Whole-Range Assessment: A Simple Method for Analysing Allelopathic Dose-Response Data

Abstract

Based on the typical biological responses of an organism to allelochemicals (hormesis), concepts of whole-range assessment and inhibition index were developed for improved analysis of allelopathic data. Examples of their application are presented using data drawn from the literature. The method is concise and comprehensive, and makes data grouping and multiple comparisons simple, logical, and possible. It improves data interpretation, enhances research outcomes, and is a statistically efficient summary of the plant response profiles.

INTRODUCTION

When investigating the allelopathic potential of a plant or an allelochemical, a test species is often employed whose responses are measured and recorded. The data are then analysed by a conventional statistical method, eg regression analysis. Subsequently, a conclusion is made based on such analysis. In those cases where a test species responds to all treatments identically or nearly so (eg. Figure 1), employment of a conventional statistical method imposes few problems. However, owing to the biological variability of a plant test species, together with its non-linear responses to a set of treatments, ie. hormesis, the majority of observed responses fluctuate and are erratic (Figures 3–5). Conventional statistical analyses in these cases often fail to fully utilize the information contained in such a data set, and may deliver an unsatisfactory outcome, even controversial results, particularly in the case of multiple comparisons made under a set of concentration treatments or equivalent (Dawkins 1983). This dilemma is well known and various analytical approaches have been developed to assess allelopathic dose-response data for different purposes (Williamson and Richardson, 1988; Dornbos and Spencer, 1990; Wu et al., 2000; Dias 2001; Liu et al., 2003; Macias et al., 2005). However, those approaches either fail to recognise the hormesis phenomenon, are statistically inefficient for multiple comparisons, or are too complicated for practical use. As a result, a simple, universal, and statistically efficient method for allelopathic multiple comparisons with dose-response data is still not available in the literature.

FIGURE 1.

Effect of BOA, MBOA, DIBOA, and DIMBOA on the activity of H⁺ -ATPase (after Friebe et al., 1997)

FIGURE 3.

Dose-responses of ryegrass seedlings to four benzoxazinoids (data from Huang et al., 2003)

FIGURE 5.

Effect of ferulic acid on seedling growth (data from Schulz et al., 1994)

This paper aims to overcome this deficiency by providing a better analysis method for allelopathic dose-response data, thus helping to ease data assessment and interpretation.

METHODS

The dose-response relationship known as hormesis (stimulation at low concentrations of allelochemicals and inhibition as the concentration increases) is well recognised in allelopathy (Lovett et al., 1989; An et al., 1993; Belz and Hurle, 2002). The extent of stimulation and inhibition is not balanced. In general, over a normal biologically realistic range of concentration levels, inhibition dominates biological responses and increases with increasing concentration levels. Stimulation occurs at the very low levels of the range and only represents a small portion of the overall biological response. Inhibition has a far more practical meaning in assessing allelopathic dose-response data. Therefore, in the following whole-range assessment method, only inhibition is considered for facilitating practical use.

Whole-range assessment. Instead of assessing the effect of individual allelochemical concentrations on test species, the overall effect/response across the whole range of allelochemical rates is considered. The approach is first to normalize biological responses by taking the control as a reference, and then to calculate the inhibition area between the control response (i.e., 100%) over the whole range of treatments (i.e., allelochemical concentrations or equivalent on the X axis) and the dose-response curve (i.e., test species responses), as generated by allelochemical concentrations or equivalent (Figure 2).

FIGURE 2.

Diagrammatic representation of biological response to allelochemical concentrations or equivalent. The shaded section represents the inhibition area. C_T — the threshold concentration for causing inhibition.

Thus

$$\text{Inhibition area }= {\int }_{C_T}^{100}[\mathrm{100 }- f(C)]dC\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}(\text{i.e.},\text{ hatched zone in Figure 2)}$$

where C is the allelochemical concentration or equivalent with arbitrary units, C_T is the threshold concentration for causing inhibition in the test species. f(C) can be any mathematical function describing the nonlinear dose-response relationships and only has theoretical meaning here. Overall biological activity across the whole range of concentrations or equivalent is then summarised, calculated, and represented by a single value, the “inhibition index”, which is defined as the percentage of the inhibition area to the total area.

Therefore:

$$\text{Inhibition index }=\frac{\text{Inhibition area}}{\text{Total area}}\times 100$$

where the total area is defined as

$${\int }_0^{100}100dC$$

The actual computation of the inhibition area and the inhibition index can be easily done with any software with a mathematical integration function, such as MicroCal Origin. By calculating the inhibition area across the whole range, account is taken of variation of all treatments. The inhibition index is a summary of the overall biological response of an organism to a tested allelochemical or equivalent and provides a relative strength indicator of biological response. Large values indicate that the species is sensitive or the allelochemical possesses strong allelopathic potential/biological activity, whilst small values indicate tolerance or weak potential/biological activity. The index can also be subject to a conventional statistical method for further analysis, such as analysis of variance, for easy grouping or significance testing of multiple comparisons.

APPLICATION AND DISCUSSION

(i) Biological activity of multiple allelochemicals assessed by a single testing species

Benzoxazinones are important secondary metabolites found in cereal rye, wheat, and maize. They are involved in plant defence against pests and diseases, and are thought to be primarily responsible for allelopathic weed suppression in rye and wheat (Niemeyer and Perez, 1995; Copaja et al., 1999; Wu et al., 2000; Sicker and Schulz, 2002). Friebe et al. (1997) investigated the sensitivity of plasma membrane H⁺-ATPase from roots of Avena sativa to four benzoxazinoids (DIBOA, DIMBOA, MBOA, and BOA). H⁺-ATPase is an important enzyme playing a central role in plant cell physiology. Its activity is closely related to plant radicle growth (Friebe et al., 1997). The results showed that H⁺-ATPase demonstrated a characteristic dose-response to the four allelochemicals tested (Figure 1). Significant inhibitory as well as stimulatory effects were found with those compounds and there were marked differences in their allelopathic potentials (Figure 1). The results in Figure 1 present ideal dose-response data. For such a data set, regardless of what statistical methods are employed for analysis, consistent outcomes are expected, one of which is demonstrated by the assessment of inhibition index in Table 1.

TABLE 1.

Biological activities of benzoxazinoids as assessed by inhibition index

Chemical name	Inhibition index by H+-ATPase activity (Friebe et al., 1997)	Allelopathic potential	Chemical name	Inhibition index by ryegrass seedlings (Huang et al., 2003)	Allelopathic potential
DIMBOA	33.3	Strong	DIMBOA	53.5	Strong
DIBOA	16.5		DIBOA	51.3
BOA	6.7		HMBOA	10.9
MBOA	3.9	Weak	HBOA	9.6	Weak

However, due to the biologically variable nature of plants, the majority of biological responses are fluctuating and erratic. Results such as in Figure 3 are not uncommon. Multiple comparisons for such a set of data are chanllenging, particularly if the potential allelochemical list is long. The original data in Figure 3 contained 11 allelochemicals. Only four are presented here, to facilitate comparison with Figure 1. The data set in Figure 3 is about the biological activities of four benzoxazinoids against a weed, ryegrass (Lolium rigidum). If one takes a conventional statistical assessment, it is highly likely that controversial outcomes will be produced. However, the whole-range assessment of this set of data yields a consistent outcome in comparison with the literature (Friebe et al., 1997; Sicker and Schulz, 2002), which shows that the significant differences in suppression of ryegrass seedlings is clear and explicit (Table 1). It is clear in this case that the whole-range assessment has delivered a simple, concise, yet satisfactorily comprehensive outcome.

Vulpia is a significant allelopathic weed in Australia (An et al., 1997a, 2000). Twenty allelochemicals identified in Vulpia residues were individually and collectively tested for their biological activity using wheat as a test species (An et al., 2001). Each compound exhibited characteristic allelochemical behaviour towards the test plant, ie., inhibition at high concentrations, and stimulation or no effect at low concentrations. Assessment of those data by the whole-range assessment method revealed that individual activities of these allelochemicals varied. In general, allelochemicals present in large quantities, such as syringic, vanillic, and succinic acids, possessed low activity, while those present in small quantities, like catechol and hydrocinnamic acid, possessed strong inhibitory activity. The 20 compounds could be grouped according to their biological activity (Table 2). Such assessments further identified the individual contribution of each allelochemical to the overall Vulpia allelopathy, and determined the factors affecting such contributions. It was revealed that, in general, the majority of compounds possessed low or medium biological activities and, collectively, contributed most of the Vulpia allelopathy, while compounds with high biological activity were in the minority, representing a small portion of the overall allelochemical quantity found in Vulpia, and making only a minor contribution to Vulpia allelopathy (Table 2).

TABLE 2.

Biological activities of Vulpia allelochemicals as assessed by inhibition index and their relative contribution to overall Vulpia allelopathy (Data from An et al., 2000, 2001)

Chemical name	Quantity in Vulpia residue (mg/g)	Inhibition index	Allelopathic potential	Relative* contribution to Vupia allelopathy
Coniferyl alcohol	0.0044	2.60		0.57
Protocatechuic acid	0.0163	2.95		2.40
3, 4-Dimethoxyphenol	0.0021	3.85		0.40
Hydrocaffeic acid	0.0038	4.25	Weak	0.81
Syringic acid	0.0565	4.25		12.01
Succinic acid	0.0722	4.48		16.19
Hydroquinone	0.0016	4.52		0.37
p-Hydroxybenzoic acid	0.0158	4.81		3.81
3-(4-Hydroxyphenyl) propanoic acid	0.0393	5.29		10.39
Gentisic acid	0.0096	5.61		2.70
Vanillic acid	0.0731	5.98		21.87
p-Coumaric acid	0.0047	6.02		1.40
Ferulic acid	0.0072	6.24	Medium	2.24
Pyrogallol	0.0064	6.55		2.10
p-Hydroxyphenylacetic acid	0.0071	6.65		2.36
2-Hydroxy-3-phenylpropanoic acid	0.0150	6.75		5.06
Catechol	0.0003	6.87		0.12
Benzoic acid	0.0114	7.61		4.34
Salicylic acid	0.0268	7.90	Strong	10.60
Hydrocinnamic acid	0.0007	8.43		0.28

*Relative contribution = $\left(\frac{C_i}{\Sigma C_i}\right)$ × 100where i = Coniferyl . . . Hydrocinnamic acid, C = inhibition index of each individual compound multiplied by their mass found in Vulpia residue.

In this case whole-range assessment has not only made multiple comparisons possible and simple, but also enhanced the interpretation of results and provided valuable insights, which could be important in the understanding of allelopathy fundamentals, allelochemical modes of action, and in employment of allelopathy for developing natural herbicides.

(ii) Biological activity of a single allelochemical assessed by multiple testing species

Quackgrass (Agropyron repens L.), one of the most aggressive perennial weeds in northern temperate parts of the world, is known to affect crop development and to reduce crop yields. Besides inhibitory effects due to competition, a high allelopathic potential of the species is thought to be responsible for growth inhibition (Schulz et al., 1994). By developing a root exudate recirculating system Schulz et al. (1994) found DIBOA, vanillin, β-hydroxybutyric-, 4-hydroxycinnamic-, ferulic-, vanillic-, syringic-, and protocatechuic acids in the root exudates of quackgrass. Subsequently two of the compounds, DIBOA and ferulic acid were tested for their effects on Lepidium sativum, Amaranthus retroflexus, Brassica napus, Lolium perenne, Poa annua, Hordeum vulgare, Secale cereale, and Triticum aestivum radicle growth. Species and dose dependent responses of the tested plants were observed (Figures 4 & 5). However, the authors only made general conclusions on the species susceptibility and allelopathic potentials of the two compounds tested. Considering the great fluctuation and complexity of the species responses presented in Figures 4 & 5, it is not surprising to see such conclusions. In contrast, data presented in Table 3 shows the outcome by whole-range assessment of this set of data. Based on this type of outcome, explicit conclusions can be made. Results clearly indicate that over all species tested DIBOA possesses much stronger allelopathic potential than ferulic acid. There are marked differences among species in their susceptibility towards the allelopathic effects of the two compounds. It is possible to group those species into a few groups with different susceptibility. The order of increasing sensitivity to DIBOA is Secale cereale, Triticum aestivum, Hordeum vulgare, Lolium perenne, Poa annua, Lepidium sativum, Amaranthus retroflexus, and Brassica napus. With ferulic acid this order is slightly different, but the most and least sensitive species and the grouping orders remain as for DIBOA. Such explicit conclusions are unlikely to be drawn by a conventional statistical method.

FIGURE 4.

Effect of DIBOA on seedling growth (data from Schulz et al., 1994)

TABLE 3.

Biological activities of DIBOA and ferulic acid as tested by multiple species and assessed by inhibition index (Data from Schulz et al., 1994)

Species name	Inhibition index for DIBOA	Susceptibility	Species name	Inhibition index for ferulic acid	Susceptibility
Secale cereale	32.8	Tolerant	Secale cereale	6.6	Tolerant
Triticum aestivum	32.9		Triticum aestivum	8.8
Hordeum vulgare	41.7		Hordeum vulgare	11.2
Lolium perenne	47.9		Poa annua	16.9
Poa annua	57.1		Lepidium sativum	36.5
Lepidium sativum	58.3		Lolium perenne	37.4
Amaranthus retroflexus	66.8		Amaranthus retroflexus	38.6
Brassica napus	69.1	Sensitive	Brassica napus	39.7	Sensitive

(iii) Susceptibility of multiple plant species to the allelopathy of a single plant

Parthenium (Parthenium hysterophorus L.) is an annual weed native to the Americas. It is an aggressive weed of disturbed sites and commonly found in cultivated fields. The weed commonly forms huge pure stands and in such areas the vegetation seldom contains other plant species, which suggests that a possible allelopathic mechanism is operative (Mersie and Singh, 1987). Allelopathic effects of entire shoot extract, plant part extracts, and shoot residue of parthenium on corn, ryegrass, wheat, and velvetleaf (Abutilon theophrasti Medik.) growth were examined by Mersie and Singh (1987). They found that parthenium shoots contained water-soluble materials that were toxic to root growth of all species tested. There was a strong correlation between extract concentration and increased toxicity to the test species (Figure 6). By comparing the responses of all species at a single concentration level, ie. 4% concentration, the authors concluded that parthenium allelopathy is species-specific and extract concentration dependent. The order of increasing sensitivity to parthenium was ryegrass, corn, wheat, and velvetleaf. However, if one examines the results presented in Figure 6 at a different concentration level, eg. 1%, the order of sensitivity will be entirely different, ie., ryegrass, wheat, velvetleaf, and corn. Hence, multiple comparisons at a single concentration level, clearly, does not account for the biologically variable nature of plant responses. In contrast, Table 4 presents the analysed outcome by whole-range assessment. The significant differences between test species sensitivity to parthenium allelopathy is clear and convincing, and the order of increasing sensitivity is explicit.

FIGURE 6.

The effect of various concentrations of parthenium extracts on root growth of corn, ryegrass, velvetleaf, and wheat (after Mersie and Singh, 1987)

TABLE 4.

Species susceptibility to Parthenium allelopathy as assessed by inhibition index (Data from Mersie and Singh, 1987)

Species name	Inhibition index	Susceptibility
Ryegrass (Lolium multiflorum Lam.)	12.9	Tolerant
Corn (Zea mays L.)	29.0
Wheat (Triticum aestivum L.)	31.3
Velvetleaf (Abutilon theophrasti Medik.)	38.4	Sensitive

In an attempt to widen strategies for managing detrimental effects of Vulpia residues, An et al. (1997b) tested genotypic variation of 12 plant species and 12 cultivars from two plant species under a series of aqueous extracts of Vulpia residues and analysed the data by the whole-range assessment method (Table 5). It showed that all test plants exhibited characteristic responses to the Vulpia extracts. Marked differences in tolerance toward the Vulpia phytotoxicity existed among species and cultivars. Such differences were widespread among plant species, with generally cocksfoot, Vulpia spp., canola, and phalaris being relatively tolerant, while lupins and barley were the most susceptible. Wheat and subterranean clover were relatively susceptible with a few cultivars being exceptional. Employment of the inhibition index not only enabled the susceptibility of each plant to Vulpia allelopathy to be presented in a concise, comprehensive, and meaningful format, but also grouped the plant species and cultivars according to their susceptibility (Table 5). Such grouping provides a basis for widening management options and for choosing appropriate species and cultivars for minimising the negative effects of Vulpia residues.

TABLE 5.

Species and cultivars susceptibility to Vulpia allelopathy as assessed by inhibition index (After An et al., 1997b)

Species & cultivars	Overall inhibition index	Susceptibility to Vulpia allelopathy
Cocksfoot	8.1
Subclover (cv. Trikkala)	9.3
V. myuros	9.4	Tolerant
Phalaris	9.7
V. bromoides	10.8
Canola	13.4
Lucerne (cv. Trifecta)	17.1
Oats	20.0	Less tolerant
Lucerne (cv. Aurora)	21.7
Wheat (cv. Ford)	27.7
Subclover (cv. Seaton Park)	28.9
Subclover (cv. Karridale)	31.0	Medium
Field Peas	32.0
Wheat (cv. Darter)	32.2
Subclover (cv. Clare)	35.7
Subclover (cv. Woogenellup)	35.9
Wheat (cv. Dollarbird)	35.9	Less sensitive
Subclover (cv. Junee)	37.6
Wheat (cv. Rosella)	37.9
Lupins	41.3
Wheat (cv. Janz)	41.7	Sensitive
Wheat (cv. Vulcan)	47.3
Barley	56.7

(iv) Susceptibility of a single plant species to the allelopathic potential of multiple plants

One of the prospects that allelopathy holds is that allelopathic plants may be used to control weeds therefore reducing reliance on synthetic herbicides.

Rice is one of the world’s most important crops. Interest in its allelopathic potential in weed suppression has been steadily increasing (Olofsdotter 1998; Ahn and Chung, 2000; Ebana et al., 2001; Seal et al., 2004). Seal et al. (2004) screened in the laboratory twenty-eight rice varieties with different countries of origin, maturity and stage of improvement for their allelopathic potentials against arrowhead (Sagittaria montevidensis), a major weed infesting the rice crops of Australia. Using the density-based equal-compartment-agar-method (ECAM), they successfully established the optimum rice seedling density (with its level of root exudates) and found that arrowhead response to rice exudates (by seedling density) follows a typical dose-response curve (Figure 7). The degree of arrowhead root inhibition ranged from 26.6 to 99.7%. A range of allelopathic potential exists in rice germplasm and such potential is species specific (Seal et al., 2004). These authors undertook a lengthy discussion on rice variety specificity at suppressing arrowhead. Unfortunately, they didn’t define a specific suppressing order even though such information is implicit in the data presented. However, such an order can be derived from their data using the whole-range assessment method (see Table 6). It clearly shows that rice varieties tested can be placed into three groups according to their ability to suppress arrowhead, i.e., strong, medium, and weak. Varieties such as Takanenishiki and IET 1444 possess the most allelopathic potential, TN-1 and Pelde the least, and Kingmen TCM being intermediate. This outcome is consistent with their more rigorously conducted experiments (Seal et al., 2004) and with the wider literature.

FIGURE 7.

Effect of rice density and variety on arrowhead root growth (after Seal et al., 2004; Seal personal communication, 2005)

TABLE 6.

Allelopathic potential of rice varieties against arrowhead weed growth as assessed by inhibition index (Data from Seal et al., 2004; Seal Personal Communication, 2005)

Rice variety name	Inhibition index	Allelopathic potential against arrowhead
Takanenishiki	80.4	Strong
IET 1444	79.2
Kingmen TCM	70.7	Medium
TN—1	65.7
Pelde	61.4	Weak

(v) Evaluation by principal component analysis

The above Vulpia susceptibility data (Table 5) were used to evaluate the inhibition index itself by principal component analysis. The original variables for principal component analysis were germination rate, and root and coleoptile length of each species measured at seven concentrations of Vulpia extract. The first principal components in all parameters measured accounted for more than 60% of the variation in the original data. Significant correlation was found between inhibition indices and their corresponding first principal components. The correlation was 0.96 for germination, 0.87 for root, 0.94 for coleoptile, and 0.88 for the combination of germination and seedling growth (Figure 8). This demonstrates that the inhibition index is a statistically efficient summary of biological dose-response profiles.

FIGURE 8.

Comparison between inhibition index analysis and principal component analysis for a combination of germination and seedling length (after An et al., 1997). Letters denote the names of test species and cultivars.

CONCLUSION

Whole-range assessment is a new concept in analysing allelopathic dose-response data. Instead of assessing the allelopathic potential of an individual allelochemical at a single concentration level, it is assessed across the whole range of multiple allelochemical rates as an overall potential/response that is represented by a single value. By calculating the inhibition area across the whole range of allelochemical concentrations or equivalent, the whole-range assessment takes into account the variations of all treatments and therefore minimises the impact of variation caused by a single treatment. It helps avoid distortion of the value averaged from stimulation (>100% control) to the maximum inhibition (ie. zero %), as used in a conventional statistical method. By calculating the inhibition index based on the total area across all treatments, it creates a convenient platform to make multiple comparisons possible. The inhibition index is a summary of the overall biological response of an organism to a tested allelochemical or equivalent, and can be further subjected to a conventional statistical method, such as analysis of variance, for easy grouping or significance testing of multiple comparisons. From the above-presented examples it is clear that the whole-range assessment concept and the corresponding inhibition index can be used in a wide range of data analyses. The technique analyses data comprehensively and yet presents the outcomes in a concise and meaningful format, and makes data grouping and multiple comparisons simple, logical, and possible. It has proven to be a statistically efficient summary of plant response profiles and is complementary to conventional statistic methods. It enhances data outcomes and provides directions for further investigations.