Intensity of soil disturbance shapes response trait diversity of weed communities: The long-term effects of different tillage systems

Agriculture, Ecosystems and Environment 207 (2015) 101–108

Contents lists available at ScienceDirect

Agriculture, Ecosystems and Environment journal homepage: www.elsevier.com/locate/agee

Intensity of soil disturbance shapes response trait diversity of weed communities: The long-term effects of different tillage systems Eva Hernández Plaza a, * , Luis Navarrete b , José L. González-Andújar a a

Instituto de Agricultura Sostenible (CSIC), Finca Alameda del Obispo, Aptdo 4084, 14080 Córdoba, Spain Instituto Madrileño de Investigación y Desarrollo Rural, Agrario y Alimentario (IMIDRA), Finca ‘El Encín’ A2, Km. 38.2, Alcalá de Henares, Madrid 28800, Spain b

A R T I C L E I N F O

A B S T R A C T

Article history: Received 6 November 2014 Received in revised form 24 March 2015 Accepted 27 March 2015 Available online xxx

Disturbances have a prominent role in structuring plant communities. However, in agroecosystems, the long-term effect of disturbances on determining trait distributions within weed communities remains little studied. We analyzed the effect of three tillage treatments, which differ in the intensity of soil disturbance, on the mean, the range and the distribution of four response traits within weed communities. We aim to test whether tillage acts as a filter restricting the range and the distribution of response traits within weed communities and leads to reduced response trait diversity or whether tilling may have a diversifying effect, creating opportunities for more phenotypes to coexist and increasing response trait diversity. To test this idea, we used data on weed abundance recorded over 24 years from an experiment in which conventional tillage (CT), minimum tillage (MT) and no-tillage (NT) systems were compared. We selected four response traits, maximum height, specific leaf area (SLA), seed weight and seed output, and computed the community weighted mean (CWM) of each trait, as well as four multi-trait metrics related to a different aspect of functional diversity. We found that soil disturbance increases available niche opportunities for weeds especially in terms of regenerative traits. CT, the greater soil disturbance, leads to a greater range and even distribution of the studied traits and that abundant weed species from CT plots hold more divergent trait values than those from MT and NT plots. Our results may be explained by the idiosyncrasy of our disturbance treatments that affect weed seed placement in the soil layers as well as the stratification and availability of soil nutrients. We also found that NT system selected for lower CWM of seed weight (and higher seed output) than MT and CT systems. NT places weed seeds mostly on the soil surface, where having a large seed output may be necessary to avoid the risk of decay or depredation. Conversely, MT and CT systems offer some advantage to other strategies such as larger seed sizes useful to germinate from depth. CWM of SLA was higher in NT and MT than in CT plots and this could be related to greater soil nutrient content in NT systems. In addition our results showed a general trend over experimental time for weed communities to increase in height (and slightly in SLA and seed production) while reducing in seed size. These features are generally associated with intensive farming systems. ã 2015 Elsevier B.V. All rights reserved.

Keywords: Community-weighted mean of response traits Functional dispersion Functional divergence Functional evenness Seed weight Specific leaf area

1. Introduction Species assemble into local communities according to their trait values. Environmental abiotic factors, biotic interactions and stochastic processes act on functional attributes present in the regional species pool to determine the relative abundance of species at local sites (Shipley et al., 2006; Weiher et al., 2011;

* Corresponding author. Tel.: +34 957499255; fax: +34 957499252. E-mail addresses: [email protected] (E. Hernández Plaza), [email protected] (L. Navarrete), [email protected] (J.L. González-Andújar). http://dx.doi.org/10.1016/j.agee.2015.03.031 0167-8809/ ã 2015 Elsevier B.V. All rights reserved.

Laliberté et al., 2012). The relative importance and the manner in which abiotic factors affect trait distributions within communities are not fully understood. Concerning disturbances, it has been suggested that they may promote a divergent distribution of trait values, thus increasing trait diversity (Grime, 2006). Disturbances may limit the growth and competitive ability of dominant species or promote novel niche opportunities allowing for a variety of phenotypes to occur. Conversely, a high level of disturbance may filter out phenotypes which are unable to cope with the harsh conditions, resulting in a low range of phenotypes and reduced trait diversity (Diaz et al., 1998; Laliberté et al., 2012; Maire et al., 2012). Eventually, the effect of disturbances on trait diversity could depend on the general intensity of disturbance in the

102

E. Hernández Plaza et al. / Agriculture, Ecosystems and Environment 207 (2015) 101–108

studied system, on the type of disturbance and on the aspects of functional diversity considered (Bruggisser et al., 2010; Pakeman, 2011). Agroecosystems are characterized by frequent disturbances, including biomass removal and soil management. Crop type and associated management practices determine the level of disturbance and pose certain limitations on the species found within these systems. Weeds are plant species that accompany crops and that respond distinctly to management practices (Gunton et al., 2011; Fried et al., 2012; Pinke et al., 2012). Weeds are recognized to hold a key role within agroecosystems; they are the greatest contributor to plant diversity, they represent crucial resources for other taxa and they may provide crops with important ecosystem services (Marshall et al., 2003; Moonen and Bàrberi, 2008; Petit et al., 2011). Nevertheless, when occurring in large abundance, weeds are also responsible for reductions in crop yields. The goal is to facilitate a balanced trade-off between the preservation of weed diversity and the maintenance of crop production. Traditionally, weed studies have concentrated on only one or a few dominant species (e.g., Oveisi et al., 2013) but assessing how management practices affect trait distributions within entire weed communities may be an opportunity for developing community-oriented weed management strategies (Storkey et al., 2010; Gunton et al., 2011; Fried et al., 2012). In this context it is useful to refer to the response-effect framework (Lavorel and Garnier, 2002; Suding et al., 2008), and focus on response traits, i.e., those that determine the way communities respond to environmental drivers. This work analyzes the effect of tillage intensity on both the mean value of response traits in weed communities and on the range and distribution of response trait values within these communities, i.e., weed response trait diversity. Tillage intensity was used as a surrogate for soil disturbance. Three tillage treatments were compared: conventional tillage (CT; plough plus cultivator), minimum tillage (MT; chisel plus cultivator) and no tillage (NT). Tillage is commonly used as a weed management practice to remove the first emerged cohorts of weeds prior to sowing. Tillage can be performed with different implements leading to more or less soil disturbance (CT vs. MT). A similar elimination of weeds can be accomplished by applying herbicide without tilling the soil (NT). Tillage affects the structure and properties of soil as well as the distribution of weed seeds in the soil layers (Swanton et al., 2000; Spokas et al., 2007; Vogeler et al., 2009). Thus, tillage has a potential impact on weed establishment and the distribution and availability of nutrients to weeds. Specifically, the aims of this research were to test whether (1) tillage acts as a strong filter restricting the range of weed phenotypes which can persist in a field. Greater tillage intensity would therefore result in reduced response trait diversity of weed communities. Or (2) whether tillage opens up available niche space, thus promoting wider ranges of phenotypes. In this case, greater tillage intensity would result in increased response trait diversity of weed communities. In order to test these hypotheses, weed abundance data recorded during 24 years from a long-term experiment in which a cereal–legume rotation was subjected to the three tillage systems (CT, MT and NT) was used. We selected four response traits, which have been previously related to the response of plants to disturbance and competition, two related to the established phase of plants’ life cycle and two associated with the plants’ persistence. We used four metrics to characterize weed trait diversity, each related to a different facet: richness (FRic), evenness (FEve), divergence (FDiv) and dispersion (FDis). We also computed the community weighted mean (CWM) of each of the response traits considered.

2. Methods 2.1. Study site and experimental design The study was conducted at El Encín Experimental Station (40
290 N; 3
220 W, Madrid, Spain, 610 m.s.l). The experiment was initiated in 1985 and is ongoing. This paper refers to weed surveys conducted from 1985 to 2011 (24 years of data, no data for 1990 and 1997). The site has a Mediterranean climate, with mild, humid winters and dry, hot summers. Average annual rainfall during the 26-year study period was 445 mm (ranging from 264 to 759 mm). Average annual temperature was 13.8
C (ranging from 11.9 to 15.5
C). The soil of the experimental field is an alfisol xeralf, from the calciortic–molic subgroup. The experiment followed a randomized block design with four replicates. The three tillage treatments, CT, MT and NT, were randomly assigned to plots (20 m  40 m) within each block. The cropping system was a rotation of winter wheat (Triticum aestivum L.) and a leguminous crop, vetch (Vicia sativa L.) or pea (Pisum sativum L.). The wheat planting date ranged from October 30th to December 19th. Fertilizers were applied at planting time (28 kg N, 37 kg P2O5, 26 kg K2O ha1; average rates) and at mid-tillering (53 kg N ha1), and post-emergence herbicide (0.2 kg a.i. ha1 ioxynil + 0.2 kg a.i. ha1 bromoxynil + 1.012 kg a.i. ha1 mecoprop) was applied at the tillering stage. Leguminous crops were in all cases planted between November 6th and January 19th. Fertilizers were only applied at planting time. Average rates of fertilizer were 14 kg N, 14 kg P2O5 and 14 kg K2O ha1 for vetch and 19 kg N, 38 kg P2O5 and 71 kg K2O ha1 for pea. No post-emergence herbicides were applied. CT involved at least one mouldboard ploughing operation with a working depth of 25 cm, followed by a secondary tillage operation with a field cultivator (10–15 cm working depth). MT involved a primary cultivation with either a chisel plow (15– 20 cm working depth) or a field cultivator, followed by a secondary operation with a field cultivator. In NT, the only operation conducted prior to wheat planting was the application of glyphosateã (0.72 kg a.i. ha1) 4–6 days prior to planting. When sowing leguminous crops with NT treatment, straw and stubble from the previous wheat crop were destroyed by chopping, applying non-selective herbicide (e.g., glyphosate) thereafter. 2.2. Weed sampling Weed species abundance was recorded yearly (except in 1990 and 1997) in ten samples (30 cm  33 cm per plot), except the first three years when only five samples were collected and in 1995 when 20 samples were obtained. Samples were located along an M-shaped itinerary at intervals of approximately 15 m and 3 m away from field borders. Sampling always took place before herbicide application, between mid February and mid April every year. Sampling time was decided according to crop maturation stage, corresponding to mid-tillering for wheat and stem elongation for vetch and pea. 2.3. Weed response trait data We chose two response traits related to plants’ vegetative phase, specific leaf area (SLA) and maximum height, and two more related to the regeneration of plant species, seed weight and fecundity. SLA is a component of the leaf economic spectrum (Wright et al., 2004). It expresses the potential area available for light interception per unit of dry leaf mass and reflects a trade-off between construction cost, water loss through transpiration and carbon gain through high photosynthetic rates (Westoby, 1998). Plant height is a whole plant trait that, in herbaceous plants, is often related to overall plant size and competitive interactions for

E. Hernández Plaza et al. / Agriculture, Ecosystems and Environment 207 (2015) 101–108

light (Westoby et al., 2002). Seed weight and fecundity (mean number of seeds produced by an individual plant) are associated with seedling survival and the time taken for seedlings to attain reproductive maturity (Moles and Westoby, 2006). Response trait data for each of the species were obtained from the literature and from public databases (Table S1). Mediterranean species are not well represented in trait databases and for some species data were not available for all the traits. We lacked 19 trait values from a total of 176. One or two traits were missing for ten species. Ranunculus sp. could not be identified to species level so all four traits were missed in this case (Table S2). Supplementary material related to this article found, in the online version, at http://dx.doi.org/10.1016/j.agee.2015.03.031.

103

(Burnham and Anderson, 2002). Wi measured the probability that model i was the best model for the observed data. In all models, assumptions of equal variances, normal distribution and uncorrelated residuals were evaluated graphically. Additionally, we computed Spearman correlations between CWM of SLA, maximum height, seed weight and number of seeds and multi-trait diversity indices. Correlations were calculated independently for each tillage system. All analyses were conducted in the R environment (R Development Core Team, 2011). Linear mixed effects models were adjusted using the lme function from the nlme library (Pinheiro et al., 2012) 3. Results

2.4. Response trait diversity measures The diversity of functional traits within a community is difficult to comprise with a single metric. Three main aspects of functional diversity have been delineated: richness, evenness and divergence (Mason et al., 2005; Villéger et al., 2008). We calculated four multitrait diversity indices each describing a different facet of functional diversity. We used functional richness (FRic; Cornwell et al., 2006; Villéger et al., 2008) and functional dispersion (FDis; Laliberté and Legendre, 2010) as measures of the volume of functional space occupied by weed communities and of the dispersion of phenotypes within that space, respectively. FDis is the mean distance of individual species, weighted by their relative abundance, to the centroid of all species, where weights are their relative abundance. FRic expresses the range of trait values in the weed community. We employed functional evenness (FEve; Villéger et al., 2008) as a measure of the regularity in the distribution of trait values in the functional space weighted by species relative abundance, and functional divergence (FDiv; Villéger et al., 2008) as a measure of species divergence in their distance from the center of gravity in the functional space. Divergence decreases when abundant species have traits that are close to the center of functional space. As some trait values were missing, the procedure described in Laliberté and Legendre (2010) was used and implemented in the “FD” library (Laliberté and Legendre, 2010) in the R environment (R Development Core Team, 2011). In addition, in order to characterize trends of the average values of response traits across tillage treatments the “FD” library was used to calculate the community weighted mean (CWM) of each of the four selected response traits. This is the average value of the trait in the community weighted by species abundance. Every metric was computed for each plot (n = 12) in each year (n = 24, missing data for 1990 and 1997). 2.5. Statistical analyses We used linear mixed effects models to test for differences between tillage treatments and time since the beginning of the experiment on the CWM of each response trait and each of the four multi-trait diversity indices considered. In all cases, time (as a covariable), type of tillage (CT, MT or NT) and their interaction were included as fixed effects. Block was included as a random effect to take into account the nature of the experiment. For each response variable, the set of candidate models, including all the possible combinations of the variables as well as the null model, were fitted and ranked by Akaike's information criterion corrected for small sample size (AICc). AICc was used as a measure of the relative support of each of the models analyzed. Additionally, we calculated DAICc and Akaike weights (Wi) for each model. DAICc is the difference between the model AICc and the AICc of the first ranked model (the one with the minimum AICc score) and represents a measure of the loss of information with respect to the best model

A total of 44 species were recorded in 24 years of sampling (Table S2). All of them were species generally found in cereal or leguminous fields in central Spain. The majority were species of autumn emergence and flowering in early spring, though some were species that typically emerge in early spring. Some perennial species occurred in NT plots, though most of the recorded species were therophytes. Descurainia sophia,Papaver rhoeas and Veronica hederifolia were the most abundant species in all treatments. Adonis aestivalis and Agrostemma githago, two species recognized to be in decline throughout Europe, exhibited low abundance in the experiment. 3.1. Response of weed community-weighted traits to tillage intensity and time Both tillage intensity and time since the beginning of the experiment had an effect on the CWM of weed traits. The most parsimonious model for the CWM of maximum height was that which included the interaction of both factors (Table 1). For the CWM of SLA, the additive model had the lowest AICc score. For the CWM of seed weight and number of seeds, both the model including the interaction of both factors and the additive model had equal support from the data, with AICc scores differing by 1.9 and 2 points, respectively, from the second best model (Burnham and Anderson, 2002; Table 1). In these cases, we report estimates and standard errors from the model with lower AICc (Table 2). The CWM of SLA slightly increased throughout the duration of the experiment in all tillage systems, with MT and NT plots exhibiting higher values than CT plots (Table 2, Fig. 1). The CWM of maximum height also increased throughout the duration of the experiment, varying with tillage intensity. Values were higher at the beginning of the experiment in NT plots but increased more in CT and MT plots through time (Table 2, Fig. 1). The CWM of seed weight was greater in MT and CT plots than in NT plots and values tended to decrease towards low seed weights at the end of the experimental period. The CWM of the number of seeds followed an opposite trend than that of the CWM of seed weight, with higher values in NT than in the other two tillage systems, and increasing values throughout time across all tillage intensities (Table 2, Fig. 1). 3.2. Response of multi-trait diversity indices to tillage intensity and time Range and trait distribution within communities were also affected by tillage intensity and time since the beginning of the experiment (Table 1). For FEve, the model with the interaction between time and tillage intensity had the lowest AICc score. For FRic, FDis and FDiv, the additive model was the most supported (Table 1). In general, CT plots had higher values in all four multitrait diversity indices analyzed. FRic values were greater in CT plots, followed by NT and MT plots (Fig. 2). The dispersion of

104

E. Hernández Plaza et al. / Agriculture, Ecosystems and Environment 207 (2015) 101–108

Table 1 Summary of the models describing variation in community-weighted mean (CWM) of SLA, height, seed weight and number of seeds and in different measures of trait diversity – functional richness (FRic), functional dispersion (FDis), functional evenness (FEve) and functional divergence (FDiv) – in relation to tillage system and time since the beginning of the experiment. df, degrees of freedom, AICc, Akaike's information criterion corrected for small sample size, Di, the AICc differences of each model compared with the most parsimonious and Wi Akaike weights. Models

df

CWM SLA

Tillage + time Tillage  time Time Tillage Intercept only

6 8 4 5 3

1272.7 1276.1 1277.2 1286.0 1289.9

0.0 3.4 4.5 13.3 17.2

0.77 0.14 0.08 0.00 0.00

Tillage  time Tillage + time Time Intercept only Tillage

8 6 4 3 5

2084.6 2099.5 2103.2 2367.8 2368.8

0.0 14.9 18.6 283.1 284.2

1.00 0.00 0.00 0.00 0.00

CWM seed weight

Tillage  time Tillage + time Time Tillage Intercept only

8 6 4 5 3

687.8 689.7 700.8 736.6 745.4

0.0 1.9 13.1 48.9 57.6

0.72 0.28 0.00 0.00 0.00

CWM number of seeds

Tillage + time Tillage  time Tillage Time Intercept only

6 8 5 4 3

6067 6069 6074.3 6097 6103.2

0.0 2.0 7.3 30.0 36.2

0.72 0.26 0.02 0.00 0.00

Tillage + time Tillage x time Tillage Intercept only Time

6 8 5 3 4

259.6 257.1 255 110.9 111.1

0.0 2.5 4.6 370.4 370.7

0.72 0.21 0.07 0.00 0.00

Tillage + time Tillage  time Tillage Time Intercept only

6 8 5 4 3

1387.9 1384.1 1368.1 1252.6 1241.4

0.0 3.8 19.8 135.3 146.5

0.87 0.13 0.00 0.00 0.00

Tillage x time Time Tillage + time Intercept only Tillage

8 4 6 3 5

230.3 222.4 220.5 210.8 208.8

0.0 7.9 9.8 19.6 21.6

0.97 0.02 0.01 0.00 0.00

Tillage + time Tillage + time Tillage Time Intercept only

6 8 5 4 3

366.4 362.3 332.6 314 286.3

0.0 4.1 33.8 52.4 80.1

0.89 0.12 0.00 0.00 0.00

CWM height

FRic

FDis

FEve

FDiv

AICc

Di

Response variable

Wi

phenotypes within the functional space occupied by weed communities (FDis) was greatest in CT plots, followed by similar values in MT and NT plots. Both FRic and FDis values increased slightly throughout the duration of the experiment. Functional evenness increased throughout the experimental period in the CT plots, while it remained constant in MT and NT plots, resulting in larger FEve values in CT plots at the end of the experimental period (Table 2, Fig. 2). A high divergence in weed trait distribution (FDiv) was seen in the experiment, though values decreased throughout the duration of the experiment (Table 2, Fig. 2). FDiv was higher in CT and NT plots than in MT plots. 3.3. Relationships between different trait diversity measures Several significant correlations between FRic, FDis, FEve and FDiv were present, but dependent on the tillage treatment. Specifically, FDis correlated positively with FEve in CT and MT

Table 2 Parameter estimates and standard errors (SE) of most parsimonious models relating CWM of four traits considered and four multi-trait diversity metrics (FRic, FDis, FEve and FDiv) to tillage system and time since the beginning of the experiment. MT, minimum tillage, NT, no- tillage system. Conventional tillage (CT) was the reference treatment in all analyses. Dependent variable

Independent variables

CWM SLA

Intercept Time Tillage MT Tillage NT

21.4 0.07 0.8 0.8

0.4 0.02 0.3 0.3

CWM height

Intercept Time Tillage MT Tillage NT Time:tillage MT Time:tillage NT

30.7 1.8 3.3 7.7 0.1 0.7

2.1 0.1 2.7 2.7 0.2 0.2

CWM seed weight

Intercept Time Tillage MT Tillage NT Time:tillage MT Time:tillage NT

2.3 0.05 0.25 0.69 0.01 0.03

0.2 0.01 0.2 0.2 0.02 0.02

CWM number of seeds

Intercept Time Tillage MT Tillage NT

FRic

Intercept Time Tillage MT Tillage NT

0.57 0.003 0.60 0.39

0.02 0.001 0.02 0.02

FDis

Intercept Time Tillage MT Tillage NT

0.09 0.001 0.03 0.04

0.003 0.0002 0.003 0.003

FEve

Intercept Time Tillage MT Tillage NT Time:tillage MT Time:tillage NT

0.41 0.01 0.11 0.11 0.01 0.01

0.03 0.002 0.05 0.05 0.003 0.003

FDiv

Intercept Time Tillage MT Tillage NT

0.92 0.01 0.14 0.04

0.02 0.001 0.02 0.02

Estimate

20,142.1 230.2 1022.4 6649.8

SE

1443.4 75.0 1398.2 1394.6

plots, FEve correlated negatively with FDiv in CT plots and FRic correlated positively with FDis in MT and NT plots. The CWM of maximum height was negatively related to the CWM of seed weight and positively to the CWM of the number of seeds. The CWM of SLA was not related to the CWM of maximum height, but was positively related to the CWM of seed weight in MT and NT plots. Significant relationships between the CWM of response traits and FDis, FEve and FDiv were also present but dependent on the tillage system (Figs. S1–S3). Supplementary material related to this article found, in the online version, at http://dx.doi.org/10.1016/j.agee.2015.03.031. 4. Discussion We assessed the long-term effects of three tillage treatments, reflecting differing soil disturbance intensities, on four aspects of response trait diversity and on average response trait values of weed communities. We found that conventional tillage, the greatest degree of soil disturbance, allowed for the highest response trait diversity in weed communities for the traits considered. Our analysis also revealed that tillage treatments hold

E. Hernández Plaza et al. / Agriculture, Ecosystems and Environment 207 (2015) 101–108

105

Fig. 1. Changes in the community-weighted trait mean values (CWM) of SLA, height, seed weight and number of seeds for different tillage systems over the 26 years of the experiment (1986–2011). Lines are fitted values from linear mixed effects models (see text for analyses details).

distinct community mean values of weed response traits. In addition, we observed a general trend for weed communities to increase in height (and slightly in SLA and seed production) while seed size decreases over the duration of the experiment. In the following paragraphs we explain our findings and their broader implications for agroecosystem functioning. 4.1. Response of weed community-weighted traits to tillage intensity over time Weed communities from NT plots had a lower CWM of seed weight throughout the duration of the experiment than those from MT and CT plots. Conversely, the CWM of the number of seeds was higher in NT plots than with the other treatments. The promotion of distinct seed weight strategies could be explained by the differences in how the tillage systems distribute weed seeds throughout soil layers. NT system leaves seeds mostly at the surface, MT system places seeds homogenously throughout the first five centimeter of soil with some seeds reaching ten centimeter, whereas in CT system only a small amount of seeds remain at the surface, few are left in the first 10 cm of soil and most are placed deeper (Chauhan et al., 2006; Spokas et al., 2007; Marshall and Brain, 1999). Seeds at the soil surface are exposed to depredation or rapid decay, compromising species regeneration, but having large amount of seeds may allow species to circumvent this risk. Thus, in NT systems it could be more advantageous to produce a large amount of small seeds than a few larger ones.

However, large seeds are beneficial when germination is required at a greater depth or when a large initial seedling size is important for seedling establishment. Therefore, not only small seeded species but also species with larger seeds may have advantages when growing in CT or MT plots, explaining the higher CWM of seed weight and lower CWM of seed fecundity in these systems. We also found a general tendency for the CWM of seed weight to decrease over time with all tillage treatments, with a parallel increase in the CWM of maximum height, especially in CT and MT plots. Our experiment was managed in an intensive manner with the use of post-emergence herbicides and synthetic fertilizers. Large seed size and short stature are phenotypes in decline in intensive farming systems (Navas, 2012; Storkey et al., 2012). In these systems, taller plants compete more effectively for light while having a large seed output, which is concurrent with low seed sizes, which will increase the number of opportunities for seedling survival. Accordingly, in cereal fields in France, Fried et al. (2012) found that the average seed weight of species that are increasing in abundance was lower than the seed weight of species in decline. They suggested that reduced seed weight may be a trait associated with highly variable germination dates, a character that in weeds is selected under current short crop rotations and extensive use of herbicides. Apart from the general increase with time, we also found some differences between the CWM of maximum height of different treatments. Increases in height were larger in plots with some soil disturbance. Thus, on average weed communities from CT and MT plots were taller than those from NT plots at the end of the

106

E. Hernández Plaza et al. / Agriculture, Ecosystems and Environment 207 (2015) 101–108

Fig. 2. Changes in the four measures of trait diversity, functional richness (FRic), functional dispersion (FDis), functional evenness (FEve) and functional divergence (FDiv), for different tillage systems over the 26 years of the experiment (1986–2011). Lines are fitted values from linear mixed effects models (see text for analysis details).

experimental period. In contrast with our results, several works have shown that plant height is lower in communities subjected to disturbances (Kyle and Leishman, 2009; Castro et al., 2010; Laliberté et al., 2012; Peco et al., 2012). This could be occurring when disturbances are related to biomass removal, for example by grazing. In these cases, taller plants lose a larger fraction of their aboveground biomass than shorter ones and may require more time to recover. Height is also associated with plant life span and time till maturation; therefore frequent disturbances pose a limitation to the development of large plants. In our experiment, biomass removal, prior to and after sowing, was carried out similarly for all treatments (L.N. personal observation). Therefore, the placement of weed seeds and the location and availability of nutrients to plants was mainly determined by tillage intensity. However, Fried et al. (2012) found that tillage depth and number of tillage passes explained the mean height of arable weed communities across France. In their data set, the average height of weed species was lower in mouldboard plowed fields compared to minimum tillage fields. The trade-offs and relationships between all traits are not fully understood. In our case, we suggest that the lower CWM of maximum height in NT plots could be a result of the occurrence of small stature species selected indirectly via the effect of tillage over other traits. SLA is a trait related to light acquisition. Higher SLA values have been associated with higher relative growth rates, shorter leaf life spans and with the production of litter with rapid decomposition

rates (Cornelissen et al., 2003; Garnier and Navas, 2012). These features may define a competitive strategy for weeds (Navas, 2012). In addition, higher SLA values are related to resource-rich environments (Cornelissen et al., 2003). In our experiment, the CWM of SLA was slightly higher in MT and NT plots compared to CT plots. These differences may be explained by the shifts in resource availabilityand stratification caused by tillage. Data from the same experiment also showed that from 1985 to 2005 the concentration of soil organic carbon in the first 10 cm of soil was higher in NT plots than in MT plots, and that CT plots had the lowest values (Hernanz et al., 2009). Also, higher values of nitrogen content in the first 15 cm of soil were observed in NT and MT plots than in CT plots (Hernanz et al., 2001). A relationship between SLA and nutrient content has been found in other systems: in eastern Australia, Wright et al., (2002) found perennial species with lower SLAvalues in dryland and infertile soils; in herbaceous communities of Mediterranean rangelands the CWM of SLA responded positively to a soil gradient from shallow to deep (Bernard-Verdier et al., 2012); and in a 27-year study in New Zealand grasslands, large CWM values of SLA were found alongside the use of fertilizer (Laliberté et al., 2012). 4.2. Response of trait diversity of weed communities to tillage intensity over time Our results are consistent with the hypothesis suggesting that disturbances may promote the coexistence of different phenotypes,

E. Hernández Plaza et al. / Agriculture, Ecosystems and Environment 207 (2015) 101–108

increasing trait diversity within a community (Grime, 2006). Each of the analyzed metrics provided information about a different facet of functional diversity, but in all cases conventional tillage, the largest degree of soil disturbance, resulted in greater response trait diversity for the traits considered. Specifically, weed communities from CT plots occupied a larger fraction of the functional space filled by weed communities in the experiment, had a greater dispersion of phenotypes and a more even distribution of species’ abundance and traits within this functional space. Also, the divergence between the trait values of the most abundant species is higher in CT plots than in MT or NT plots (higher FDiv in CT plots). These results may be explained by the idiosyncrasy of our disturbance treatments in comparison to other studies where disturbance implied differences in biomass removal (Carmona et al., 2012; Gerisch et al., 2012; Laliberté et al., 2013). As previously explained, the rate of removal of established individuals was similar in all three treatments, whereas weed seed placement and nutrient availability differed. The analysis of FRic revealed that the breadth of environmental conditions that are suitable for weed species are much lower in NT and MT systems than in CT systems. In relation to regenerative traits, NT systems may favor low seed sizes and greater seed outputs, whereas the opposite strategies may also have an advantage in CT systems. FRic has been shown to decrease across a large gradient of disturbance from woodlands to arable lands (Pakeman, 2011). However, within arable lands our results showed that sites subjected to the highest amount of soil disturbance (CT plots) hold more strategies in terms of seed size and seed output, leading to larger FRic and FDis values. This is in agreement with the relationship between FDis and the CWM of seed weight and the CWM of the number of seeds. FDis increases in NT plots as the CWM of seed weight also increases, whereas in MT and CT plots, FDis peaks when the CWM of seed weight attains medium values. In NT plots the most advantageous regenerative strategy is to have low seed sizes but years/plots in which other strategies appear FDis increases. In our experiment there were about 10 species which represented the main species recorded over the 26-year experimental period (Table S2), while the presence and abundance of other species were more variable. FRic does not take species abundance into account, so large FRic scores could result from rare species holding extreme trait values (Laliberté and Legendre, 2010). FDis reflects the dispersion of phenotypes within the functional space and as it incorporates species abundance, a comparison with FRic suggests that the trait values of less frequent species contributed more to the differences in the variation in response trait diversity between NT and MT systems. In CT plots the abundance of strategies within the functional space is more evenly distributed that in MT or NT plots at the end of the experimental time. This is in agreement with results from Pakeman (2011) where FEve increases with a gradient of disturbance from woodlands to arable fields. In their work, this was indicative of the trade-off between disturbance and competition in structuring plant communities. This may not be the explanation in our system, where biotic interactions among the weeds and wheat were similar across treatments. In our case, the absence of soil disturbance (NT plots) could be acting as a filter to reduce the abundance of certain strategies especially in relation to regenerative traits. The simple existence of different strategies does not assure that every phenotype realizes a substantive role in the agroecosystem. An even abundance of traits allows functions and services provided by distinct phenotypes to be more fully developed. Although we have only focused on traits directly related to the growth and persistence of weeds, these may be correlated with other traits which hold a role in supporting biological diversity within agroecosystems, as we explain below. In our study FDiv relates to whether the most abundant weed species possess distinct traits. The increase in divergence of

107

functional traits between weeds and crops, related to resource use, has been proposed as a mechanism to reduce the competitive effect of weeds on crops (Navas, 2012). Our analysis did not include the crop, therefore the greater FDiv achieved in CT plots may be related to a reduction of competitive interactions between weed species, perhaps due to an exploitation of different regeneration niches. Nevertheless, the reduction in functional divergence found throughout the experimental period is the result of a convergence of trait values, leading to more homogenous (and competitive) weed communities. This occurs specifically in relation to vegetative traits, as shown by the negative correlation between FDiv and the CWM of maximum height (and a similar tendency for SLA) for all tillage systems, and resulting in taller weed communities with greater SLA. This raises the question about the direct implication of management practices on crop–weed interactions. In this work, we refer to the response-effect framework (Lavorel and Garnier, 2002; Suding et al., 2008). Communities respond to environmental conditions via response traits and resultant communities affect ecosystem functioning via effect traits. We showed that differences in tillage intensity lead to shifts in the mean, the range and the distribution of response trait values within weed communities. However, response and effect traits are not necessarily the same (Suding et al., 2008), and the next step is to discern how traits from the resultant weed communities will affect agroecosystem functioning, specifically in relation to crop–weed competition and to the provision of ecosystem services. In our case, it could be that more strategies, in term of seed size and seed output, do not represent an increase in the competitive effect of weed communities, but that species filtered by their response traits may hold values for other traits having a greater impact on the agroecosystem. Finally we want to draw attention to the methodology used in this work. As previously noted, we have obtained trait values from the literature and public databases. These values represent static measures of traits and do not take into account that selection pressures operate at the individual level, and may, in time, lead to differences in trait values between populations that are under different environmental conditions. In this work, variation in weed community composition between different tillage systems may be thought to be due to species different responses to tillage intensity as a function of their trait values. Therefore we think, in this case, that the use of trait values from the literature may be valid. Another possible drawback of using trait values from published sources is that some floras are not well represented and missing trait values may produce a bias in the results. We lack some traits values for eleven species (Table S2), but this may not be an issue for our work. Species with missing trait values were all recorded in low abundance and for only a few years of the experiment, and we used functional diversity metrics which, except for FRic consider trait values weighted by species abundance. 5. Conclusions Our results show that, for the traits included in this study, tilling acts by opening up available niche space, thus permitting a wider range of weed phenotypes to coexist. Regenerative traits (seed weight and seed output) seem to be the most responsive traits in our work. There is greater variability in seed size and seed output inweed communities from conventional tillage plots than from no-tillage plots. Under the conditions of our experiment, no tillage systems favor species with lower seed sizes and greater seed output whereas conventional tillage systems allow for a higher range of strategies to co-occur. Our results also showed that the relative abundance of weed species with higher stature and smaller seed size has increased throughout the years in all tillage systems. Considering management practices as crop abiotic factors (Navas, 2012) acting on different weed response traits may serve to

108

E. Hernández Plaza et al. / Agriculture, Ecosystems and Environment 207 (2015) 101–108

link the ecology of agricultural weeds to the broader body of ecological theory. The goal is to define which environmental conditions are influenced by each management practice and consequently, which response traits are potentially affected. Tillage systems differ in the intensity of soil disturbance, affecting weed seed placement and nutrient availability in the soil profile. If we aim to provide management advice, the final objective will be to identify the effect traits, those that affect agroecosystem functioning, from the resultant weed communities. Specifically those related to crop-weed competition and the provision of ecosystem services by weeds. Acknowledgments This work has been funded by FEDER (European Regional Development Funds) and the Spanish Ministry of Economy and Competitiveness funds (Projects AGL2012-33736 and AGL201239929-C03-01). We are grateful to the Weed Science Group from El Encín for managing the long-term experiment. We thank Jane Morrison for her careful English revision and Fernando Bastida for help with Table S2. Suggestions and commentaries from two anonymous reviewers greatly improved the former version of this article. References Bernard-Verdier, M., Navas, M.-L., Vellend, M., Violle, C., Fayolle, A., Garnier, E., 2012. Community assembly along a soil depth gradient: contrasting patterns of plant trait convergence and divergence in a Mediterranean rangeland. J. Ecol. 100, 1422–1433. Bruggisser, O.T., Schmidt-Entling, M.H., Bacher, S., 2010. Effects of vineyard management on biodiversity at three trophic levels. Biol. Conserv. 143, 1521– 1528. Burnham, K.P., Anderson, D.R., 2002. Model Selection and Multimodel Inference. A Practical Information—Theoretic Approach. Springer-Verlag, New York, USA. Carmona, C.P., Azcárate, F.M., de Bello, F., Ollero, H.S., Lepš, J., Peco, B., 2012. Taxonomical and functional diversity turnover in Mediterranean grasslands: interactions between grazing, habitat type and rainfall. J. Appl. Ecol. 49, 1084–1093. Castro, H., Lehsten, V., Lavorel, S., Freitas, H., 2010. Functional response traits in relation to land use change in the Montado. Agric. Ecosyst. Environ. 137, 183–191. Cornelissen, J.H.C., Lavorel, S., Garnier, E., Díaz, S., Buchmann, N., Gurvich, D.E., Reich, P.B., Steege, H.t., Morgan, H.D., van der Heijden, M.G.A., Pausas, J.G., Poorter, H., 2003. A handbook of protocols for standardised and easy measurement of plant functional traits worldwide. Aust. J. Bot. 51, 335–380. Cornwell, W.K., Schwilk, D.W., Ackerly, D.D., 2006. A trait-based test for habitat filtering: convex hull volume. Ecology 87, 1465–1471. Chauhan, B.S., Gill, G., Preston, C., 2006. Influence of tillage systems on vertical distribution: seedling recruitment and persistence of rigid ryegrass (Lolium rigidum) seed bank. Weed Sci. 54, 669–676. Diaz, S., Cabido, M., Casanoves, F., 1998. Plant functional traits and environmental filters at a regional scale. J. Veg. Sci. 9, 113–122. Fried, G., Kazakou, E., Gaba, S., 2012. Trajectories of weed communities explained by traits associated with species’ response to management practices. Agric. Ecosyst. Environ. 158, 147–155. Garnier, E., Navas, M.-L., 2012. A trait-based approach to comparative functional plant ecology: concepts, methods and applications for agroecology. A review. Agron. Sustain. Dev. 32, 365–399. Gerisch, M., Agostinelli, V., Henle, K., Dziock, F., 2012. More species, but all do the same: contrasting effects of flood disturbance on ground beetle functional and species diversity. Oikos 121, 508–515. Grime, J.P., 2006. Trait convergence and trait divergence in herbaceous plant communities: mechanisms and consequences. J. Veg. Sci. 17, 255–260. Gunton, R.M., Petit, S., Gaba, S., 2011. Functional traits relating arable weed communities to crop characteristics. J. Veg. Sci. 22, 541–550. Hernanz, J.L., López, R., Sánchez-Girón, V., Navarrete, L., 2001. Evolución de la materia orgánica y otras propiedades químicas del suelo en experimentos a largo plazo con tres sistemas de laboreo. I Congreso de Agroingeniería, Valencia, Spain. Hernanz, J.L., Sánchez-Girón, V., Navarrete, L., 2009. Soil carbon sequestration and stratification in a cereal/leguminous crop rotation with three tillage systems in semiarid conditions. Agric. Ecosyst. Environ. 133, 114–122. Kyle, G., Leishman, M.R., 2009. Plant functional trait variation in relation to riparian geomorphology: the importance of disturbance. Austral Ecol. 34, 793–804. Laliberté, E., Legendre, P., 2010. A distance-based framework for measuring functional diversity from multiple traits. Ecology 91, 299–305. Laliberté, E., Shipley, B., Norton, D.A., Scott, D., 2012. Which plant traits determine abundance under long-term shifts in soil resource availability and grazing intensity? J. Ecol. 100, 662–677.

Laliberté, E., Norton, D.A., Scott, D., 2013. Contrasting effects of productivity and disturbance on plant functional diversity at local and metacommunity scales. J. Veg. Sci. 24, 834–842. Maire, V., Gross, N., Börger, L., Proulx, R., Wirth, C., Pontes, L.d.S., Soussana, J.-F., Louault, F., 2012. Habitat filtering and niche differentiation jointly explain species relative abundance within grassland communities along fertility and disturbance gradients. New Phytol. 196, 497–509. Lavorel, S., Garnier, E., 2002. Predicting changes in community composition and ecosystem functioning from plant traits: revisiting the Holy Grail. Funct. Ecol. 16, 545–556. Marshall, E.J.P., Brain, P., 1999. The horizontal movement of seeds in arable soil by different soil cultivation methods. J. Appl. Ecol. 36, 443–454. Marshall, E.J.P., Brown, V.K., Boatman, N.D., Lutman, P.J.W., Squire, G.R., Ward, L.K., 2003. The role of weeds in supporting biological diversity within crop fields. Weed Res. 43, 77–89. Mason, N.W.H., Mouillot, D., Lee, W.G., Wilson, J.B., 2005. Functional richness, functional evenness and functional divergence: the primary components of functional diversity. Oikos 111, 112–118. Moles, A.T., Westoby, M., 2006. Seed size and plant strategy across the whole life cycle. Oikos 113, 91–105. Moonen, A.C., Bàrberi, P., 2008. Functional biodiversity: an agroecosystem approach. Agric. Ecosyst. Environ. 127, 7–21. Navas, M.L., 2012. Trait-based approaches to unravelling the assembly of weed communities and their impact on agro-ecosystem functioning. Weed Res. 52, 479–488. Oveisi, M., Rahimian Mashhadi, H., Yousefi, A.R., Alizade, H., Baghestani, M.A., Gonzalez-Andujar, J.L., 2013. Predicting maize yield in a multiple species competition with Xanthium strumarium and Amaranthus retroflexus: comparing of approaches to modeling herbicide performance. Crop Prot. 45, 15–21. Pakeman, R.J., 2011. Functional diversity indices reveal the impacts of land use intensification on plant community assembly. J. Ecol. 99, 1143–1151. Peco, B., Carmona, C.P., de Pablos, I., Azcárate, F.M., 2012. Effects of grazing abandonment on functional and taxonomic diversity of Mediterranean grasslands. Agric. Ecosyst. Environ. 152, 27–32. Petit, S., Boursault, A., Le Guilloux, M., Munier-Jolain, N., Reboud, X., 2011. Weeds in agricultural landscapes. A review. Agron. Sustain. Dev. 31, 309–317. Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., R Development Core Team, 2012. nlme: Linear and Nonlinear Mixed Effects Models. R package version 3.1–103. http:// CRAN.R-project.org/package=nlme. Pinke, G., Karacsony, P., Czucz, B., Botta-Dukat, Z., Lengyel, A., 2012. The influence of environment, management and site context on species composition of summer arable weed vegetation in Hungary. Appl. Veg. Sci. 15, 136–144. R Development Core Team, 2011. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. http:// www.R-project.org/. Shipley, B., Vile, D., Garnier, É., 2006. From plant traits to plant communities: a statistical mechanistic approach to biodiversity. Science 314, 812–814. Spokas, K., Forcella, F., Archer, D., Reicosky, D., 2007. SeedChaser: vertical soil tillage distribution model. Comput. Electron. Agric. 57, 62–73. Storkey, J., Moss, S.R., Cussans, J.W., 2010. Using assembly theory to explain changes in a weed flora in response to agricultural intensification. Weed Sci. 58, 39–46. 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. Lond. Ser. B 279, 1421–1429. Suding, K.N., Lavorel, S., Chapin, F.S., Cornelissen, J.H.C., Díaz, S., Garnier, E., Goldberg, D., Hooper, D.U., Jackson, S.T., Navas, M.-L., 2008. Scaling environmental change through the community-level: a trait-based responseand-effect framework for plants. Global Change Biol. 14, 1125–1140. Swanton, C.J., Shrestha, A., Knezevic, S.Z., Roy, R.C., Ball-Coelho, B.R., 2000. Influence of tillage type on vertical weed seedbank distribution in a sandy soil. Can. J. Plant Sci. 80, 455–457. Villéger, S., Mason, N.W.H., Mouillot, D., 2008. New multidimensional functional diversity indices for a multifaceted framework in functional ecology. Ecology 89, 2290–2301. Vogeler, I., Rogasik, J., Funder, U., Panten, K., Schnug, E., 2009. Effect of tillage systems and P-fertilization on soil physical and chemical properties, crop yield and nutrient uptake. Soil Till. Res. 103, 137–143. Weiher, E., Freund, D., Bunton, T., Stefanski, A., Lee, T., Bentivenga, S., 2011. Advances, challenges and a developing synthesis of ecological community assembly theory. Proc. R. Soc. Lond. Ser. B. 366, 2403–2413. Westoby, M., 1998. A leaf-height-seed (LHS) plant ecology strategy scheme. Plant Soil 199, 213–227. Westoby, M., Falster, D.S., Moles, A.T., Vesk, P.A., Wright, I.J., 2002. Plant ecological strategies: some leading dimensions of variation between species. Ann. Rev. Ecol. Syst. 33, 125–159. Wright, I.J., Westoby, M., Reich, P.B., 2002. Convergence towards higher leaf mass per area in dry and nutrient-poor habitats has different consequences for leaf life span. J. Ecol. 90, 534–543. Wright, I., Reich, P., Westoby, M., Ackerly, D., Baruch, Z., Bongers, F., Cavender-Bares, J., Chapin, F., Cornelissen, J., Diemer, M., Flexas, J., Garnier, E., Groom, P., Gulias, J., Hikosaka, K., Lamont, B., Lee, T., Lee, W., Lusk, C., Midgley, J., Navas, M.-L., Niinemets, U., Oleksyn, J., Osada, N., Poorter, H., Poot, P., Prior, L., Pyankov, V., Roumet, C., Thomas, S., Tjoelker, M., Veneklaas, E., Villar, R., 2004. The worldwide leaf economics spectrum. Nature 428, 821–827.