Contemporary Evolution of Plant Growth Rate Following Experimental Removal of Herbivores: Notes on Turley et al. (2013)

Theoretical Background and Conceptual Framework

Plant-herbivore interactions have persisted for over $400 \times 10^6\,\text{years}$ , serving as a primary evolutionary driver for the diversification of plant defense strategies across macroevolutionary timescales.
Plant defensive strategies against herbivores are broadly categorized into two mechanistic classes:
- Resistance traits: Morphological, physical, or chemical traits that deter herbivore feeding and directly reduce the amount of damage sustained by the plant.
- Tolerance traits: Physiological or developmental traits that limit the negative fitness consequences of herbivore damage without necessarily reducing damage itself.
Macroevolutionary diversification requires microevolutionary selection: for herbivory to drive diversification across geological epochs, herbivores must exert natural selection that produces evolutionary changes in plant populations over contemporary, short-term timescales spanning one to hundreds of generations.
The prerequisites for contemporary microevolutionary adaptation are widespread in plant systems:
- Ubiquitous heritable genetic variation ( $h^2 > 0$ ) exists within plant populations for both chemical resistance and tolerance to defoliation.
- Herbivore feeding imposes intense directional or stabilizing natural selection on defensive phenotypes.
Despite theoretical expectations, direct empirical links between herbivore presence or absence and contemporary evolutionary responses remain scarce and are historically dominated by observational surveys:
- Prior observational studies demonstrated positive correlations between herbivory intensity and defensive investment in wild populations (e.g., wild parsnip and invasive weeds), suggesting that herbivore removal relaxes selection and prompts the evolutionary loss of costly defenses.
- Observational field correlations are inconclusive because confounding evolutionary mechanisms, including spatial sorting, environmental clines, genetic drift, or abiotic selection agents, can produce identical phenotypic gradients in the absence of herbivore-mediated selection.
Vertebrate herbivores (such as mammalian grazers) exert fundamentally distinct selective regimes compared to insect herbivores:
- Vertebrate grazers crop entire shoots closely to the ground, dramatically reducing vegetative biomass and altering community architecture.
- Vertebrates impose both direct selective forces via biomass removal and tissue destruction, and indirect selective forces via altered competitive hierarchies, enhanced light penetration, and shifted soil nutrient dynamics.

Study System: Lowland Acidic Grasslands at Silwood Park

Rabbit grazing and exclosure fence at Silwood Park, England

The experimental study was conducted in species-poor lowland acidic grasslands located at Silwood Park, Imperial College, England ( $51.409^\circ\,\text{N}, 0.642^\circ\,\text{W}$ ), characterized by sandy, nutrient-poor, acidic soils containing a mixture of monocot and eudicot species.
The European rabbit (Oryctolagus cuniculus) represents the dominant vertebrate herbivore and a keystone species in this system:
- Rabbits have inhabited the British Isles for thousands of years and have maintained high population densities at Silwood Park since the 1950s.
- Rabbit grazing prevents ecological succession by inhibiting the colonisation and development of woody shrubland and deciduous forest.
- Close cropping by rabbits restricts plant vertical growth and eliminates reproductive flowering shoots in heavily grazed pastures.
Ecological impacts of rabbit grazing at Silwood Park:
- Community vegetation height is reduced by $56\%$
- $\text{Ungrazed mean height} = 18\,\text{cm}$
- $\text{Grazed mean height} = 8\,\text{cm}$
- Standing aboveground plant biomass is reduced by $25\%$
- $\text{Ungrazed mean biomass} = 80\,\text{g dry weight per } 0.125\,\text{m}^2$
- $\text{Grazed mean biomass} = 60\,\text{g dry weight per } 0.125\,\text{m}^2$
- Following $14\,\text{years}$ of rabbit exclusion, ecosystem net primary productivity (measured as total seasonal aboveground biomass accumulation) increases by $\sim 60\%$ , and soil nutrient availability increases by $\sim 180\%$ .
Herbivore-mediated community sorting patterns among co-occurring grassland plant species:
- Grazing "increaser" species: Species whose relative abundance increases under intense grazing pressure (e.g., Holcus lanatus, Senecio jacobaea, Trifolium repens); these taxa generally display rapid growth rates and superior defoliation tolerance.
- Grazing "decreaser" species: Species whose abundance declines markedly under grazing (e.g., Achillea millefolium, Vicia sativa, Festuca rubra).
- Grazing-neutral species: Species whose population cover remains static regardless of grazing pressure (e.g., Rumex acetosa, Stellaria graminea, Veronica chamaedrys). Unchanging abundance in neutral taxa suggests potential microevolutionary adaptation to the altered biotic and abiotic regimes following herbivore removal.
Biology of the focal model plant, Rumex acetosa L. (Polygonaceae; common garden sorrel):
- A dioecious perennial herbaceous forb common in British grasslands.
- Relative ground cover remains equivalent inside exclosures ( $\text{mean cover} = 4.4\%$ ) and outside exclosures ( $\text{mean cover} = 3.6\%$ ), with standard error $\text{SE} = 3.0$ \ and sample size $N = 1002$ .
- Reproduction under continuous grazing occurs almost exclusively through clonal vegetative proliferation because regular mammalian grazing removes emergent floral meristems.
- Although flowering and seed production occur inside rabbit exclosures, seedling recruitment in undisturbed swards is minimal due to dense competitive exclusion by established perennial vegetation.
- Evolution within local patches is dominated by clonal lineage selection.

Experimental Design and Methodological Procedures

Long-Term Field Manipulation:
- A total of 22 rabbit exclosures were established between 1984 and 2009 across approximately $100\,\text{ha}$ of grassland meadows at Silwood Park.
- Exclosures generated a chronosequence representing a cessation of grazing ranging from less than 1 year ( $4\,\text{months}$ ) up to $26\,\text{years}$ .
- The solely manipulated ecological variable was the exclusion of rabbits; all other environmental conditions, climate variations, and background community processes operated naturally.
- Each exclosure is treated as an evolutionarily discrete population, an approach validated by known microgeographical local adaptation occurring over spatial scales of meters in British grasslands.
- In 2009, newly erected short-term exclosures permitted grazed R. acetosa populations to set seed without altering their long-term selection regime.
Common-Garden Greenhouse Setup:
- A cross-sectional experimental design was implemented to separate phenotypic plasticity from genetic divergence.
- Seeds harvested in 2009 from 22 exclosure populations were germinated synchronously on moist filter paper to eliminate maternal environmental effects mediated by germination phenology.
- Seedlings were potted individually into $1.3\,\text{dm}^3$ ( $15.2\,\text{cm}$ diameter) round pots filled with a $50:50$ sand:potting soil mixture (Fafard no. 2, Agawam, MA), supplemented with $\sim 1.8\,\text{g}$ of slow-release fertilizer (Osmocote $14:14:16$ N:P:K, Scotts, Marysville, OH).
- Target sample size per population was 24 plants (realized range: 4 to 24 individuals per exclosure; grand mean: 22 individuals).
Factorial Evaluation of Tolerance and Competitive Ability:
- Experimental design: A full factorial arrangement consisting of two damage levels ( $\text{undamaged}$ vs. $\text{damaged}$ ) crossed with two competition levels ( $\text{no competitor}$ vs. $\text{competitor}$ ), replicated 6 times across all 22 populations.
- Simulated herbivory protocol: Aboveground plant tissue was severed using a razor blade to a residual height of $1\,\text{cm}$ above the substrate surface at two developmental milestones: day 23 and day 42 post-transplantation.
- Interspecific competition assay: Plants were grown alongside a single individual of the common evening primrose (Oenothera biennis, Onagraceae) planted $6.5\,\text{cm}$ away in the shared pot. A single North American genotype of O. biennis served as a standardized ecological phytometer.
- Harvest procedures: Destructive harvesting occurred 68 days post-planting. Aboveground and belowground biomass were separated, washed to remove substrate (fine roots collected for R. acetosa; taproots collected for O. biennis), oven-dried to a constant weight, and quantified to the nearest milligram.
- Relative tolerance index: $\text{Tolerance} = \frac{\text{Biomass}_{\text{damaged}} - \text{Biomass}_{\text{undamaged}}}{\text{Biomass}_{\text{undamaged}}}$
- Relative competitive ability index: $\text{Competitive Ability} = \frac{\text{Biomass}_{\text{with competitor}} - \text{Biomass}_{\text{without competitor}}}{\text{Biomass}_{\text{without competitor}}}$
Quantification of Chemical Resistance Metabolites:
- Plants grown in a separate common greenhouse experiment under identical soil and climate parameters were harvested at the $\sim 12\text{--}14$ true leaf stage and lyophilized (VirTis Genesis freeze dryer, SP Industries, Warminster, PA).
- Soluble protein-binding tannins:
- Assayed via radial diffusion into bovine serum albumin (BSA) protein gels.
- Tissue samples ( $100\,\text{mg}$ homogenized powder) were extracted 3 times in $70\%$ aqueous acetone at room temperature, evaporated under vacuum centrifugation, and resuspended in deionized water.
- Aliquots of $8\,\mu\text{L}$ extract were loaded into $4\,\text{mm}$ wells punched in $1\%$ agarose gels containing $0.1\%$ BSA dissolved in $9\,\text{cm}$ petri dishes ( $10\,\text{cm}^3$ volume).
- Gels were incubated for 2 days at $30^\circ\,\text{C}$ , after which precipitation rings formed by tannin-protein complexes were measured with digital calipers and converted to tannic acid equivalence.
- Total leaf oxalates:
- Homogenized dry tissue ( $20\,\text{mg}$ ) was suspended in $1.5\,\text{cm}^3$ deionized water and heated at $85^\circ\,\text{C}$ for $50\,\text{min}$ .
- Supernatant was extracted, diluted to a $0.3\%$ aqueous dilution, and plated on a 96-well microplate.
- Reaction mixture: $240\,\mu\text{L}$ of $2.125 \times 10^{-5}\,\text{M}$ crystal violet reagent, $30\,\mu\text{L}$ of diluted sample, and $30\,\mu\text{L}$ of a catalytic reagent ( $0.40\,\text{M } H_2SO_4$ and $0.004\,\text{M } K_2Cr_2O_7$ ).
- Kinetic catalytic oxidation was measured at an absorbance of $630\,\text{nm}$ exactly $4\,\text{min}$ post-initiation (BioTek PowerWave X model UPRWIE).
Maternal Investment Control:
- Fifty individual seeds from each sampled exclosure were weighed to test for differences in maternal provisioning across population age.
- Linear regression revealed no relationship between exclosure age and seed weight ( $t = -1.1$ , $df = 12$ , $P = 0.3$ ).
Statistical Analysis:
- Linear regression models evaluated evolutionary trajectories for 5 focal traits against exclosure age (time elapsed since grazer exclusion).
- Linear mixed-effects models (nlme library in R) tested the main fixed effects of exclosure age, simulated herbivory, competition, and their interactive terms, assessed using Likelihood Ratio Tests (LRT).
- Pairwise evolutionary trade-offs were examined via Pearson product-moment correlation coefficients ( $r$ ).
- Evolutionary rates were calculated in Haldanes ( $h$ , proportional phenotypic shift per generation in standard deviation units) and Darwins ( $d$ , proportional rate of change per million years), assuming an annual generation time of 1 year based on early reproductive competence.

Evolved Changes in Plant Growth Rate

Growth rate of Rumex acetosa populations across 26 years of herbivore exclusion under combinations of competition and damage

Plant intrinsic growth rate (defined as total accumulated dry biomass divided by 68 growth days) decreased significantly and linearly with increasing duration of rabbit exclusion across nearly all microenvironments:
- Undamaged plants grown without competition:
- Total biomass accumulation rate decreased linearly by $30\%$ over $26\,\text{years}$ of grazer exclusion ( $t = -3.7$ , $df = 20$ , $P = 0.001$ ).
- Damaged plants grown without competition:
- Growth rate declined by $30\%$ over $26\,\text{years}$ ( $t = -3.3$ , $df = 19$ , $P = 0.004$ ).
- Damaged plants grown with competition:
- Growth rate declined by $35\%$ over $26\,\text{years}$ ( $t = -2.5$ , $df = 20$ , $P = 0.02$ ).
- Undamaged plants grown with competition:
- Displayed a $21\%$ phenotypic reduction in growth rate over $26\,\text{years}$ , representing a consistent downward trend ( $t = -1.66$ , $df = 20$ , $P = 0.1$ ).
Rate of contemporary evolution:
- The maximum phenotypic evolutionary rate of change in intrinsic plant growth rate was $0.08\,\text{Haldanes}$ .
- This equates to a sustained directional shift of at least $0.08$ population standard deviations per generation over a quarter of a century.
Adaptive interpretation of growth rate changes:
- High baseline growth rate functions directly as an adaptive defense against persistent, severe vertebrate grazing. Rapid growth permits continuous tissue replenishment, counteracting repeated biomass consumption and preventing grazer-mediated extinction.
- Relaxation of grazing selects against high growth rates. In the absence of mammalian herbivores, fast vegetative growth represents a net metabolic or physiological liability, leading to directional clonal selection favoring slow-growing, resource-conservative genotypes.
- The reduction in growth rate contradicts conventional abiotic resource predictions: despite documented increases in soil nutrients ( $+180\%$ ) and standing biomass in ungrazed swards, plants did not evolve faster growth to compete for light, demonstrating that direct grazing pressure exerted stronger selection on growth rate than competition-induced resource gradients.

Stability of Chemical Defenses, Tolerance, and Competitive Ability

Rumex acetosa chemical defense concentrations, tolerance to herbivory, and competitive ability as a function of exclosure age

Secondary Metabolite Concentrations:
- Protein-binding tannins: Concentrations in dried foliage varied from $0.5\%$ to $1.8\%$ tannic acid equivalence across populations, but showed no directional evolutionary relationship with exclosure age ( $t = -0.8$ , $df = 13$ , $P = 0.4$ ).
- Total foliar oxalates: Concentrations ranged from $80\,\mu\text{g}/\text{mg}$ to $190\,\mu\text{g}/\text{mg}$ dry tissue, exhibiting no evolutionary correlation with time since herbivore removal ( $t = 0.3$ , $df = 11$ , $P = 0.8$ ).
- Rates of evolutionary divergence for constitutive secondary chemistry were statistically indistinguishable from zero.
Tolerance to Herbivory:
- Experimental clipping reduced total final biomass of R. acetosa across all populations by $68\%$ ( $\chi^2 = 669.1$ , $df = 1$ , $P < 0.001$ ).
- Tolerance to defoliation did not evolve across the 26-year chronosequence ( $t = 0.2$ , $df = 19$ , $P = 0.9$ ).
- Tolerance remained stable regardless of the presence or absence of interspecific competitors, confirmed by a non-significant age-by-damage interaction in the mixed-effects framework ( $\chi^2 = 0.04$ , $df = 1$ , $P = 0.8$ ).
Competitive Interactions:
- The presence of the phytometer Oenothera biennis severely suppressed R. acetosa growth, reducing final dry biomass by $80\%$ ( $\chi^2 = 857.4$ , $df = 1$ , $P < 0.001$ ).
- Competitive ability of R. acetosa was unrelated to the age of the ancestral exclosure ( $t = 0.7$ , $df = 20$ , $P = 0.5$ ).
- The competitive interaction was asymmetric: the presence of R. acetosa caused no detectable change in the dry biomass of O. biennis ( $t = 0.4$ , $df = 63$ , $P = 0.7$ ).
- Exclosure age did not interact with competition treatment ( $\chi^2 = 1.2$ , $df = 1$ , $P = 0.3$ ).
Pairwise Trait Correlations and Evolutionary Trade-offs:
- A significant negative evolutionary trade-off occurred between intrinsic plant growth rate and tolerance to herbivory (Pearson's $r = -0.44$ , $df = 19$ , $P = 0.04$ ).
- Phenotypic correlations between growth rate and secondary chemical defenses were non-significant ( $P > 0.1$ ).
- No relationship was observed between tolerance to herbivory and competitive ability ( $P > 0.1$ ).

Synthesis of Vertebrate Herbivore Exclusion Across Grassland Systems

A systematic review of 12 common-garden studies manipulating vertebrate herbivores reveals pervasive evolutionary shifts in grass and forb populations, with variable trajectories across systems:
- Carman and Briske (1985):
- Herbivore: Domestic cattle
- Plant taxon: Schizachyrium scoparium (Poaceae)
- Trait changes following herbivore exclusion: Growth rate ( $0$ ), Competitive ability (No data), Tolerance (No data)
- Polley and Detling (1990):
- Herbivore: Black-tailed prairie dog
- Plant taxon: Agropyron smithii (Poaceae)
- Trait changes following herbivore exclusion: Growth rate ( $0$ ), Competitive ability (No data), Tolerance ( $0$ )
- Tomas et al. (2000):
- Herbivore: Domestic cattle
- Plant taxon: Piptochaetium napostaense (Poaceae)
- Trait changes following herbivore exclusion: Growth rate ( $0$ ), Competitive ability (No data), Tolerance (No data)
- Turley et al. (2013; this study):
- Herbivore: European rabbit
- Plant taxon: Rumex acetosa (Polygonaceae)
- Trait changes following herbivore exclusion: Growth rate ( $-30\%$ ), Competitive ability ( $0$ ), Tolerance ( $0$ )
- Briske and Anderson (1992):
- Herbivore: Domestic cattle
- Plant taxon: Schizachyrium scoparium (Poaceae)
- Trait changes following herbivore exclusion: Growth rate ( $-$ , decrease), Competitive ability ( $-$ , decrease), Tolerance (No data)
- Jefferies (1988):
- Herbivore: Lesser snow goose
- Plant taxon: Puccinellia phryganodes / Carex subspathacea (Poaceae / Cyperaceae)
- Trait changes following herbivore exclusion: Growth rate ( $-$ , decrease), Competitive ability (No data), Tolerance (No data)
- Loreti et al. (2001):
- Herbivore: Domestic cattle
- Plant taxon: Paspalum dilatatum (Poaceae)
- Trait changes following herbivore exclusion: Growth rate ( $-$ , decrease), Competitive ability (No data), Tolerance ( $0$ )
- Polley and Detling (1988):
- Herbivore: Black-tailed prairie dog
- Plant taxon: Agropyron smithii (Poaceae)
- Trait changes following herbivore exclusion: Growth rate ( $-$ , decrease), Competitive ability (No data), Tolerance (No data)
- Detling and Painter (1983):
- Herbivore: Black-tailed prairie dog
- Plant taxon: Agropyron smithii (Poaceae)
- Trait changes following herbivore exclusion: Growth rate ( $+$ , increase), Competitive ability (No data), Tolerance ( $+$ , increase)
- Jaindl et al. (1994):
- Herbivore: Domestic cattle
- Plant taxon: Festuca idahoensis (Poaceae)
- Trait changes following herbivore exclusion: Growth rate ( $+$ , increase), Competitive ability (No data), Tolerance ( $0$ )
- Jaramillo and Detling (1988):
- Herbivore: Black-tailed prairie dog
- Plant taxon: Bouteloua gracilis (Poaceae)
- Trait changes following herbivore exclusion: Growth rate ( $+$ , increase), Competitive ability ( $0$ ), Tolerance ( $0$ )
- Painter et al. (1989):
- Herbivore: Black-tailed prairie dog
- Plant taxon: Agropyron smithii (Poaceae)
- Trait changes following herbivore exclusion: Growth rate ( $+$ , increase), Competitive ability ( $-$ , decrease), Tolerance ( $-$ , decrease)
- Painter et al. (1989):
- Herbivore: Black-tailed prairie dog
- Plant taxon: Bouteloua gracilis (Poaceae)
- Trait changes following herbivore exclusion: Growth rate ( $+$ , increase), Competitive ability ( $0$ ), Tolerance ( $-$ , decrease)
Distribution of evolutionary responses in growth rate across vertebrate exclusion literature:
- Growth rate increased upon grazer removal in $40\%$ of studies.
- Growth rate decreased upon grazer removal in $40\%$ of studies.
- Growth rate exhibited no significant evolutionary change in $20\%$ of studies.
Directional discordance across empirical systems stems from competing ecological trade-offs:
- Systems where growth decreases upon herbivore exclusion (e.g., Rumex acetosa at Silwood Park) are governed by selection where fast continuous growth acts as a primary compensatory mechanism against chronic tissue loss, but entails a severe ecological or physiological cost when defoliation ceases.
- Systems where growth increases upon herbivore exclusion are governed by allocation trade-offs: herbivory selects for high allocation to physical or qualitative chemical defenses, which restricts intrinsic elongation rates; removal of herbivores relaxes defense allocation, enabling evolution toward maximal competitive growth.

Theoretical Implications for Plant Defense and Eco-Evolutionary Dynamics

Challenges to the Resource Availability Hypothesis (RAH):
- The Resource Availability Hypothesis (Coley et al. 1985) posits that environmental resource availability (e.g., nutrients, water, light) determines a plant population's optimal growth rate, which subsequently governs allocation constraints between somatic growth and anti-herbivore defenses.
- A fundamental foundational postulate of RAH is that intrinsic plant growth rate represents an adaptive specialization to the abiotic physical environment, rather than an anti-herbivore defense.
- The documented microevolution of plant growth rate in Rumex acetosa invalidates this foundational premise: growth rate itself evolved rapidly as a constitutive adaptation to herbivory.
- If mammalian grazing directly selects for elevated intrinsic growth rates, RAH paradoxically predicts that plants experiencing intense grazing will evolve lower constitutive chemical resistance due to growth-defense allocation constraints, demonstrating that classical defense theories fail when growth rate is itself an evolving defense mechanism.
Constraints Imposed by Negative Genetic Correlations:
- The observed negative evolutionary correlation between growth rate and tolerance ( $r = -0.44$ ) provides a mechanistic explanation for why tolerance failed to evolve following rabbit exclusion.
- Inherent trade-offs between rapid baseline biomass accumulation and cellular or architectural tolerance mechanisms prevent both traits from undergoing simultaneous positive directional selection.
Disconnect Between Microevolution and Ecological Feedbacks:
- Current eco-evolutionary theory hypothesizes that rapid phenotypic evolution within a focal species will cause reciprocal feedbacks that alter competitive interactions, community structure, and ecosystem function.
- Despite a $30\%$ evolutionary drop in intrinsic growth rate in R. acetosa, zero detectable change occurred in interspecific competitive outcomes with Oenothera biennis.
- Microcosm and chemostat experiments with simplified, low-diversity trophic webs frequently demonstrate rapid eco-evolutionary feedbacks, but species-rich natural communities with high ecological redundancy buffer against evolutionary feedbacks.
- Rather than disrupting interspecific competitive dynamics, local microevolution in persistent, grazing-neutral grassland perennials serves to expand species' realized ecological niches, maintaining community-level diversity under shifting disturbance regimes without altering competitive hierarchies.