻¹ or 0.71, 1.24, and 1.69 mS·cm⁻¹ in Virginia or Tennessee when using rescue treatment; a rescue drench (RD), topdressed controlled release Al₂(SO₄)₃ (FST), or topdressed ground Al₂(SO₄)₃ (AST) (Table S1). By the end of year 1 average EC was 0.10 to 0.23 mS·cm⁻¹ in all treatments unless continually receiving Al as a drench (CD; 0.3–0.5 mS·cm⁻¹).

3.1.2. Flower Color Measurements

Cluster analysis was performed on RGB color data to determine differences between treatments. Flowers of plants receiving no aluminum (CO) from both locations clustered together and were different from all other treatments receiving Al (Figure 2). Although the rest of the treatments were a separate cluster, locations further grouped separately (CVF vs. TAH), with each having two sub-clusters. For Virginia (CVF), the drenches, RD and CD, and controlled release Al₂(SO₄)₃ treatments FST and FSL clustered. In the TAH location, all soil incorporated Al₂(SO₄)₃ treatments and rescue treatments clustered together, respectively. The continuous drench (CD) performed most similarly to the topdressed, high rate of controlled release Al₂(SO₄)₃ (FST). Percent blue calculations were similar to the cluster analysis in that the CO was different from all treatments, except the rescue drench (RD) in CVF (

Table 2. Percent blue ¹ for the first H. macrophylla ‘Bailmer’ PP15,298 Endless Summer^® Original flower from nine aluminum treatments across two locations in year 1.

Treatment	CVF 2	TAH
Non-Rescue Treatments
CO	34.34 b3	34.04 b
Drench
CD	40.48 a	40.16 a
Incorporated
FSL	39.45 a	38.77 a
FSH	38.65 a	37.96 a
ASL	38.54 a	38.33 a
ASH	38.33 a	38.96 a
Rescue Treatments
RD	38.17 ab	40.09 a
FST	38.65 a	39.96 a
AST 4	.	.
	p < 0.0001	p < 0.0019

¹ Percent blue = (B/∑RBG ∗ 100). ² CVF = Coastal Virginia Flats; TAH = Tennessee Appalachian Highlands, CO = untreated control, CD = bi-weekly drench of Al₂(SO₄)₃, FSL = low rate of controlled release Al₂(SO₄)₃ incorporation, FSH = high rate of controlled release Al₂(SO₄)₃ incorporation, ASL = low rate of granular Al₂(SO₄)₃ incorporation, ASH = high rate of granular Al₂(SO₄)₃ incorporation, RD = Al₂(SO₄)₃ rescue drench, FST = rescue treatment with controlled release Al₂(SO₄)₃ topdressed, AST = rescue treatment with granular Al₂(SO₄)₃ topdressed. ³ Data for each location followed by the same letters are not significantly different, according to one-way analysis of variance and Tukey’s multiple comparison test. ⁴ AST treatment did not flower by the end of the experiment.

). However, none of the treatments were distinctly blue (Figure 3).

3.1.3. Plant Growth, Quality and Flower Index Measurements

No differences in root:shoot ratio were observed in Tennessee (TAH), ranging from 0.30 to 0.58; however, in Virginia (CVF) plants receiving a bi-weekly drench of Al₂(SO₄)₃ (CD; 0.34) differed (p = 0.0008) from those receiving no Al (CO; 0.26). Root:shoot of all Virginia plants ranged from 0.23 to 0.34. At both locations, no treatment had a greater growth index than CO, which ranged from 46.8 to 50.3, however, RD was similar to the CO at both locations, 46.7 at TAH and 46.0 at CVF. Growth index was the lowest when ground Al₂(SO₄)₃ was topdressed (AST), ranging from 5.6 to 12.8, regardless of location, or when continually receiving Al₂(SO₄)₃ as a drench in CVF, 27.4 (Figure 4).

Quality ratings were similar to growth index. The lowest quality plants, 0.0 in CVF and 0.2 in TAH, occurred when applying a topdressed granular Al₂(SO₄)₃ (AST), regardless of location. Low quality ratings were also observed in both locations when plants received bi-weekly Al₂(SO₄)₃ drench, 1.0 in CVF and 1.2 in TAH, or controlled release Al₂(SO₄)₃ (FST), 1.4 in CVF and 1.5 in TAH (Figure 4). Plants receiving no Al (CO) in Virginia (CVF) had better quality than all other treatments (3.6 versus 0.2 to 3.1, respectively), while plants in Tennessee (TAH) receiving only a rescue drench (RD) had better quality than all other treatments (4.15 versus 0.0 to 3.5, respectively). Topdressing granular Al₂(SO₄)₃ (AST) resulted in plant death for the majority of the plants in both locations, causing both low growth index and quality ratings.

There was no effect of Al treatment on flower index at either location (p ≥ 0.0876; Figure 4). However, flower number was higher if plants received no Al (CO), 4.5 compared with 0 to 3.5 at CVF and 6.0 compared with 0 to 4.7 at TAH (p < 0.0001).

Pore-water Al levels in the substrate were initially measured at 1.19 ± 0.12 mg·L⁻¹. At the end of the experiment, Al concentration was greatest when receiving bi-weekly Al₂(SO₄)₃ drenches (CD) when compared to all other treatments, 1.82 mg·L⁻¹ compared with 0.12 to 0.40 mg·L⁻¹ at CVF (p < 0.0001) and 11.59 mg·L⁻¹ compared with 0.10 to 0.34 mg·L⁻¹ at TAH (p < 0.0001) (

Table 3. Aluminum (Al) concentration of substrate pore-water extract via Virginia Tech Extract Method and recently matured leaves for H. macrophylla ‘Bailmer’ PP15,298 Endless Summer^® Original produced in a pine bark substrate and subjected to nine aluminum treatments in two locations during year 1.

Treatment	Al, Extractant (mg·L−1)		Al, Foliar Tissue (µg·g−1 Dry Weight)
	CVF 1	TAH	CVF	TAH
Non-Rescue Treatments Control
CO	0.12 b 2	0.10 b	105 d	121 d
Drench
CD	1.82 a	11.59 a	1237 a	1633 a
Incorporated
FSL	0.37 b	0.26 b	444 c	577 c
FSH	0.30 b	0.23 b	523 bc	501 cd
ASL	0.26 b	0.31 b	534 bc	569 c
ASH	0.29 b	0.24 b	567 bc	492 cd
Rescue Treatments
RD	0.29 b	0.29 b	587 bc	538 c
FST	0.40 b	0.26 b	783 b	1147 b
AST 3	0.39 b	0.34 b	.	.
	p < 0.0001	p < 0.0001	p < 0.0001	p < 0.0001

¹ CVF = Coastal Virginia Flats; TAH = Tennessee Appalachian Highlands, CO = untreated control, CD = bi-weekly drench of Al₂(SO₄)₃, FSL = low rate of controlled release Al₂(SO₄)₃ incorporation, FSH = low rate of controlled release Al₂(SO₄)₃ incorporation, ASL = low rate of granular Al₂(SO₄)₃ incorporation, ASH = high rate of granular Al₂(SO₄)₃ incorporation, RD = Al₂(SO₄)₃ rescue drench, FST = rescue treatment with controlled release Al₂(SO₄)₃ topdressed, AST = rescue treatment with granular Al₂(SO₄)₃ topdressed. ² Data for each location followed by the same letters are not significantly different, according to one-way analysis of variance and Tukey’s multiple comparison test. ³ Plants did not survive to end of experiment.

). Treatment effects were reflected in foliar Al at the end of the experiment with the bi-weekly Al₂(SO₄)₃ drenched (CD) plants having Al concentrations higher than all other treatments, 1237 versus 105 to 783 µg·g⁻¹ dry weight in CVF and 1633 vs. 121 to 1136 µg·g⁻¹ dry weight in TAH (Table 3, p < 0.0001). Tissue levels in plants never receiving Al (CO) in CVF were lower than all other treatments (105 µg·g⁻¹ dry weight, p < 0.0001), however, in TAH, the high level of substrate incorporated Al regardless of source (FSH, ASH) was not different from treatment receiving no Al (CO).

In Virginia (CVF), there were no differences in foliar nitrogen (N) levels (p = 0.3914) at the end of the season; however, plants receiving no Al (CO; 2.21%) or a low rate of incorporated granular Al₂(SO₄)₃ (ASL, 1.69%) were below the recommended range (

Table 4. Foliar analysis of H. macrophylla ‘Bailmer’ PP15,298 Endless Summer^® Original subjected to nine aluminum treatments in two locations in year 1.

Treatment		N (%)	K (%)	P (%)	Mg (%)	Ca (%)
Recommended rate [26]		2.24–5.6%	2.2–7.8%	0.25–0.70%	0.22–0.61%	0.60–2.00%
CVF 1	CO	2.21	1.30	0.17 bc 2	0.31 a	0.91 a
	CD	2.40	1.29	0.12 d	0.18 d	0.50 c
	FSL	2.39	1.38	0.23 a	0.29 ab	0.74 ab
	FSH	2.66	1.32	0.23 a	0.30 a	0.77 a
	ASL	1.69	1.38	0.20 ab	0.25 abc	0.68 abc
	ASH	2.42	1.35	0.23 a	0.28 abc	0.75 ab
	RD	2.61	1.31	0.14 cd	0.23 bcd	0.67 abc
	FST	2.71	1.49	0.14 cd	0.21 cd	0.52 bc
	AST 3	.	.	.	.	.
	p-value	0.3914	0.2705	<0.0001	<0.0001	<0.0001
TAH	CO	1.59 c	0.96	0.17 ab	0.31	1.29
	CD	2.28 a	1.06	0.12 b	0.26	1.08
	FSL	1.94 abc	0.84	0.18 ab	0.32	1.35
	FSH	1.83 abc	0.86	0.18 ab	0.32	1.26
	ASL	1.85 abc	0.81	0.18 ab	0.36	1.56
	ASH	2.11 abc	0.96	0.22 a	0.30	1.25
	RD	1.80 bc	0.92	0.14 ab	0.32	1.24
	FST	2.44 ab	1.07	0.14 ab	0.29	1.08
	AST	.	.	.	.	.
	p-value	0.0087	0.5181	0.0765	0.3150	0.1694

¹ CVF = Coastal Virginia Flats; TAH = Tennessee Appalachian Highlands, CO = untreated control, CD = bi-weekly drench of Al₂(SO₄)₃, FSL = low rate of controlled release Al₂(SO₄)₃ incorporation, FSH = high rate of controlled release Al₂(SO₄)₃ incorporation, ASL = low rate of granular Al₂(SO₄)₃ incorporation, ASH = high rate of granular Al₂(SO₄)₃ incorporation, RD = Al₂(SO₄)₃ rescue drench, FST = rescue treatment with controlled release Al₂(SO₄)₃ topdressed, AST = rescue treatment with granular Al₂(SO₄)₃ topdressed. ² Data for each location followed by the same letters are not significantly different, according to one-way analysis of variance and Tukey’s multiple comparison test. ³ AST treatment did not survive to the end of the experiment.

). Similarly, the majority of treatments in Tennessee (TAH) were below recommended foliar N except plants receiving a bi-weekly Al₂(SO₄)₃ drench (CD), 2.28%, and controlled release Al₂(SO₄)₃ topdress (FST), 2.44%, both being greater than plants receiving no Al (CO), 1.59% (p = 0.0087). End of season foliar potassium (K) and phosphorus (P) levels were low in all treatments in both locations [26]. No differences were seen between treatments in either location for K (p ≥ 0.2705). Differences were seen in Virginia (CVF) for P, with those receiving a bi-weekly Al₂(SO₄)₃ drench (CD; 0.12%) being lower than the CO (0.17%). Substrate amended with controlled release or granular Al₂(SO₄)₃ (ASH, FSH, FSL; all 0.23%) were greater than the plants receiving no Al (CO; p < 0.0001). No treatments were different from the CO for P in Tennessee (TAH), but plants that received a bi-weekly Al₂(SO₄)₃ drench (CD, 0.12%) differed from those with a high substrate incorporated ground Al₂(SO₄)₃ (ASH; 0.22%). Magnesium (Mg) and calcium (Ca) levels were below the recommended range when receiving bi-weekly Al₂(SO₄)₃ drench (CD), 0.18 and 0.50, respectively, and controlled release Al₂(SO₄)₃ topdress (FST), 0.21 and 0.52, respectively, in CVF, and lower than plants receiving no Al (CO; 0.31% and 0.91%, respectively; p < 0.0001). Mg and Ca levels were not different in Tennessee (TAH) (p ≥ 0.1694) and within recommendations.

3.1.4. SME Measurements

The saturated media extract technique was performed on the upper, middle, and lower strata at the end of the year 1 season. pH was not affected by strata (p ≥ 0.2128) at either location. In Virginia (CVF), EC was affected by strata with EC decreasing with increasing depth (p < 0.0001); however, EC was ≤0.18 mS·cm⁻¹ across all strata. Al was greatest for the upper and middle strata, 0.98 mg·L⁻¹ compared with 0.78 mg·L⁻¹ in the lowest strata. In Tennessee (TAH), substrate EC and Al were not affected by strata (p = 0.3023 and 0.0502, respectively).

3.2. Bluing a Crop Held Over: Year 2 Results

3.2.1. Substrate Pore-Water Chemical Properties

In both locations, pore-water pH was lower in rescue treatments (ASD, TDL, and TDH), ranging from 3.0 to 3.3, compared 3.5 to 4.0 at 47 WAI (p < 0.0001) of the control (UC) that received no additional Al, only Al₂(SO₄)₃ incorporated in the prior year (Figure 5). In Virginia (CVF), average pH for UC decreased from 3.5 to 3.2 by the end of the experiment (56 WAI) but remained higher than rescue treatments. In Tennessee (TAH), pH increased in all rescue treatments. At the end of the experiment (56 WAI), rescue treatment pH was equal to that of UC pore-water (p = 0.0984).

Average substrate pore-water EC was higher in the rescue treatments (ASD, TDL, and TDH) after application than the UC in both locations (p < 0.0001). The differences between treatments decreased and at the end of the experiment there were no differences in Virginia (CVF) (p = 0.1966), while differences remained in Tennessee (TAH; p = 0.0154) at 56 WAI.

3.2.2. Flower Color Measurements

Cluster analysis showed no differences between the treatments and control in both locations, but the locations clustered separately (Figure 6). Percent blue was not different in CVF (p = 0.6519), while plants from Tennessee (TAH) receiving topdressed Al₂(SO₄)₃, had a higher percentage of blue, 44.42% for TDH and 45.75% for TDL, compared to the control (UC; 39.44%) (p = 0.0006;

Table 5. Percent blue for the first H. macrophylla Endless Summer^® flower from four aluminum treatments across two locations in year 2.

Treatment	CVF 1	TAH
UC	40.97	39.44 c 2
ASD	41.43	41.28 bc
TDL	41.60	45.75 a
TDH	42.29	44.42 ab
p-value	0.6519	0.0006

¹ CVF = Coastal Virginia Flats, TAH = Tennessee Appalachian Highlands, UC = untreated control, ASD = rescue treatment with Al₂(SO₄)₃ drench, TDL = rescue treatment with low rate of 90-day controlled release Al₂(SO₄)₃ topdress, TDH = rescue treatment with high rate of 90-day controlled release Al₂(SO₄)₃ topdress. ² Data for each location followed by the same letters are not significantly different, according to one-way analysis of variance and Tukey’s multiple comparison test.

).

3.2.3. Plant Growth, Quality and Flower Index Measurements

Root:shoot ratios ranged from 0.33 to 0.62 in Virginia (CVF) and 0.51 to 0.82 in Tennessee (TAH). In both locations only the high topdress rate of 90-day controlled release Al₂(SO₄)₃ (TDH) was different from the control (UC), 0.62 and 0.82, respectively (p = 0.0008 and = 0.0046).

In both locations, the high topdress rate of 90-day controlled release Al₂(SO₄)₃ (TDH) had a reduced growth index (p < 0.0049) and lower quality rating (p < 0.0226) than the UC (Figure 7). In Virginia (CVF), the two other rescue treatments, Al₂(SO₄)₃ drench (ASD) and low rate of controlled release Al₂(SO₄)₃ topdress (TDL) were not different from the UC or TDH for growth index or plant quality. Growth index and quality rating for Al₂(SO₄)₃ drench (ASD) were not different from the control (UC) plants in Tennessee (TAH); however, they were both greater than both topdressed rescue treatments (Figure 7). Flower size was not affected by treatments in either location (p = 0.4280 in CVF and p = 0.2064 in TAH), however, flower number was highest in when plants received Al only the prior year (UC having 7.1 in CVF and 7.8 in TAH compared with 3.6 to 5.3 for all other treatments in CVF and 3.1 to 6.8 in TAH).

Initial Al concentrations in extractant were high for the rescue treatments, ranging from 38.8 to 113.5 mg·L⁻¹ in CVF and 61.1 to 186.6 mg·L⁻¹ in TAH (Figure 8). When pooled across time, pore water Al in treatments receiving the drench (ASD) and topdressed with the high rate of 90-day controlled release Al₂(SO₄)₃ (TDH) were greater than the control (UC) in both locations (p = 0.003 and p < 0.0001 in CVF and TAH, respectively). Al applied as a drench remained mobile and was easily extracted when compared to the coated Al topdress treatments. More than 75% of the Al was leached out within three weeks of application, compared to twelve weeks for the topdress applications (Figure 8). The control (UC) had a low level of Al in extractant despite not having any Al applied since the previous season. At the end of the experiment foliar Al was lower in the UC compared with all other treatments in both locations (p < 0.0001).

In Virginia (CVF), foliar Al was 237 µg·g⁻¹ dry weight in the control (UC) that received no Al the second year and other treatments ranged from 752 to 1283 µg·g⁻¹ dry weight. Foliar Al of UC in Tennessee (TAH) was approximately double that of CVF (418 µg·g⁻¹ dry weight) and other rescue treatment were similar, ranging from 897 to 1230 µg·g⁻¹ dry weight Al. The highest Al foliar concentration was observed in the two controlled release topdress treatments.

Foliar N was sufficient at both locations, ranging from 2.93–3.99% in Virginia (CVF) and 2.77–3.68% in Tennessee (TAH). In Virginia (CVF), foliar K, P, and Ca levels were low compared to recommended levels [26] (

Table 6. Foliar analysis of H. macrophylla ‘Bailmer’ PP15,298 Endless Summer^® Original subjected to four aluminum treatments in two locations during spring of the 2^nd production season (year 2).

Treatment		N (%)	K (%)	P (%)	Mg (%)	Ca (%)
Recommended rate [26]		2.24–5.6%	2.2–7.8%	0.25–0.70%	0.22–0.61%	0.60–2.00%
CVF 1	UC	2.93 b 2	1.49 b	0.17	0.22 ab	0.34 ab
	ASD	3.76 a	1.80 ab	0.17	0.26 a	0.42 a
	TDL	3.58 a	1.64 b	0.12	0.22 ab	0.32 b
	TDH	3.99 a	2.05 a	0.19	0.21 b	0.31 b
	p-value	<0.0001	0.0013	0.0526	0.0317	0.0069
TAH	UC	2.77 b	0.87 b	0.14	0.38 a	0.78 a
	ASD	2.99 ab	1.14 b	0.12	0.37 a	0.69 ab
	TDL	3.54 ab	1.58 a	0.13	0.32 ab	0.57 bc
	TDH	3.68 a	1.65 a	0.14	0.25 b	0.49 c
	p-value	0.0287	<0.0001	0.1533	0.0005	<0.0001

¹ CVF = Coastal Virginia Flats, TAH = Tennessee Appalachian Highlands, UC = untreated control, ASD = rescue treatment with Al₂(SO₄)₃ drench, TDL = rescue treatment with low rate of controlled release Al₂(SO₄)₃ topdress, FSH = rescue treatment with high rate of controlled release Al₂(SO₄)₃ topdress. ² Data for each location followed by the same letters are not significantly different, according to one-way analysis of variance and Tukey’s multiple comparison test.

). Foliar K was greater when Al was topdressed with the high rate of 90-day controlled release Al₂(SO₄)₃ (TDH; 2.05%) than for UC (1.49%) and TDL (1.64%; p = 0.0013). P ranged from 0.12 to 0.19% and no difference was observed among treatments (p = 0.0526). Foliar Ca was highest in the drench (ASD; 0.42%), but not different from control (UC; 0.34%). Mg levels were below the recommended levels for the high Al₂(SO₄)₃ topdress rate (TDH; 0.21%), but sufficient for all other treatments.

In Tennessee (TAH), foliar K and P levels were below recommended ranges [26]. The topdress treatments had a higher level of K than the control (UC; p < 0.0001), but no differences were seen between treatments for foliar P (p = 0.1533) Topdress treatments had low Ca levels compared to recommended levels and were significantly lower than hydrangeas receiving no Al (UC) in year 2 (p < 0.0001).

4. Discussion

4.1. Bluing the Initial Year of Production: Year 1 Discussion

Hydrangea macrophylla is a relatively pH, salt, and Al-tolerant species; especially within the varied growing conditions suggested by literature and nursery crop producers. Once in the plant, Al binds with citrate to form an Al-citrate complex in the sap that is transported and sequestered in the vacuoles [27,28,29]. However, Al has been shown to alter root morphology and mineral nutrient uptake of Al tolerant corn and wheat, creating stubby roots [30] and could in turn be toxic or growth limiting if concentration of Al is too high over long exposure times [31], impacting carbon allocation and root and shoot biomass. This negative impact was observed in the differences of hydrangea root to shoot ratio in Virginia during year 1 and in both locations in year 2, when a high rate of 90-day controlled release Al₂(SO₄)₃ was applied as a topdress. Application of a rescue topdress treatment or bi-weekly drench (CD) generally resulted in the most plant deterioration or death, yet the color was not bluer. There was a 61.1–94.4% and 56.5–100.0% decrease in plant quality from these Al treatments in Virginia (CVF) and Tennessee (TAH), respectively. Topdress rescue treatments also generally had fewer flowers than plants receiving no Al (CO). Additionally, observations that root growth was limited largely or entirely to the top 3 cm of the container, i.e., “pancaking” in addition to the aforementioned phytotoxicity after Al treatments were applied, prompted further investigation of pH, EC, and Al concentration-including but not limited to understanding the electrochemical properties of Al occurring in the top, middle, and bottom strata of the container profile.

Aluminum sulfate is a strongly acidic compound with a pKa of −3. As such, application of Al₂(SO₄)₃ lowers substrate pH; thus, obtaining and maintaining the desired pH becomes even more of a challenge, yet critical to ensuring a blue flower color and a healthy hydrangea plant. Al³⁺ is considered available for horticultural purposes at pH ≤ 5.5, while most plants require a pH above 5.0 for optimal growth, creating a very narrow target range for Hydrangea macrophylla producers. The recommendation for H. macrophylla ‘Bailmer’ PP15,298 Endless Summer^® Original on culture sheets is to maintain a substrate extract pH between 4.8 to 5.0 [32]. However, blue flowering hydrangeas are being produced in substrates with pH ranging from 4.0 to 5.7 (Jeff Stoven, Bailey Nursery, personal communication).

Root zone pH of less than 4.0 can cause a decline in hydrangea root health and hinder growth when produced in soilless culture [10,17]. In a hydroponic system, Naumann et al. [33] demonstrated damage in the form of lesions on hydrangea root tips when grown in the presence of Al at a pH > 4.0 (4.3). In the present study, substrate pH for the four treatments that incorporated Al remained < 4.0 until the end of the experiment, and then only exceeded 4.0 in Tennessee (TAH). In our study, the most dramatic decline in pH was observed when the substrate was continuously drenched with dissolved Al₂(SO₄)₃ during the first year, resulting in a pH ≤ 3.5 beginning 5 WAI and remaining at or below that pH for duration of the first year (Supplementary Table S1). Therefore, we hypothesize root health was also affected following application of Al₂(SO₄)₃ in rescue treatments when initial pore-water pH ranged from 4.2 to 5.2 before decreasing to 3.1 to 3.52 (Figure 1 and Supplementary Table S1). Thus, plants may have been subjected to a sub-optimal pH in the presence of high concentrations of Al³⁺; although pH did not differ across substrate strata (p-value ≥ 0.2128) averaging 3.7 in Virginia (CVF) and 4.12 in Tennessee (TAH). An exception may be the upper 2.5 cm, which was inhabited by roots; however, pH was not measured for this sub-strata. Tharpe et al. [34] elicited similar low pH and concomitant decline in crop health among H. macrophylla Bloomstruck^® when grown in soilless substrate in the absence of exogenous lime or Al and topdressed with 90% sulfur. Conversely, Wallace and Wieland [35] demonstrated no deleterious effect on plant dry weight when producing greenhouse grown H. macrophylla, with blue sepals, using containerized soil at a pH 3.5 (versus 4.8 or 7.5) [35]. To counter the effects of low pH on the health of the plants, dolomitic lime is often applied to the substrate. Application of dolomitic lime also supplies Ca and Mg while increasing substrate pH to optimal levels for plant health; however, the increased pH decreases Al³⁺ availability causing H. macrophylla flowers to be more purple than blue [20] even when the pH was reportedly low enough for Al uptake. We did not include lime in order to determine the maximum, unadulterated effect of different types of Al fertilization on substrate solution pH, plant health, and flower color. Ca and Mg levels were sufficient, i.e., not limiting, in both locations [26] and commensurate with other research in which hydrangea was grown at a low pH and in the presence of Al [16,36]. However, our pH remained well below the acceptable range for most plants, even among controls, and some treatments led to crop loss. Thus, our conclusion is that growers who are not incorporating lime are strongly cautioned about using similar rescue treatments. Future studies could include applying rescue treatments in conjunction with, or closely following, incorporating lime or using liquid lime application to maintain pH > 4 [10,17] and determine if they can be effective and safe.

Upon application of Al treatments, an increase in EC was observed in conjunction with the decrease in pH. Current recommendations for H. macrophylla suggest an EC from 1.5 to 2.7 mS·cm⁻¹ [32]; however, pore-substrate water EC as high as 13.1 mS·cm⁻¹ have been observed in waste water suitability trials with no detrimental effect on H. macrophylla growth [36]. Conversely, average leachate EC of 6.9 or 7.4 mS·cm⁻¹ resulted in a 6% or 16% decrease, respectively, in H. macrophylla ‘Leuchtfeuer’ shoot dry weight when being continuously irrigated with a saline water, with or without flushing, and compared to a control (freshwater EC 3.23 mS·cm⁻¹) [37]. After application of Al₂(SO₄)₃, EC in all treatments was within the recommended range, except at 11 WAI immediately following the rescue treatments in Virginia (CVF) where Al₂(SO₄)₃ was top-dressed, regardless of controlled release (FST) or water soluble (AST) availability, resulting in an EC of 2.26 and 3.21 mS·cm⁻¹, respectively, or EC of 1.69 mS·cm⁻¹ after an application of ground Al₂(SO₄)₃ (AST) in Tennessee (TAH) (Supplementary Table S1). By 15 WAI, EC had decreased to below recommended levels in all treatments except the bi-monthly drench (CD) in Virginia (CVF). Individual mineral nutrient levels were not measured as substrate extracts at the end of the first year, but measured Al levels were < 0.40 mg·L⁻¹ in all treatments except the substrate receiving a bi-monthly drench (CD) regardless of location. Average pore-water extract Al concentrations of plants receiving Al₂(SO₄)₃, a proxy for potential leaching, ranged from 0.23 to 11.59 mg·L⁻¹. With an average 1.10 mg Al·L⁻¹. The upper concentration measured in this experiment is 1.4-fold greater than the upper limit in the recently revised EPA Al freshwater aquatic criteria, 4.8 mg·L⁻¹ for protection of 95% of aquatic genera (criteria is water quality dependent). This criterion may limit future unconstrained use of Al₂(SO₄)₃ as an agrichemical if regulatory action is enacted.

Foliar Al levels were affected by treatment both directly, i.e., rate, application method, and timing, and indirectly through root damage that impaired growth and subsequent uptake from the bulk substrate solution. With a root system limited to the upper 3 cm, it is expected that overall Al uptake would be reduced but that treatments applied to that zone would be better able to supply Al. In fact, there were lower levels of Al in the plant tissue from soil incorporated treatments than when Al₂(SO₄)₃ was continuously drenched on a bi-weekly schedule (CD) (Table 3). With each drench application, Al³⁺ would become abundant in the substrate solution, allowing for plant uptake at every application. In Tennessee (TAH), there was a corresponding increase in foliar Al for FST, 1147 µg·g⁻¹ dry weight, which was only less than 1633 µg·g⁻¹ dry weight foliar Al when bi-weekly drenched (CD), (Table 3). Greater uptake into the plant from coated Al topdress treatment is likely due to Al being released over time, rather than all at once as with the rescue drench treatment. However, it must be emphasized that the topdress rescue treatments caused crop damage and death (Figure 4). Like Blom and Piott [16], in the present study we also saw low concentrations of K and P in the substrate; however Osaki et al. [38] reported stimulation of plant growth, indicating an increase in N, P, and K uptake by Al-accumulators such as H. macrophylla grown in soil, with some of the increase in P concentrations linked to Al-P precipitation in the roots. Due to the low pH observed in experiments herein, the substrate pore-water concentration of neither P nor Al were affected by Al-P precipitates prior to hydrangea uptake [39].

Al uptake was sufficient to change the sepal color from pink to purple in all Al treatments; however, no treatments achieved true blue color. Recommendations for minimum foliar or sepal Al concentration associated with blue sepals include 1000 µg·g⁻¹ sepal dry weight (several cultivars) [40], 1300–2000 µg·g⁻¹ foliar dry weight (‘Early Blue’) [18], or 320 µg·g⁻¹ sepal dry weight, (‘R.F. Felton’) [10]. Schreiber et al. [8] noted there were no differences when sampling leaf tissue compared to sepals, i.e., if sepals were blue, Al concentrations in the leaves and sepals were greater than 40 µg·g⁻¹ fresh weight. In the present study, all foliar Al concentrations besides the CO were greater than 250 µg·g⁻¹ dry weight as recommended by Allen [1]; however, only one treatment in one location, CD in TAH, yielded the minimum foliar Al level suggested by Landis et al. [18], 1300 µg·g⁻¹ dry weight. When Al concentrations in the foliar tissue were greater than 150 µg·g⁻¹ dry weight, as was the case with all Al treatments, sepal color was purple. Schreiber et al. [41] noted that extractable anthocyanin content varied for H. macrophylla cultivars with Endless Summer^® Original being in the 2^nd lowest bluing group. Cultivars in the vivid and vibrant groups typically had 2.0 to 4.1 (up to 6.7)-fold greater extractable anthocyanin levels. Optimum foliar Al levels will be greater in cultivars with higher anthocyanin content and may partially explain the variation in Al concentration associated with blue sepals in the literature [41].

Other possible causes of poor root development and subsequent decreased shoot dry mass in Al treated plants is the possible lack of a co-factor that aids in Al accumulation or tolerance when produced in soilless substrates or undesirable edaphic conditions that occur when Al₂(SO₄)₃ was applied. Je-Ju Do Island, South Korea (formerly Quelpaert Island) and Chiba Peninsula, Japan and the surrounding islands were the original location of the hybrid parent plant materials; subsp. Macrophylla and subsp. Serrata, respectively [42] of H. macrophylla ‘Bailmer’ PP15,298 Endless Summer^® Original. All are maritime locations that provide a well-drained volcanic soil high in Al and appear to be abundant in silicon (Si). For example, on Je-Ju Do Island, Si from dissolved monosilicic acid (H₄SiO₄) has been reported to be 33 to 63 mg·L⁻¹ in water and average available soil Si content up to 150 mg·kg⁻¹ [43,44]. Silicon is a known co-factor in ameliorating Al toxicity in root cell walls of wheat, dependent on soil pH, even when already exhibiting tolerance through organic acid (malate) root exudates [45].

Soilless substrates, including sphagnum peat and pine bark have been reported to have low amounts of plant available Si. Foliar Si was ~900 mg·kg⁻¹ Si in Zinnia elegans grown in a soilless substrate, regardless of total extractable content. However, when compared to uptake of zinnia grown in a Si-enriched hydroponic solution, foliage contained 12,682 mg·kg⁻¹ Si [46]. In support of this hypothesis, Wallace and Wieland [35] compared foliar and flower mineral nutrient concentration of blue and pink flowering H. macrophylla in the Los Angeles, California landscape. They found little difference in foliar Al concentration that ranged from 510 to 480 µg·g⁻¹ Al in blue or pink flower clusters, respectively. However, Wallace and Wieland observed significant differences in silicon (Si), iron (Fe), manganese (Mn), titanium (Ti) as well as molybdenum (Mo), chromium (Cr), and tin (Sn), possibly a result of unreported differences in soil pH. In the same paper, Wallace and Wieland reported blue flower clusters occurred in H. macrophylla when containerized soil pH was 3.5 (versus 4.8 or 7.5) in a companion greenhouse experiment, observing an increase in foliar and flower Al, Mn, Mg, and Si concentration without any significant difference in plant dry weight [35]. In our study we saw little differences in Mg or Mn when producing plants across treatments and all were within allowable ranges, leaving Si as possible synergist for Al in H. macrophylla. Furthermore, increased Si availability, not reported by authors, could also be a reason success in bluing H. macrophylla has been seen when using aluminosilicate clays versus solely being another source of Al [10,11,12,13,14,15].

4.2. Bluing a Crop Held Over: Year 2 Discussion

Few studies have looked at Al uptake in hydrangea when held over from a previous year. Blom and Piott [16] examined containerized, summer-grown hydrangea that were treated with Al₂(SO₄)₃ in late summer before being brought into a cooler for spring forcing. They found Al was stored mainly in the roots of dormant plants with some translocation to the stems and buds. Field-treated plants forced for spring bloom were similar in color to plants treated with low Al₂(SO₄)₃ at forcing. Okada and Okawa [47] reported uptake of Al from peat:clay media when growing hydrangea prior to forcing, however 70–80% of Al was lost due to defoliation during cold storage. We incorporated Al₂(SO₄)₃ into the substrate at potting and grew plants for a full season before overwintering. No additional Al was applied to control plants the second spring, yet bluing of the sepals occurred. Extractant analysis through the second season showed low levels of Al were present in the substrate. Color analysis in Virginia (CVF) showed no differences between treatments and control for both percent blue and cluster analysis, while percent blue in Tennessee (TAH) was different between plants receiving no Al (UC) and the rescue topdress treatments. However, all plants had purple sepals, indicating the even when plants received no additional Al (UC), Al was either stored in the roots or taken up from the low levels in the substrate.

Once the experiment was complete, plants from TAH were held in the bowhouses in Tennessee (TAH) and observed for an additional two years. Bluer flowers were observed on some of the plants when compared to the original study. Extractant from plants receiving Al in year 1 (UC) or topdressed with Al₂(SO₄)₃ in year 2 (TDL and TDH) had a low level of residual Al (0.6, 0.7, and 0.8 µg·ml⁻¹, respectively), whereas when a rescue drench of Al₂(SO₄)₃ was applied in the second year (ASD), slightly higher levels of Al (1.5 µg·ml⁻¹) were recorded. These Al values are similar to the low level of Al that remained at 56 WAI in year 2 when UC, TDL TDH and ASD, were 0.4, 2.1, 1.6, and 1.6 µg·ml⁻¹, respectively. Foliar Al for UC, ASD, and TDL had decreased by 40–50% in all treatments, regardless of their original level, which ranged from 418 µg·g⁻¹ dry weight to 1230 µg·g⁻¹ dry weight. However, foliar Al for plants receiving the high rate of controlled release Al₂(SO₄)₃ topdress (TDH), which was visually the bluest increased by 8%, from 1089 µg·g⁻¹ dry weight to 1180 µg·g⁻¹ dry weight. Blue component (B) increased 12.5%, from 166.9 (2019) to 187.7 (2021), though there was no increase in calculated % blue, 44% at both timepoints. Substrate pH for the rescue drench (ASD) and half of the plants receiving low rate of controlled release Al₂(SO₄)₃ (TDL) remained below 5.0, while UC, half of the TDL, and TDH had a substrate pH above 5.0.

5. Conclusions

None of the treatments yielded blue flowers and most foliar Al levels remained below the highest range in the literature, 1300–2000 µg·g⁻¹ dry weight, recommended in greenhouse forced hydrangea production [18]. However, all treatments except the control achieved foliar Al levels commensurate with those found on blue flowering plants [1]. There seems to be no definitive, singular reason that explains why the hydrangeas in the present study were not a “true blue”, nor is there a clear definition of what a “true blue” is in consumers eyes’ or the literature. Similarly, the issue of root system “pancaking” a known phenomenon in H. macrophylla production and observed in these experiments, has not been investigated. Some treatments caused severe plant damage or death. Growers are advised to carefully consider the role of lime and the balance between fostering Al availability and maintaining a substrate pH conducive to H. macrophylla health and growth. Observing the role that Al application may have in raising the pH at which damage occurs is of particular significance. Future treatments exploring different application timings or forms of Al, combinations of Al₂(SO₄)₃ drench and controlled release applications in combination with lime and Si treatments would complement this research. More research is needed on predicting flower color in outdoor production systems and the potential to incorporate nonsulfate-based Al sources such as clays pre-charged with Al and possibly incorporation of wollastonite to simultaneously deliver Al and supplemental Si to promote blue flowers at a higher pH, thus avoiding the risk of an extremely low pH, anaerobic environment, or Al toxicity.