Scholarly article on topic 'Herbicide impact on non-target plant reproduction: What are the toxicological and ecological implications?'

Herbicide impact on non-target plant reproduction: What are the toxicological and ecological implications? Academic research paper on "Biological sciences"

Share paper
Academic journal
Environmental Pollution
OECD Field of science
{"Non-target plant reproduction" / "Herbicide effect" / "Ecological risk assessment" / "Ecotoxicological implications" / "Non-crop habitats"}

Abstract of research paper on Biological sciences, author of scientific article — C. Boutin, B. Strandberg, D. Carpenter, S.K. Mathiassen, P.J. Thomas

Abstract Declining plant diversity and abundance have been widely reported in agro-ecosystems of North America and Europe. Intensive use of herbicides within cropfields and the associated drift in adjacent habitats are partly responsible for this change. The objectives of this work were to quantify the phenological stages of non-target plants in in-situ field situations during herbicide spray and to compare plant susceptibility at different phenological stages. Results demonstrated that a large number of non-target plants had reached reproductive stages during herbicide spray events in woodlots and hedgerows, both in Canada and Denmark where vegetation varies considerably. In addition, delays in flowering and reduced seed production occurred widely on plants sprayed at the seedling stage or at later reproductive periods, with plants sprayed at reproductive stages often exhibiting more sensitivity than those sprayed as seedlings. Ecological risk assessments need to include reproductive endpoints.

Academic research paper on topic "Herbicide impact on non-target plant reproduction: What are the toxicological and ecological implications?"

Contents lists available at ScienceDirect

Environmental Pollution

journal homepage:

Herbicide impact on non-target plant reproduction: What are the toxicological and ecological implications?^

C. Boutin3,*, B. Strandbergb, D. Carpentera, S.K. Mathiassenc, P.J. Thomas3

a Environment Canada, Science & Technology Branch, 1125 Colonel By Drive, Raven Rd., Carleton University, Ottawa, ON K1A 0H3, Canada b Aarhus Universitet, Department of Bioscience, Vejlsovej 25, 8600 Silkeborg, Denmark cAarhus Universitet, Department of Agroecology, Forsogsvej 1, 4200 Slagelse, Denmark



Article history: Received 31 May 2013 Received in revised form 23 August 2013 Accepted 4 October 2013


Non-target plant reproduction Herbicide effect Ecological risk assessment Ecotoxicological implications Non-crop habitats


Declining plant diversity and abundance have been widely reported in agro-ecosystems of North America and Europe. Intensive use of herbicides within cropfields and the associated drift in adjacent habitats are partly responsible for this change. The objectives of this work were to quantify the phenological stages of non-target plants in in-situ field situations during herbicide spray and to compare plant susceptibility at different phenological stages. Results demonstrated that a large number of non-target plants had reached reproductive stages during herbicide spray events in woodlots and hedgerows, both in Canada and Denmark where vegetation varies considerably. In addition, delays in flowering and reduced seed production occurred widely on plants sprayed at the seedling stage or at later reproductive periods, with plants sprayed at reproductive stages often exhibiting more sensitivity than those sprayed as seedlings. Ecological risk assessments need to include reproductive endpoints.

Crown Copyright © 2013 Published by Elsevier Ltd. All rights reserved.

1. Introduction

Fertilizers and herbicides are the most widely used chemicals in farmlands and have been instrumental in the tremendous increase in crop productivity since World War II (Boutin, 2013). However, there has also been growing concerns about declining plant species richness, abundance and diversity (Fried et al., 2009) both within cropfields and in adjacent habitats including field margins, hedgerows, ditches, as well as small woodlots and wetlands (Andreasen and Stryhn, 2008; Crone et al., 2009; Romero et al., 2008; Storkey et al., 2012; Sutcliffe and Kay, 2000). Many plant species associated with agroecosystems have become rare to the extent that they are registered in the Red Data Books (International Union for Conservation of Nature) of several countries, including several arable species considered agricultural weeds (Albrecht and Mattheis, 1998; Türe and Böcük, 2008; Wilson, 1994). Failure to adequately assess and properly regulate herbicide effects can have important ecological implications for plant survival, seed

q This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial-No Derivative Works License, which permits non-commercial use, distribution, and reproduction in any medium, provided the original author and source are credited.

* Corresponding author.

E-mail addresses: (C. Boutin), (D. Carpenter).

production, long-term seedbank replenishment and eventual species composition of not only primary producers, but also species at other trophic levels.

Although fertilizer use is of great concern (Kleijn and Verbeek, 2000), this paper will primarily address herbicide effects and assessment. Herbicides used in agriculture for weed control in major crops are primarily sprayed in May or June in Canada (as per pesticide labels In most European countries, herbicides are sprayed several times in any given year depending on the crops (Strandberg et al., 2012). In Denmark, spring sown crops are usually sprayed with herbicides in April and May while autumn sown crops are sprayed in September and October. In the Netherlands, an average of 5.7 herbicides are sprayed on food crops (between three and nine depending on the crops) and 10.3 (between six and 15) herbicides per year are applied on field cultivated flower crops (EFSA, 2012). Though it has not yet been quantified, it is likely that herbicides will reach weeds and non-target plants at all phenological stages depending on the application time.

When plants are sprayed in cropfields and sublethal doses of herbicides reach non-target plant species in adjacent habitats through drift, runoff and/or volatilisation, resultant effects on sensitive species can be observed in any of four ways: a) Plants at the seedling stage during spray will have their vegetative parts affected, b) the same plants could express the effect through negative impacts on seed production at later stages, c) plants at the

0269-7491/$ — see front matter Crown Copyright © 2013 Published by Elsevier Ltd. All rights reserved.

reproductive phase during spray have their seed production impacted or d) the vegetative parts of the F1 generation are affected (Fig. 1). Therefore, it appears that seedlings and plant species at late vegetative and reproductive stages may be affected differently, and this is most likely influenced in turn by the type of herbicide applied.

For regulatory purposes, greenhouse tests utilizing species growing singly in pots or in monoculture are required to assess the potential undesirable effects of herbicides on non-target, wild plants found within the vicinity of croplands. These tests are performed on emerging seedlings or on plants at the 2—6 leaf stage (usually using crops as surrogates for wild species) with effects recorded 14—28 days after the spray event (OECD, 2006; USEPA, 2012 — Fig. 1a). Several greenhouse studies have effectively shown that the seedling stage was more sensitive to herbicides than later growth stages, at least for some species (Boutin et al., 2000; Zwerger and Pestemer, 2000). However, other studies have shown that some species that have reached the reproductive stage following exposure exhibited negative herbicidal effects at doses below those observed for the seedling or vegetative stages (Boutin et al., 2000; Carpenter and Boutin, 2010; Carpenter et al., 2013; Riemens et al., 2008, 2009; Strandberg et al., 2012). In some cases, reproductive endpoints (seed production or measurable equivalent) may be more appropriate to assess than aboveground vegetative biomass, for instance when plants are exposed at later developmental stages when growth has ceased (Steadman et al., 2006; Strandberg et al., 2012; Walker and Oliver, 2008). The ISO protocol (2005; Fig. 1b) was developed to examine both the inhibition of growth and the reproductive capacity of plants following soil contamination (not specific to pesticides) under controlled conditions using two test species: a rapid-cycling variant of turnip rape (Brassica rapa CrGC syn. Rbr) and oat (Avena sativa L.). Though this test assesses both the vegetative and reproductive effects of contaminants on plants, it is not usually conducted for pesticide registration. There is no known protocol to determine effects when plants are sprayed at maturity (see Fig. 1c, d).

The objectives of this work were multi-faceted and aimed to address some of the above-mentioned issues. Our first objective was to quantify the numbers and types of species present at the vegetative and reproductive stage in non-crop habitats during herbicide spray events to indicate potential risks to non-crop plant

c (H ■C CL

Fig. 1. Representation of phenological stage at spray (or testing) time and stage of recorded herbicide effects. Veg = Vegetative, Rep = Reproductive period. Letters within quadrants are used for reference purposes in the text. OECD (Organisation for Economic Co-operation and Development) 2006, USEPA (United States Environmental Protection Agency) 2012 and ISO (International Organization for Standardization) 2005 refer to standard guidelines for plant toxicity testing.

reproduction. Results from two experiments with woodlots (Canada) and hedgerows (Denmark) are presented. Our second objective was to measure effects of herbicides on the initiation of flowering. This information was obtained from four field and greenhouse studies conducted in Canada and Denmark. Our third objective was to compare endpoints (vegetative and reproductive) and phenological stages at spray with the purpose of developing a more realistic estimation of effects which could be used in risk assessment. To meet this third objective, new and existing data in the published literature were compiled and analysed.

2. Material and methods

This article presents work from seven experimental studies, including unpublished experiments as well as unpublished data that were part of experiments already published.

2.1. Vegetation/Phenology surveys

Two experiments were conducted to determine plant phenology at the time of herbicide spraying. In EXP1, three woodlots (S1—S3) were selected in south-western Ontario, Canada (480971E—4757140N). The S1 woodlot was fairly open and dry and had been grazed a few years prior to the study (only surveyed in 1993). The S2 woodlot sustained some disturbance due to a cabin access path. It contained a wide mix of spring ephemeral vegetation typical of rich, well-drained soils. The S3 woodlot was characterised by a small stream and ponds. The main tree species found in the three woodlots included iron wood (Ostrya virginiana (Mill.) K. Koch), bitternut hickory (Carya cordiformis (Wangenh.) K. Koch), sugar maple (Acer sac-charum Marshall), blue beech (Carpinus caroliniana Walter), American beech (Fagus grandifolia Ehrh.) and silver maple (Acer saccharinum L.).

The woodlots were adjacent to three different fields, all planted with soybean (Glycine max (L.) Merr.) in 1993, corn (Zea mays L.) in 1994 and wheat (Triticum aestivum L.) and corn in 1996. Herbicides (imazethapyr in 1993, dicamba in 1994 and MCPA in 1996) were sprayed under normal operational conditions by the farmer in May of each year (no trial was conducted in 1995). Herbicide application occurred in the early morning or evening, when no precipitation was forecasted; wind speed was at 8 km/h or less in the direction of the woodlot from across the planted field.

Quadrats were established along ten transects (at 10 m distances) per woodlot positioned perpendicular to the field. Five transects were abutted to an in-field 15 m buffer zone where no spray occurred and five transects had no buffer zone. Each transect consisted of 1 m2 quadrats placed at 1,2,4,8,16 and 32 m distances into the woodlots. All vegetation below 2 m in height was surveyed for species composition. The phenological stage (vegetative or flowering) was recorded prior to the spray operation while symptoms of herbicidal impact (comparing qualitative visual assessment prior to and after the spray) were recorded four times between May and July. As long as one plant of a given species in a quadrat was flowering the species was considered at the reproductive stage for that quadrat.

In EXP2, 40 hedgerows were surveyed in Denmark in organic and conventional farming systems. In the first trial, starting in 2007, ten hedgerows on conventional farms were surveyed for three years. In the second trial in 2008, a further ten hedgerows were selected in conventional farms along with 20 hedgerows in organic sites, and were surveyed for four years (UTM coordinates: 517126E to 594023E— 6168669N to 6259952N). Hedgerows were selected as pairs of organic and conventional to eliminate landscape effects and were similar in terms of woody species composition, management, orientation, age (80—150 years old) and crops on the neighbouring fields (cereals). All hedgerows (at least 400 m in length) had one to three rows of deciduous trees and shrubs along the entire length and a 0.5—1 m wide zone covered with herbaceous species at the field's edge. The main hedgerow trees and shrubs were oneseed hawthorn (Crataegus monogyna Jacq.), sweet cherry (Prunus avium (L.) L.), European mountain-ash (Sorbus aucuparia L.), Swedish whitebean (S. intermedia (Ehrh.) Pers.), European alder (Alnus glutinosa (L.) Gaertn.) and dwarf honeysuckle (Lonicera xylosteum L.).

Sampling of the flowering ground flora was performed in the herbaceous zone on the west-facing side. In each hedgerow sampling was conducted within fifteen 0.5 by 0.5 m permanent quadrats placed on a 100 m transect with a distance of 6.5 m between successive quadrats. Sampling was carried out monthly from May to mid-September. A survey of herbicide usage near ten hedgerows examined from 2007 to 2009 confirmed that applications occurred from the end of March until late October, with few treatments happening in June and July. At each sampling day the pheno-logical stage of all vascular plants found within each quadrat was recorded. Any given species was recorded as flowering in a given month (May—September) if it was in flower in a specific hedgerow in any year, meaning that the maximum count for each species for a given month is twenty for organic and conventional farming types.

EXP2 data were also used to assess delays in flowering in relation to herbicide usage. The total number of species was tabulated for overall phenological assessment, and a list of 57 species preferably used by pollinators in Denmark was built based on expert knowledge and a thorough literature review (Benton, 2006;

Period of recorded effects

Veg Rep

aJ Routine regulatory testing, OECD 2006, USEPA 2012 b Effects on seed production, ISO 2005

d Effects on F1 generation, No known protocol c Effects on seed production, No known protocol

Calabuig, 2000; Calabuig and Madsen, 2009; Goulson, 2010; Madsen and Calabuig, 2008, 2010; Strandberg et al., 2011; Westrich, 1990).

2.2. Greenhouse experiment for effects on reproduction

Three experimental greenhouse studies examined the effects of herbicide exposure on flowering times. In Canada, time to first flowering was recorded for nine species in a greenhouse dose—response study (see Carpenter and Boutin (2010) for experimental details) (EXP3). Plants (six replicates) were exposed to nine doses of glufosinate ammonium ranging from 0 to 667.5 g-ai ha 1 (0—89% of label rate) at the 3—6leaf-stage. A second experiment was performed using upland and wetland species (see Carpenter et al. (2013) for experimental details) (EXP4). Six plant replicates were exposed to eight doses of chlorimuron-ethyl ranging from 0 to 9.63 g-ai ha 1 (0—107% of label rate) at the 4—6 leaf-stage and time to first flowering was recorded for 11 species. In Denmark, red clover (Trifolium pratense L variety 'Merula') and dandelion (Taraxacum vulgare Web.) were exposed at the bud stage to four doses (0, 5, 25,100% of recommended 144 gai ha 1 label rate) of the herbicide fluroxypyr (EXP5). Onset of flowering and number of flowers were recorded for 44 days after exposure.

2.3. Comparing phenological stage and endpoints at time of spray

Several published studies were used in an attempt to assess the most sensitive growth stage at spray and the most sensitive endpoints of plants when sprayed with herbicides. Furthermore, two additional studies were available to us (Holst et al., 2008; Strandberg et al., 2012). The first study (EXP6) focused on the susceptibility of six test species at various phenological stages: Geranium molle L., Geranium rob-ertianum L., Silene noctiflora L., Silene vulgaris (Moench) Garcke, Tripleurospermum inodorum (Merat) M. Lainz and Achillea millefolium L to the herbicides mecoprop-P (MCPP), metsulfuron methyl and glyphosate. In the second study (EXP7), an additional four species, Festuca ovina L., Agrostis tenuis Sibth., Solanum nigrum L and Echinochloa crus-galli (L.) P. Beauv. were tested with glyphosate and foramsulfuron + iodosulfuron. In the two experiments, seeds were directly sown in 2 L pots or were sown in trays and transplanted into pots as small seedlings. Pots were placed outdoors and were watered several times daily. Herbicides were applied at two growth stages (different plants for different growth stages). The early growth stage was characterized by plants with three to ten leaves and the late growth stage was classified as either at elongation (grasses) or at the flower bud

Table 1

List of species in flower during spray time each year (three woodlots were surveyed in 1993, two woodlots in 1994 and 1996) (EXP1). Family, growth habit and lifespan are presented for each species.

stage (broadleaved species). Five to seven doses of each herbicide were applied separately to six replicate pots. Plants from three replicates of each treatment were harvested three to four weeks after spray and biomass fresh weights were recorded. The remaining three replicates were used for assessing seed production.

2.4. Statistical analysis

In EXP1 and EXP2, two-way ANOVAs were used to assess the number and proportion of species at flowering stages during spray events. In EXP3, EXP4 and EXP5, Kruskal—Wallis and Conovan-Inman post-hoc tests were used for significance between doses. Chi-square tests, Student t-tests and Mann—Whitney U tests were conducted to assess differences in the number of species affected by herbicide spray (EXP1) and in flowering between organic and conventional systems (EXP2). In all cases, the assumptions of normality and homogeneity of variance were tested.

The EC50 (effective concentration resulting in a 50% decrease as compared to the controls) was used to examine herbicide effects. In EXP3 and EXP4, EC50s were calculated using non-linear regressions when the data met the assumptions of normality and homogeneity of variance, else the nonparametric ICPIN method was used (Norberg-King, 1993). Dry vegetative biomass and reproductive parameters (seed production or measurable equivalent) were used separately in each calculation. Similarly in EXP6 and EXP7, the EC50 using fresh weight and seed production (number of seeds) were analysed with non-linear regressions using log-logistic dose response models (Seefeldt et al., 1995). For each herbicide, dose—response curves were estimated for each plant species and growth stage using two endpoints (biomass and seed production). Fitness of model was verified using an F-test for lack of fit, comparing the residual sum of squares.

3. Results

3.1. Vegetation/phenology surveys

A total of 104 species were identified across the three woodlots during the three years of the study (EXP1), of which 87% were perennial species. Of these, 34 species from 21 families were found flowering prior to spray operations (Table 1). Between 13.1% and

Scientific name


Growth habit


Year present

Alliaria petiolata (M. Biev.) Cavara & Grande Brassicaceae Forb A/B U U U

Arisaema atrorubens (L.) Schott Araceae Forb P U U U

Barbarea vulgaris W.T. Aiton Brassicaceae Forb B U

Bromus ciliatus L. Poaceae Graminoid P U

Caltha palustris L. Ranunculaceae Forb P U

Carex nigra (L.) Reichard Cyperaceae Graminoid P U

Carex laxiflora Lam. Cyperaceae Graminoid B U

Carex rosea Schkuhr ex Willd. Cyperaceae Graminoid B/P U

Cerastium fontanum Baumg. Caryophyllaceae Forb P U

Claytonia virginica L Portulacaceae Forb P U U U

Cornus stolonifera Michx. Cornaceae Subshrub P U U U

Dactylis glomerata L Poaceae Graminoid P U

Equisetum arvense L. Equisetaceae Forb P U

Erigeron philadelphicus L. Asteraceae Forb B/P U

Erythronium americanum Ker Gawl. Liliaceae Forb P U U U

Euonymus obovatus Nutt. Celastraceae Subshrub P U U U

Fragaria virginiana Duchesne Rosaceae Forb P U U U

Geranium maculatum L Geraniaceae Forb P U U

Hydrophyllum virginianum L. Hydrophyllaceae Forb P U U U

Medicago lupulina L. Fabaceae Forb P U

Poa pratensis L. Poaceae Graminoid P U U U

Podophyllum peltatum L. Berberidaceae Forb P U

Prunus virginiana L. Rosaceae Shrub P U

Ranunculus abortivus L. Ranunculaceae Forb P U

Ribes americanum Mill. Grossulariaceae Subshrub P U U

Rubus idaeus L. Rosaceae Subshrub P U U U

Taraxacum officinale F.H. Wigg. Asteraceae Forb P U U U

Trillium erectum L. Liliaceae Forb P U U

Veronica serpyllifolia L. Scrophulariaceae Forb P U U U

Viola labradorica Schrank Violaceae Forb P U U U

Viola cucullata Aiton Violaceae Forb P U U U

Viola sororia Willd. Violaceae Forb P U U U

Viola pubescens Aiton Violaceae Forb P U U U

Vitis riparia Michx. Vitaceae Vine P U U U

Percent of flowering species in three Canadian woodlots surveyed in May during three years prior to herbicide spray at six distances (pooled) from cropfields (EXP1). Analyses of variance for differences between woodlots and surveyed distances are presented.

Canadian woodlots Analysis of variance (p values)

S1 S2 S3 Site Distance Site*distance

1993 1994 1996 22.7 ± 1.1 18.2 ± 3.9 Not surveyed 20.6 ± 2.8 Not surveyed 21.6 ± 2.8 17.3 ± 2.4 13.8 ± 2.3 13.1 ± 2.1 0.114 0.250 0.027 0.036 0.621 0.108 0.008 0.617 0.116

22.7% of the species were in flower in the woodlots at different years, with no consistent trends between distances, except for a significant interaction between sites and distances in 1993 (Table 2) due to variability in flowering of S2 woodlots across distances. Significantly more species were in flower in S2 than in S3 in 1994 and 1996 (Table 2).

A total of 35 species (34% of total species, Appendix A) were found with marked herbicide effects including epinasty, leaf mottling, withering, yellowing, leaf and stem twisting, necrosis and bud malformations. Of these, 13 species in flower during herbicide application exhibited symptoms consistent with pesticide poisoning (Appendix A). The buffer zone was reasonably protective of the off-field plants: between 4 and 11 species each year exhibited symptoms of herbicide damage in the buffer zone transects compared to 10—23 species in the no buffer zone transects (p < 0.05, Table 3). Most affected plants were located in quadrats within the 1 to 4 m range, though some effects were noticeable up to 32 m (not shown).

Hedgerows surveyed in Denmark comprised 192 species from 34 families of which 56% were perennial species (EXP2). The number of species in flower each month during the growing season was always significantly higher in hedgerows adjacent to organic fields than near conventional fields (Fig. 2). Many species were in flower for several months. The peak of flowering occurred in July with 64.0 and 57.3% of species in flower in organic and conventional hedgerows respectively, while 10.9—44.3% of species were in flower during normal spray applications in May, June and September.

3.2. Delay in flowering

The total number of species and, to a lesser extent, families were greatly reduced in hedgerows abutted to conventional fields (EXP2), with 187 and 113 species in organic and conventional hedgerows, respectively (Table 4a). Of all the species in common in the hedgerows of both types of farming systems (n = 108), onset of flowering was more rapid for 37 species in organic hedgerows, while a mere 12 species flowered earlier in conventional hedgerows (Table 4a). The average number of hedgerows in which species

o 20 u

£ 15 o

June July August September

Fig. 2. Number of species in flower in hedgerows adjacent to organic (n = 20) or conventional (n = 20) fields surveyed in Denmark during the growing season from May to September (EXP2). Error bars represent standard error. A two-way ANOVA (square root transformation) found significant differences between organic and conventional sites (p < 0.000), months (p < 0.000) and the interaction (p < 0.000). Mann— Whitney U tests indicate significant differences between organic and conventional sites for each month (p < 0.05 all cases).

flowered (only using common species to both farming types) was significantly higher in organic (6.45) than in conventional (3.88) systems (t-test, df = 108, p < 0.000). The average length of flowering (number of months) for each species was also significantly higher in organic (2.89) than in conventional (2.24) farming (t-test, df = 108, p < 0.000) (Table 4a). The same trend was observed for species that are used by pollinators (n = 57). These trends continued to be apparent when both total and pollen species were considered on a monthly basis (Table 4b).

In controlled experiments, flowering time was delayed for most species with increasing doses of both glufosinate ammonium (EXP3; Fig. 3a, Table 5a) and chlorimuron-ethyl (EXP4; Fig. 3b, Table 5b). Significant delays in flowering were observed at doses below 5% (three species) and 25% (most species) of label rate (Table 5a, b). Likewise, in EXP5 data showed that the average cumulative number of flowers produced by T. pratense was severely impaired at all doses of fluroxypyr (5—100%, 7.2—144.0 g-ai ha-1) while T. vulgare experienced effects at higher label rates (25—100%, 36.0—144 g-ai ha-1; Fig. 4). Similarly, the average number of days for the onset of flowering was also significantly delayed in both plants except at the 5% dose (Fig. 4).

3.3. Effects of time of spray and period of recorded effects

The reproductive endpoint of plants sprayed with herbicides at early vegetative stages, in many cases, showed more

Table 3

Total number of species present per site and number of species affected (total and in flower) by herbicides in 1993 (n = 3 woodlots pooled), 1994 (n = 2) and 1996 (n = 2) (EXP1). Results of Mann—Whitney U tests (p values) are presented for differences between buffer zone (bz) and no buffer zone (no bz) transects for each year separately. Tests were conducted for species presence/absence and species occurrence or number of quadrats in which species were affected [within square brackets]. Separate analyses were conducted for species affected while in flower._

Year of study Herbicide used Number of species [occurrence] Mann-Whitney U tests p values

Total species affected Affected in flower

Total in sites Total affected Affected in flower Presence Occurrence Presence Occurrence

With bz No bz With bz No bz

1993 Imazethapyr 90 10 [25] 19 [45] 3 [9] 7 [17] 0.001 0.056 0.023 0.115

1994 Dicamba 70 11[17] 23 [58] 2 [4] 8 [18] 0.001 0.001 0.003 0.014

1996 MCPA 80 4 [12] 10 [15] 1 [1] 7 [10] 0.015 0.018 0.002 0.002

Plant species characteristics of 20 organic (Org) and 20 conventional (Conv) hedgerows surveyed in Denmark (EXP2). Pollen species are plants preferably used by pollinators (see Material and Methods). T-test and Chi-square test were used for differences between organic and conventional sites. a) Overall comparisons = pooled across months, b) Comparisons for each month separately._

a) Overall comparisons All species (n = 192) Pollen species (n = 57)

Org Conv Org Conv

Total number of hedgerows recorded 20 20 20 20

from May to Sept

Total number of species 187 113 55 29

Number of annuals 65 36 13 6

Number of biennials 16 10 6 4

Number of perennials 106 67 37 25

Number of unique species 78 5 27 1

Number of common species 108 28

Total number of families 35 11

Number of families 33 31 11 9

Number of unique families 4 2 2 0

Using species common to both types of hedgerows All species (n = 108) Pollen species (n = 28)

No. species with faster (early) onset of flowering 37 12 11 4

Average maximum number of hedgerows 6.45 3.88 4.93 2.46

in which species flowered

Standard error 0.52 0.39 0.76 0.46

T-test p value 0.000 0.008

Average length of flowering (# of months 2.89 2.24 2.71 1.93

flowered) in each species

Standard error 0.13 0.11 0.23 0.18

T-test p value 0.000 0.009

b) Monthly comparisons All species (n = 192)

May June July Aug Sept All months

Org Conv Org Conv Org Conv Org Conv Org Conv Org Conv

Initiation of flowering — new 25 20 89 40 61 38 8 12 4 3 187 113

species per month

Chi-square p value 0.456 0.000 0.021 0.371 0.705 0.000

Total no. of species flowering 25 20 104 50 146 70 110 64 87 43

in each month

Chi-square p value 0.456 0.000 0.000 0.000 0.000

% Species flowering (out ot total 13.4 17.7 55.6 44.2 78.1 61.9 58.8 56.6 47.1 38.0

number of species)a

Pollen species (n = 56)

Initiation of flowering — new 7 6 24 7 24 12 0 4 0 0 55 29

species per month

Chi-square p value 0.781 0.002 0.045 0.079 0.005

Total no. of species flowering 7 6 31 8 46 17 32 17 22 8

in each month

Chi-square p value 0.781 0.000 0.000 0.032 0.011

% Species flowering (out ot total 12.7 20.7 56.4 27.6 83.6 58.6 58.2 58.6 40.0 27.6

number of pollen species)b

a n = 187 and 113 for organic and conventional hedgerows, respectively. b n = 55 and 29 for organic and conventional hedgerows, respectively.

sensitivity (lower EC50) by an average factor of 3.1 to the various herbicides than the vegetative endpoint (aboveground biomass) measured three to four weeks after spray (Table 6a). These patterns were found to be herbicide dependent. For chlorimuron-ethyl, only three species showed more sensitivity in their reproductive measure (out of 11 species; endpoints were equally sensitive for two species) while for glufosinate ammonium, it was seven out of 12 species and for 2,4-D, all species displayed more or equal sensitivity in reproduction. When plants were sprayed during their reproductive period (flower bud stage), all nine species were found to be more susceptible to herbicides in their seed production (or equivalent) than in their vegetative parts by a factor of 7.2 (Table 6b). Overall, reproductive endpoints were more sensitive in 58% of all cases (34 out of 59 species) whereas vegetative measures were more sensitive in 32% (19 out of 59 species). There was equal sensitivity in the remaining six cases (Table 6a and b).

It was also found that the same plant species sprayed at the young vegetative stages exhibited more sensitivity than when sprayed at a later phenological stage (S. noctiflora, S. vulgaris and Geranium molle in Table 6a and b). Herbicide sensitivity varied with the species tested. For example, S. noctiflora and G. robertianum did not respond in a similar manner with the different herbicides used. Not surprisingly, exposed young plants were also usually more sensitive than exposed older plants in terms of vegetative endpoints (Table 6c).

4. Discussion

Herbicide application frequencies vary from single spray events in Canada to repeated applications throughout the growing season in Denmark and a majority of Europe (EFSA, 2012; Strandberg et al., 2012). Drift from cropfields into adjacent non-target habitats is the most likely scenario for exposing non-target plants to herbicides.

1 1.9 3.4 6.9 13 24.8 47

Doses of glufosinate ammonium (% of label rate)

2 0.4 -

£ 0.2 H

£2 u=

.2 0.1 -a>

1 1.95 3.8 7.4 14.5 28.2

Doses of chlorimuron ethyl (% of label rate)

Fig. 3. Time to first flower (inverse transformation +1) produced by different plant species exposed to doses of the herbicides a) glufosinate ammonium (EXP3) and b) chlorimuron ethyl (EXP4). The inverse transformation was used as it allows the incorporation of zeros — higher numbers thus represent shorter times to first flower. See Table 5 for species codes.

Between 5% (commonly) and 25% (occasionally) of the applied herbicide dose is expected to reach the vegetation in field margins and boundaries (e.g. hedgerows, woodlots, etc.) (Holterman et al., 1997; Weisser et al., 2002). Data presented in this study revealed that species would be at various phenological stages, including reproductive stages, at spray times, in both Canada and Denmark. It is therefore essential to assess the sensitivity to herbicides not only when plants are sprayed at young vegetative stages as per current guidelines (OECD, 2006; USEPA, 2012) but also at later stages when reproduction is occurring in order to assess the impacts on plant reproduction.

4.1. Delays in flowering

Our work showed that herbicides can cause marked delays in flowering times and reductions in flower production in many species, both under greenhouse and field conditions. Flowering was clearly impeded in Capsella bursa-pastoris, Anagallis arvensis, Heli-anthus strumosus, Lobelia inflata, Trifolium pratense and Taraxacum

vulgare under greenhouse conditions (Table 5, Figs. 3, 4). In contrast, the annual species Chenopodium album appears to have responded to herbicide injury by lessening the time period for first flowering at higher doses (Fig. 3b), although the number of seeds produced at higher doses was reduced (Carpenter et al., 2013). This is a strategy adopted by some short-lived ruderal species in the presence of stressors to insure that at least some progeny are produced (Harper, 1977).

Most importantly, effects of herbicides on timing of flowering and seed production have seldom been measured under field conditions. In comparing the phenology of species present in organic and conventional hedgerows, remarkable differences were noticed in the current study (EXP2). Herbicide use was one of the most important differences between the studied organic and conventional hedgerows in Denmark, although other agricultural practices differed. Nevertheless, organic farming promoted not only plant diversity but also plant flowering capacity whereas conventional farming inhibited flower production of the fewer plants found in adjacent hedgerows and resulted in a shift in flowering. This in turn

Results of Kruskal—Wallis non-parametric tests and Conover—Inman post hoc tests conducted on flowering time (measured as days after exposure) for each species and the herbicides a) glufosinate ammonium (EXP3) and b) chlorimuron ethyl (EXP4) separately. Day 1 for flowering was the first day that a flower was observed on any given plant of that species. Different letters for each species indicate where significant differences occurred when compared to control. Doses in grams of active ingredient per hectare (gai ha-1) are shown with % of label rate within square brackets. Shadows indicate where significant differences begin. Species codes are used in Fig. 3._

a) Species Glufosinate doses (g-ai ha 1) [% label rate]

Species 0 7.5 14.25 25.5 51.75 97.5 186 352.5 667.5

Code [0%] [1%] [1.9%] [3.4%] [6.9%] [13.0%] [24.8%] [47%] [89%]

Avena sativa L. AS a ab a a b b c c c

Capsella bursa-pastoris (L.) Medick. CBP a a b ab b c c c c

Cucumis sativus var. "Pot Luck Hybrid" L. CU a a a a b bc cd d d

Elymus canadensis L. EC ab ab ab a bcd abc cd cd d

Fagopyrum esculentum Moench FE a a a a a a b b b

Helianthus annuus var. "Teddybear" L. HA ab a ab ab ab bc cd d d

Juncus dudleyi Wiegand JD abc abc ab a cd bcd d cd d

Phytolacca americana L. PA ab a ab b ab c d d d

Solanum lycopersicum var. "Tiny Tim" L. SL a ab ab b cd d e ef f

b) Chlorimuron doses (g-ai ha ) [% label rate]

0 0.09 0.18 0.34 0.67 1.31 2.54 4.95 9.63

[0%] [1%] [1.95%] [3.8%] [7.4%] [14.5%] [28.2%] [55%] [107%]

Anagallis arvensis L. AA a ab ab ab b b c c c

Capsella bursa-pastoris (L.) Medick. CBP a a ab bc bc cd de e e

Centaurea cyanus L. CC a a a ab a ab bc bc c

Chenopodium album L. CA a a a a a a a a a

Cleome serrulata Jacq. CS a ab a ab ab bc b bc c

Elymus canadensis L. EC a a a a a a a a a

Helianthus strumosus L. HS a ab a ab bc ab cd d d

LI a a ab ab bc cd de ef f

Elymus virginicus L. EV a a a a a a a a a

Lycopus americanus Muhl. Ex W. Bartram LA ab a ab ab abc bc cd d d

Polygonum pensylvanicum L. PP a a a a a a a a a

may cause disharmony with pollinator activities as pollinators can be very sensitive to flowering events (Santandreu and Lloret, 1999). Effects on timing of flowering can have consequences on pollinating insects as they may be less able to survive in non-crop habitats during periods when crop plants are unavailable for pollination (Carvalheiro et al., 2010). Alternatively, delays in flowering time may expose flowers to unfavourable weather conditions (e.g. frost or drought). Herbicide effects appear to constitute yet another stressor affecting plant—insect interactions, adding to other stressors including land-use modifications at the landscape scale (Kremmen et al., 2007) that are increasingly impacting agro-ecosystems.

The overall plant composition in a community can be modified with repeated use of sublethal doses of herbicides reaching semi-natural habitats at the margins of cropfields. Boutin and Jobin (1998) found that hedgerows and woodlots abutted to intensively managed cropfields contained more annual species and grasses than those next to less intensively managed cropfields that contained primarily perennial forbs. In their study however, effects of herbicides could not be dissociated from the effects of increased fertilizer drift into marginal habitats near intensively managed farms. Gove et al. (2007) demonstrated the long-term implications of herbicide use on six woodland species. The experimental component (greenhouse and outdoor experiments) revealed that several species were affected by glyphosate treatment and that sensitivity differed between species. A complementary field survey found that the most sensitive species documented in the greenhouse experiments were also the least abundant in fields with high agricultural herbicide inputs. Although effects on reproductive output were not measured, the observed modifications in plant composition may have been largely due to a decrease or failure to reproduce in species most affected by herbicides.

Effects of herbicides on reproduction can be observed both soon after spray and at later dates. In a field study conducted in Denmark,

it was found that berry production in hawthorn (Crataegus monogyna Jacq.) was severely impaired by average spray drift concentrations higher than 2.5% of the label rate of metsulfuron methyl (0.1 g-ai ha-1), a sulfonylurea herbicide, and that the effect was still observed one year later (Kjar et al., 2006a, 2006b. Therefore, risk assessment may be underestimated when only short term acute exposures (<28 days) are considered, neglecting chronic sublethal impacts of herbicides.

4.2. Importance of measuring reproduction

Plant response to herbicides can be observed during the vegetative and/or reproductive period and is dependent on the life-stages at which herbicides reach plants. It was shown in the studies presented that plants sprayed during early vegetative stages were affected in their vegetative parts (as in Fig. 1a), and this is a period where plants appear to be very sensitive. However, plant responses may be delayed and subsequently very pronounced (ex. by affecting seed output — see Fig. 1b). We demonstrated that assessing the effects of herbicides on reproductive outputs (regardless of timing of spray) was of prime importance since reproduction was frequently more sensitive than the corresponding vegetative endpoint that would normally be evaluated in routine regulatory testing (i.e. aboveground biomass at 14—28 days post-spray; OECD, 2006; USEPA, 2012). Plants at reproductive stages invest resources towards seed production and thus largely decrease their vegetative growth. It is therefore not surprising that herbicides applied during the reproductive phase will greatly impact seed output as compared to vegetative growth (Table 6b).

Apart from the work presented with measurable EC50s (Table 6), a suite of additional studies also found that plants sprayed at the vegetative leaf-stage suffered negative impacts on reproduction. Riemens et al. (2008) found that the seed production of

control 7.2 36.0 144.0

Doses (g active ingredient per hectare)

20 ° C>

nj 10 ü o

control 7.2 36.0 144.0

Doses (g active ingredient per hectare)

Fig. 4. Average number of days following herbicide application for flowering to occur (histograms) and average cumulative number of flowers (lines) in a) Trifolium pratense Web. (red clover) and b) Taraxacum vulgare L. (dandelion) exposed to increasing doses of fluroxypyr (Starane 180S) during the bud stage formation (EXP5). Doses correspond to 0, 5, 25 and 100% of the recommended 144 g-ai ha-1 label rate. Significant differences between doses were detected using Kruskal—Wallis non-parametric tests applying the Conover—Inman test for post hoc comparisons. Differences between doses are presented with capital letters above histograms for number of days after exposure for flowering to occur, and with small letters above lines for cumulative number of flowers. All error bars represent standard error.

Stellaria media (L.) Vill. was a more sensitive endpoint than the vegetative endpoint measured on plants sprayed with glufosinate ammonium as seedlings. Likewise, effects of the herbicide tepra-loxydim on seed production were greater than those observed for fresh vegetative weight in three grasses (Riemens et al., 2009). Rinella et al. (2010) tested three auxinic herbicides (2,4-D, dicamba and picloram) applied at field rates to Bromus japonicus Thunb. to examine questions related to the control of this invasive species. They observed a dramatic effect on the germinable seeds produced when the herbicides were sprayed at all life-stages. This further reinforced the notion that reproductive endpoints can often be more sensitive to herbicide applications than vegetative endpoints (biomass). In an experiment with a mixture of two sulfonylurea herbicides used on two crop species, Gealy et al. (1995) found that at doses below 10% label rate, vegetative symptoms were very pronounced, flowering was delayed and final seed yield was significantly reduced.

In a few studies, plants were only sprayed at the reproductive phase (i.e. at onset of flowering or during seed formation) with considerable effects on the reproductive measures (as in Fig. 1c and d). Guo et al. (2009) tested the effects of three herbicides (paraquat, 2,4-D and glyphosate) on Solidago canadensis L. (an introduced, invasive species in China's natural and cropped areas). All herbicides significantly decreased pollen germination and pollen tube growth in S. canadensis at relatively low doses (300 g-ai ha-1 or 15— 25% of label rate depending on usage and country). In another study, the effects of glyphosate were tested on plants treated during early and late reproduction (Clay and Griffin, 2000). Effects on seed production and seedling emergence were very pronounced on the three species tested. Blackburn and Boutin (2003) found effects on seed germination in plants from several families sprayed with glyphosate at the reproductive stage. Additionally, a series of experiments conducted with low doses of sulfonylurea herbicides on plants at the onset of reproduction revealed a considerable effect on reproduction in several species (Al-Khatib and Tamhane, 1999; Fletcher et al., 1993,1996; Gealy et al., 1995). Overall, these studies demonstrate the importance of measuring herbicide impact at various phenological stages.

Several studies have sought to demonstrate mechanisms by which herbicides may affect reproductive structures. Ratsch et al. (1986) found that for the herbicides DGME, dalapon and TCA, flower deformations at even the lowest doses tested precluded pollination and seed formation. Other studies have uncovered herbicide effects on male structures, especially with glyphosate

Table 6

Data showing the most sensitive endpoint measured for herbicide dose—response studies. Plant species were sprayed either at a young or late vegetative stage or during reproduction. EC50s correspond to the dose causing a 50% reduction in biomass of vegetative parts or reproduction (seed production or measurable equivalent). Factors are ratios calculated as EC50 of vegetative parts/EC50 of reproductive parts or late vegetative stage. Refer to published articles for details on the methodology and reproductive endpoints measured for each species. a) Plants sprayed during young vegetative period EC50 (g-ai ha1)

Herbicides and species Lifespan Growth stage Vegetative parts Reproduction Factor Source reference

Glufosinate ammonium (g-ai ha 1) (EXP3)

Avena sativa L A 3 -6 leaf stage

Fagopyrum esculentum Moench A 3 -6 leaf stage

Helianthus annuus var. "Teddybear" L A 3 6 leaf stage

Solanum lycopersicum var. "Tiny Tim" L. A/P 3 6 leaf stage

Bouteloua gracilis (Willd. Ex Kunth) Lag. Ex Griffiths P 3 6 leaf stage

Elymus canadensis L. P 3 6 leaf stage

Juncus dudleyi Wiegand P 3 6 leaf stage

Capsella bursa-pastoris (L.) Medick. A 3 6 leaf stage

Hypericum perforatum L. P 3 6 leaf stage

Melilotus officinalis (L.) Lam. A/B/P 3 6 leaf stage

Phytolacca americana L. P 3 6 leaf stage

Solanum dulcamara L. P 3 6 leaf stage

216.77 149.31 1.45 Carpenter and Boutin (2010)

56.02 113.6 0.49

117.3 145.25 0.81

65.37 145.89 0.45

115.95 101.09 1.15

165.04 43.08 3.83

154.31 49.11 3.14

33.37 41.49 0.80

81.68 40.99 1.99

36.08 31.49 1.15

97.17 62.74 1.55

40.68 94.28 0.43

Table 6 (continued )

a) Plants sprayed during young vegetative period EC50 (g-ai ha1)

Herbicides and species Lifespan Growth stage Vegetative parts Reproduction Factor Source reference

Chlorimuron ethyl (g-ai ha1) (EXP4)

Capsella bursa-pastoris (L.) Medick. A 4—6 leaf stage 1.53 0.66 2.32 Carpenter et al. (2013)

Centaurea cyanus L. A 4—6 leaf stage 6.95 >9.6 0.72

Elymus canadensis L. P 4-6 leaf stage >9.6 >9.6 1.00

Chenopodium album L. A 4—6 leaf stage >9.6 6.65 1.44

Helianthus strumosus L. P 4—6 leaf stage 1.85 2.49 0.74

Lobelia inflata L. A 4—6 leaf stage 0.66 3.74 0.18

Anagallis arvensis L. A/B 4—6 leaf stage >9.6 1.92 5.00

Glyceria striata (Lam.) Hitchc. P 4—6 leaf stage 0.63 1.54 0.41

Lycopus americanus Muhl. Ex W. Bartram P 4—6 leaf stage 2.61 3.59 0.73

Polygonum pensylvanicum L. A 4—6 leaf stage 1.67 3.36 0.50

Elymus virginicus L. P 4—6 leaf stage >9.6 >9.6 1.00

Mecoprop (g-ai ha-1) (EXP6)

Silene noctiflora L A 6—8 leaves 69 38.1 1.81 Strandberg et al. (2012)

Silene vulgaris (Moench) Garcke P 6—8 leaves 154.1 <20 7.71

Geranium molle L. A/B/P 6 leaves 137.1 <75 1.83

Geranium robertianum L. A/B 6 leaves 54.6 <150 0.36

Glyphosate (g-ai ha-1) (EXP6 and EXP7)

Silene noctiflora L. A 6—8 leaves 74.4 87.2 0.85

Silene vulgaris (Moench) Garcke P 6—8 leaves 70.8 37.6 1.88

Geranium molle L. A/B/P 6 leaves 33.9 <22 1.54

Geranium robertianum L. A/B 6 leaves 108.2 <180 0.60

Echinochloa crus-galli (L.) P. Beauv. A 6—8 leaves 44.4 46.6 0.95

Metsulfuron methyl (g-ai ha-1) (EXP6)

Silene noctiflora L. A 6—8 leaves 0.6 0.34 1.76

Silene vulgaris (Moench) Garcke P 6—8 leaves >2 1 2.00

Geranium molle L. A/B/P 6 leaves 0.07 <0.03 2.33

Geranium robertianum L. A/B 6 leaves 0.33 0.25 1.32

Foramsulfuron + iodosulfuron (g-ai ha-1) (EXP7)

Echinochloa crus-galli (L.) P. Beauv. A 6—8 leaves 0.48 2 0.24

Taraxacum officinale F.H. Wigg. ssp. officinale P 6 leaves 7.7 0.1 77.00

Tribenuron (g-ai ha-1)

Rapistrum rugosum (L.) All. A 4—6 leaf stage 0.51 0.84 0.60 Rotchés-Ribalta et al. (2012)

Neslia paniculata (L.) Desv. A 4—6 leaf stage >7.5 >7.5 1.00

Galium aparine (spurium) L. A 4—6 leaf stage 5.89 >7.5 0.79

Asperula arvensis L. A 4—6 leaf stage >7.5 >7.5 1.00

Papaver rhoeas L. A 4—6 leaf stage 0.93 0.17 5.46

Papaper argemone L. A 4—6 leaf stage 0.18 0.25 0.72

2-4-D (g-ai ha-1)

Rapistrum rugosum (L.) All. A 4—6 leaf stage 189.11 84.17 2.25

Neslia paniculata (L.) Desv. A 4—6 leaf stage 204.12 198.11 1.03

Galium aparine (spurium) L. A 4—6 leaf stage >564 >564 1.00

Asperula arvensis L. A 4—6 leaf stage >564 >564 1.00

Papaver rhoeas L. A 4—6 leaf stage >564 402.00 1.40

Papaper argemone L. A 4—6 leaf stage 480.95 69.15 6.96

Overall ratio for spray at young vegetative stage 3.1

b) Plants sprayed during reproductive period

Mecoprop (g-ai ha-1) (EXP6)

Silene noctiflora L A Flower buds 668.20 48.80 13.69 Strandberg et al. (2012)

Silene vulgaris (Moench) Garcke P Flower buds 313.90 49.20 6.38

Geranium molle L. A/B/P Flower buds 830.00 <75 11.07

Glyphosate (g-ai ha-1) (EXP6 and EXP7)

Silene noctiflora L. A Flower buds 160.90 43.10 3.73

Silene vulgaris (Moench) Garcke P Flower buds 79.60 <45 1.77

Geranium molle L. A/B/P Flower buds 63.20 <22 2.87

Metsulfuron methyl (g-ai ha-1) (EXP6)

Silene noctiflora L. A Flower buds >4.0 0.31 12.90

Silene vulgaris (Moench) Garcke P Flower buds >16.0 5.80 2.76

Geranium molle L. A/B/P Flower buds 0.60 <0.06 10.00

Overall ratio for spray at reproduction 7.2

c) Plants sprayed at two vegetative stages Early vegetative Late vegetative

Foramsulfuron + iodosulfuron (g-ai ha-1)

Solanum nigrum L. A/P 4 & 8—10 leaves 0.20 (0.15-0.24) 1.23 (0.91-1.55) 0.16 Strandberg et al. (2012)

Echinochloa crus-galli (L.) P. Beauv. A 4 & 6—8 leaves 0.25 (0.13-0.38) 0.48 (0.17-0.80) 0.52

Taraxacum officinale F.H. Wigg. ssp. officinale P 6 & 9 leaves 7.7 (4.0-11.5) 330.8 (162.3-823.9) 0.02

Agrostis tenuis Sibth. P 1 tiller + 3 leaves 1.4(1.0-1.7) 2.4 (2.0-2.8) 0.58

& elongation

(continued on next page)

Table 6 (continued)

c) Plants sprayed at two vegetative stages

Herbicides and species

EC50 (g-ai ha1)

Lifespan Growth stage

Early vegetative Late vegetative

Factor Source reference

Festuca ovina L.

Glyphosate (g-ai ha-1) Solanum nigrum L Echinochloa crus-galli (L.) P. Beauv. Taraxacum officinale F.H. Wigg.

ssp. officinale Agrostis tenuis Sibth.

Festuca ovina L.

Festuca ovina L.

Overall ratio for spray at young and old vegetative stages

6 tillers & 30 cm diam

4 & 8-10 leaves 4 & 6-8 leaves 6 & 12 leaves

6.7 (5.3-8.0) >124

39.4 (32.6-46.1) 86.3 (75.7-96.8) 30.6 (22.139.1)

1 tiller + 3 leaves 50.8 (35.1—66.6) 29.4 (21.5—37.2) & elongation

6 tillers & 30 cm 116.8 (91.2—142.3) 230.7 (168.9—292.6) diam

2 tillers + 4—6 leaves & elongation

94.5 (75.0-94.9) 59.0 (43.8-74.2)

99.2(81.0-117.4) 0.40

44.4 (32.0-56.8) 1.94

1042.6 (653.9-1431.4) 0.03

1.73 0.51 1.60

tested on crops (Baucom et al., 2008; Pline et al., 2002; Thomas et al., 2004; Yasuor et al., 2006). To date, studies on wild species are lacking.

4.3. Conclusion

We have demonstrated that non-crop plants in habitats abutting cropfields can be at various phenological stages during herbicide spray. We also established that plants in hedgerows adjacent to organic fields flowered earlier and for longer periods of time during the growing season than the same plants adjacent to conventional hedgerows where herbicides are regularly applied in adjacent cropfields. Based on EC50 values, in 58% of the cases recorded with a broad range of species and herbicides, reproductive parameters were more sensitive endpoints than corresponding vegetative measures (biomass) at both early and late stages of growth (compared to 32% for vegetative endpoints). Thus, plants should be tested for longer periods (to assess seed output) than currently recommended in guidelines (OECD, 2006; USEPA, 2012). Failure to adequately assess

and properly regulate herbicide effects can have important ecological considerations for plant survival, seed production, long-term seed-bank replenishment and eventual species composition of not only primary producers, but also species at other trophic levels. Measures to reduce herbicide use and alleviate drift effects to non-target habitats should be adopted (de Jong et al., 2008). In terms of toxi-cological implications, if testing with plants continues to only be conducted on a short-term basis, the use of an extrapolation factor appears to be justified to ensure that risks to plant communities within agro-ecosystems are not underestimated.


We wish to thank R. Rotches-Ribalta for providing some of the data inTable 6. We would also like to thankJ. Allison, N. De Silva,J. Parsons, L. Lauridsen and M. Thompson for assistance in the data collection and compilation. This research was funded by Environment Canada and the Danish Environmental Protection Agency and the International Centre for Research in Organic Food Systems (ICROFS).

Appendix A. Total number of quadrats in which woodlot plant species were affected by herbicide spray (EXP1). Woodlots (n = 3 in 1993 and n = 2 in 1994 and 1996) were surveyed in five transects perpendicular to a 15 m spray-free buffer zone (with bz) and in five transects where spray no buffer zones (no bz) were established (6 quadrats per transect - see text). Species in flower at the time of spray are marked with an asterisk. Plants were assessed qualitatively by visual assessment. Within brackets are the data for the woodlot only surveyed in 1993 (see Material and methods for additional detail).

Species Families Growth habit Lifespan Number of quadrats in which species were affected Total 1993 1994 With bz No bz With bz No bz 1996 With bz No bz

Solidago canadensis L. Asteraceae Forb P 33 4(1) 5(1) 3 8 9 2

Rubus idaeus L.* Rosaceae Subshrub P 23 5 7 2 7 0 2

Fraxinus americana L Oleaceae Sappling tree P 21 2 8(1) 2 8 - -

Acer saccharum Marshall Aceraceae Sappling tree P 11 2 1 1 5 0 2

Symphyotrichum lateriflorum (L.) Asteraceae Forb P 8 2 4 0 2 - -

A. Love & D. Love

Cornus stolonifera Michx.* Cornaceae Subshrub P 8 1 3 2 2 - -

Circaea lutetiana L. Onagraceae Forb P 7 1 1 1 4 - -

Crataegus spp. Rosaceae Shrub P 7 2 1 2 2 - -

Viola spp.* Violaceae Forb P 7 3 3 0 1 - -

Alliaria petiolata (M. Biev.) Brassicaceae Forb A/B 5 0 1 0 2 0 2

Cavara & Grande*

Hydrophyllum virginianum L.* Hycrophyllaceae Forb P 4 0 1 0 1 0 2

Geum spp. Rosaceae Forb P 4 2 2 - - - -

Taraxacum officinale F.H. Wigg.* Asteraceae Forb P 3 - - 0 2 0 1

( continued )

Species Families Growth habit Lifespan Number of quadrats in which species were affected

Total 1993 1994 1996

With bz No bz With bz No bz With bz No bz

Viburnum lentago L. Caprifoliaceae Shrub P 3 0 1 1 0 1 0

Geranium robertianum L. Geraniaceae Forb P 3 - - 0 2 0 1

Symphyotrichum lanceolatum Asteraceae Forb P 2 - - 0 2 - -

(Willd.) G.L. Nesom

Ribes americanum Mill.* Grossulariaceae Subshrub P 2 - - 0 2 - -

Rhamnus cathartica L. Rhamnaceae Shrub P 2 0 1 0 1 - -

Fragaria virginiana Duchesne* Rosaceae Forb P 2 0 2 - - - -

Geum canadense Jacq. Rosaceae Forb P 2 - - 1 0 1 0

Malus pumila Mill. Rosaceae Sappling tree P 2 0 1 0 1 - -

Prunus virginiana L.* Rosaceae Shrub P 2 - - - - 1 1

Arctium spp. Asteraceae Forb B 1 - - 0 1 - -

Erigeron philadelphicus L.* Asteraceae Forb B/P 1 0 (1) - - - -

Podophyllum peltatum L.* Berberidaceae Forb P 1 - - - - 0 1

Carpinus caroliniana Walter Betulaceae Sappling tree P 1 - - 0 1 - -

Euonymus obovatus Nutt.* Celastraceae Subshrub P 1 - - - - 0 1

Dipsacus fullonum L. Dipsacaceae Forb B 1 - - 0 1 - -

Quercus alba L. Fagaceae Sappling tree P 1 - - 0 1 - -

Prunella vulgaris L. Lamiaceae Forb P 1 - - 1 0 - -

Unknown Grass spp. Poaceae Grass 1 0 1 - - - -

Geum aleppicum Jacq. Rosaceae Forb P 1 - - 0 1 - -

Geum macrophyllum Willd. Rosaceae Forb P 1 - - 1 0 - -

Solanum dulcamara L. Solanaceae Forb P 1 0 (1) - - - -

Vitis riparia Michx.* Vitaceae Vine P 1 - - 0 1 - -


Albrecht, H., Mattheis, A., 1998. The effects of organic and integrated farming on rare arable weeds on the Forschungsverbund Agrar-okosysteme Munchen (FAM) research station in southern Bavaria. Biol. Conserv. 86, 347-356.

Al-Khatib, K., Tamhane, A., 1999. Dry pea (Pisum sativum L.) response to low rates of selected foliar- and soil-applied sulfonylurea and growth regulator herbicide. Weed Technol. 13, 753-758.

Andreasen, C., Stryhn, H., 2008. Increasing weed flora in Danish arable fields and its importance for biodiversity. Weed Res. 48,1-9.

Baucom, R.S., Mauricio, R., Chang, S.-M., 2008. Glyphosate induces transient male sterility in Ipomoea purpurea. Botany 86, 587-594.

Benton, T.G., 2006. Bumblebees. The Natural History & Identification of the Species Found in Britain. Collins, London.

Blackburn, L.G., Boutin, C., 2003. Subtle effects of herbicide use in the context of genetically modified crops: a case study with glyphosate (Roundup®). Ecotox-icology 12, 271-285.

Boutin, C., 2013. Herbicides: non-target species effect. In: Jorgensen, S.E. (Ed.), Encyclopedia of Environmental Management, vol II. Taylor and Francis, New York, pp. 1406-1417.

Boutin, C., Lee, H.-B., Peart, E.T., Batchelor, S.P., Maguire, R.J., 2000. Effects of the sulfonylurea herbicide metsulfuron methyl on growth and reproduction of five wetland and terrestrial plant species. Environ. Toxicol. Chem. 19 (10), 25322541.

Boutin, C., Jobin, B., 1998. Intensity of agricultural practices and effects on adjacent habitats. Ecol. Appl. 8 (2), 544-557.

Calabuig, I., 2000. Solitary Bees and Bumblebees in Danish Agricultural Landscape (PhD thesis). University of Copenhagen, Dept. of Population Ecology.

Calabuig, I., Madsen, H.B., 2009. Kommenteret ckeckliste over Danmarks bier - Del 2: Andrenidae (Hymenoptera, Apoidea). Entomol. Medd. 77 (2), 83-113.

Carpenter, D., Boutin, C., 2010. Sublethal effects of the herbicide glufosinate ammonium on crops and wild plants: short-term effects compared to vegetative recovery and plant reproduction. Ecotoxicology 19, 1322-1336.

Carpenter, D., Boutin, C., Allison, J., 2013. Effects of chlorimuron ethyl on terrestrial and wetland plants: levels of, and time to recovery following sublethal exposure. Environ. Pollut. 172, 275-282.

Carvalheiro, L.G., Seymour, C.L., Veldtman, R., Nicolson, S.W., 2010. Pollination services decline with distance from natural habitat even in biodiversity-rich areas. J. Appl. Ecol. 47, 810-820.

Clay, P.A., Griffin, J.L., 2000. Weed seed production and seedling emergence responses to late-season glyphosate applications. Weed Sci. 48 (4), 481-486.

Crone, E.E., Marler, M., Pearson, D.E., 2009. Non-target effects of broadleaf herbicide on a native perennial forb: a demographic framework for assessing and minimizing impacts. J. Appl. Ecol. 46, 673-682.

deJong, F.M.W., de Snoo, G.R., van de Zande, J.C., 2008. Estimated nationwide effects of pesticide spray drift on terrestrial habitats in the Netherlands. J. Environ. Manage. 86, 721-730.

European Food and Safety authority, 2012. Scientific opinion on the science behind the development of a risk assessment of plant protection products on bees (Apis mellifera, Bombus spp. and solitary bees). EFSA J. 10. j.efsa.2012.2668, 2668 [275 pp.], Available online: efsajournal.

Fletcher, S.J., Pfleeger, T.G., Hillman, C.R., 1993. Potential environmental risks associated with the new sulfonylurea herbicides. Environ. Sci. Technol. 27 (10), 2250-2252.

Fletcher, J.S., Pfeeger, T.G., Ratsch, H.C., Hayes, R., 1996. Potential impact of low levels of chlorosulfuron and other herbicides on growth and yield of nontarget plants. Environ. Toxicol. Chem. 15, 1189-1196.

Fried, G., Petit, S., Dessaint, F., Reboud, X., 2009. Arable weed decline in Northern France: crop edges as refugia for weed conservation? Biol. Conserv. 142, 238243.

Gealy, D.R., Boerboom, C.M., Ogg Jr., A.G., 1995. Growth and yield of pea (Pisum sativum L.) and lentil (Lens culinaris L.) sprayed with low rates of sulfonylurea and phenoxy herbicide. Weed Sci. 43, 640-647.

Goulson, D., 2010. Bumblebees: Behaviour, Ecology and Conservation. Oxford University Press, Oxford.

Gove, B., Power, S.A., Buckley, G.P., Ghazoul, J., 2007. Effects of herbicide spray drift and fertilizer overspread on selected species of woodland ground flora: comparison between short-term and long-term impact assessments and field surveys. J. Appl. Ecol. 44, 374-384.

Guo, S.L., Jiang, H.W., Fang, F., Chen, G.Q., 2009. Influences of herbicides, uprooting and use as cut flowers on sexual reproduction of Solidago canadensis. Weed Res. 49 (3), 291-299.

Harper, J., 1977. Population Biology of Plants. Academic Press, London.

Holst, N., Axelsen, J.A., Bruus, M., Damgaard, C.F., Kudsk, P., Lassen, J., Madsen, K.H., Mathiassen, S.K., Strandberg, B., 2008. Sprojtepraksis i ssdskifter med og uden lyphosattolerante afgroder. In: Effekter pá floraen i mark og hegn. Pesticide Research 121. Danish Ministry of the Environment EPA.

Holterman, H.J., van de Zande, J.C., Porskamp, H.A.J., Huijsmans, J.F.M., 1997. Modelling spray drift from boom sprayers. Comput. Electron. Agric. 19,1-22.

International Organization for Standardization (ISO), 2005. Soil Quality — Biological Methods - Chronic Toxicity in Higher Plants. ISO 22030.

Kjsr, C., Strandberg, M., Erlandsen, M., 2006a. Metsulfuron spray drift reduces fruit yield of hawthorn (Crataegus monogyna L.). Sci. Total Environ. 356, 228-234.

Kjsr, C., Strandberg, M., Erlandsen, M., 2006b. Effects on hawthorn the year after simulated spray drift. Chemosphere 63, 853-859.

Kleijn, D., Verbeek, M., 2000. Factors affecting the species composition of arable field boundary vegetation. J. Appl. Ecol. 37, 256-266.

Kremmen, C., Willians, N.M., Aizen, M.A., Gemmill-Herren, B., LeBuhn, G., Minckley, R., Packer, L., Potts, S.G., Roulston, T., Steffan-Dewenter, I., Vázquez, D.P., Winfree, R., Adams, L., Crone, E.E., Greenleaf, S.S., Keitt, T.H., Klein, A.-M., Regetz, J., Ricketts, T.H., 2007. Pollination and other ecosystem services produced by mobile organisms: a conceptual framework for the effects of land-use change. Ecol. Lett. 10, 299-314.

Madsen, H.B., Calabuig, I., 2008. Kommenteret ckeckliste over Danmarks bier - Del 1: Colletidae (Hymenoptera, Apoidea). Entomol. Medd. 76,145-163.

Madsen, H.B., Calabuig, I., 2010. Kommenteret ckeckliste over Danmarks bier - Del 3: Milittidae & Megachilidae (Hymenoptera, Apoidea). Entomol. Medd. 78, 73-99.

Norberg-King, T.J., 1993. A Linear Interpolation Method for Sublethal Toxicity: the Inhibition Concentration (ICp) Approach (Version 2.0). U.S. Environmental Protection Agency, Environmental Research Laboratory, Duluth (MN). Technical Report 03-93.

Organisation for Economic Co-operation and Development (OECD), 2006. Terrestrial Plants Test: Seedling Emergence and Seedling Growth Test (No. 208) and Vegetative Vigour Test (No. 227). In: OECD Guidelines for Testing Chemicals. Paris (France).

Pline, W.A., Viator, R., Wilcut, J.W., Edmisten, K.L., Thomas, J., Wells, R., 2002. Reproductive abnormalities in glyphosate-resistant cotton caused by lower CP4-EPSPS levels in the male reproductive tissue. Weed Sci. 50, 438-44 .

Ratsch, H.C., Johndro, D.J., McFarlane, J.C., 1986. Growth inhibition and morphological effects of several chemicals in Arabidopsis thaliana (L.) Heynh. Environ. Toxicol. Chem. 5, 55-60.

Rinella, M.J., Haferkamp, M.R., Masters, R.A., Muscha, J.M., Bellows, S.E., Vermeire, L.T., 2010. Growth regulator herbicides prevent invasive annual grass seed production. Invasive Plant Sci. Manag. 3 (1), 12-16.

Riemens, M.M., Dueck, T., Kempenaar, C., 2008. Predicting sublethal effects of herbicides on terrestrial non-crop plant species in the field from greenhouse data. Environ. Pollut. 155,141-149.

Riemens, M.M., Dueck, T., Kempenaar, C., Lotz, L.A.P., Kropff, M.J.J., 2009. Sublethal effects of herbicides on the biomass and seed production of terrestrial non-crop plant species, influenced by environment, development stage and assessment date. Environ. Pollut. 157, 2306-2313.

Romero, A., Chamorro, L., Sans, F.X., 2008. Weed diversity in crop edges and inner fields of organic and conventional dryland winter cereal crops in NE Spain. Agric. Ecosyst. Environ. 124, 97-104.

Rotchés-Ribalta, R., Boutin, C., Blanco-Moreno, J.M., Carpenter, D.J., Sans, X.F., 2012. Effects of herbicide use on ecological patterns of characteristic and rare arable weeds. In: British Ecological Society Meeting, Birmingham, UK (Poster).

Santandreu, M., Lloret, F., 1999. Effect of flowering phenology and habitat on pollen limitation in Erica multiflora. Can. J. Bot. 77 (5), 734-743.

Seefeldt, S.S., Jensen, J.E., Fürst, E.P., 1995. Log-logistic analysis of dose-response relationships. Weed Technol. 9, 218-227.

Steadman, K.J., Eaton, D.M., Plummer, J.A., Ferris, D.G., Powles, B., 2006. Late-season non-selective herbicide application reduces Lolium rigidum seed numbers, seed viability, and seedling fitness. Aust. J. Agric. Res. 57,133-141.

Storkey, J., Meyer, S., Still, K.S., Leuschner, C., 2012. The impact of agricultural intensification and land-use change on the European arable flora. Proc. R. Soc. B Biol. Sci..

Strandberg, B., Mathiassen, S.K., Bruus, M., Kjaer, C., Damgaard, C., Andersen, H.V., Bossi, R., Lofstrom, P., Larsen, S.E., Bak, J., Kudsk, P., 2012. Effects of Herbicides on Non-target Plants: How Do Effects in Standard Plant Tests Relate to Effects in Natural Habitats?. Pesticide Research No 137 Danish Ministry of the Environment, EPA, p. 114.

Strandberg, B., Axelsen, J., Kryger, P., Enkegaard, A., 2011. Best0vning og bio-diversitet. Faglig rapport fra DMU nr. 83 .

Sutcliffe, O.L., Kay, Q.O.N., 2000. Changes in the arable flora of central southern England since 1960s. Biol. Conserv. 93,1—8.

Thomas, W.E., Pline-Srnie, W.A., Thomas, J.F., Edmisten, K.L., Wells, R., Wileut, J.W., 2004. Glyphosate negatively affects pollen viability but not pollination and seed set in glyphosate-resistant corn. Weed Sci. 52 (5), 725—734.

Türe, C., Böcük, H., 2008. Investigation of threatened arable weeds and their conservation status in Turkey. Weed Res. 48, 289—296.

United States Environmental Protection Agency (USEPA), 2012. Ecological Effects Test Guidelines: Seedling Emergence and Seedling Growth (OCSPP 850.4100), Early Seedling Growth Toxicity Test (OCSPP 850.4230), Vegetative Vigor (OCSPP 850.4150), Background and Special Considerations — Tests with Terrestrial, Aquatic Plants, Cyanobacteria, and Terrestrial Soil-core Microcosms (OCSPP 850.4000).

Walker, E.R., Oliver, L.L., 2008. Weed seed production as influenced by glypho-sate applications at flowering across a weed complex. Weed Technol. 22 (2), 318—325.

Weisser, P., Landfried, M., Koch, H., 2002. Off-crop drift sediments on plant surfaces — exposure of non-target organisms. Aspect Appl. Biol. 66, 225—230.

Westrich, P., 1990. Die Wildbienen Baden-Württembergs. Verlag Eugen Ulmer, Stuttgart, Germany.

Wilson, P.J., 1994. Managing field margins for the conservation of the arable flora. Br. Crop Prot. Counc. Monogr. 58, 253—258.

Yasuor, H., Abu-Abied, M., Belausov, E., Madmony, A., Sadot, E., Riov, J., Rubin, B., 2006. Glyphosate-induced anther indehiscence in cotton is partially temperature dependent and involves cytoskeleton and secondary wall modifications and auxin accumulation. Plant Physiol. 141,1306—1315.

Zwerger, P., Pestemer, W., 2000. Testing the phytotoxic effects of herbicides on higher terrestrial non-target plants using a plant life cycle test. Z. PflKrankh. PtlSchutz. Sonderh. 17, 711—718.