New Phytol. (1990), 115,

A new method which gives an objectivemeasure of colonization of roots byvesicular-arbuscular mycorrhizal fungi

BY T. P. M C G O N I G L E S M. H. MILLERS, D. G. EVANS^G. L. FAIRCHILD^ AND J. A. SWAN^^ Department of Land Resource Science, University of Guelph, Guelph, Ontario,NIG 2W1, Canada^Department of Statistics, Oregon State University, CorvalUs, OR 97331-4606, USA^ International Institute of Tropical Agriculture, PMB 5320, Oyo Road, Ihadan, Nigeria* Organization Studies and Applied Psychology Division, Aston University, Birmingham,B4 7ET, UK{Received 19 October 1989; accepted 1 March 1990)

SUMMARY

Previously described methods to quantify the proportion of root length colonized by vesicular-arbuscular (VA)mycorrhizal fungi are reviewed. It is argued that these methods give observer-dependent measures of colonizationwhich cannot be used to compare, quantitatively, roots examined by different researchers. A modified method isdescribed here to estimate VA mycorrhizal colonization on an objective scale of measurement, involving inspectionof intersections between the microscope eyepiece crosshair and roots at magnification x 200; it is referred to asthe magnified intersections method. Whether the vertical eyepiece crosshair crosses one or more arbuscules isnoted at each intersection. The estimate of colonization is the proportion of root length containing arbuscules,called the arbuscular colonization (AC). The magnified intersections method also determines the proportion ofroot length containing vesicles, the vesicular colonization (VC), and the proportion of root length containinghyphae, the hyphal colonization (HC). However, VC and HC should be interpreted with caution because vesiclesand hyphae, unlike arbuscules, can be produced in roots by non-mycorrhizal fungi.

Key words: Magnified intersections, percent infection, fractional infection, method, arbuscules.

INTRODUCTION

Procedures to determine visually the proportion ofroot length colonized by VA mycorrhizal fungi canbe divided into three groups: (i) subjective es-timation, (ii) calculation of the percentage of rootsegments or microscope fields of view that contain anycolonization, and (iii) grid-line intersect methods.Techniques using chemical estimation of chitin(Hepper, 1977; Bethlenfalvay, Pacovsky & Brown,1981) are not considered here.

Subjective visual estimation has been used fre-quently (Mosse, 1973; Sanders et aL, 1977; Abbott &Robson, 1978; Buwalda & Goh, 1982; Hopkins,1987), and has been modified to estimate theintensity of arbuscular proliferation in colonizedregions (Trouvelot, Kough & Gianinazzi-Pearson,1986). By repeatedly scoring root systems,Giovanetti & Mosse (1980) calculated a standard

error of between 2 and 5 % colonization for sub-jective visual estimation. However, no investigationhas yet been undertaken to assess the accuracy of thismethod. In addition, this method has not yet beenconsidered in terms of the relationship betweenperceived magnitude (^) and stimulus magnitude{(j)), known as Stevens' power law (Teghtsoonian &Teghtsoonian, 1971):^ = hij/, (1)where /J is a constant which depends on the choice ofunits, and ^ is dependent on the type of stimuluscharacteristic that is perceived (Stevens, 1975). Partof the work presented here was designed to in-vestigate the accuracy of subjective estimation and toconsider it in terms of Stevens' power law.

Several studies have calculated colonization as thenumber of root segments with any colonization,divided by the total number of segments examined

496 T. P. McGonigle and others

20%

30%

60%

Figure 1. A schematic drawing of three 10 cm lengths ofroot. Each one has the same proportion of root lengthcolonized, which expressed as a percentage is 20 %. Thestippled regions represent colonized portions of the roots,and dashed lines indicate where the root lengths are to becut into ten 1 cm root segments. When each of thesegments is rated as a plus or a minus for colonized or notcolonized respectively, the estimates of colonization will be20, 30 or 60%, as shown for the three arrangements.

(Daft & Nicolson, 1972; Khan, 1974; Read,Koucheki & Hodgson, 1976; Biermann &Linderman, 1981; Malibari, Al-Fassi & Ramadan,1988). This is the same in principle as calculating thenumber of microscope fields of view with anycolonization divided by the total number of fields ofview examined, a method also used (Baylis, 1967;Sutton, 1973; Newman, Heap & Lawley, 1981;Plenchette, Furlan & Fortin, 1982; Dodd & Jeffries,1986). These methods almost always overestimatepercentage colonization, the degree of overestimationdepending on the lengths of the segments, and on thelengths of the regions of colonization (Fig. 1).

The grid-line intersect method (Giovannetti &Mosse, 1980) or various modifications of it, is thetechnique most frequently used to estimatecolonization. We found that over two-thirds of asample of 80 papers, published after 1980 andreporting colonization data, used this technique.The procedure is to note the presence or absence ofcolonization at each intersection of root and grid-line, after dispersing the roots above a grid of squaresdrawn on a petri dish, and observing under adissecting microscope at x 40 magnification. Inmany cases colonization by VA mycorrhizal fungican readily be determined. However, even at x 80magnification it is not possible to ascertain if theroots are mycorrhizal at all intersections. This isbecause cortical cells or parts of the stele can becomestained (Dodd & Jeffries, 1986), the roots may becrowded with hyphae, and because arbuscules can bediflicult to detect when they are small. Structuresformed by other fungi may also be confused witharbuscules at low magnification. Differentresearchers are unlikely to be consistent in the waythey record these difficult intersections, and mayarrive at different answers. The grid-line intersectmethod using the dissecting microscope can there-

fore be expected to give a relative measure ofcolonization.

To determine unequivocally whether arbusculesare present in all cases requires examination at x 200magnification. Ambler & Young (1977) described agrid-line intersect method involving the compoundmicroscope, but this still has the difficulty that someintersections must be classified as colonized or notwhen hyphae but no arbuscules are seen. Sincearbuscules are the only unique feature of VAmycorrhizas, decisions as to whether hyphae seenalone are mycorrhizal may vary from person toperson. This method is therefore vulnerable to biasand probably generates a relative measure ofcolonization.

We describe below a technique which enablescolonization to be scored objectively because theonly decision necessary at each intersection iswhether or not arbuscules are present. The methodrequires that the roots are observed at a magnificationsufficient to easily discern the presence of arbuscules,and that the intersection line is thin relative to thewidth of the arbuscules. We refer to this technique asthe magnified intersections method.

MATERIALS AND METHODSInvestigation into the subjective methodThe accuracy of subjective estimation was assessedusing test diagrams so that the true value for theproportion being judged by eye was known. Seventystudents were asked to participate as subjects. Thediagrams (a) and (b) in Figure 2 were chosen torepresent schematically in two dimensions, rootsunder a microscope. All subjects were allowed 20 s tovisually examine each of (a) and (b) one at a time andin that order. Subjects were instructed to estimate,during the 20 s allocated to each diagram, theproportion of the total length of the straight line,enclosed within the circle, that is thick.

Collecting and processing roots

Roots were collected from a field of Zea mays L. cv.Pioneer 3949 at the Elora Research Station(43 31' N, 80 14' W) at 15 days after 50 % silking inthe summer of 1988. Sampling was effected bytaking 5 cm diameter soil cores directly over the cutstems of randomly selected plants. To obtain rootswith a reasonably broad range of colonization,samples were taken from areas within the maize fielddiffering in phosphorus fertilizer history and fromdepths up to 30 cm. Roots were washed from thesoil, and stored in formyl acetic alcohol. Forsubsampling the roots were cut into 1 cm fragmentsand dispersed in water. The water was stirredvigorously and a subsample collected in a beakerimmersed in the water for a few seconds. Rootsamples were cleared for 20 min at 121 C in 10%

Objective measurement of colonization by VA mycorrhizal fungi 497

Root

Directionof

travelacrossslide

Perimeterof fieldof view

Verticalcrosshair

Horizontalcrosshair

Position to whichvertical crosshair

is moved

ib)

Figure 2. The two diagrams (a) and {b) shown to the 70subjects in the investigation into subjective visual es-timation. See text for further details.

KOH, and stained in Chlorazol Black E (Brundrett,Piche & Peterson, 1984). Roots of a selection of 12-week-old leek {Allium porrum L.) seedlings were alsoexamined. Leek seeds were germinated for 1 week onmoist paper towels. For the 5th-7th week inclusivethey were transplanted into mature pot cultures ofGlomus versiforme Karsten (Berch). At other timesthey were grown in inoculum free calcined clay andgiven water and nutrients as described previously(Brundrett, Piche & Peterson, 1985).

Estimating colonization using the magnifiedintersections method

Roots were mounted in glycerin on microscopeslides and covered with 40 x 22 mm coverslips.Between 2 and 4 slides were used for each subsample,but all slides for a subsample were treated as a singleunit, and not as subsubsamples. Roots were alignedparallel to the long axis of the slides and observed atmagnification x 200 as follows. The field of view ofthe microscope was moved using the stage graticuleto make four, six or eight complete passes across eachslide perpendicular to its long axis. The number anddistance between passes is not critical, but should beconstant for a subsample. Except where the cortex

Figure 3. A diagram to show how a magnified intersectionperpendicular to the long axis of the root can be madewhen the root is aligned with its long axis at an angle to thevertical crosshair. The stage is moved until the centre ofthe crosshairs is contiguous with the first edge of the rootreached. To make the perpendicular intersection, thevertical crosshair is then rotated as shown.

was missing, all intersections between roots and thevertical eyepiece crosshair were considered. Theposition on the root surface at which the centre of theeyepiece crosshairs entered through the side of theroot was taken as the point of intersection. Rotationof the vertical crosshair ensured each intersectionwas at right angles to the long axis of the root (Fig.3). Where the centre of the crosshairs entered a rootthrough an end rather than a side, the point of exitfrom the root through its side was taken as the pointof intersection. Roots too wide to fit into the field ofview at x 200 magnification were examined in two ormore width portions.

To examine each intersection, the plane of focuswas moved completely through the root and a notemade of whether the vertical crosshair actually cutany arbuscules, vesicles and hyphae. Intersectionswere counted in the following categories; ' negative'(no fungal material in root), 'arbuscules', 'vesicles',and ' hyphae only'. If the vertical crosshair cut one ormore arbuscules or vesicles, the appropriate categorywas incremented by one, and similarly forintersections where hyphae only were crossed. Whenboth vesicles and arbuscules were scored at anintersection, the total number of intersections wasonly increased by one. The arbuscular colonization(AC) and vesicular colonization (VC) were calculatedby dividing the count for the 'arbuscules' and' vesicles' categories respectively by the total numberof intersections examined. Hyphal colonization (HC)was calculated as the proportion of non-negativeintersections.

All of the data collected using the magnifiedintersections method was for subsamples examined ina random order with the identity of the rootsunknown to the observer.

33 A N P 115

498 T. P. McGonigle and others

Table 1. A summary of the variables in the investigation into subjective visual estimation using test diagrams {a)and {b)

VariableTruevalue

Subjectmean

Standarddeviation

Coefficient ofvariation (%)

Score for {a)Score for {b)Score for (6)

score for {a)Score for {b)

-r-score for (a)

0-450-630-18

1-40

0-490-660-17

1-37

0-110-110-09

0-23

221753

17

Assessment of the variability of data from themagnified intersections methodTo determine the variability associated with thescore for a given subsample, one subsample wasscored ten times using different starting points forthe first pass of the field of view in each case. Thisensured difTerent intersections were examined foreach of the ten estimates. To assess the variabilitybetween subsamples, 10 subsamples were scored foreach of two root samples. To determine variabilitybetween observers within subsamples, onesubsample was scored by eight observers. Thesevaried in experience of examining mycorrhizas from5 years to none at all. For this comparison eachobserver used the same stage graticule locations forthe passes across the slides. To assess the number ofintersections required for a root subsample, one rootsample (analysed by three observers using differentroot subsamples) was recorded so that a runningmean of AC could be calculated for each incrementof 10 intersections.

0-8

07

0-6

o 0-5o

CO

0-4

0-3

0-2 04 0-6Score for (b)

08 10

Figure 4. A scatterplot of the scores for diagrams {a) and(6) in Figure 2 in the investigation into subjective visualestimation. Each of the 70 points on the graph correspondsto the estimates made by one subject. The dashedhorizontal and vertical lines indicate the true values for (a)and (b).

Comparison of the magnified intersections method tothe dissecting microscope grid-line intersect method

Prior to mounting on microscope slides, 30 rootsubsamples (one for each of 30 samples) were scoredfor colonization under the dissecting microscopeusing the grid-line intersect method (Giovannetti &Mosse, 1980). Roots were dispersed in glycerin in a9 cm Petri dish marked into 1 cm grid-squares. Atleast 100 intersections were examined for eachsubsample. Each intersection was counted in thefollowing categories; colonized, not colonized andunclear. Two values for proportion of root lengthcolonized were calculated. For the lower estimate,the unclear scores were treated as not colonized; forthe upper estimate, they were treated as colonized.

variation, almost all subjects (64 out of 70) were ableto tell that the proportion in {b) was larger than thatin {a). The scores for {a) and {b) were positivelycorrelated (correlation coefficient = 0-648, P