RETRACTED: Phytoremediation of Heavy-Metals-Contaminated Soils: A Short-Term Trial Involving Two Willow Species from Gloucester WillowBank in the UK

Abstract

Phytoremediation, as a bioremediation process in which plants are used to remove contaminants from an environment, has proved to be a practical and low-cost strategy for recovering mining-affected areas. This study aims to assess the potential for use in phytoremediation of two willow species, Salix viminalis and Salix dasyclados, by testing their potential for cleaning-up a range of soils with differing heavy metal concentrations: Pb (111, 141, 192 and 249 mg /kg), Zn (778.6, 1482, 2734 and 4411 mg/kg) and Cd (3.00, 5.03, 9.14 and 16.07 mg/kg). The extracted metals were preferentially translocated to the leaves with considerably higher concentrations and relative BAFs in the case of S. viminalis. The highest recorded Zn concentration of over 0.5% was found in the leaves of S. viminalis growing in soil 4. However, under the conditions of the experiments, S. dasyclados showed greater potential for use in phytoremediation, especially if coupled with use of biomass for energy production. An assessment of the suitability of willow species in this role, with regard to wider aspects involved, such as use of resultant biomass and/or waste management, revealed good potential. Willows are fast growing, grow vigorously from coppiced stumps and have extensive root systems. Therefore, their use in bioenergy production through pyrolysis or combustion, coupled with flue gas screening, is strongly advised.

1. Introduction

Anthropogenic activity, combined with natural element concentration rises in certain cases, has resulted in significant sections of land worldwide becoming contaminated with potentially dangerous compounds, both organic and inorganic, over the previous two to three centuries. Metal mining and processing has been, and continues to be, a key contributor to this. According to reports, there are more than five million places worldwide where heavy metals have contaminated the soil. The Environment Agency [1] listed over 1300 ex-mining sites in the UK, which have led to significant contamination of both land and waters. Engineering-based strategies for cleaning up contaminated land are often ex situ (dig and dump, soil washing and solidification for example) and therefore involve considerable transport costs along with the costs of treatment and/or disposal [2]. The cost of excavation and burial of one acre of contaminated soil at a hazardous waste site in the USA was approximately USD 1,000,000 in 1997, with the cost of remediating US sites at this time estimated at over USD 300 billion [3]. Problems of land contamination are somewhat elevated in the developing world where legislation is often not in place to regulate the emissions from industrial processes and power generation. Phytoremediation has not been widely embraced in the emerging economies, mainly due to social and economic constraints and because these countries have less control over their regulations [4]. Annual human inputs to the environment in European countries are estimated to be 5580 tonnes for Pb and 181 tonnes for Cd, according to estimates. Heavy metals and mineral oils are the most common contaminants, accounting for approximately 60% of soil pollution and 53% of groundwater contamination [5,6,7].

Bioremediation techniques have therefore been developed as a cost-effective alternative and have provided additional benefits to the overall remediation of contaminated sites. Bioremediation can be described as the utilisation of biological processes for remediation of a medium via removal, immobilisation or degradation of the contaminants [8]. One major advantage of bioremediation techniques over traditional methods is that natural soil systems and micro-organisms tend to be left intact [9]. Fungi, bacteria, biological waste materials, plants and trees have all been used as biological media and processes in the remediation area. The use of plants and trees, often known as phytoremediation, is the subject of this research. There are a number of ways in which the processes involved in plant growth and the characteristics of individual species have been utilised. Phytodegradation and phytovolatilisation apply in general to organic contaminants, which can often be broken down to less toxic constituents, or volatilised [10]. The latter can apply in the case of some inorganic substances, such as Hg and Se, which can also be volatilised. Phytostabilisation involves the immobilising of contaminants by conversion to less mobile oxidation states or by adsorption to root material [11], and rhizofiltration involves the removal of contaminants from waters by root processes. Phytoextraction is the process of contaminant uptake and subsequent translocation to harvestable plant fractions [12]. Scientists consider phytoextraction a subprocess of phytoremediation in which plants remove dangerous elements from soil or water, most usually heavy metals [7,13,14].

Initially, phytoremediation research focused mostly on a group of plants known as hyperaccumulators. These are plants that can collect metals in their tissues at hundreds of times the amounts found in other plants [15,16]. Species that were capable of such levels of accumulation were discovered by observing plants, which colonised contaminated areas. A commonly quoted example of a hyperaccumulator, Thlaspicaerulescens, has been shown to accumulate Zn to concentrations of over 3% and Cd up to 0.1% [17]. Numerous studies were carried out investigating the accumulation potential of hyperaccumulators. However, the potential for their use in practical field-scale applications is limited by a number of factors. They tend to be herbaceous plants with very little biomass and shallow rooting systems [18]. Cultivation and harvesting on a large scale are, therefore, difficult, and the plants would only remove contaminants from the surface layer of the soil. The focus of research interests in the field of phytoremediation has shifted towards plants with higher biomass and deeper rooting zones, which can also accumulate metals and other contaminants, albeit at considerably lower concentrations than hyperaccumulators. Research has also been carried out, which investigates the potential for increasing accumulation levels, either by soil amendments [19] or by genetically engineering the accumulation characteristics of the plants themselves [20].

Willows have a number of qualities, which are favourable when considering their use in this way. They exhibit a very vigorous growth pattern and respond well to short-term coppicing, growing back abundantly from stumps [21]. They have an extensive root network, including a considerable amount of fine fibrous root material [22], and they have been shown to accumulate some metals and translocate them to harvestable plant fractions [23].

In respect to real field-size phytoremediation applications, there are a number of challenges that must be recognized and handled. These include cost, timescale and disposal, and/or potential use of the resultant contaminated biomass. The properties and extent of contamination as well as site characteristics also play a major role in determining the feasibility of potential phytoremediation projects.

This study aims to assess the potential for use in phytoremediation of two willow species, Salix viminalis and Salix dasyclados, by the process of pot experiments using soils with a range of levels of contamination with Zn, Cd and Pb. Other factors relating to the suitability of the two willow species for use in this role, such as those listed above, are also assessed with regard to current research in the field.

2. Materials and Methods

2.1. Preparation of Plant Species

In order to compare the potential for use in phytoremediation applications of the two willow species, Salix viminalis and Salix dasyclados, three cuttings of each species were planted in each range of four soils, contaminated to a varying degree with Zn, Cd and Pb. The willow cuttings were obtained from a commercial supplier (The Willow Bank, Gloucestershire, UK) in 30 cm lengths of equal thickness. One centimetre was removed from each cutting to give a composite sample for each species. These samples were then ashed and analysed for Pb and Zn content using an atomic absorption spectro-photometer (AAS) using a 10% nitric acid extraction. No Pb was detected for either species, and Zn concentrations were 63.8 mg/kg for S. viminalis and 65.7 mg/kg for S. Dasyclados. At this time, the scope of the study did not include Cd, but it can be assumed, due to low Zn levels, that the initial Cd content was negligible.

2.2. Preparation of Soils Samples

The four soils were made up of a mixture of commercial compost (Levington’s All-purpose compost, Frimley, UK) and material from a spoil tip at Dyffryn Castell zinc mine at grid reference SN-77274 81365. The ratios of compost to spoil material were estimated to give an appropriate range of concentrations using results from a previous analysis of the 5 mm fraction of spoil from this location [24], which showed concentrations of 29,675.98, 32.91 and 94.89 mg/kg for Zn, Cd and Pb, respectively. Samples of each soil were then air dried, disaggregated and sieved to give sub-samples of the <2 mm and <120 µm fractions. These were analysed for Zn, Cd and Pb content with an AAS using a concentrated nitric acid extraction, and the initial soil concentrations are shown below in

Table 1. Initial metal concentrations in the four soils showing results obtained for both the <2 mm and <120 µm soil fractions.

	Zn Content (mg/kg)		Cd Content (mg/kg)		Pb Content (mg/kg)
Soil	<2 mm	<120 µm	<2 mm	<120 µm	<2 mm	<120 µm
1	503	778	2.04	3.00	90	110
2	1163	1482	4.07	5.03	138	141
3	2401	2734	7.86	9.14	154	192
4	3660	4411	12.58	16.07	182	249

. All soils exceed the EA soil guidance value for Cd (1.8 mg/kg) in land intended for food cultivation [25]. All soils exceed both the Zn intervention level (720 mg/kg) and the optimum Pb level (85 mg/kg) on the New Dutch List (National Institute for Public Health and the Environment https://www.rivm.nl/en accessed 16 November 2021). Heavy metals found in our soil are non-biodegradable and have remained in the soil for a long time, creating a long-term environmental concern. The complexity of dissolved organic and inorganic ligands influences the total quantity of accessible metal in soil.

Duplicate samples were analysed to determine the level of precision of the analyses. Due to the different levels of precision achieved in analysing the two different soil fractions, shown in full in

Table 2. Results from duplicate analyses for determining precision levels. Standard deviation (sd) is calculated using the range of values multiplied by 0.4. Precision within this study is expressed as ± two coefficients of variance (cv), i.e., at a confidence level of 95%.

	Zn Content (mg/kg)		Cd Content (mg/kg)		Pb Content (mg/kg)
Sample	<2 mm	<120 µm	<2 mm	<120 µm	<2 mm	<120 µm
Soil 4.a	4440	3871	14.02	13.74	187.9	254.9
Soil 4.b	2777	4528	9.36	16.64	162.4	253.7
Soil 4.c	4637	4545	15.08	15.14	193.1	256.7
Soil 4.d	3795	4586	13.40	17.57	171.7	249.7
Soil 4.e	2650	4525	11.03	17.25	195.2	231.6
Mean	3660	4411	12.58	16.07	182.1	249.3
SD	794.8	285.8	2.29	1.53	13.16	10.01
CV (%)	21.72	6.48	18.19	9.52	7.23	4.02

, a decision was made to use only the <120 µm fraction in further analyses within this study.

2.3. Experimental Design and Analytical Methods

The cuttings were planted in individual plant pots, and each set of six (three of each species), which were planted in the same soil, were placed together in a plastic box, which was covered to reduce excess water flow and subsequent soil constituent leaching. Small holes were made in the cover for the cuttings to grow through. The six plants in the same soils were placed together so that any mixing due to excess water and leaching, which did occur, would at least be between soils with the same initial metal concentrations. The plants were placed in a south-facing location, and growing conditions were monitored to ensure maximum possible homogeneity. Over a four-month period, plant growth was monitored, and care was taken to keep moisture levels in the soils as equal as possible.

On removal, the plants were separated into three portions: roots, shoots and leaves. The shoots were fully recovered, as they had remained intact over the growth period. Some leaves were shed, especially by the plants showing most adverse defects, so recovery was near total. It proved very difficult to remove all root material from the soil. Therefore, a representative sample was taken for each, and any figures relating to root mass are considerably less than the total (on average, approximately 40% recovery was achieved, but there was a high level of variability). Root material was washed to remove excess soil. Composite samples were taken of each of the soils using sub-samples from each of the three plants of each species growing in each soil.

In the laboratory, the root samples were washed more thoroughly using a pressure hose. All plant samples were then oven dried, weighed and cut into small pieces before being placed in a furnace at 500 °C to ash overnight. A wet digestion using concentrated nitric acid was then carried out. The soil samples were dried and sieved to retrieve the <120µm fractions and, again, a concentrated nitric acid wet digestion was used.

All samples were then analysed for Zn, Cd and Pb content using an AAS with further dilutions where necessary to remain within the working limits of the instrument. AAS is an analytical technique that measures the concentration of elements qualitatively and quantitively. The detection limits for Cd, Pb and Zn were, respectively, 0.66–0.10 mg/kg, 3.33–0.5 mg/kg and 0.33–0.05 mg/kg (Zn). As part of the quality control programme for the study, a standard reference material was used and analysed with each set of samples [23]. Duplicate samples were analysed to determine the precision levels of each method for each element, and the data are given in

Table 3. Results from duplicate analyses to determine precision levels. Standard deviation (SD) is calculated using the range of values multiplied by 0.4. Precision within this study is expressed as ± two coefficients of variance (cv), i.e., at a confidence level of 95%. Plant sample Pb concentrations were below practical detection limits of the analytical procedure.

Soil Samples				Plant Samples
Sample ID	Concentration mg/kg			Sample ID	Concentration (mg/kg)
	Zn	Cd	Pb		Zn	Cd
V4.1	3805	9.18	292	d4cL 9	758	3.40
V4.2	3909	9.04	294	d4cL 25	883	3.63
V4.3	3808	8.64	294	d4cL 26	918	3.70
				d4cL 27	840	3.23
Mean	3841	8.95	293	Mean	849.75	3.49
SD	62.2	0.33	1.05	SD	80.0	0.24
CV (%)	1.62	3.65	0.36	CV (%)	9.42	6.74

.

A relevant reference material was used for each method (Canmet Till 2 for the soils and IEAE413 algae for the plant material), and the extraction data are given below in

Table 4. Data used for determination of accuracy of analytical methods considering relevant reference materials (Canmet Till 2 and IEAE413 algae).

	Canmet Till 2			IAEA 413 Algae
Determinant	Published Con. mg/kg	Measured Con. mg/kg	Extraction Rate %	Published Con. mg/kg	Measured Con. mg/kg	Extraction Rate %
Zn	116	110	95	169	120	71
Cd	0.3	0.76	252	204	142	70
Pb	21	16	76	242	177	73

. Note that the analytical Cd concentration is considerably higher than the published data. However, as the reference material has a low Cd content, this can simply be regarded as near-total extraction [26].

3. Results

3.1. Biomass Levels

Plant growth was visibly affected in all samples of S. viminalis, with a general trend of decreasing plant quality with increasing soil contamination. This was evident in overall biomass levels and discoloured and wilted leaves due to chlorosis, which is a process by which Zn and Cd inhibit chlorophyll production due to induced deficiencies in Fe, Mn and Mg [27]. In contrast, S. dasyclados showed little or no evidence of detrimental growth effects. All plants were of a similar size with healthy-looking green leaves. Dry weights of the three fractions (roots, shoots and leaves) harvested from each of the plants are given in

Table 5. Dry weights of plant fractions prior to laboratory preparation.

	S. viminalis				S. dasyclados
	Dry Weight (g)				Dry Weight (g)
	Roots	Shoots	Leaves	Total	Roots	Shoots	Leaves	Total
Soil 1
Plant a	1.07	16.80	5.96	23.83	2.95	28.17	10.72	41.84
Plant b	1.35	16.66	5.99	24.00	4.16	39.48	14.22	57.86
Plant c	1.13	13.75	7.53	22.41	1.82	29.64	13.35	44.81
Mean	1.18	15.74	6.49	23.41	2.98	32.43	12.76	48.17
Soil 2
Plant a	1.56	23.88	13.13	38.57	1.41	9.34	7.05	17.80
Plant b	0.91	5.81	3.87	10.59	1.73	32.98	17.39	52.10
Plant c	0.74	2.77	0.60	4.11	1.42	19.03	9.95	30.40
Mean	1.07	10.82	5.87	17.76	1.52	20.45	11.46	33.43
Soil 3
Plant a	0.58	2.77	0.96	4.31	1.89	24.19	13.39	39.47
Plant b	0.31	3.25	2.41	5.97	2.77	33.08	17.60	53.45
Plant c	0.57	3.32	0.72	4.61	2.75	31.27	14.57	48.59
Mean	0.49	3.11	1.36	4.96	2.47	29.51	15.19	47.17
Soil 4
Plant a	0.65	4.25	0.96	5.86	1.57	7.41	6.29	15.27
Plant b	1.46	4.56	1.69	7.71	6.48	45.75	23.57	75.80
Plant c	1.12	2.86	1.13	5.11	5.81	36.68	17.26	59.75
Mean	1.08	3.89	1.26	6.23	4.62	29.95	15.71	50.27

below. Figures for the shoots are totals, as all shoot material was retrieved. Those for leaves are near total, as a small amount of leaf material was lost due to poor growth and subsequent shedding, and those for roots are considerably below total due to difficulties in recovering root material.

3.1.1. Heavy Metals Concentration in Plant Samples

Lead

The results obtained for Pb concentrations in the plant samples were close to, or below, the detection limit of the analytical method used and, in many cases, especially with S. dasyclados, Pb was not detected at all. To illustrate this, values that were not corrected for dilutions are given in

Table 6. Raw results for Pb concentrations in plant samples. Values are not corrected for dilution steps within the analytical process. “nd” denotes that Pb was not detected.

	S. viminalis			S. dasyclados
	Pb Concentration (mg/kg)			Pb Concentration (mg/kg)
	Roots	Shoots	Leaves	Roots	Shoots	Leaves
Soil 1
Plant a	0.315	0.109	0.284	nd	nd	nd
Plant b	0.147	0.010	0.184	nd	0.464	0.278
Plant c	0.227	0.075	0.279	0.041	nd	nd
Soil 2
Plant a	0.195	0.005	0.243	nd	nd	0.085
Plant b	0.219	0.042	0.305	nd	nd	0.085
Plant c	0.246	0.126	0.333	0.012	nd	nd
Soil 3
Plant a	0.262	0.001	0.178	0.022	nd	nd
Plant b	0.318	0.037	0.216	nd	nd	nd
Plant c	0.279	0.003	0.283	nd	nd	0.049
Soil 4
Plant a	0.332	0.078	0.278	nd	nd	0.131
Plant b	0.248	nd	0.435	nd	nd	0.032
Plant c	0.448	nd	0.292	0.246	nd	0.029

below. The theoretical detection limit for Pb using the analytical method practised in the laboratory at Aberystwyth University is given as 0.4 mg/kg [28]. In practice, the theoretical limit will not be achieved, and therefore, a higher limit should be implemented. Cantle [28] states that a figure ten times the theoretical value should be used; however, a value of 1.0 mg/kg is deemed to be reasonable (pers. comm., Brown, A., laboratory supervisor). Any results obtained below this level are not thought to be accurate and should be regarded as meaningless, which is the case for all plant samples.

Zinc

The results obtained for Zn concentrations in plant samples, corrected for dilution steps within the analytical process, are given in

Table 7. Zinc concentrations from analysis of plant samples, corrected for dilution steps.

	S. viminalis			S. dasyclados
	Zn Concentration (mg/kg)			Zn Concentration (mg/kg)
	Roots	Shoots	Leaves	Roots	Shoots	Leaves
Soil 1
Plant a	302	112	613	56	207	400
Plant b	133	129	718	36	190	325
Plant c	302	240	1408	74	182	303
Mean	245	160	913	55	193	343
Soil 2
Plant a	488	136	1730	134	338	623
Plant b	293	366	2250	140	335	488
Plant c	314	408	3477	143	295	588
Mean	365	303	2486	139	323	566
Soil 3
Plant a	592	396	3375	208	368	948
Plant b	564	401	2100	164	325	640
Plant c	543	423	3289	152	320	580
Mean	566	407	2921	174	338	723
Soil 4
Plant a	819	500	4050	245	545	1340
Plant b	405	371	5300	173	390	1035
Plant c	348	479	2825	132	460	849
Mean	524	450	4058	183	465	1075

, with an error margin, at the 95% confidence level, of ± 18.8%. Mean values are also given for each fraction of each plant.

Cadmium

The results obtained for Cd concentrations in plant samples, corrected for dilution steps within the analytical process, are given in

Table 8. Cadmium concentrations from analysis of plant samples, corrected for dilution steps.

	S. viminalis			S. dasyclados
	Cd Concentration (mg/kg)			Cd Concentration (mg/kg)
	Roots	Shoots	Leaves	Roots	Shoots	Leaves
Soil 1
Plant a	1.13	0.39	1.33	0.53	1.25	1.48
Plant b	0.74	0.59	1.38	0.38	1.18	1.38
Plant c	1.25	0.76	2.13	0.40	1.13	0.90
Mean	1.04	0.58	1.61	0.43	1.18	1.25
Soil 2
Plant a	2.45	0.81	3.88	1.08	3.90	2.58
Plant b	1.52	1.66	4.88	0.68	2.23	1.65
Plant c	2.75	2.49	11.61	0.85	2.58	1.85
Mean	2.24	1.65	6.79	0.87	2.90	2.03
Soil 3
Plant a	4.76	2.56	8.63	0.98	3.50	4.23
Plant b	3.57	2.54	4.03	0.98	2.85	2.55
Plant c	4.51	2.89	8.26	0.73	2.50	1.83
Mean	4.28	2.66	6.97	0.89	2.95	2.87
Soil 4
Plant a	5.59	3.00	11.00	2.53	5.15	6.73
Plant b	2.05	2.56	10.58	0.78	3.38	3.48
Plant c	1.45	3.73	7.48	0.75	4.55	3.49
Mean	3.03	3.10	9.68	1.35	4.36	4.56

, with an error margin, at the 95% confidence level, of ±13.5%. Mean values are also given for each fraction of each plant.

3.1.2. Heavy Metals Concentration in Soil Samples

The concentrations of each element determined for each of the four soils at the beginning of the experiment were given in Table 2. A composite sample of each of the four soils was created for each of the two willow species. Sub-samples were then taken for analysis, and the results are given in

Table 9. Final soil concentrations of Zn, Cd and Pb. Separate values given for soils that were supporting each of the two willow species.

Soil	S. viminalis			S. dasyclados
	Final Zn Conc. (mg/kg)	Final Cd Conc. (mg/kg)	Final Pb Conc. (mg/kg)	Final Zn Conc. (mg/kg)	Final Cd Conc. (mg/kg)	Final Pb Conc. (mg/kg)
1	519	1.65	79	566	1.85	83
2	919	2.80	93	1061	3.27	113
3	2301	6.59	167	2216	5.83	195
4	3841	8.95	293	3467	9.79	189

. The error margins at the 95% confidence level are: ±3.2% for Zn, ±6.3% for Cd and ±0.7% for Pb.

4. Discussion

The primary objective of this study was to investigate the extraction potential and translocation characteristics of the two willow species with reference to the three potentially harmful elements: Pb, Zn and Cd. Each element is considered in turn below, relating the results of the pot experiments with those reported in the relevant literature, including both laboratory and field studies. Any shortcomings of the current study, which have come to light as the project progressed, are also highlighted with suggestions for improvement in any future work of this kind. The latter part of the discussion deals with the wider issues relating to the use of willow species in the bioremediation of contaminated soils, focusing particularly on the subjects of further usage potential and waste management of the resultant contaminated biomass.

4.1. Zinc

As can be seen in Table 7, appreciably high concentrations of Zn were present in both of the willow species, with by far the greatest levels recorded in the leaves in both cases. The highest recorded Zn concentration of over 0.5% was found in the leaves of S. viminalis growing in soil 4. Concentrations can also be seen to increase steadily with soil contamination levels for both species, as is shown graphically below in Figure 1. The concentrations reported for S. viminalis are consistently higher than those reported for S. dasyclados, the only exceptions being the Zn levels in the shoots of plants grown in soils 1 and 2.

Direct comparisons with related studies are complicated by variations in soil characteristics, which, other than metal contamination levels, are often not stated. However, the Zn concentrations found in the plant samples, and the level of translocation to the foliage, are consistent with findings of Refs [29,30,31,32], which reported maximum leaf Zn concentrations of 5061 ppm. In a broad review of the phytoremediation of metal contamination using willows, Witters [33] reported leaf Zn concentrations in the range of 2663 to 4249 ppm, with shoot variations ranging from 549 to 766 ppm.

There are also differences in the overall trends, which can be noted for the two species. S. dasyclados reported concentrations increase from the roots to shoots to leaves, and each of the three plant fractions show a steady increase with soil contamination level, as is illustrated by the trendline gradients in Figure 2 below. The leaves show a response rate three times greater than the shoots and six times greater than the roots. In the S. viminalis samples, shoot concentrations are slightly lower than those in the roots, with leaf concentrations considerably higher, determined as being between 4 and 10 times greater than those for roots or shoots. Both root and shoot concentrations increase steadily with soil contamination levels at a similar rate. Leaf concentrations are evidently more responsive, with the rate of increase being greater by an order of magnitude (Figure 2). This indicates that the proportion of Zn translocated to the leaves, as well as plant Zn concentrations, increases with greater soil contamination levels. The extraction capabilities and translocation characteristics of S. viminalis can, therefore, be seen to be more favourable than those for S. dasyclados.

Phytoremediation capabilities of plants are often characterised and expressed in the literature in the form of bioaccumulation factors (BAFs), as in Refs [34,35], and are also sometimes called bioconcentration factors (BCFs), as in Refs [36,37]. These values are established by dividing the concentration of an element found in a plant sample by the concentration of that element in the soil (or other growth medium in the case of laboratory-based experiments). A BAF value greater than 1 indicates that the plant actively concentrates metals within its tissues and can, therefore, be termed a bioaccumulator [38]. The range and mean value of BAFs relating to Zn from this study are reported in

Table 10. BAF values, range and mean value (in brackets) for leaf fractions of plant samples with regard to Zn.

Bioaccumulation Factor (BAF) Range and (Mean) for Zn
	S. viminalis	S. dasyclados
Soil 1	0.79–1.81 (1.17)	0.39–0.51 (0.44)
Soil 2	1.17–2.35 (1.68)	0.33–0.42 (0.38)
Soil 3	0.77–1.23 (1.07)	0.21–0.35 (0.26)
Soil 4	0.64–1.20 (0.92)	0.19–0.30 (0.24)

below. Values are given based on leaf fraction concentrations. It is evident that, by the criteria stated above, S. viminalis can be termed a bioaccumulator, whereas S. dasyclados cannot.

The overall picture is somewhat different than that given by recording plant fraction Zn concentrations. S. dasyclados generally extracted a greater amount of Zn by mass (Figure 3). In the two less contaminated soils, the total extraction is of a similar magnitude, but in the two soils with greater levels of contamination, S. dasyclados removed considerably more (by a factor of approximately four). For S. viminalis, the extraction capabilities appear to have peaked at a contamination level equal to that of soil 2 (1482 mg/kg), whereas for S. dasyclados, the level of extraction is still increasing at the level of contamination of soil 4 (4411 mg/kg). Additionally, in the case of S. dasyclados, the accumulation in the shoot fraction of the plant can be seen to be of more importance than was evident when displaying merely the plant fraction Zn concentrations.

These results differ from those of Ref [39], which reported that S. viminalis extracted slightly more total Zn than S. dasyclados. However, only one of the three soils utilised in their study had a Zn contamination level similar to the one used here (Litavka at 3718 mg/kg is comparable to soil 4), and other soil properties, which are not reported, could vary greatly. Additionally, the growth period in this study, due to project time constraints, was only four months long, whereas the Ref [39] study was carried out over two full vegetation seasons. This difference in growth periods could have an additional effect on the amounts extracted, other than merely due to a longer timescale. This is because the levels of extraction and translocation characteristics have been shown to vary throughout the plant’s growth season [10]. Total remediation factors, the extraction amounts as a percentage of total Zn present, are reported by Ref [39]. This is not possible in this study, as the total amounts of Zn added to each soil were not measured. Soil Zn concentrations dropped considerably (

Table 11. Soil Zn concentrations (mg/kg). Initial and final values given for each soil along with decrease in Zn concentration expressed as a percentage, total extraction for each set of three plants (mg) and a calculated initial Zn content implied by these figures (mg).

S. viminalis
Soil	Initial Conc. (mg/kg)	Final Conc. (mg/kg)	Conc. Decrease (%)	Total Extracted (mg)	Implied Initial Zn (mg)
1	779	519	33.4	26.7	80
2	1482	919	38.0	41.3	109
3	2734	2301	15.8	15.3	97
4	4411	3841	12.9	22.7	176
S. dasyclados
Soil	Initial Conc. (mg/kg)	Final Conc. (mg/kg)	Conc. Decrease (%)	Total Extracted (mg)	Implied Initial Zn (mg)
1	779	566	27.4	32.1	117
2	1482	1061	28.4	39.2	138
3	2734	2216	18.9	63.3	334
4	4411	3467	21.4	88.5	414

), but how much of that decrease is attributable to phytoremediation cannot be certain. Despite prevention measures, some leachate accumulated in the plastic containers, which was not collected and analysed and could potentially have contained appreciably high concentrations of Zn. In a further study of this type, any such leachate would be investigated.

In calculating the values for initial Zn content, the total amount extracted was taken to account for a fraction of the initial content proportional to the degree of soil concentration decrease. This figure could then be scaled up to give an implication of the initial Zn content. The values given are totals for three plant pots, and there is little similarity between the figures obtained for the two species. One possible explanation for this could be the differences in leaching levels due to the considerably lower amount of root material in the S. viminalis plants, particularly in the two more contaminated soils. The figures obtained for S. dasyclados do follow a similar trend to that of the initial soil concentrations. Taking soil 4 as an example, an initial Zn content of 138 mg per pot is suggested, as the implied initial Zn content of 414 mg is a total of three pots. This appears very low, given the initial concentration of 4411 mg/kg, and noting that each pot initially contained at least 1 kg of soil, but it must be remembered that all concentration measurements relate to the fine (<120 µm) soil fraction. The implied mass of the initial fine soil fraction can therefore be calculated as 31 g (as 138 mg/31 g equates to 4411 mg/kg). This seems to be a reasonable figure, but as soil grain size distribution analysis was not carried out, this is somewhat speculative. However, if this speculative calculation is taken to be broadly accurate, it can be implied that, within this study, phytoremediation accounts for a considerable amount of the decrease in soil Zn concentration for S. dasyclados, whereas for S. viminalis, other factors, such as leaching, must be assumed to be important factors as well.

4.2. Cadmium

The results obtained for plant fraction Cd concentration and translocation characteristics exhibit a broadly similar pattern (although with considerably lower concentrations) as those for Zn, as can be seen in

Table 12. BAF values, range and mean (in brackets) for leaf fraction plant samples with regard to Cd.

Bioaccumulation Factor (BAF) Range and (Mean) for Cd
	S. viminalis	S. dasyclados
Soil 1	0.44–0.71 (0.54)	0.30–0.49 (0.42)
Soil 2	0.77–2.31 (1.35)	0.33–0.51 (0.40)
Soil 3	0.44–0.94 (0.76)	0.20–0.46 (0.31)
Soil 4	0.47–0.68 (0.60)	0.22–0.42 (0.28)

and Figure 4. Shoot concentrations are somewhat greater relative to leaf concentrations in the case of S. dasyclados when compared to Zn trends. Plant fraction Cd concentrations generally rise with increasing levels of soil Cd contamination (Figure 5), and the low R² values shown are probably due to the relatively small number of plants growing in each soil. Further studies would involve growing a greater number of plants to obtain more representative averages.

Actual concentration levels are considerably lower than reported in a number of relevant published studies. For example, Ref [39] reported maximum Cd concentrations in willow leaves of 204 mg/kg, and Ref [40] reported Cd concentrations in shoots of 63 mg/kg. However, Refs [41,42,43] reported Cd concentrations similar to those found in this study in S. viminalis grown in field trials in both calcareous and acidic soils. Concentrations were quoted of between 3.7 and 6.0 mg/kg for leaves and between 1.5 and 3.6 mg/kg for shoots. There are a number of possible reasons for such a broad range of Cd concentrations being reported in related studies, some of which are considered below.

BAF values for the leaf fractions of the two species are given in Table 12. By the criteria given above, again, only S. viminalis can be termed a bioaccumulator and then only when grown in soil with characteristics similar to soil 2. In the case of S. dasyclados, the BAF values relating to the accumulation of Cd are very similar to those relating to Zn accumulation, whereas in the case of S. viminalis, the values relating to Cd are considerably lower than those relating to Zn. This implies that S. dasyclados has an equal propensity for accumulating both Cd and Zn, whereas S. viminalis preferentially accumulates Zn over Cd. In its review of Cd phytoextraction using willow, Ref [15] reports BAF values between 0.05 and 16.8 for shoots and between 0.17 and 27.9 for leaves, with values for plants grown on mine spoil being 0.30 for shoots and 0.17 for leaves [44,45,46].

Again, as can be seen in Figure 6, a somewhat different trend is seen when plant biomass production is taken into account. The trends in total extraction by mass are very similar to those relating to Zn extraction. The totals for S. viminalis peak in soil 2 and drop down at soil 3. The totals for S. dasyclados rise steadily with soil Cd contamination levels and are greater than those for S. viminalis in all four soils. The total Cd extraction values reported by Ref [23] are greater than those reported here, but the ratios of extraction by the same two species are the same. They also found that S. dasyclados extracted considerably more Cd by mass than S. viminalis.

Soil concentrations of Cd dropped considerably, as can be seen in

Table 13. Soil Cd concentrations (mg/kg). Initial and final values given for each soil along with drop in Cd concentration expressed as a percentage, total extraction for each set of three plants (mg) and a calculated initial Cd content implied by these figures (mg).

S. viminalis
Soil	Initial Conc. (mg/kg)	Final Conc. (mg/kg)	Conc. Decrease (%)	Total Extracted (mg)	Implied Initial Cd (mg)
1	3.00	1.65	45.1	0.06	0.1
2	5.03	2.80	44.4	0.12	0.3
3	9.14	6.59	27.9	0.06	0.2
4	16.07	8.95	44.3	0.08	0.2
S. dasyclados
Soil	Initial Conc. (mg/kg)	Final Conc. (mg/kg)	Conc. Decrease (%)	Total Extracted (mg)	Implied Initial Cd (mg)
1	3.00	1.85	38.5	0.16	0.4
2	5.03	3.27	35.1	0.22	0.6
3	9.14	5.83	36.2	0.39	1.1
4	16.07	9.79	39.1	0.54	1.4

. Again, as with Zn, there is little similarity between the initial amounts of Cd implied by considering the figures relating to the two species. S. dasyclados values appear to correlate well with initial soil concentrations, whereas S. viminalis values do not.

If the value for soil 4 is considered as an example, as it was above, an initial Cd content of 0.46 mg (1/3 of 1.4 mg, which is a total for three pots) implies an initial fine (<120 µm) fraction mass of 29 g (as 0.46 mg/29 g equates to 16.07 mg/kg). This figure is very close to the 31 g implied by the same calculation relating to Zn. However, this should be taken only as a rough estimation, as it fails to take into account any other factors that would affect concentration values, such as the activity of other fine fraction constituents during the course of the study, which would affect mass balancing. The implication here is the same as with Zn, in that the phytoremediation of Cd appears to account for an appreciable amount of the decrease in soil Cd concentration in the case of S. dasyclados, whereas factors such as leaching seem to be more implicated in the case of S. viminalis, again, probably due to considerably less root material.

In considering the possible reasons for a wide range of Cd extraction capabilities of willow species reported in this and other comparable studies, it should first be noted that contamination levels of soils used in this study were assessed only on a near-total concentration basis (using a concentrated nitric acid extraction) with no separate extraction method used to establish the bioavailable fraction. There could, therefore, be differences in bioavailable proportions of soils used in this and comparative studies. A determination of bioavailable fractions of element concentrations would be a useful improvement that could be implemented in the procedure if further studies were to be carried out. However, Cd present in soils, which originates from inorganic sources, as is the case in this study, has been shown to be the most readily bioavailable, with Cd present predominantly as free Cd²⁺ ions in acid soils [36,37,38]. The major factor affecting the mobility of an element, and therefore its bioavailability within a soil, is the soil pH [39,40,41]. However, the values for soil pH are not given in many comparative studies [20,23].

The accumulation of elements in plants is also strongly governed by the behaviours of other elements present and their interactions [36,38]. In the case of Cd, Zn content has been shown to have the greatest effect with high Zn concentrations, resulting in a decrease in Cd uptake [46,47]. Therefore, Zn/Cd ratios of the soils need to be considered. The initial Zn/Cd ratios in the soils used in this study range from 260 to 299, with an average of 282. Ratios in the soils of comparative studies are usually considerably lower, with examples ranging from 38 to around 200 [20]. This difference could partly account for the difference in overall levels of Cd accumulation, with high relative Zn concentrations inhibiting Cd uptake.

4.3. Lead

As stated earlier, concentrations of Pb found in the plant samples were below the practical limits of the analytical method used. Other methods could have been employed to determine the concentrations accurately but, as these were well below levels at which phytoremediation on any practical scale would be possible, this was not deemed necessary. The reasons as to why concentrations were so low in this study when other comparable projects reported relatively high Pb concentrations [24] and willows have been used in the field in phytoextraction processes [22]. Although Ref [24] reported elevated Pb concentrations, these were predominantly in the roots, with only limited translocation to the above-ground harvestable plant fractions. This suggests that a more appropriate form of phytoremediation of Pb contaminated soils using willows would be through phytostabilisation rather than phytoextraction [3,44,48].

Lead is relatively immobile in soils with a pH within the range of the soils in this study (4.9–5.2). For plants to take up immobile soil-bound materials, they must actively mobilise them by one of several possible root processes, such as the release of metal-chelating agents [3,49,50]. Plants exist, which will do this for Pb—sunflowers, for example [51]—but the results of this and other studies imply that willows are not among them.

Another factor affecting the bioavailability of Pb in soil is the level of organic matter present. This is because Pb has a strong propensity for adsorbing to soil organic matter [36,52,53]. Due to the method employed to create a range of soils for this project, mixing commercial compost with mine spoil material, the soils all had a considerably high organic content. This will have further decreased the bioavailability of Pb. If further studies are to be carried out, a more realistic method of soil/contaminant mixing should be employed, with the most useful results provided by field trials or pot experiments using a range of soils collected from the field.

5. Conclusions

Both species of willow performed well in the pot experiments within this study with regard to Zn, moderately well with regard to Cd and poorly with regard to Pb. The extracted metals were preferentially translocated to the leaves, with concentrations and relative BAFs considerably higher in the case of S. viminalis. Thus, by the criteria which are usually cited in the related literature, S. viminalis is a better bioaccumulator than S. dasyclados under the conditions of the experiments. However, when biomass production is taken into account, it becomes evident that S. dasyclados extracted more Zn by mass in the two more contaminated soils and more Cd by mass in all four soils, with similar amounts of Zn by mass removed from the two less contaminated soils by both species. It is therefore clear that under the conditions of the experiments, S. dasyclados showed greater potential for use in phytoremediation, especially if coupled with use of biomass for energy production.

Willows have several other qualities, which add to their overall potential in facilitating land remediation, with extensive root systems reducing soil erosion and contaminant leaching problems. The most significant additional consideration is that of the resultant biomass use and/or disposal. Numerous approaches to this have been researched, and the most promising solutions involve use of biomass for energy production, with flue gas treatment technologies employed to capture the contaminants. The resultant flue ash is greatly reduced in volume from the original waste material and, if the formation of ashcrete building materials proves to be effective and is regarded as an acceptable waste disposal pathway, a somewhat rounded scenario can be described: using phytoremediation by short-term rotation willow coppice producing fuel in the form of biomass for pyrolysis or by combustion in a combined heat and power plant, from which the flue ash is made into ashcrete blocks and locked up in future buildings. Advances in metal recovery technologies resulting in considerably lower processing costs could lead to a sustainable pathway for the flue ash and a more widespread use of phytomining for metal recovery.