Herbicide Rotations and Mixtures - American Chemical Society

Chapter 30

Herbicide Rotations and Mixtures Effective Strategies to Delay Resistance 1

2

J. Gressel and L. A. Segel 1

Department of Plant Genetics, Weizmann Institute of Science, Rehovot 76 100, Israel Department of Applied Mathematics, Weizmann Institute of Science, Rehovot 76 100, Israel

Downloaded by UNIV OF PITTSBURGH on April 4, 2017 | http://pubs.acs.org Publication Date: February 23, 1990 | doi: 10.1021/bk-1990-0421.ch030

2

Herbicide resistant populations have evolved only in monoculture and/or mono-herbicide conditions at the rates we have previously predicted. Contrary to our original model, no populations of atrazine-resistant weeds have appeared in corn where rotations of crops and herbicides or herbicide mixtures were used. This is due to the previous lack of information about the greatly reduced fitness of the resistant individuals, which could be expressed only during rotational cycles, and also to the greater sensitivity of resistant individuals to other herbicides, pests and control practices. Our new model, presented here, describes how these factors reduce the resistant individuals to extremely low frequencies during rotation. It is not yet clear whether such factors will delay the rate of appearance of other types of resistance; e.g. the multiple resistances to grass-killing herbicides in wheat, or to inhibitors of acetolactate synthase, where thefitnessof resistant biotypes may be higher. Weeds have only evolved resistant populations to the herbicides described in the previous chapters where there was mono-culture with a single family of herbicides. The only exceptions have been where different herbicides having the same site of action were used (e.g. a rotation of photosystem II inhibiting herbicides). It is expected that resistant populations will soon evolve in wheat monoculture where high selection pressure herbicides with different sites of action but the same mode of degradation are rotated (1, 2). The appearance of resistance in mono-culture and/or mono-herbicide usage followed a simple population genetics model that integrated: (a) the selection-pressure of the herbicide (based on the rate used and its persistence); (b) the germination dynamics of the weeds (over the season and from the soil seed-bank); (c) the initial frequency of resist ants due to mutation of susceptible individuals; (d) thefitnessof the resistant mutants in competition with the wild type under 0097-6156/90/0421-0430$08.25A) © 1990 American Chemical Society

Green et al.; Managing Resistance to Agrochemicals ACS Symposium Series; American Chemical Society: Washington, DC, 1990.


30. GRESSEL AND SEGEL

431

Herbicide Rotations and Mixtures

field conditions and (e) the number of generations (seasons) the herbicide was used (3, 4). This model was helpful in understanding why resistance evolved to the high selection pressure s-triazines in corn but not to lower selection pressure thiocarbamates, chloroacetamides and phenoxy herbicides in this crop. It is consistent with our model that there was no evolution of resistant populations of weeds to 2,4-D and MCPA in wheat. The model led one to expect the appearance of weed biotypes resistant to the sulfonylureas, chlorotoluron, diclofop-methyl and mecoprop in weeds of wheat. Resistant biotypes evolved first in those weeds where the herbicides exert the highest selection pressure. For example diclofop-methyl exerts a much higher selection pressure on Lolium spp than on Avena spp; in agronomic terms, this herbicide is more effective on Lolium, and Lolium has indeed evolved resistance with greater rapidity (Powles et al., this volume). Because of a lack offieldecology data, the early models inaccurately predicted evolution of resistance where herbicide mixtures and rotations were used. Vast areas of corn have received such rotations with s-triazine herbicides for 30 years, and resistant populations have not appeared. The models did not consider the then unknown extreme lack offitnessin many resistant biotypes in the seasons the triazines were not used. The model (3, 4) correctly predicted that mixtures could considerably delay the evolution of resistance but the lack offielddata on selection pressure of mixtures left it to the reader to insert the correct data into the equations and the accompanying figures. The revised model presented below considers what happens during the "off" years when a given herbicide is not used, and predicts that some high selection pressure herbicides can be sparingly used, in rotation, possibly even after resistance has appeared. The data show the urgent need of further research concerning the physiological-ecology of resistant weed populations. The Original Model The original model (3, 4) described different possible rates of enrichment of resistant individuals in populations, until the populations were predominantly resistant. It was assumed that (different) constant proportions of susceptibles and resistants germinated, survived to the end of the season, etc., and that susceptible and resistant individuals had (different) constant seed yields. It was further assumed that resistants initially formed an exceedingly small fraction of the population, certainly a well-warranted assumption during the years of resistance enrichment, until there were sufficient numbers of individuals for resistance to be evident. The following simplified equation described the proportion of resistant individuals of a given weed species, N in the n** year following continuous application of herbicide (3,4): n

N

(*)

=

N

(«)

i 4^

n

+

(Eq. 1)

n

Strictly speaking, JVW gives the proportion of resistant seeds deposited per unit area in year n. However, other quantities such as the number of resistant



432

MANAGING

RESISTANCE TO AGROCHEMICALS

plants that germinate or that become established can be obtained from by multiplication with suitable constants. What is essential is that all these various measures of resistance change by a constant factor each year, giving rise to an exponential increase in resistant individuals (Fig, 1A.). In the case of Eq. 1 this factor is 1 + (fa/n). Various parameters appear in the simple algebraic Eq. 1 for the frequency (proportion) of resistant weeds in a species that is preponderantly sensitive to a given herbicide. is the very low frequency of resistant individuals in the population before it is exposed to the herbicide. In the absence of herbicide, resistance is sustained in the population by a balance between mutation and depletion of the proportion of resistant individuals by their lesserfitnessin the absence of selection, resulting in a resistance fraction somewhat lower than the mutation frequency. If mutants were more fit than the wild type, they would be the wild type. Mutantfitnesscan be near neutrality, and the mutants would then be found in different proportions at various geographical areas due to "drift". The factor / is a reproductivefitnessmeasure that compounds the relative ability of resist ants, compared to susceptibles, in germination, establishment, growth, seed production and survival. Fitness is thus always measured when the resistant and susceptible individuals are in competition with each other in the absence of selection for the mutants. When they are grown separately the resistant individuals are often less 'productive', and in competitive situations the resistant individuals are more affected (Table I). (See Appendix 1 for exact mathematical definitions.) The factor a describes the selection pressure. It is the ratio of the fraction of resistants that survive herbicide application during a season to the corresponding fraction of susceptibles. Thus, early or late germinating susceptible individuals that produce seeds are considered in this "effective-kill'', in contrast with the initial control usually measured by weed scientists. Thus, for example, if the herbicide kills 95% of the susceptibles and none of the resistants, then a = 20. The factor n is the average number of years that a seed remains viable in the seed bank. According to Eq. 1, the frequency of resistance in the population starts at a very low value and then increases by a constant factor each year. In spite of the exponential increase, it will be hard to detect resistant individuals in a field until resistance is at a level of 10-30% of the population. Eq. 1 allowed the plotting of rates of enrichment of resistance, with different scenarios of selection pressure, seed bank size,fitness,and initial mutation frequency (Fig. 1A). The values that could be "plugged" into the equation to generate the scenarios were based on a very limited data-base, mostly from corollary systems, such as heavy-metal tolerance. We knew too little about weed-herbicide interactions at that time to make precise estimates. With the experience of hindsight, we can see where the model was clearly correct, and where it needed modification.





IQ-o y Resistance visible in the field

..-1.

0

•

1

1

1

.

i

.

.

N

.

-1-4

1

1—J..

10 20 SEASONS REPEATED TREATMENT

0 15 Seasons : 2 on • • 0 20 Seasons 1 on

; 1 off

0 Seasons

; 2 off

30 1 on

;

433

I off

i

I

30

30

45

40

60

60

90

Seasons

Figure 1. Prediction of the effect of herbicide rotation in the original model. A. Overall average effect of scenarios with different selection pressures (a = 0.1= 90% effective kill (EK); a = 0.01 = 99% EK, a = 0.33 = 75% EK, a = 0.5 = 50% EK), seed bank dynamics (n; See Table III) and differentialfitness.The different scales give the different rotational scenarios, from mono-herbicide to one treatment in 3 seasons. B. Calculated effect of herbicide stoppage, and restarting at various intervals. Source: Redrawn from equations andfiguresin refs. (3) and (4).


434

MANAGING RESISTANCE TO AGROCHEMICALS

Table I. Lower Productivity and Competitive Fitness of Resistant Biotypes Herbicide

Notes

Productivity Competitive Fitness (1:1)*

Species resistant

s-triazines resistant -r susceptible Amaranthus hybridus (seed) .11 (seed) .90 .18 Senecio vulgaris .32 (seed) .47 .43 (DW) .60 Chenopodium album (seed) .75 .08 C. strietum (seed) 1.00 1.78 Brassica napus (seed) .76 .28 Solarium nigrum (seed) .67 Trifluralin Eleucine indica (DW) .68 Diclofop-methyl Lolium rigidum (DW) 0.59-0.74


0

Ref (5) (6) (5) (7) isogenic' (8) (9) (9) isogenic (10) isogenic (11) (12) density (13) dependent

Productivity is measured by growing resistant and susceptible biotypes separately; competitivefitnessis measured by growing them in a mixture. °DW is the dry weight of above ground parts of plants; "seed" may include fruit or wholeflowerdepending on study cited. Fitness of 1:1 mixture; isogenic means that nearly nuclear isogenic material or reciprocal F hybrids were used. 6

x

Affirmation of the model (a) The importance of selection pressure. The atrazine and chlorsulfuron levels required to kill different weeds vary over more than an order of magnitude. The selection-pressure of both herbicides is greatest on broadleaf species requiring the lowest rates for weed-control. The selection pressure is least for those weeds requiring the highest herbicide levels, i.e. the grasses. The first weeds to evolve atrazine resistant populations were Senecio, Amaranthus spp





435

and Chenopodium spp; the last were the grasses, as the model predicted. At least two broad-leaved weeds and one crop have already evolved chlorsulfuron resistance in thefield,under selection pressure of this recently introduced and highly persistent herbicide. The herbicides with the greatest persistence will exert the highest selection pressure. The triazines, dinitroanilines, and sulfonylureas meet this criterion with season-long control, and resistance has evolved to all these groups. By contrast, 2,4-D and other phenoxy herbicides, thiocarbamates, etc. have much shorter biological-persistence in the soil and resistance has not evolved to them. Paraquat resistance has evolved, and may seem to contradict the theory, as paraquat loses biological activity upon touching the soil, due to binding to colloids. The lack of residual persistence was balanced by farmer persistence. All cases of paraquat resistance occurred where this herbicide was employed 5-10 times during each season in mono-herbicide usage. (b) Seed bank life. The longer the life in the seedbank, the greater the buffering effect of susceptible seed from previous years, decreasing the rate of resistance. Senecio vulgaris has evolved resistance in orchards, nurseries and roadsides where there was no mechanical cultivation, but not in cornfields. The Senecio seed is incorporated into the soil seed bank in cultivated cornfields, where it is viable for many years (14). All Senecio seed falling on undisturbed soil on roadsides or orchards germinates or dies during the following season (15). Resistance thus evolved where there was the lowest average seed bank life time n, as predicted. Such information must be considered in formulating strategies for resistance management. Many other species do not have a seedbank, i.e. in no-till agriculture where n = 1. The model predicted that the rotation or mixing of herbicides with the same site of action (and thus similar mode of resistance) will result in the same effect as using a single herbicide. It was therefore suggested that use of corn/soybean; atrazine/metribuzin or sulfonylurea/imidazolinone rotations with various crops would be contraindicated (1, 2, 4). For the same reason, the genetic engineering of atrazine-resistant soybean for use in monoherbicide culture would be a mis-directed project, unless the atrazine usage in corn were replaced by other herbicides. Atrazine-resistant lambsquarters appeared in corn/sugarbeet; atrazine/pyrazon rotations (16) following this prediction. A case of decreased sensitivity to all oxidant generating herbicides has appeared in Conyza canadensis in orchards treated with paraquat after this weed evolved atrazine resistance (17).

Cases where the model was or may be inadequate (a) Use of other photosystem II herbicides against atrazine-resistant w The first atrazine-resistant Chenopodium appeared after 5-10 years of continuous herbicide use. According to the model, it should take 5-10 years for resistance to occur to each other photosystem II herbicide not having cross resistance (e.g. diuron, pyridate, phenolic types), if used on atrazine-resistant weeds, as further mutations would be required. Instead there is evidence that resistances to other photosystem II herbicides evolve more quickly.



436

MANAGING


When the model was formulated, the precise mutation frequency of chloroplast genome mutations was unknown and could only be estimated as being an exceedingly low number, much lower than nuclear mutation frequencies. There are now some data to indicate that the mutation frequency of chloroplast genes is orders of magnitude higher in certain individual plants than the normal plastid mutation frequency. This is due to a nuclear "plastome mutator" gene which increases the frequency of chloroplast mutations (18). Triazine resistance seems to have evolved in the individuals bearing the plastome mutator gene, and triazine-resistant populations will carry a much higher frequency of this mutator gene than susceptible populations. This would allow other chloroplast inherited photosystem II herbicide resistances to evolve at much higher rates in triazine-resistant populations than in wild type populations (19). If all cases of triazine resistance occurred where there was an active plastome mutator gene, it is highly likely that in most cases where there is a mutation conferring triazine resistance, there will be other mutations in the chloroplast genome as well. These would also lower fitness. (b) Multiple-resistances to herbicides with vastly different modes of action.

Multiple-resistances to insecticides and drugs are rather common, and are documented to the level of molecular biology. They are a rather recent occurrence with herbicides. Lolium rigidum that evolved resistance to diclofop-methyl was found to have a surprising cross resistance to chlorsulfuron as well as to all other wheat selective herbicides (20). Alopecurus myosuroides, which evolved resistance to chlorotoluron, was cross resistant to chlorsulfuron, pendimethalin and diclofop-methyl (21). Similar surprising multiple cross resistances to insecticides and drugs were often traced to the evolution of higher levels of nonspecific esterases, hydrolases or monooxygenases. The resistance in Alopecurus to chlorotoluron can be abolished by adding specific monooxygenase inhibitors along with the herbicide (22). Thus, we may have problems similar to those of the entomologists and pharmacologists. We can learn from their experience that multiple resistance can be both alleviated and delayed or prevented from occurring by the addition of synergists, to the herbicides (22) in our case. (c) Lack of triazine resistance in rotation with herbicides that do not di-

rectly inhibit photosynthesis. The original model did not adequately account for events in the "off-years" during rotations. Resistance was held to evolve at afixedrate that was a function of the number of generations or seasons a weed was treated with a particular herbicide. According to the old model, this meant that if it would take 6-10 years for resistance to occur in monoculture corn with atrazine as the sole herbicide, it would take 9-15 years in a corn/corn/wheat (or soybean) rotation where atrazine was used 2 of 3 years; or 12-20 years in a corn/wheat (or soy) rotation where atrazine was used every other year; or 18-30 years in a corn/wheat/soy rotation where atrazine was used once in 3 years (Fig. 1A). When the model was formulated 10 years ago, it appeared that the time was just about ripe for triazine resistance to break out in vast areas of the cornbelt where such rotations were used, as there had




437


been 6-10 years use of atrazine since it was introduced. Resistant populations did not appear. Very few farmers in the cornbelt grow monoculture corn, unlike in areas to the east of the cornbelt where atrazine resistance appeared. Some farmers use rotations. Some employ mixtures of atrazine with chloroacetamide herbicides, which allow the use of less atrazine (lowering the selection pressure) and which also kill Amaranthus and Chenopodium, as well as grass weeds. Such mixtures substantially delay resistance both according to the old model and from practice, although the magnitude is yet unclear. With the chloroacetamide herbicides used in these mixtures under attack and their use restricted in many areas and forbidden in others, herbicide rotation may be the only strategy remaining to delay the evolution of triazine resistance in corn. Why has rotation been a better strategy than predicted? We now think we know some of the answers and can incorporate them into an updated model, using similar equations, but with better data inserted. The newer data mainly consider the highly reducedfitnessesof the resistant biotypes, which are of greater magnitude and importance than we had expected. We discuss below some of these indications; more data are clearly needed, especially on intraspecific competition. Revised model: problems to be considered

The original model used an averagefitnessdifferential for all the generations treated (Fig. IA). Thefitnessdifferential between resistant and susceptible individuals essentially disappears with herbicides such as triazines, giving season long control, as there is no time for this differential to be expressed. Only the resistant biotypes can survive when the herbicide is present. Thus, the fitness differential seemed to be unimportant with triazines, but was possibly an important factor in delaying resistance to other herbicides with more ephemeral action. Thefitnessdifferential was only considered important when herbicide usage was stopped for a long duration and was responsible for the decay in the resistant populations (Fig. IB) (3, 4). We now think we know more about how fitness should be measured, and we have learned that resistant biotypes are often more susceptible to some of the herbicides and cultivation procedures used in the rotational years (negative cross-resistance). Based on this we now have to modify the model to consider what happens to resistant individuals in the "off" years when the herbicide in question is not used. Fitness of resistant weeds

The very low frequencies of resistant individuals in thefieldmust compete with the crop, with resistant members of other weed species, and (when the herbicide is not present) with susceptible members of the same and other species. If triazine resistance evolved only in populations with plastome mutator genes (see above), there is the strong possibility of multiple mutations in these plastid genomes, including deleterious mutations giving unfit alleles of other genes. These may explain much of the unfitness of resistant plants, as well as the published variabilities of plastid fitness. Deleterious nuclear



438

MANAGING


mutations can be bred or selected out of populations, because of chromosomal segregation and through somatic and meiotic crossing over. This is not as easy with chloroplasts, where such recombinations as negligible. Thus, the unfitness in atrazine-resistant plants may not be due to the psbA gene mutation, as has already been argued on biophysical grounds (e.g. 7, 24-26). During the evolution of resistant populations only intraspecific competition has been considered, except for one study (9). We await more data from the agro-ecologists on the importance of interspecific competition. The earliest and classical studies on competitivefitnesswere performed by pregerminating seedlings of resistant and susceptible individuals, interplanting them atfixeddistances, and allowing them to grow to maturity (5). The yields of the resistant and susceptible biotypes were measured. In almost all cases where this was done, the susceptible individuals outyielded the resistant ones (Table I). Some of the triazine-resistant grasses have been reported to be more productive (27), but this must be checked under more rigorous conditions. Most competition experiments have not been made with material that has nuclear isogenicity, where resistant and susceptible alleles are in otherwise identical nuclear backgrounds. Nuclear isogenicity is easy to achieve by using reciprocal hybrids in cytoplasmically inherited triazine resistance. There is no way at present to guarantee an identical plastome (except for the resistance gene) in such crosses until plastomes are engineered by site directed mutagenesis to have only psbA gene mutations. Repeatedly backcrossed material also provides near-nuclear isogenicity with a large differential in competitive fitness (Table I). In Chenopodium strictum there seems to be nofitnessdifference between resistant and susceptible individuals (9). C. strictum is a very slow-growing species and it is unlikely that photosynthetic electron transport limits its growth. Other cases are more complicated and may be due to various interrelated functions: (a)fitnesswas not measured from germination on; (b) there may be density dependent functions; (c) there may be different reactions to environmental conditions; (d) there may be different germination characters and seed bank dormancies; (e) the narrow genetic base of resistant individuals (28-30) is probably detrimental compared to the broad base in the susceptible wild type. (a) Fitness from germination. Weeds produce hundreds to thousands of seeds to replace one plant. Clearly most must perish before maturity, many during overwintering, early germination, and establishment. The competition prior to establishment can well be the fiercest. None of the reported competitive fitness studies measured competition at this stage. The simplest way would be to plant mixtures of resistant and susceptible seed and assay (using small leaf discs) which plants are resistant and susceptible. The seeding should be done at a variety of depths and densities (see below), and preferably under field conditions to best mimic the natural environment. As seeds from resistant plants are often smaller than those from sensitive plants, there surely should be a definite competitive disadvantage to resistance when the selecting herbicide is not present.





439

(b) Density dependence of fitness has not been adequately measured. For example, the reasons that Poa annua biotypes resistant to triazines evolved only in genotypes that were prostrate and not in those that are erect (31) are not clear. Possibly, the lack of competition allowed the prostrate types to spread. There are variations in the density dependence of diclofop-methyl resistant Lolium (13). The resistant plants were morefitunder sparse than dense spacing. Thefitnessalso varies with ratio of resistant to susceptible individuals in competition. In Table I, only the data from a 1:1 ratio are given. A perusal of much of the original data show that the resistant biotypes are even lessfitthan in the 1:1 mixture when they are in a lower proportion in the population. The data suggest thatfitnessshould be measured at various densities and ratios of sensitive to resistant individuals. (c) Environment and fitness. A lowered optimal temperature for growth and photosynthesis for resistant biotypes is one of the common (32) but not universal (32) pleiotropic (33) effects concomitant with triazine resistance. Nuclear isogenic lines of the same species were used in (11, 32) but not in (33). The triazine-resistant biotypes often germinate at lower temperatures (34, 35). Interpretation of thesefindingscan be complicated: the earlier germination and 'head start can be highly advantageous under many greenhouse conditions but in the field it can be devastating. A late frost will decimate the earlier germinating resistant population and leave the later germinating susceptible population. This shows thatfitnessmust be measured under field conditions to be indicative of what happens in the field. (d) Changing seedbank characters. Repeated and strong selection for resistant weeds under monoculture can easily abolish the "spaced out" germination typical of weeds during the season and over many seasons. The selection pressure being applied is similar to that used by our ancestors who turned weeds into crops by selecting for germination uniformity, thus abolishing weed germination characters. Under manyfitnesstests, this higher immediate germination of resistant individuals versus the susceptibles (e.g. 36) may give real but misleading results that will not mimic long termfitnessproperties under field conditions. (e) Narrow genetic base. Electrophoretic studies of resistant populations have always shown that they possess a much narrower base than adjacent susceptible populations (28-31). This is due to the 'founder effect' of mutants in diverse populations measured soon after evolution. In genetic evolutionary terms this suggests that under certain narrow conditions the resistant populations may be more fit than the wild type, but under broad and varying environmental conditions they will be less fit. In many cases this might mean that thefitnesswill slowly increase given the chance of repeated crossing with the wild type susceptible population due to 'nuclear compensation'. Because of the fundamental biochemical lesion caused by triazine resistance it is doubted thatfitnesscould really increase. One can increase the yield of triazine-resistant species by intercrossing. Still the reciprocal intercrosses with the sensitive biotype as female parent always outyielded the resistant offspring in such crosses (37). The effect of interbreeding increasingfitnesswill have to be checked with other types of herbicide resistance. 9


440

MANAGING



Negative cross resistance

One of the most important factors that the early model did not consider, because it was then unknown, is that triazine-resistant individuals were often less fit under the agronomic procedures used during the 'off' years when triazines were not used. Unfortunately we have no accurate counts of the demise of resistant populations when triazine usage was stopped due to high resistance levels; these would have been most telling. What has been shown is as follows: (i) Standard mechanical cultivations of mixed resistant and sensitive Senecio vulgaris populations reduced the resistant individuals by a far greater extent than the susceptible ones (14). (ii) The lack offitnessoften can be due to other biotic factors. It was found that triazine-resistant rutabagas, which nominally produced yields as high as the near iso-nuclear susceptible biotype, were totally and selectively decimated by a viral infection (Souza-Machado, 1987, pers. comm.). Triazine-resistant Amaranthus hybridus was selectively eaten by beetle larvae, and triazineresistant Chenopodium album was more susceptible to fungal disease than the wild-types (R. Ritter, 1988, pers. comm.). (iii) Many herbicides are more toxic to resistant individuals than to susceptible ones (Table II). Table II only contains data for negative cross-resistances, but these are not the preponderant case and are clearly not universal. Still, they can clearly be elucidated and incorporated into strategies for managing resistant weeds. The negative cross-resistances in atrazine-resistant weeds include herbicides that act at or near the same site in photosystem II (DNOC and dinoseb) as well as herbicides acting on other photosystems (paraquat) or at totally different sites. There was negative cross-resistance to other tubulin binding herbicides in dinitroaniline resistant Eleucine indica (Table II), but not to six commercial herbicides on this weed (12). The negative cross-resistance to imazaquin (Table II) occurred in only one of 21 chlorsulfuron resistant mutants. The other mutants had varying levels of co-resistance to imazaquin. Resistant biotypes sometimes grow better in the presence of the herbicide than without the herbicide. For instance, Lipecki (pers. comm.) found that a triazine-resistant biotype of Amaranthus hybridus had double the dry weight per plant at 5 kg/ha simazine than without the herbicide. It would be useful to have more such quantitative experiments. This lower resistant biotype productivity when the herbicide is not present results in a stronger lack of competitivefitnessin the "off" years. How the model must be modified

In the seasons when herbicides are used, the enrichment for resistance will be as in Fig. 1A for lines with nofitnessdifferential (/=i). In the "off years (when the herbicide is not used) thefitnessof the resistant individuals may be exceedingly low. The overallfitnessis a compoundedfitnessof each stage or condition where measured: germination, establishment, growth, effects of environment, agronomic procedures and herbicides. Both inter- and intraspecific




441

Table II. Negative Cross-Resistance of Herbicide Resistant Biotypes Primary resistance species

Negative crossresistance

Dinitroaniline Eleucine indica

I50

ref

R/S 6 (38)

chlorpropham

triazines Amaranthus retroflexus


parameter measured

Chenopodium album Brassica napus Senecio vulgaris Conyza canadensis Kochia scoparia Epilobium ciliatum

mecoprop Stellaria media MSMA-DSMA Xanthium stramonium

dinoseb fluometuron DNOC dinoseb dinoseb dinoseb DNOC 2,4-D oxyfluorfen paraquat pyridate chlorpropham

FW thylakoids thylakoids

benazolin

FW

thylakoids FW FW FW FW FW

paraquat bentazon

.27 .22 .5 .27 .66 .21 .1

.46

(39) (40) (40) (39) (39) (39) (41) (42) (43) (43) (43) (44)

.53

(45)

.50 .65

(46) (46)

b b b b

Chlorsulfuron Datura stramonium

imazaquin

FW

0.03

(47)

Paraquat Conyza canadensis

glufosinate

PS

.26

(17)

0.

FW = fresh weight. PS = Photosynthetic C 0 fixation.Thylakoids = photosystem II activity of isolated thylakoids. The I is the concentration lowering the parameter measured by 50%. R = resistant; S = susceptible. Strain CSR2 only. Where no I R/S ratio is given, there was a large degree of negative cross-resistance at a single herbicidal rate. 2

50

0

6

50

competitions should be considered at all stages. We have no information on differential resistant vs. susceptible survival in the seed bank. As no measurements have been made of the loss of resistant individuals during the off years, we can only generate a series of curves based on expectedfitnessdifferentials (Table I) or even higherfitnessdifferentials based on negative cross-resistances (Table II) and the negative effects of the agronomic procedures used with other crops.


442

MANAGING

RESISTANCE

TO AGROCHEMICALS


The Revised Model The previous model kept track of the various influences affecting resistant and susceptible weeds. In particular, a seed bank was presumed to contain fully viable seeds for n years which then died. We neglected the loss of seeds from the seed bank due to germination. In the appendix to our earlier paper (3), an alternative model was developed based on the hypothesis that a constant fraction of seeds in the seed bank perish each year (6^ for resistants and 6^ for susceptibles). In this alternative model the effects of germination on the seed bank are tallied. The two models agreed and yielded Eq. 1 when the factors / and a of Eq. 1 satisfied fa 1, i.e., when the herbicide selection pressure was sufficiently high. We now know that this will not be met in rotational years withfitnessnear 0.1 and a < 10. Our revised model starts with the model in the appendix of ref. (3), as it is somewhat more accurate and biologically realistic. Our previous discussion of the preferred model was terse, so we rederive the equations in Appendix 1 of this chapter. The somewhat more accurate counterpart to Eq. 1 is N

W

=

N

i*)

{(x _

(! _

+

/

a

[

l

_ (x _

(x _

(Eq. 2) (5)

This equation explicitly depends on the fraction of susceptible (7 ) and resistant (V*)) seeds that germinate. It also depends on all other aspects, including relative germination ability of resistants in competition with susceptibles. See Eq. A1.20 for the exact definition of thefitnessfactor / . For simplicity, and in the absence of the detailed information that would permit use of Eq. 2, we shall generally assume that the fraction germinating in a given season is small (y«) < i ^(s) ^ y j ^at resistant and susceptible seeds perish to an equal extent in the seed bank: ?

a n (

= R)

=6

(Eq. 3)

R)

With this, Eq. 2 becomes N< = N^ {1 + 6{fa - l)}

n

(Eq. 4a)

Another important special case of Eq. 2 corresponds to a "no till" situation. Here all seeds germinate. Thus ^W = "y = 1 so that Eq. 2 reduces to 5

R)

n

JVJ*> = N$ [fa]

(Eq.4b)

As shown in Ref. (3), the average lifetime of seeds in the seedbank, n, is related to the factor 6 by n = \ o Thus, Eq. 4a, reduces to Eq. 1, in cases when fa » 1.


(Eq. 5)



443

Perhaps more meaningful than the average lifetime, n, of seeds in the seedbank is their half-life, n i . If there are N seeds in the bank initially, after n years iV (l - S) seeds remain. The half-life, n = | , is thus calculated from N {1 — 6) = |iV"o> so that 6 and rti are related by 0

n

0

n


0

Table III. Relationship Between Seed Bank Survival Fraction (1-5), Seed Bank Half-Lives, m , and Average Seed Bank Residence Time, (n) 2 n

6

1-6

0.50 0.29 0.21 0.16 0.13 0.07

0.50 0.71 0.79 0.84 0.87 0.93

2.0 3.4 4.8 6.3 7.7 14.3

1 2 3 4 5 10

(From Eqs. 5, 6).

0.7 For 6 < 1 n± « —

(Eq. 7)

which should be compared to Eq. 5. Table III provides values of 6 for several different half-lives. We previously (3) made calculations with Eq. 1, the counterpart of Eq. 4, for a number of parameter values, showing the enrichment of resistance when various levels of herbicides with different selection pressures are applied annually. We also estimated the effects of periodically stopping herbicide use by setting a = 1 in calculating the resistance enrichment factor (Fig. IB). We now realize that thefitnesscoefficient may differ in "on" years from its value in "off" years. Thus we will be working with the following enrichment factor (IT), obtained from Eq. 4a, that gives the increase in resistance following a period of p "on" years of herbicide application and q ofT years without herbicide: tt

H , = [1 + 6{ f p q

a

m

- 1)]" [1 - 6(1 - /.„)]«


(Eq. 8)

MANAGING

444


Note that the p "on" years and the q "off* years can occur in any order during the p + q year period that is under study. Note also that the "off factor 1 —6(1 — f ff) = 1 — 6 + Sf f can be approximated by 1 — 6 if f f Q


c o

5 Q>

0) 1 on,3off

-

- 8 8 5:

Q)

83 c

Ion,2off

75 &

Ion,1 off

CO 1

0

I

1

1 on, 1 off i

1 1

0.2 0.4 10 5 3 2

0 1

Q> U

\ on, 2 off

i

i

1

-50

CL

l

0.2 0.4 I 10 5 3 2

8 n (yrs) l/2

Figure 4. Lack of enrichment for resistance under various selection pressures (a), under different rotation strategies, with different weed seed dynamics in the seed bank. The selection pressure is also shown as "effective kill" (the percent reduction in sensitive propagules over a whole season) with the assumption that the rare resistant individuals are totally unaffected by the herbicide. Here f = 1 while f takes a relatively large value (A) and a (B) value so small that f has a negligible effect. on

off

off

T

0.1

0.2 5

3

0.3 2

—I

0.4

T

0.5 S i

n (yrs) |/2

Figure 5. Selection pressures that cause a doubling of the proportion of resistant individuals in the population every 3 years. The data are given for a 1 on: 2 off rotational strategy, with different fitnesses in the off years (/on = 1), and different seed bank dynamics.



30, GRESSEL AND SEGEL


449

Figure 6. Amplification of resistance in a 3 year period as a function of fitness in the off years (f = 1) and seed bank dynamics. A 1 on: 2 off rotational scheme is used and the "on" herbicide has afixedselection pressure of a = 10 (effective kill = 90 %). on


MANAGING

450 The derivative of H

lt2


with respect to 6 vanishes at tt-l-2(l-/.„) „ 1 3(1 -/.//)(

Herbicide Rotations and Mixtures - American Chemical Society

Recommend Documents