Repeated exposure to enhanced UV-B radiation in successive generations increases developmental instability (leaf fluctuating asymmetry) in a desert annual


G. F. Midgley Fax: +21 797 6903; e-mail:


Populations of the desert annual Dimorphotheca sinuata, derived from a common seed stock, were exposed concurrently over four successive generations to either ambient (representing no stratospheric ozone depletion) or elevated (representing 20% stratospheric ozone depletion) UV-B levels during their complete life cycle. Leaf fluctuating asymmetry (FA) was measured in populations of plants grown from seeds of selected generations which had experienced different UV-B exposure histories, and from seeds collected from a wild population of this species which grows in a naturally enhanced UV-B environment. These measured plants had been grown in a greenhouse under essentially UV-B-free conditions. Leaf FA was significantly increased by greater numbers of enhanced UV-B exposures in the parentage of the seed. There was a linear to exponential dose–response relationship between number of UV-B exposure iterations in seed parentage and leaf FA, suggesting that damage to DNA caused by UV-B exposure during plant development may not be fully repaired, and thus be inherited by offspring and accumulated over successive generations in this species. Leaf FA of plants grown from seed from the wild population was not significantly greater than that of control plants whose parentage experienced only ambient UV-B exposures, although this negative result may have been due to low sampling intensity and measurement resolution, and the relatively low UV-B enhancement experienced by the wild population. We conclude that leaf FA may constitute a relatively sensitive yet inexpensive means of quantifying UV-B damage to plants.


Stratospheric ozone depletion and the resulting increase in solar ultraviolet-B radiation (UV-B, 280–320 nm) impinging on our world (Kerr & McElroy 1993) continue to elicit concern about the sensitivity of the Earth's biota to this stress (Tevini 1991). Direct effects of enhanced UV-B radiation on the growth and metabolic processes of plants grown under realistic field irradiance conditions seem to be relatively minimal and transient (Rozema et al. 1997), though elevated UV-B may directly affect some ecosystem processes such as decomposition (Gehrke et al. 1995). In experiments which use realistic levels of photosynthetically active radiation (PFD, 400–700 nm), such as outdoor as opposed to many controlled chamber experiments, UV-B protection mechanisms are enabled, and are thought to operate efficiently (Caldwell & Flint 1994). Current understanding of realistic field-based results is that elevated UV-B causes photomorphogenic changes at the individual plant level, rather than other inhibitory effects and productivity losses (Rozema et al. 1997). However, DNA damage in plants has not been eliminated as an important result of elevated UV-B, as relatively low fluence rates may result in many disruptive effects at the molecular level, notably dimer formation in DNA (Strid, Chow & Anderson 1994). Indeed, even UV-A radiation has been shown to induce the formation of dimers (Quaite, Sutherland & Sutherland 1992), but effective repair mechanisms are thought to be able to reverse most of these, and other UV-B induced lesions (Sancar & Sancar 1988), especially in the presence of visible light of sufficient intensity. Thus it is presently unclear whether damage at the molecular level is a cause for concern in natural ecosystems.

Findings of UV-B induced reductions in pollen viability in several South African annual species grown under enhanced UV-B (Musil 1995) indicate that this ecologically critical developmental stage, at least, is potentially vulnerable to genetic damage by UV-B. Recently it has been suggested that, even under experimental treatments using natural light, damage to the plant genome caused by elevated UV-B may also be inherited by successive generations of the desert annual Dimorphotheca sinuata DC. (Asteraceae) and thus accumulate in the genetic material (Musil 1996). This form of damage may be extremely important in plant populations which have rapid turnover of generations, such as annual species which are common in high-radiation desert environments. Furthermore, populations which are isolated by habitat fragmentation may be further at risk to this form of damage due to limited outcrossing opportunities.

Testing the hypothesis of accumulating genetic damage using genetic techniques would be extremely expensive. However, an inexpensive alternative measure does exist. Fluctuating asymmetry (FA) is a measure used to infer phenotypic quality, or fitness (Polak & Trivers 1994; Watson & Thornhill 1994), and has been used widely in biological studies to reveal both direct and inherited effects of environmental stress on organism developmental stability. FA comprises small departures from the perfect symmetry expected, for example, in each member of a bilateral trait such as leaf lamina width, which results from the expression of the same genes. Thus FA in a bilateral organ is a form of asymmetry, calculated as the absolute difference between the right (R) and left hand (L) sides:


which shows a normal distribution around a mean of zero, with zero representing optimal genetic expression and providing the benchmark for the assessment of FA (Polak & Trivers 1994). This measure has not yet been applied in UV-B stress studies, and may be well suited for identifying inherited and accumulating developmental instability as distinct from photomorphogenic effects. Annual species such as D. sinuata are ideal test cases, as they can be grown to provide material with different UV-B exposure histories relatively rapidly.

In this study we measured leaf morphological parameters to test whether the exposure of successive generations of the annual D. sinuata leads to accumulating developmental instability. In particular, we aimed to test whether longer histories of UV-B exposure were related to increasing FA. Although FA measures have been used successfully in biology for some time, only recently has the dose–response of FA to applied stress been quantified, with some success (e.g. Kozlov et al. 1996). We also tested whether the experimental changes in FA are reflected in native populations growing naturally in a high UV-B radiation environment.


The derivation of the experimental seed populations of D. sinuata is described by Musil (1996). The seed populations were obtained from ancestors established originally from a common seed source, which had been exposed to different combinations of ambient and elevated UV-B radiation during their entire growth cycle over multiple generations (Fig. 1). In the first two generations, populations were grown in a polycarbonate-clad greenhouse (no transmission below 400 nm). Peak daily photosynthetic photon flux densities (PFD) in the greenhouse ranged seasonally (spring to midsummer) from 600 to 1800 μmol m–2 s–1 (typically 90% of that outdoors). UV-B radiation at ambient and enhanced levels was supplied by fluorescent sun lamps (Phillips TL/12 40 W UV-B, The Netherlands) suspended above the plants, during an 8 h period centred on solar noon (Musil 1995). Lamps were filtered with 0·075-mm-thick cellulose acetate film (Courtaulds Chemicals, Derby, UK, with transmission down to 290 nm) which was replaced weekly. Spectral irradiances of filtered lamps, measured after sunset with a monochromator spectroradiometer (IL-1700, International Light Inc., Newburyport, USA), were weighted with the action spectrum for intact plant DNA (Quaite, Sutherland & Sutherland 1992), normalized at 290 nm, and integrated over wavelength to obtain biologically effective UV-B irradiances as a function of distance from the lamp source. Lamps also supplied low amounts of supplemental UV-A radiation (biologically effective doses were about 13% of corresponding ambient values outdoors, as calculated according to Quaite et al. 1992). Control plants received levels of UV-B radiation which correspond with weighted doses presently received at the southerly distribution limit of D. sinuata (33°56′S, Cape Town, South Africa) over its natural growing period (varying seasonally from 2·55 to 8·85 kj m–2 d–1 for clear skies). Treatment plants received UV-B radiation which approximates the weighted doses that would be received simultaneously at the northerly distribution limit of this species (26°38’S, Aus, Namibia) under an anticipated 20% ozone depletion (seasonal range: 4·70–11·41 kj m–2 d–1 for clear skies) as calculated according to an empirical model (Green 1983). UV-B radiation treatments were adjusted weekly to account for seasonal changes in radiation intensity.

Figure 1.

. Diagram of UV-B treatments applied to the parentage of seeds used in this study. Each growth cycle represents the full life cycle of each parent generation; boxes represent plant populations which received unique UV-B treatments (clear = ambient UV-B and shaded = enhanced UV-B). Boxes with heavy lines represent those populations which were the source for seeds used in this study, and their corresponding numbers give the number of iterations of enhanced UV-B treatments.

Plants of the third and fourth generations were grown outdoors, and supplemental UV-B radiation was provided by lamps suspended above plants (as described above) in a step-wise manner during an 8 h period centred on solar noon (Musil 1996). For control plants (receiving ambient UV-B levels), lamps were filtered with 0·12-mm-thick Mylar-D film (DuPont De Nemours, Wilmington, Delaware, USA, with no transmission below 316 nm). Lamps above treatment plants were filtered with 0·075-mm-thick cellulose acetate film (as described above). All filters were replaced and UV-B radiation treatments adjusted weekly to account for seasonal changes in radiation intensity. Control plants received only ambient levels of UV-B radiation (i.e. doses presently received at the southerly distribution limit of D. sinuata of 33°56’S). Treatment plants received both ambient and supplemental UV-B radiation to achieve doses expected simultaneously at the northerly distribution limit of this species (26°38’S) under an anticipated 20% ozone depletion, as described above. For all four generations, populations were grown for similar lengths of time at similar times of year, such that total seasonal UV-B doses were comparable.

Plants were allowed to set seed under the experimental conditions, providing seed for the successive iterations of UV-B treatments (Fig. 1). Collected seeds were stored dry and at room temperature. At each iteration, a number of seeds were set aside in a reserve bank for later work, including this study. In this study, plants were also grown from seeds of a wild population collected near Augrabies Falls, South Africa (29°S, seasonal UV-B dose range: 3·30–9·02 kj m–2 d–1), to provide a comparison with a native population from a somewhat higher UV-B irradiance environment than that in the south-western Cape.

For this study, seed was germinated in a medium of coarse sand, leaf mould and loam (2:1:1) in pots (10 seeds per pot) on tables in a polycarbonate-clad greenhouse. Thus the plants germinated and developed in an essentially UV-free environment, revealing the accumulated effects of UV-B radiation history, and not direct UV-B effects. The polycarbonate used was highly transparent to visible light, attenuating only about 10% of PFD. After the plants had grown for 3 months, and were in flower, four leaves were collected from five individuals of similar sizes (each from a different pot, and always taller than 10 cm) which had developed from each seed source representing a unique UV-B history.

Leaves of D. sinuata in this study were generally between 50 and 70 mm long and between 6 and 20 mm wide (median width 13·5 mm). The leaf margins are sinuate-dentate, with between four and six teeth positioned on each side of the leaf, more or less oppositely. A small but distinct hardened point marks the apex of each tooth. Leaves were placed on a flat surface with a consistent orientation so that widths could be measured on left- and right-hand sides of the midrib. Consistent differentiation of left- and right-hand sides is critical for discerning fluctuating asymmetry from the two alternative asymmetrical patterns (discussed below). Widths were measured with a magnifying lens and a graduated scale with a resolution of 0·1 mm. Measurements were made at the broadest part of the leaf, from the midrib to the point of a tooth, and from the midrib to the narrowest part of the sinus (the leaf margin between the teeth).

From the width measurements, FA was calculated for both sinus (FAs) and tooth width (FAt) measurements. Following Kozlov et al. (1996), we adjusted FA for leaf size, as follows:

FA = 2 ×|RL|/(R + L),

where R and L represent widths measured on the right- and left-hand sides of the midrib, respectively.

Apart from FA, two alternative forms of asymmetry exist, namely anti-symmetry (where asymmetry is the norm, but the side containing the larger character varies), and directional asymmetry (where a character on one side usually has a value larger than that on the other). It is important to distinguish FA statistically from these possible alternative forms. Absence of antisymmetry was demonstrated by the normality of distribution of leaf-specific values of (RL) (Kolmogorov–Smirnov normality test: P = 0·64), and the absence of directional symmetry shown by the zero mean test of leaf specific values of (RL) (t-test: two-tailed P = 0·17).

Individual leaf area and leaf perimeter were measured with an image analyser (Delta T Devices Ltd, Cambridge, UK), and a shape factor calculated. The shape factor is a ratio of the perimeter of the leaf to the perimeter of a circle with the same area as the leaf, that is, it represents leaf non-circularity due to leaf elongation and margin shape so that progressively more elongated and complex leaves have shape factor values progressively greater than 1.

Leaf FA data for control populations (i.e. those from the original seed batch and those which had been subjected only to consecutive ambient UV-B treatments; see Fig. 1) were first subjected to one-way analysis of variance (ANOVA, Unistat 4·5, Unistat Ltd, London, UK) to test for increased homozygosity due to inbreeding from an initially limited seed stock. Although FA tended to increase very slightly with increasing number of growth cycles, this effect was far from significant (F2,12 = 1·32, P = 0·3 for FAt, F2,12 = 0·9, P = 0·43 for FAs), and all control data were thus treated as a homogenous set.

One-way analysis of variance (ANOVA) was used to test for UV-B radiation history effects on all leaf measures. The values for four individual leaves per plant were averaged to derive a plant-specific value, giving replication of five plant-specific values per UV-B radiation history (use of leaf-specific values as independent replicates would constitute pseudo-replication). To test for potential loss of information by averaging at the individual plant level in the case of FA values, an alternative, valid statistical design was also employed. Here individual leaf values were treated as repeated measures per plant in a repeated measures ANOVA, yielding almost identical results to the plant-averaged approach. We present only the results for plant-averaged data. Where UV-B history significantly altered a leaf measure (at the 5% level), a multiple range test (LSD method, 5% level) was employed, and three regression models (linear, exponential and second-order polynomial) used to detect a dose–response (plants grown from field-collected seed at Augrabies Falls were excluded from the regression analysis).


All leaf morphological measures other than FA showed a non-significant response to UV-B exposure history (Table 1), suggesting that gross leaf characteristics such as individual area, length of perimeter and leaf shape factor (together representing leaf shape complexity) and leaf sinuation (represented by tooth height) were not heritably affected by repeated exposure to this stress. Very few previous UV-B stress studies have explicitly quantified the heritable negative effects of UV-B radiation. One previous experiment which used seeds from the same source, but in which plants were grown under ambient and enhanced UV-B conditions (Musil 1996), showed that UV-B history had a greater relative negative effect on gross plant characteristics (total leaf area, leaf dry mass and stem mass) than did direct UV-B exposure effects. In combination with the present study, this suggests that any inherited effects which accumulate during repeated exposures to UV-B radiation may interact with and aggravate the direct effects of UV-B exposure. These studies also highlight real difficulties in differentiating between accumulated and direct UV-B impacts.

Table 1.  . Effects on leaf morphological measures of repeated exposure to enhanced UV-B radiation in successive generations of the desert annual Dimorphotheca sinuata. Results are given as F ratios and corresponding probabilities (one–way ANOVA) Thumbnail image of

Leaf FA was the only morphological measure altered by UV-B exposure history (Table 1) appearing in development of both leaf teeth (FAt) and sinuses (FAs). The lower significance level for UV-B effect on FAs is almost certainly due to its lower measurement resolution [FA is prone to measurement error (Palmer & Strobeck 1986), and resolution of FAs was roughly only 75% of that of FAt]. The significant effect of UV-B history on leaf FA strongly suggests a negative impact of this stress on plant developmental stability. Its heritability and its expression under essentially UV-B free conditions together constitute very good evidence that DNA damage occurred during exposure of the parent material to enhanced UV-B conditions. This may have happened at one or many of several sensitive stages, including apical meristem development during floral induction (e.g. Feldheim & Conner 1996), ovule development (but see Day & Demchik 1996), anther and pollen sac development, pollen development (e.g. Jackson 1987; Demchik & Day 1996), and even in immature and mature seeds (e.g. Musil 1996). UV-B induced DNA damage in seeds has already been implicated in the alteration of subsequent adult plant photochemical and reproductive performance in a related annual species, D. pluvialis (Musil 1994), and it has been suggested that damage to DNA while partially dehydrated, i.e. during either or both pollen and seed stages, may be of particular importance (Musil 1996). Knowledge of the relative impact of UV-B exposure on key developmental stages would be of great use in predicting the vulnerability of plant populations in the field, particularly in southern Africa where UV-B levels vary seasonally due to changes in stratospheric ozone associated especially with the breakdown of the Antarctic polar vortex in spring (Bodeker, Scourfield & Barker 1992). Effects of such fluctuations could be dependent on the developmental stage of terrestrial plants (Jordan et al. 1994).

The relationship between FA and UV-B exposure history appeared to follow a simple dose–response pattern (Fig. 2). The multiple range test differentiated two homogenous groups which followed the sequence of enhanced UV-B iterations, but clearly separated only the extreme treatments along this axis (i.e. control plants from plants with a history of both three and four generations of enhanced UV-B exposures). For FAt, linear regression yielded the same correlation coefficient as both exponential and polynomial regressions, and is the most parsimonious description of the dose–response. For FAs, exponential and polynomial regressions gave slightly better fits to the data (exponential: r2 0·30, polynomial: r2 0·28, linear: r2 0·27), and in this case the linear model is a conservative interpretation. The significant linear relationship between FA and UV-B exposure iterations, combined with the results of the multiple range test, suggests a progressively accumulating effect of this stress at the genetic level, which is, at best, only partially repaired. This study therefore raises the possibility of a heritable, accumulating effect of UV-B radiation on plants, as has previously been proposed (Stapleton 1992), as distinct from a direct irradiation effect. The lack of repair is surprising, given that all the generations were grown under light conditions thought to be of sufficient intensity to ameliorate damaging UV-B effects (Caldwell & Flint 1994; Strid et al. 1994), and suggests that light-independent repair mechanisms (such as excision–repair and those dependent on DNA glycosylases; Hoeijmakers & Lehmann 1994) may not have been completely effective. The potential impact of heritable and accumulating DNA damage in plant species on the functioning of natural ecosystems has not received due attention in recent reviews (e.g. Caldwell & Flint 1994; Teramura & Sullivan 1994; Caldwell et al. 1995; Rozema et al. 1997), due to the paucity of information on the topic.

Figure 2.

. Dose–response relationship of leaf FAt and FAs to the number of iterations of exposure to enhanced UV-B radiation in the desert annual Dimorphotheca sinuata. Values for the Augrabies Falls population were excluded from the regression to determine dose–response. UV-B exposure histories with letters in common (placed along the X-axis) are not significantly different at the 5% level (LSD multiple range test).

Mean FA values for control plants in this study (roughly 0·05) compare well with those measured for birch leaf controls in a pollution study (Kozlov et al. 1996). Furthermore, the mean maximum FAt value reported here (roughly 0·1), is only slightly lower than maximum values measured for birch leaves growing under high atmospheric pollution loads (Kozlov et al. 1996), suggesting that FA could serve as an indicator of UV-B damage. However, plants from the wild population sourced at Augrabies Falls showed FAs values which grouped them in an intermediate position, overlapping with the control plants and those which had received all but the greatest number of enhanced UV-B iterations. This suggests that this wild population, theoretically exposed to higher natural ambient UV-B levels than the control population grown at higher latitude, shows only weak evidence for developmental instability. It may require far greater sampling intensity and better measurement resolution in this species to test for changes in FA caused by UV-B differences of latitudinal or geographical origin. It may be that other defects associated with excessive developmental instability would be selected against under natural conditions, given a large enough gene pool, thus masking this effect in natural populations. It is possible that rising UV-B levels would increase FA more markedly in isolated populations with limited potential for outcrossing, and in perennials which may accumulate damage to meristems over successive years.

In conclusion, this paper demonstrates simply that a measure of developmental instability, leaf FA, is increased by a history of greater numbers of UV-B exposures in the parentage of populations of the desert annual D. sinuata. This is manifested even when plants are grown under UV-B free conditions, suggesting that the accumulated UV-B damage is heritable. The measurement of FA may provide a sensitive and inexpensive means to quantify developing UV-B stress in plants.


We are grateful to two anonymous referees for useful comments, and to the Roland and Leta Hill Trust, administered by WWF South Africa, which funded all greenhouse facilities.