Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

A Functional Characterisation of a Wide Range of Cover Crop Species: Growth and Nitrogen Acquisition Rates, Leaf Traits and Ecological Strategies

  • Hélène Tribouillois,

    Affiliation INRA, UMR AGIR, 24 Chemin de Borde Rouge – Auzeville, CS 52627, 31326 Castanet-Tolosan Cedex, France

  • Florian Fort,

    Affiliation INRA, UMR AGIR, 24 Chemin de Borde Rouge – Auzeville, CS 52627, 31326 Castanet-Tolosan Cedex, France

  • Pablo Cruz,

    Affiliation INRA, UMR AGIR, 24 Chemin de Borde Rouge – Auzeville, CS 52627, 31326 Castanet-Tolosan Cedex, France

  • Raphaël Charles,

    Affiliation Agroscope, Institute of Plant Production Sciences, 50 Route de Duillier, CP 1012, CH-1260 Nyon 1, Switzerland

  • Olivier Flores,

    Affiliation Université de la Réunion/CIRAD, UMR—Peuplements Végétaux et Bioagresseurs en Milieu Tropical, 97410 Saint Pierre, France

  • Eric Garnier,

    Affiliation Centre d’Ecologie Fonctionnelle et Evolutive (UMR 5175), CNRS - Université de Montpellier - Université Paul-Valéry Montpellier – EPHE, 1919, route de Mende, 34293 Montpellier Cedex 5, France

  • Eric Justes

    Affiliation INRA, UMR AGIR, 24 Chemin de Borde Rouge – Auzeville, CS 52627, 31326 Castanet-Tolosan Cedex, France


Cover crops can produce ecosystem services during the fallow period, as reducing nitrate leaching and producing green manure. Crop growth rate (CGR) and crop nitrogen acquisition rate (CNR) can be used as two indicators of the ability of cover crops to produce these services in agrosystems. We used leaf functional traits to characterise the growth strategies of 36 cover crops as an approach to assess their ability to grow and acquire N rapidly. We measured specific leaf area (SLA), leaf dry matter content (LDMC), leaf nitrogen content (LNC) and leaf area (LA) and we evaluated their relevance to characterise CGR and CNR. Cover crop species were positioned along the Leaf Economics Spectrum (LES), the SLA-LDMC plane, and the CSR triangle of plant strategies. LA was positively correlated with CGR and CNR, while LDMC was negatively correlated with CNR. All cover crops could be classified as resource-acquisitive species from their relative position on the LES and the SLA-LDMC plane. Most cover crops were located along the Competition/Ruderality axis in the CSR triangle. In particular, Brassicaceae species were classified as very competitive, which was consistent with their high CGR and CNR. Leaf functional traits, especially LA and LDMC, allowed to differentiate some cover crops strategies related to their ability to grow and acquire N. LDMC was lower and LNC was higher in cover crop than in wild species, pointing to an efficient acquisitive syndrome in the former, corresponding to the high resource availability found in agrosystems. Combining several leaf traits explained approximately half of the CGR and CNR variances, which might be considered insufficient to precisely characterise and rank cover crop species for agronomic purposes. We hypothesised that may be the consequence of domestication process, which has reduced the range of plant strategies and modified the leaf trait syndrome in cultivated species.


Cover crops are defined as plant covers in agrosystems whose ecosystem services reduce negative environmental impacts of agriculture, e.g. avoid groundwater nitrate pollution, protects soils against erosion or improve their fertility. Currently, cover crops are sown to i) reduce nitrate leaching by maximising nitrogen (N) acquisition before the drainage period (i.e. the “nitrate catch crop effect”) [1,2]; ii) produce “green manure” by N recycling for later release to the subsequent cash crop [3]; iii) produce biomass, which is important for soil carbon sequestration; and iv) reduce weed growth and soil seed-bank formation during the fallow period of “bare soil” between two main cash crops [1,4]. The amount of N available for the subsequent cash crop is influenced by the amount of N acquired by the cover crop, the residue quality and its decomposition rate [4]. Biomass production and N acquisition determine the C:N ratio of cover crop residues which controls the dynamics of N release from residues incorporated into the soil and then the green manure effect [5,6]. Despite the current strong interest in cover crops in cropping systems only little scientific-based information is available to choose cover crop species in a given cash crop succession on objective and sound bases. To our knowledge, only few scientific studies have been conducted to propose a method to select cover crop according to the ecosystem functions targeted. An example is the study of Damour et al. [7] which deals with tropical cover crops in banana. Traditional approaches used in agronomy are based on field experiments involving labour and time-consuming measurements (i.e. aerial and root biomass, nutrient acquisition). Since such approaches are intractable when screening a large number of species is required, we assessed whether a trait-based approach widely used in plant ecology [810] could be relevant for characterising the behaviour of cover crops species. We focused on leaf functional traits because they are robust, easily and quickly measurable on a wide range of species and environments.

Ecological theories [11,12] and empirical works [13,14] showed that, in wild species, relative growth rate (RGR) is related to various leaf functional traits [15,16] and it is one of the components underlying plant strategies [11,12]. Ecologists have established the existence of a fundamental trade-off between traits allowing rapid resource capture and efficient resource conservation. Specific leaf area (SLA) and leaf N concentration (LNC) appear to be effective predictors of a species’ position along the “Leaf Economic Spectrum” (LES) that captures the gradient from “acquisitive” (high SLA and LNC) to “conservative” (low SLA and LNC) species [17]. As proposed by Wilson et al. [18], leaf dry matter content (LDMC) is complementary to SLA for assessing plant functioning and resource acquisition/conservation strategies. Indeed acquisitive strategies are associated with high SLA and low LDMC whereas conservative strategies are associated with low SLA and high LDMC [1921]. Moreover, LDMC is more strongly related to litter decomposition than SLA: litters with high LDMC tend to decompose more slowly than litters with low LDMC, leading to litter accumulation [22], thereby potentially delaying N release and then the green manure effect from cover crop residues. As emphasised above, another important potential service attributed to cover crops is limiting weed proliferation by a rapid growth, soil coverage and N acquisition, inducing a strong competition for resources which relates to their competitive abilities. Grime (1977) identified three main avenues for species’ adaptive specialisations that reflect trade-offs between their capacities to be competitive (C), stress tolerant (S) and ruderal (R). Locating species within the CSR space, and thereby assessing their relative competitive ability, normally requires measuring a large number of traits [13,23]. However, Pierce et al. [24] recently developed a new method that uses only three leaf traits-SLA, LDMC and leaf area (LA)- to assess a species’ position in the CSR space.

The aim of our study was to assess whether conceptual and methodological developments of the trait-based approach to plant functioning initially developed for wild species in natural ecosystems, could be used to characterise and classify the resource-use and ecological strategies of domesticated species such as cover crops. Species from various botanical families were analysed all together as recommend by Diaz et al. [25]. We made the hypothesis that these ecological strategies could be assessed at the cover level for cultivated species in monocrop. We hypothesised that, over cover crop species, a range of resource acquisition/conservation strategies corresponding to a range of growth and N acquisition rates could be investigated using leaf functional traits. It can be assumed that species with acquisitive strategies having high SLA and low LDMC would ensure rapid development of plant growth, N acquisition and soil covering after summer sowing, while conservative species having low SLA and high LDMC would ensure slower growth and N acquisition during and after the fallow period.

To test these hypotheses, we performed a functional characterisation of a range of cover crop using i) leaf functional traits and ii) agronomical indicators to evaluate growth and N acquisition of species such as the crop growth rate (CGR) and the crop N acquisition rate (CNR). The quantity of biomass produced during the growing period strongly depends on the CGR and on the amount of N acquired by the plants depends on the CNR [16,26]. We assumed that CGR and CNR are two relevant complementary indicators reflecting the ability of cover crops to function as a nitrate catch crop and a producer of green manure for the subsequent cash crop [1,2]. Leaf traits were measured at two experimental locations to test the sensitivity of the leaf functional trait approach to differences in environmental conditions (soils and climates in particular). To better understand where these cover crop species lie in previously identified axes of resource-use strategy variation [27] and verify the robustness of the method, these experimental data were compared to those from two large published data sets established for a large number of wild species (S1 Text): (i) one corresponding to the well-known Leaf Economics Spectrum [17] and (ii) the second one describing the SLA and LDMC relationship corresponding to the trade-off between acquisitive and conservative resource [28]. Finally, we used the methodology of Pierce et al. [24] to locate the cover crop species on the CSR space [11] and characterize differences in their ecological strategies.

Materials and Methods

Growth conditions

Two field experiments were conducted from August to October 2012 on research public areas at the INRA site in Auzeville-Tolosane (43°31’ N, 1°30’ E), southwest France and the Agroscope site in Nyon (46°22’ N, 6°14’ E), Switzerland. This study was permitted by the both institutions of INRA and Agroscope and field studies did not involve endangered or protected species. The French and Swiss sites had significant different mean temperatures of 18.4°C and 16.7°C during this period, respectively, and both have deep clay loam soils. Species were sown in monocrops at their recommended plant density and to constitute homogeneous dense plant covers, in plots of 10 m2 on 3rd August 2012 in Switzerland and on 16th August 2012 in France. We assumed that the two experiments conducted in non-limiting water and N availability in France and Switzerland constitute a true complete fully randomized design with two replicates corresponding to the two environmental conditions. Nonetheless, we took care to obtain very homogeneous plant covers in order to have the same range of plant intraspecific competitions for each species. Moreover, because species were grown in non-water- and non-N-limiting conditions, as proposed by Grime & Hunt (1975) [29], we can assume that these conditions allowed to both i) minimize “genotype x environment interactions” and ii) compare the potential abilities of species independently of environmental conditions. Irrigation was performed regularly and 100 kg ha-1 of N from fertilizer was provided to non-N fixing species at sowing. It was assumed that the symbiotic fixation of N2 of Fabaceae was effective thanks to the presence in soils of efficient Rhyzobium and observed good nodulation in roots for all legumes which insured optimal growth conditions and N acquisition.

Species characteristics, crop growth and N acquisition rates

Thirty-six taxa (34 species and two varieties of Vicia faba and Pisum sativum) were selected as cover crop species based on their supposed ability to provide ecosystem services during fallow period [1,4,30]. These species are adapted to grow in French and European soils and climates and match the range of species commonly used as cover crops by farmers. Most of these species are not used as main crop in current arable cropping systems and are non-host species for pests and diseases of the main cash crops. The species were also chosen to represent a diversity of botanical families: Fabaceae, Brassicaceae, Poaceae, Asteraceae, Polygonaceae and Hydrophyllaceae. Species from the Fabaceae family are commonly used to provide green manure effect whereas Brassicaceae or Poaceae are more often used to catch nitrate from soil [3,4]. These taxa were also chosen to ensure a wide range of growth and N acquisition rates, phenology, shoot architecture and taxonomic diversity. All species had an annual life cycle except four biannual or triennial grassland species: Secale multicaule, Medicago lupulina, Melilotus officinalis and Onobrychis viciifolia. The Poaceae family was separated into two groups according to photosynthetic pathway (C3 versus C4).

We measured CGR and CNR as two complementary simple indicators-conversely to relative growth rate or RGR which is more complex to measure- to estimate plant abilities to growth and acquire N rapidly of a wide range of different species and at a single sampling date (corresponding to approximately the middle of the vegetative phase of all cover crop species). Furthermore, we measured these two indicators at the cover level, and not at the plant level, in order to take into account, in actual agronomical conditions, intraspecific competitions between plants for abiotic resources acquisition. The calculation of CGR makes it possible to compare homogeneous covers all species over a given interval of thermal time, as used in some agronomic studies [31,32]. Degree-days (DD), defined as the sum of degrees by which each days’ mean air temperature was higher than the base temperature, were calculated from the date of sowing (T0) until the date of the sampling (T1) (12 October and 28 September 2012 for the French and Swiss sites, respectively). We assumed a same base of temperature of 0°C for all species because the growth occurred in summer and early autumn, when the minimum temperature remains higher than 10°C, a temperature probably higher that the base temperature of all species. At T1, DD equalled 1160 DD and 1118 DD in the French and Swiss sites, respectively, allowing us to consider the two sites as two true statistical replicates.

Finally, CGR (in kg DM ha-1 DD-1) and CNR (in g N ha-1 DD-1) were calculated for each species at both sites using the following equations: where DM1 and DM0 are shoot dry matters (DM) (kg ha-1) and Nacq1 and Nacq0 are the amounts of N acquired by plants in shoots (g ha-1) at T1 and T0, respectively. DM0 and Nacq0 are equal to 0 because they correspond to DM and Nacq at the sowing date. For each site and species, DM was harvested on one square meter and measured on two independent samples. N acquired was estimated by calculating total shoot N content as shoot biomass (DM) x total-shoot N concentration measured at T1 by elemental analyses based on the Dumas method (Elementar MicroVario Cube, Germany).

Measurement of leaf traits

At each experimental site, leaf traits were measured at the vegetative stage [33] on 20 individual plants per species. Immediately after harvest, the 20 plant samples were placed in demineralised water and stored in the dark at 4°C for at least seven hours. Water was regularly sprayed on leaves to ensure that they were water-saturated [28,33]. Traits were measured on the youngest mature leaf (lamina and petiole) free from herbivore or pathogen damage. Leaves were weighed to obtain the water-saturated leaf mass and scanned with a manual scanner (Scanjet 3770, Hewlett-Packard) to measure their surface area. Leaves were then oven-dried at 60°C for 48h and weighed to measure their dry mass. Finally, leaves were ground and total-N content was measured.

LDMC was calculated as the ratio of dry mass (mg) of a leaf to its water-saturated fresh mass (g) and is expressed in mg g-1. SLA was calculated as LA (m2) divided by its dry mass (kg) and is expressed in m2 kg-1. LNC was calculated as the total amount of N (mg) per unit leaf dry mass (g) and is expressed in mg g-1.

Data analysis

Co-variations among traits were evaluated using principal component analysis (PCA). We assessed the position of cover crop species along the LES axis of variation using the relationship between SLA and LNC established by Wright et al. [17]. We did the same for the SLA-LDMC relationship established for 1392 wild herbaceous species based on a compilation of published data (S1 Text). The position of species on the CSR triangle was assessed with ternary coordinates calculated from LA, LDMC and SLA values, as described by Pierce et al. [24]. Predictive regressions of this method translated PCA coordinates in multidimensional space into coordinates in a strategy triangle.

As a first analysis of the possible effect of domestication and breeding on species traits, we compared the trait values measured on cover crops at the experimental sites to those for annual wild species of the same genus available in the TRY database (Kattge et al. 2011 [34] and S2 Text). Leaf traits were extracted from the database for all native species of the same genera as those used in our experiment. For each trait, the mean value for all species within the wild genus was calculated and compared to the values obtained for the corresponding genus in our experiment. On average, four congeneric wild species were available within a wild genus for that comparison. For buckwheat (Polygonum fagopyrum or Fagopyrum esculentum), we used the wild species from the Polygonum genus.

Statistical analyses, correlations and generalized linear modelling (GLM) with stepwise regression procedure (stepAIC, P<0.05) were performed using the statistical software package R (version 2.14.0).


Trait values and rankings are maintained between the two sites

There were no significant differences between the mean values of the four leaf traits measured in France and Switzerland (Student’s t-tests P-values: 0.10 for LNC, 0.42 for LDMC, 0.89 for LA and 0.90 for SLA). Consequently, species ranking based on these traits did not differ between the two sites: Spearman correlation coefficients between values obtained in France and Switzerland were 0.93, 0.87, 0.77 and 0.91 (P<0.0001) for SLA, LDMC, LNC and LA, respectively. Similar results were observed for CGR and CNR: there was no significant difference either in mean values (P = 0.23 and 0.21 respectively) or in species ranking between both sites, with Spearman correlation coefficients of 0.67 (P = 0.0001) and 0.62 (P = 0.0003), respectively.

Species’ functional characterisation and growth and N acquisition rates

CGR and CNR varied respectively from 0.50 kg DM ha-1 DD-1 and 12.4 g N ha-1 DD-1 for Medicago lupulina to 5.66 kg DM ha-1 DD-1 and 155.8 g N ha-1 DD-1 for Helianthus annuus. Overall, Brassicaceae and Asteraceae had high CGR and CNR values, except Camelina sativa which had lower values (Table 1). CGR (in kg DM ha-1 DD-1) were significantly positively correlated (R2 = 0.73 and P<0.0001) with CNR (in g N ha-1 DD-1) using a logarithmic regression model (data not shown).

Table 1. Botanical family, taxonomic status, leaf functional traits values, crop growth rate (CGR) and crop N acquisition rate (CNR) of the 36 taxa studied.

Leaf traits also spanned a wide range of values: SLA varied from 14.3 m² kg-1 for Brassica rapa to 52.2 m2 kg-1 for Vicia villosa, with a mean of 31.9 m2 kg-1. LDMC varied from 78.9 mg g-1 for Brassica rapa to 209.6 mg g-1 for Sorghum bicolor var. sudanense, with a mean of 127.7 mg g-1. Although all species were grown under non-limiting N conditions, LNC also varied over a wide range, from 25.2 mg g-1 for Trigonella foenum-graecum to 63.4 mg g-1 for Vicia villosa. Finally, LA varied from 6.1 cm2 for Lupinus angustifolius to 222.8 cm2 for Helianthus annuus (Table 1). All families with more than two species (i.e. Brassicaceae, Fabaceae and Poaceae) had large variability in traits values.

Co-variations among plant traits

No significant correlation was detected between CGR and SLA, LDMC or LNC, the only significant positive relationship was found between LA and CGR, but the correlation was relatively weak (R2 = 0.42; P<0.001) (Table 2 and Fig. 1). For CNR, significant positive relationships were found but also relatively weak with LA (R2 = 0.35; P<0.001) and also with LDMC (R2 = 0.27; P = 0.001) (Table 2 and Fig. 1). These significant relationships were also obtained independently for both experimental sites (Table 2). The generalized linear models combining several leaf traits increased significantly R2 until 0.47 for CGR and 0.58 for CNR (Fig. 2a-b) compared to simple correlations.

Table 2. Pearson’s correlations between crop growth rate (CGR) and crop N acquisition (CNR) of cover crop species (n = 36) and leaf functional traits: specific leaf area (SLA), leaf dry matter content (LDMC), leaf nitrogen content (LNC) and leaf area (LA).

Fig 1. Relationships between crop growth rate (CGR) and leaf functional trait values: (a) leaf dry matter content (LDMC), (b) leaf area (LA); and relationships between crop N acquisition rate (CNR) and leaf functional trait values: (c) leaf dry matter content (LDMC), (c) leaf area (LA).

The significance of differences was assessed by Student’s t-tests. Values of traits and CGRs are means of both experimental sites.

Fig 2. Generalized linear models based on significant leaf traits (P<0.05) as input variables to fit a) crop growth rate (CGR) and b) crop nitrogen acquisition rate (CNR).

The equation of the CGR model is: CGR = 4.8E-2*SLA + 1.5E-2* LA and that for CNR is: CNR = 63 + 0.4*LA+ 1.5*SLA0.3*LDMC, where LDMC is leaf dry matter content (mg g-1), LA is leaf area (cm2), and SLA is specific leaf area (m2 kg-1).

The first two axes of the PCA of functional traits accounted for 80% of the variability in the data (49% and 31%, respectively). Axis 1 opposed LA to SLA and LNC (Fig. 3a). These traits strongly correlated with Axis 1 (Loading coefficients were -0.91 for SLA, -0.75 for LNC, and 0.74 for LA; all P<0.001). As a result, species with negative coordinates on this axis, such as Vicia villosa, V. benghalensis, V. sativa, Lens nigrican, Camelina sativa, had high SLA and LNC but low LA. In contrast, species with positive coordinates, such as Brassicaceae (e.g. Brassica rapa, Raphanus sativus, B. napus) had the opposite syndrome (Fig. 3b). Axis 2 was strongly and significantly positively correlated with LDMC (Pearson’s correlation coefficient was 0.89; P<0.001). Sorghum bicolor var. sudanense, Trigonella foenum-graecum and Onobrychis viciifolia had the highest positive coordinates on this axis, and Brassica rapa, Raphanus sativus and Vicia faba cv SSNS the lowest. All C3 Poaceae were located around the PCA centre and inside the Fabaceae ellipse, and the scatter of C3 Poaceae species in the former group was low. Both C4 Poaceae were located at a high positive position on Axis 2. Correlations between PCA axes and cover crops growth and N acquisition rates (CGR and CNR not active in PCA) were assessed (S1 Fig.): Axis 1 was significantly positively correlated with CGR but not with CNR, whereas Axis 2 was weekly negatively correlated with CGR and was more strongly correlated with CNR.

Fig 3. Principal component analysis (PCA) based on four functional traits measured on 36 cover crop species.

(a) Correlation circle between SLA (specific leaf area), LDMC (leaf dry matter content), LNC (leaf nitrogen content) and LA (leaf area) and loadings; coordinates position of CGR (crop growth rate) and CNR (crop nitrogen acquisition rate); (b) Botanical families and cover crop species position along the first two axes of the PCA; A: Asteraceae; B: Brassicaceae; F: Fabaceae; H: Hydrophyllaceae; P: Polygonaceae; PC3: C3 Poaceae; PC4: C4 Poaceae.

Functional strategies of cover crop species

Data from the worldwide data set used to establish general relationships involved in the leaf economics spectrum [17] show that for herbaceous species, SLA values range from 1.5 to 69.4 m2 kg-1 (mean: 19.7 m2 kg-1) and LNC values range from 4.5 to 63.6 mg g-1 (mean: 26.4 mg g-1) (Fig. 4). Cover crop species spanned 34% of the log SLA range and 35% of the log LNC range, and were located in the high SLA—high LNC zone of the LES relationship (Fig. 4). A significant positive correlation was found for cover crops species between log SLA and log LNC (R2 = 0.36 and P = 0.0001), and species from different cover crop families were not distinguishable from one another. Brassicaceae had the lowest SLA and the highest LNC, except for one species (Camelina sativa) (Fig. 4).

Fig 4. Position of cover crop species along grass and herbs of the Leaf Economic Spectrum (LES) (Wright et al. [17]) based on specific leaf area (SLA) and leaf nitrogen content (LNC) axes (the latter logarithmic).

Cover crop species were positioned on the SLA-LDMC plane based on a data set composed of wild herbaceous species (Fig. 5). LDMC of wild species composed of grasses and forbs varied between 42.2 and 714 mg g-1 (mean: 221 mg g-1), while their SLA varied from 1.7–94.4 m2 kg-1 (mean: 23.2 m2 kg-1). Cover crop species spanned 48% of this range for SLA and only 25% of the LDMC range. They were mostly located in the low LDMC zone of the plane (Fig. 5), and no relationship was found between SLA and LDMC in our data set (R2 = 0.0006 and P = 0.88). Species of Fabaceae, C3 Poaceae, Asteraceae, Polygonaceae and Hydrophyllaceae tended to be mixed in this space. In contrast, most Brassicaceae species had relatively low SLA and LDMC and were surprisingly close to succulent species. The two C4 Poaceae species clearly had higher LDMC than the other cover crop species.

Fig 5. Cover crop species position compared to the relationship between leaf dry matter content (LDMC) and specific leaf area (SLA) established for 1392 grass, forbs and succulents wild species (S1 Text).

Cover crop strategies corresponded to a restricted region of the CSR triangle: 21 taxa were located in the C-R range in a gradient from C to R strategies (Fig. 6). All Brassicaceae were classified as competitive, with C-axis coordinates from 60–96%, except Camelina sativa (35% on this axis). Helianthus annuus (Asteraceae) was also classified as competitive, at 71% on the C-axis, as expected due to its high CGR and CNR. Species from other families were mostly positioned between competitive and ruderal (47% and 46%, on average, along the C- and R-axes, respectively). Some species were positioned as ruderal, scoring up to 65% on the R-axis (e.g. Camelina sativa, Lens nigricans and Vicia villosa). Species generally scored low on the stress tolerance axis, with the highest values reached by the two C4 Poaceae (Sorghum bicolor var. sudanense and Setaria italica), but with a maximum value of only 22%. Correlations between Grime’s triangle axes and CGR and CNR showed that these two indicators were positively and significantly related to the C axis (R2 = 0.22; P = 0.004 for both) but negatively and significantly related to the S axis, especially for CNR (R2 = 0.18; P = 0.050 for CGR and R2 = 0.35; P<0.001 for CNR). Both CGR and CNR were negatively related to the R axis (R2 = 0.11 and R2 = 0.04 respectively), but only the relationship with CGR was weakly significant (P = 0.049 and P = 0.223 respectively).

Fig 6. Cover crop species position on Grime’s CSR triangle (Pierce et al. [24]).

C: competition, S: stress-tolerance, R: ruderality. Letters correspond to species identification (Table 1).


The functional-trait approach is robust for cover crop species

Results showed that leaf functional trait values and rankings were closely maintained among cover crop species grown under non-limiting water and N availabilities at two experimental sites with different soil and climate characteristics. These results are consistent with those of Al Haj Khaled et al. [35] and Kazakou et al. [36] who demonstrated the robustness of this approach for various types of herbaceous species. In our study, in which N fertiliser was added to both sites and legumes showed good nodulation allowing efficient N2 fixation, LNC ranking was maintained, which might not always be the case (Kazakou et al. [36]). The results also showed that significant relationships between traits and indicators of functions (CGR and CNR) were maintained for both experimental sites. This stability in rankings, trait values and the relationships between traits and functions confirms the robustness of leaf functional traits approach for characterising functional strategies of domesticated species such as cover crops.

Leaf traits related to cover crop growth and N acquisition rates

Axis 1 of the PCA of leaf traits showed a gradient ranging from species with small leaves, high LNC and high SLA to those with large leaves, low LNC and low SLA. This axis was positively correlated with CGR, suggesting that species with high LA, low SLA and low LNC are actually growing more quickly than those with low LA, high SLA and high LNC. However, we demonstrated that leaf area was the only leaf functional trait significantly correlated with CGR, showing that species with large leaves are growing more quickly than those with small leaves. This is consistent with the study of Hajek et al. [37] based on species used is forestry system. Contrary to our initial hypothesis, we did not find any correlation between CGR and SLA or LNC. This result suggests that, for cover crop species, there is no direct relationship between net photosynthesis expressed per unit leaf mass, captured by a combination of SLA and LNC [17], and the absolute growth rate of the crop cover (CGR). Leaf traits that influence growth rates of the aboveground cover (g ha-1 DD-1) therefore appear to differ from those that influence relative growth rate expressed on a plant biomass basis (RGR: g g-1 DD-1).

Axis 2 distinguished species according to their LDMC and was negatively correlated with CGR and especially with CNR. This axis indicates a gradient of species ability to acquire N (fast N uptake or high N2 fixation). As a result, the LDMC appeared to be the most efficient traits to estimate species ability to acquire N. Finally, the cover crop species positioned positively with axis 1 and negatively with axis 2 would be the species the most efficient as cover crops: they grow fast and acquire N rapidly after sowing and during crop establishment. This position corresponds to Brassicaceae and Asteraceae which is consistent with their highest CGR and CNR.

Finally, approximately half of the variance of CGR and CNR could be explained by a combination of several leaf functional traits. This value is relatively low and might be considered too coarse to characterize the properties of cover crop species for agronomic purposes.

Cover crop species position along functional and ecological strategies

With their high SLA and LNC, cover crop species are located in the “acquisitive” part of the LES, which is generally associated with a high photosynthetic rate, high respiration rate and rapid leaf turnover [17]. Although the cover crop species follow the relationship between log SLA and log LNC, the LES-plane did not allow to classify them into a gradient of growth and N acquisition abilities. Although cover crops spanned a relatively large range of the SLA-LDMC distribution we did not find any significant relationship between the two traits for the cover crop species screened (cf. [28]). Brassicaceae from our data set are located close to succulents, which was not expected because of the large difference in growth rates between the two groups of species (high for Brassicaceae and low for succulents). Despite their low SLA, Brassicaceae species had high CGR and CNR. They were therefore very efficient acquisitive species that produce much biomass and are specifically sought after as cover crops adapted to short fallow periods, in relation with their early growth during autumn. However, this is consistent with their low LDMC values, showing that LMDC seems more informative than the SLA to discriminate cover crop species strategies.

We showed that positioning species on the Grime’s triangle [13] was consistent for exploring differences between cover crop species; in fact, CGR and CNR were positively correlated with the C axis. As a consequence the most competitive cover crop species are those having the highest CGR and CNR. All Brassicaceae (except Camelina sativa) and Helianthus annuus were identified as highly competitive and thus corresponds to their ability to rapidly grow and acquire N after sowing (high CGR and CNR) and occupy the space needed to acquire light. Again according to the CSR triangle, most of the cover crop species seem not to tolerate stress, except for C4 Poaceae. This is in agreement with the fact that C4 plants, especially sorghum, are often tolerant to water stress [38,39]. Cover crop species with low CGR are closer to the ruderal strategy and those with low CNR which are more stress tolerant, based on the negative correlations with R and S axes. However, we found that Sorghum bicolor var. sudanense, which should be the most stress-tolerant, had one of the highest CGR and CNR, pointing to a relative lack of precision of the method to accurately predict the strategies of cover crop species.

A first hypothesis explaining a cover crop domestication syndrome

The hypothesis tested here is that domestication or breeding modified the values of leaf functional traits of cover crops and thus their leaf traits syndrome, i.e. the combination of leaf traits values for a given species. Examination of Fig. 5 suggests that constraints exerted by domestication have been stronger on LDMC than on SLA, as shown by the narrowest range of variation of the former trait in cover crops. We tested this hypothesis further by comparing leaf traits values of cover crops with those of their homologous wild annual species from the same genus taken from the TRY database. The analysis was done at the genus level because no data was available to confidently identify the wild ancestor of each cover crop species (Table 3). The comparison indicated that cover crop species presented systematically and significantly higher SLA and LNC but lower LDMC than wild species. Moreover, Fabaceae which did not receive any fertilizer in our experiment also presented a significant higher LNC than wild species (Table 3), suggesting that the comparison of leaf trait values between the TRY database and our experimental data is valid. This is also consistent with results found by Kazakou et al [36] who showed that SLA, LDMC and LA are not very sensitive to N fertilisation. The impact of domestication seems to have been stronger on LDMC and LNC than on SLA, based on the distance in trait values between cover crops and wild species: LDMC and LNC were respectively 37% lower and 48% higher in cover crops, while SLA was only 24% higher. This suggests domestication and breeding “forced” cover crop species towards low LDMC values, but not towards high SLA values, as also illustrated by the deformation of the SLA-LDMC relationship for cover crop species (Fig. 5). This could be related to cover crop species adaptation to different environments and reflect fundamental differences in the strategies they use to acquire, invest and use resources [40,41]. Similar results were illustrated for certain species by Gambino & Vilela (2011) [42], who found that domesticated primroses (Primula vulgaris) accumulated more biomass than wild ones and had higher SLA. In another study, domesticated cassava (Manihot esculenta) had a higher SLA and CO2 exchange rate and lower LDMC than wild ones [43]. Further tests of this potential differential effect of domestication on leaf trait values should be conducted on a broader range of species, with a particular emphasis on the comparison between crops and their wild ancestor.

Table 3. Leaf trait values of cover crop species obtained in the present study compared to trait values of congeneric wild species obtained from the TRY plant-trait database (Kattge et al. [34] and S2 Text).

Cover crop traits and ecosystem functions related to N management

Overall, the fact that LDMC of cover crops was reduced while at the same time LNC was increased in comparison to wild species indicates that cover crops are effective acquisitive species inducing rapid biomass production, ability for photosynthesis and N accumulation in their tissue. This leaf trait syndrome of cover crops illustrates their ability to provide both i) an efficient “nitrate catch crop” function by trapping rapidly most of the nitrate available in the soil (high CGR and CNR), and ii) a green manure function by increasing the amount of N acquired and accumulated in the tissues of the cover crop (high CNR). Moreover, the relatively low LDMC values of cover crops indicate that the N release for the subsequent cash crop will be favoured, since litters coming from leaves with low LDMC tend to decompose rapidly [22,44]. If so, cover crop species would tend to produce higher-quality residue, increasing the rate of N release after residue incorporation into the soil and then produce N green manure effect. Considering the decomposition of crop residues, our results seem consistent with García-Palacios et al. [45], who worked with some of the species tested in our experiments (Avena sativa, Helianthus annuus, Secale cereale, Sorghum sudanense). These authors showed that domesticated species produced a litter of higher quality than wild species of the same genus. Among the cover crop species considered in this study, Phacelia tanacetifolia, some Brassiceae and Fabaceae presented low LDMC suggesting that these species would be efficient to fulfil the N green manure function. In a modelling study, Justes et al. [6] demonstrated that the lower the C:N ratio and then the highest LNC, the more rapidly N is released from cover crop residues after their incorporation into the soil, reinforcing the green manure function of cover crops, in particular for legumes having high LNC and low LDMC.


Leaf functional traits allowed us to explore differences in the ability of cover crop species to grow fast and rapidly acquire N after sowing and during the cover establishment. Despite the wide range of CGR and CNR spanned by the studied species, they can be considered as belonging to the acquisition-use strategy as indicated by converging results from the three ecological methods tested. We demonstrated that LA and LMDC were correlated to crop growth and N acquisition rates and that combining several leaf traits allowed us to explain approximately half of the CGR and CNR variances. This might be considered insufficient to precisely characterise and rank the cover crop species for agronomic purposes. Furthermore, this study pointed out that LDMC values of cover crops was substantially lower than those of wild species. We hypothesised that domestication and breeding would have modified the leaf trait values of cultivated species as cover crops. Further studies are necessary to better understand the links between functional traits and the functions of cover crops, so as to improve the use of trait based approach in the perspective of choosing species in relation to targeted ecosystem functions.

Supporting Information

S1 Fig. Relationships between crop growth rate (CGR) and crop N acquisition rate (CNR) and strategy axes from principal component analysis (PCA): axis 1 and 2.

The significance of differences was assessed by Student’s t-tests. Values of traits and CNR and CGR are means of both experimental sites.


S1 Text. List of references of published data used to establish the SLA-LDMC relationship on 2176 wild species.


S2 Text. List of original references via the global TRY plant trait database.



The authors would like to thank M. Labarrère, L. Bossut and U. Steiner for their efficient technical help. We also thank A. Gavaland, P. Bataillon and D. Campergue at the INRA Auzeville experimental unit for technical assistance. We are grateful to J. Kattge, coordinator of the TRY database, and all of its contributing authors (see S2 Text) for allowing us to use their data from the TRY database.

Author Contributions

Conceived and designed the experiments: HT EJ PC EG RC. Performed the experiments: HT EJ PC RC. Analyzed the data: HT EJ PC FF EG OF RC. Wrote the paper: HT EJ PC FF EG RC OF.


  1. 1. Justes E, Beaudoin N, Bertuzzi P, Charles R, Constantin J, Dürr C, et al. The use of cover crops to reduce nitrate leaching: Effect on the water and nitrogen balance and other ecosystem services. The International Fertiliser Society meeting. Cambridge; 2012. pp. 4–43.
  2. 2. Kristensen HL, Thorup-Kristensen K. Root Growth and Nitrate Uptake of Three Different Catch Crops in Deep Soil Layers. Soil Sci Soc Am J. 2004;68: 529–537.
  3. 3. Tonitto C, David MBB, Drinkwater LEE. Replacing bare fallows with cover crops in fertilizer-intensive cropping systems: A meta-analysis of crop yield and N dynamics. Agric Ecosyst Environ. 2006;112: 58–72.
  4. 4. Thorup-Kristensen K, Magid J, Jensen LS. Catch crops and green manures as biological tools in nitrogen management in temperate zones. Adv Agron. 2003;79: 227–302.
  5. 5. Jensen ES. Nitrogen Accumulation and Residual Effects of Nitrogen Catch Crops. Acta Agric Scand. 1991;41: 333–344.
  6. 6. Justes E, Mary B, Nicolardot B. Quantifying and modelling C and N mineralization kinetics of catch crop residues in soil: parameterization of the residue decomposition module of STICS model for mature and non mature residues. Plant Soil. 2009;325: 171–185.
  7. 7. Damour G, Dorel M, Tran Quoc H, Meynard C, Risède J. A trait-based characterization of cover plants to assess their potential to provide a set of ecological services in banana cropping systems. Eur J Agron. Elsevier B.V.; 2014;52: 218–228.
  8. 8. Violle C, Navas M-L, Vile D, Kazakou E, Fortunel C, Hummel I, et al. Let the concept of trait be functional! Oikos. 2007;116: 882–892.
  9. 9. Garnier E, Navas M-L. A trait-based approach to comparative functional plant ecology: concepts, methods and applications for agroecology. A review. Agron Sustain Dev. 2011;32: 365–399.
  10. 10. Grime JP. Plant strategies and vegetation process. Wiley. London, UK; 1979. p. 222.
  11. 11. Grime JP. Evidence for the existence of three primary strategies in plants and its relevance to ecological and evolutionary theory. Am Nat. 1977;111: 1169–1194.
  12. 12. Westoby M. A leaf-height-seed (LHS) plant ecology strategy scheme. Plant Soil. 1998;199: 213–227.
  13. 13. Grime JPP, Thompson K, Hunt R, Hodgson JGG, Cornelissen JHCHC, Rorison IHH, et al. Integrated screening validates primary axes of specialisation in plants. Oikos. 1997;79: 259–281.
  14. 14. Reich PB, Walters MB, Ellsworth DS. Leaf life-span in relation to leaf, plant, and stand characteristics among diverse ecosystems. Ecol Monogr. 1992;62: 365–392.
  15. 15. Reich PB, Walters MB, Tjoelker MG, Vanderklein D, Buschena C. Photosynthesis and respiration rates depend on leaf and root morphology and nitrogen concentration in nine boreal tree species differing in relative growth rate. Funct Ecol. 1998;12: 395–405.
  16. 16. Poorter H, Garnier E. Ecological Significance of Inherent Variation in Relative Growth Rate and Its Components. In: Pugnaire FI, Vallardares F, editors. Functional Plant Ecology. 2nd ed. CRC press, Boca Raton; 2007. pp. 67–100.
  17. 17. Wright IJ, Reich PB, Westoby M, Ackerly DD, Baruch Z, Bongers F, et al. The worldwide leaf economics spectrum. Nature. 2004;428: 821–7. pmid:15103368
  18. 18. Wilson PJPJ, Thompson K, Hodgson JGJG. Specific leaf area and leaf dry matter content as alternative predictors of plant strategies. New Phytol. 1999;143: 155–162.
  19. 19. Cruz P, De Quadros FLF, Theau JP, Frizzo A, Jouany C, Duru M, et al. Leaf Traits as Functional Descriptors of the Intensity of Continuous Grazing in Native Grasslands in the South of Brazil. Rangel Ecol Manag. 2010;63: 350–358.
  20. 20. Poorter H, Bergkotte M. Chemical composition of 24 wild species differing in relative growth rate. Plant, Cell Environ. 1992;15: 221–229.
  21. 21. Garnier E, Lavorel S, Ansquer P, Castro H, Cruz P, Dolezal J, et al. Assessing the effects of land-use change on plant traits, communities and ecosystem functioning in grasslands: a standardized methodology and lessons from an application to 11 European sites. Ann Bot. 2007;99: 967–85. pmid:17085470
  22. 22. Fortunel C, Garnier E, Joffre R, Kazakou E, Quested H, Grigulis K, et al. Leaf traits capture the effects of land use changes and climate on litter decomposability of grasslands across Europe. Ecology. 2009;90: 598–611. pmid:19341132
  23. 23. Hodgson JG, Wilson PJ, Hunt R, Grime JP, Thompson K. Allocating C-S-R plant functional types: a soft approach to a hard problem. Oikos. 1999;85: 282–294.
  24. 24. Pierce S, Brusa G, Vagge I, Cerabolini BEL. Allocating CSR plant functional types: the use of leaf economics and size traits to classify woody and herbaceous vascular plants. Thompson K, editor. Funct Ecol. 2013;27: 1002–1010.
  25. 25. Diaz S, Hodgson JG, Thompson K, Cabido M, Cornelissen JHC, Jalili A, et al. The plant traits that drive ecosystems: Evidence from three continents. J Veg Sci. 2004;15: 295–304.
  26. 26. Wilson JW. Analysis of Growth, Photosynthesis and Light Interception for Single Plants and Stands. 1981; 507–512.
  27. 27. Westoby M, Falster DS, Moles AT, Vesk PA, Wright IJ. Plant Ecological Strategies: Some Leading Dimensions of Variation Between Species. Annu Rev Ecol Syst. 2002;33: 125–159.
  28. 28. Garnier E, Shipley B, Roumet C, Laurent G. A standardized protocol for the determination of specific leaf area and leaf dry matter content. Funct Ecol. 2001;15: 688–695.
  29. 29. Grime BYJP, Hunt R. Relative Growth-Rate: Its Range and Adaptive Significance in a Local Flora. J Ecol. 1975;63: 393–422.
  30. 30. Justes E, Beaudoin N, Bertuzzi P, Charles R, Constantin J, Dürr C, et al. Réduire les fuites de nitrate au moyen de cultures intermédiaires: Conséquences sur les bilan d’eau et d'azote, autres services écosystémiques. Rapport d’étude, INRA France; 2012 p. 404.
  31. 31. Bedoussac L, Justes E. Dynamic analysis of competition and complementarity for light and N use to understand the yield and the protein content of a durum wheat–winter pea intercrop. Plant Soil. 2010;330: 37–54.
  32. 32. Karimi M, Siddique K. Crop growth and relative growth rates of old and modern wheat cultivars. Aust J Agric Res. 1991;42: 13–20.
  33. 33. Cornelissen JHC, Lavorel S, Garnier E, Díaz S, Buchmann N, Gurvich DE, et al. A handbook of protocols for standardised and easy measurement of plant functional traits worldwide. Aust J Bot. 2003;51: 335.
  34. 34. Kattge J, Diaz S, Lavorel S, Prentice IC, Leadley P, Bönisch G, et al. TRY—a global database of plant traits. Glob Chang Biol. 2011;17: 2905–2935.
  35. 35. Al Haj Khaled R, Duru M, Theau JP, Plantureux S, Cruz P. Variation in leaf traits through seasons and N-availability levels and its consequences for ranking grassland species. J Veg Sci. 2005;16: 391.
  36. 36. Kazakou E, Violle C, Roumet C, Navas M-L, Vile D, Kattge J, et al. Are trait-based species rankings consistent across data sets and spatial scales? Pillar V, editor. J Veg Sci. 2014;25: 235–247.
  37. 37. Hajek P, Hertel D, Leuschner C. Intraspecific variation in root and leaf traits and leaf-root trait linkages in eight aspen demes (Populus tremula and P. tremuloides). Front Plant Sci. 2013;4: 415. pmid:24155751
  38. 38. Ludlow MM, Muchow RC. A Critical Evaluation of Traits for Improving Crop Yields in Water-Limited Environments. Adv Agron. 1990;43: 107–153.
  39. 39. Lopes MS, Araus JL, van Heerden PDR, Foyer CH. Enhancing drought tolerance in C4 crops. J Exp Bot. 2011;62: 3135–53. pmid:21511912
  40. 40. Lambers H, Poorter H. Inherent Variation in Growth Rate Between Higher Plants: A Search for Physiological Causes and Ecological Consequences. Adv Ecol Res. 1992;23: 187–261.
  41. 41. Aerts R, Chapin FS. The mineral nutrition of wild plants revisited: a re-evaluation of processes and patterns. Adv Ecol Res. 2000;30: 1–67.
  42. 42. Gambino P, Vilela A. Morphological traits and allocation patterns related to stress-tolerance and seed-yield in wild and domesticated evening primrose (Oenothera L. Onagraceae). Ind Crops Prod. 2011;34: 1269–1276.
  43. 43. Pujol B, Salager J-L, Beltran M, Bousquet S, McKey D. Photosynthesis and Leaf Structure in Domesticated Cassava (Euphorbiaceae) and a Close Wild Relative: Have Leaf Photosynthetic Parameters Evolved Under Domestication? Biotropica. 2008;40: 305–312.
  44. 44. Kazakou E, Violle C, Roumet C, Pintor C, Gimenez O, Garnier E. Litter quality and decomposability of species from a Mediterranean succession depend on leaf traits but not on nitrogen supply. Ann Bot. 2009;104: 1151–1161. pmid:19710073
  45. 45. García-Palacios P, Milla R, Delgado-Baquerizo M, Martín-Robles N, Alvaro-Sánchez M, Wall DH. Side-effects of plant domestication: ecosystem impacts of changes in litter quality. New Phytol. 2013;198: 504–13. pmid:23356416