Low genetic variation in tolerance to defoliation in a long-lived tropical understorey palm

Defoliation is a ubiquitous stressor that can strongly limit plant performance. Tolerance to defoliation is often associated with compensatory growth. Genetic variation in tolerance and compensatory growth responses, in turn, play an important role in the evolutionary adaptation of plants to changing disturbance regimes but this issue has been poorly investigated for long-lived woody species. We quantified genetic variation in plant growth and growth parameters, tolerance to defoliation and compensatory responses to defoliation for a population of the understorey palm Chamaedorea elegans. In addition, we evaluated genetic correlations between growth and tolerance to defoliation. We performed a greenhouse experiment with 731 seedlings from 47 families with twelve or more individuals of C. elegans. Seeds were collected in southeast Mexico within a 0.7 ha natural forest area. A two-third defoliation treatment (repeated every two months) was applied to half of the individuals to simulate leaf loss. Compensatory responses in specific leaf area, biomass allocation to leaves and growth per unit leaf area were quantified. We found that growth rate was highly heritable and that plants compensated strongly for leaf loss. However, genetic variation in tolerance, compensation, and the individual compensatory responses was low. We found strong correlations between family mean growth rates in control and defoliation treatments. We did not find indications for growth-tolerance trade-offs: genetic correlation between tolerance and growth rate were not significant. The low genetic variation in tolerance and compensatory responses observed here suggests a low potential for evolutionary adaptation to changes in damage or herbivory, but high ability to adapt to changes in environment that require different growth rates. The strong correlations between family mean growth rates in control and defoliation treatments suggest that performance differences among families are also maintained under stress of disturbance.


44
Defoliation due to herbivory, pathogens, physical damage or harvesting is an ubiquitous 45 stressor that can strongly limit individual plant performance (i.e. growth, reproduction and 46 survival) as it entails a reduction in photosynthesis and resources, and thus in future growth. 47 Performance reductions due to defoliation are often proportionately smaller than expected 48 based on the fraction of leaf area that is being removed [1,2] and in some cases plants even 49 increase their performance under defoliation [3,4]. In that sense plants can be tolerant to 50 defoliation, and this tolerance is often associated with compensatory growth, a mechanism by 51 which negative effects of leaf loss are mitigated [5]. There are three types of compensatory 52 growth responses: plants can compensate for growth by allocating more new assimilates to 53 leaves, by allocating new assimilates more efficiently to leaf area (i.e. by increasing specific 54 leaf area), or by growing faster with existing leaf area (i.e. by increasing net assimilation rate 55 [6]). 56 Many plant species have evolved tolerance to leaf loss [e.g. 5, 7, 8, 9], which indicates that 57 plants have evolved compensatory growth responses. However, relatively little work has been 58 done to study genetic variation in these compensatory growth responses [8]. Furthermore, 59 tolerance can only evolve when there is heritable variation in compensatory mechanisms within performance under no disturbance [5]. However, plants can also tolerate defoliation without 4 67 investing in reserves: by increased photosynthetic activity due to less self-shading, or by higher 68 stomatal conductance due to changed root-shoot ratio [7,8]. If this is the case, growth under 69 undamaged conditions and tolerance would be expected to be uncorrelated or even positively 70 correlated. The trade-off between growth and tolerance is believed to be a significant factor in 71 determining species habitat adaptation [12]. If tolerance and performance under unstressed Many studies have evaluated genetic variation in performance in short-lived species (mostly 78 annuals and bi-annuals), and some genetic variation in tolerance and genetic correlations 79 between performance and tolerance to leaf-loss [13]. However, for long-lived woody plant 80 species much less is known about these issues [14]. Haukioja & Koricheva [15] argue that 81 tolerance to defoliation might be just as important for long-lived species as it is for short-lived 82 species, but this has not been empirically tested. Defoliation tolerance might be especially 83 relevant for understorey species because shade tolerance is often associated with storage of 84 reserves that allow recovery after damage [12,16]. More information on the existence of genetic 85 variation in performance, tolerance and genetic correlations between these two, would increase 86 our understanding of the adaptive ability of long-lived plant populations to environmental 87 changes.

88
In this study we analyzed the extent to which growth and tolerance to defoliation are heritable 89 and if these two variables are genetically correlated. We did this for the long-lived, shade 90 tolerant, tropical understorey palm Chamaedorea elegans. Leaf loss due to herbivory and 5 91 physical damage is high and an important factor limiting the performance of this species [17,92 18]. C. elegans has been shown to compensate for leaf loss, by changing net assimilation rate 93 (NAR) and allocation of biomass to leaf mass [6]. Furthermore, the leaves of this species are a 94 non-timber forest product, and populations of this species are under pressure due to increased 95 harvesting activities [19]. 96 Specifically, we answered the following questions for our study population: 97 1. Is there evidence of genetic variation in plant growth and related parameters? 98 2. Is there evidence of genetic variation in tolerance to defoliation (in terms of growth 99 rate), compensatory growth, and compensatory growth responses (i.e. changes in net 100 assimilation rate (NAR), specific leaf area (SLA) and biomass allocation to leaves)? 101 3. Are growth rate and tolerance to defoliation genetically correlated?

102
To answer these questions, we performed a greenhouse experiment with seedlings in which a 103 defoliation treatment was applied. We choose to use seedlings because (1) tropical forest 104 seedlings are strongly affected by damage from falling debris and herbivory [16] (2) growing 105 seedlings from collected seeds of mother plants ensured that seedlings were half-sibs (3) using 106 seedlings allowed to increase sample size and obtain results within 1.5 years. We estimated 107 genetic variation in growth parameters, tolerance (in terms of growth), compensatory growth 108 and compensatory growth responses. We used an iterative growth model [6, 20] to estimate 6 114 Species and site of seed collection 115 The experiment was performed with the forest understorey palm species Chamaedorea elegans 116 Mart, which naturally occurs in rainforest in Mexico, Guatemala,and Belize [21]. It is single 117 stemmed, produces a single cluster of leaves and is dioecious. It naturally occurs mostly on 118 karstic outcrops. Herbivory and falling canopy debris are both major causes of leaf loss in this 119 species [6,17]. Furthermore, leaves are harvested as a Non-Timber Forest Product (NTFP) for 120 use in the floral industry, causing many populations to be under pressure [19,22]. with a stem length <10 cm were mapped and tagged for a similar experiment in 1997 (using the 129 same methods as in our experiment [17]). In total 3009 seeds from 207 different mother plants 130 were collected, with number of seeds per mother plant ranging from one to 95 seeds. Seeds 131 were cleaned (mesocarp was removed), air-dried and weighed, and they were kept in zip-lock 132 bags that allowed some gas exchange. greenhouse. The experiment started for each seedling six months after germination (6 months 138 is an age at which C. elegans seedlings growing under the conditions of this experiment have 139 been depleted, S1 File). Plant size was measured non-destructively at the start of the 140 experiment, and a 2/3 defoliation treatment was applied to half of the individuals from each 141 family. The defoliation treatment was repeated every 8 weeks, up to the age of 12.5 months, 142 when plant biomass and other parameters were measured destructively. Details on 143 measurements are provided below in the Data collection and curation section. The timeframe 144 of the experiment (i.e. 6.5 months) is similar to other experiments studying tolerance-145 performance trade-offs in seedlings of long-lived species [11,16], and was considered to likely 146 be long-enough to reveal differences in allocation of assimilates to storage rather than growth 147 (one of the main mechanisms explaining growth-tolerance trade-offs) [16,23].

149
In January 2013, seeds were planted at approximately 0.5 cm depth in large trays filled with 150 potting soil. The tray was placed in a growth chamber, where the temperature was kept constant 151 at 30°C day and night, air humidity at 90%. Germination of individual seeds was recorded two 152 times a week. One and a half weeks after emergence, seedlings were transplanted into small 153 pots of 8.5 x 8.5 x 9.5 cm (l x w x h), filled with low nutrient soil (40% peat moss peat, 20% 154 Nordic fraction 2, 20% Baltic peat agent, 20% normal garden peat, 1% pg mix, 0.2% Micromax) 155 and moved to a greenhouse where they were placed in a cage covered with 75% shade cloth to 156 allow for adjustment to changed climatic conditions. After one week, they were moved to a 157 table with flood system allowing a nutrient solution to be absorbed from below into the pots 158 (pH 5.0, EC 0.8, NPK ratio 12-14-24). Seedlings stayed on the table with flood system for the 159 duration of the experiment (see the Experimental setup section below). To simulate forest 160 conditions, temperature in the greenhouse was kept at a minimum day/night temperature of 8 161 24/22°C, air humidity at 80%, day length was reduced to a maximum of twelve hours using 162 automatically closing black screens. Light levels were in summer months reduced using 163 (depending on the month) either 25% or 50% shade cloth, such that plants received 164 approximately 2 mol per day, which is the average light intensity in the forest understorey at 165 the site where seeds were collected [24]. Monthly target shade levels were based on the 10-year 166 monthly average light intensities recorded at the location of the greenhouses.

167
Experimental design and treatment 168 The experiment was laid out as a randomized block design with six blocks. To this end, the 169 table was divided into six equal parts lengthwise to create the blocks. Seedlings from the same 170 mother (half-sib families) were randomly distributed over the blocks and over position within 171 the block. Because families differed in number of seedlings, sometimes a family was only 172 present in one block (this was the case for families with only one seedling), and sometimes in 173 all six (which was the case for families with at least six seedlings).

174
To assign the seedlings to control or defoliation treatments, we ranked all plants in a family 175 according to age (i.e. date of emergence). We then randomly assigned a treatment (i.e. 176 defoliation or control) to the oldest one, giving the other treatment to the second oldest plant 177 and alternating in this way across the age hierarchy. Of all seedlings that were assigned to the 178 defoliation treatment, two out of every three leaflets were cut off at six months of age. This 179 treatment was repeated (for newly produced leaves) every eight weeks.

181
At six months of age, we measured seedling stem length and diameter. In addition, we measured 182 leaf width, lamina length, rachis length, rachis diameter, leaflet width, and number of leaflets 183 of all leaves, as well as the length of unopened leaf. With this information, seedling biomass 9 184 (per plant part) and leaf area of the seedlings of six months of age were estimated using an 185 allometric model, that we constructed based on data of a destructive harvest of extra seedlings 186 of six months of age from the same experimental conditions (see S1 File for details).

187
Surviving seedlings were destructively harvested at 12.5 months of age (1387 in total). Plants  To estimate growth and several growth-related variables (net assimilation rate (NAR), fraction 207 of newly assimilated mass that is allocated to lamina growth (f lam ), fraction in daily change in 208 mean specific leaf area (γ) and relative growth rate (RGR)), we used an iterative growth model  tolerance is a more common measure, but compensation gives more insight in the underlying 230 mechanisms.

231
To be able to estimate genetic variation in tolerance and compensation, information on 232 differences in tolerance within families, and therefore per individual is required. In order to be 233 able to calculate tolerance and compensation per individual, each individual in the defoliation 234 treatment was paired with a family member from the control treatment, based on rank order of 235 estimated biomass at six months of age (i.e. seedling age at the beginning of the experiment).  For tolerance in biomass growth, we calculated biomass change between 6 months and 12 241 months of age, for which the values were obtained from direct measurements. We excluded leaf 242 mass in this calculation. 243 We estimated compensatory growth per individual using the approach of Anten,et al. [6]. We

281
Genetic variation in growth parameters 282 We found large variation among different families in biomass growth and RGR (Fig 1). We 283 determined within and among family variance components for biomass growth rate, RGR, and 284 the growth parameters NAR, biomass allocation (f lam ), and SLA change (γ) that were estimated 285 by the iterative growth model ( NAR, f lam , and γ, were much lower, especially for the control individuals (Table 1). that grow relatively fast without the stress of defoliation do not necessarily grow relatively fast when they suffer 296 leaf loss. The changes in rank between biomass growth rate and RGR indicate that families that grew fast in 297 absolute terms did not necessarily grow fast in relative terms. We compared family mean control and defoliation treatment values of all growth parameters 309 (Fig 2). Family mean biomass growth rate was as expected, lower in the defoliation treatment between brackets.

321
We tested whether families responded differently to defoliation, and therefore whether there  (Table 2). This suggests that families do not respond significantly different to leaf 326 loss in terms of biomass growth, RGR, NAR, allocation to leaf mass nor SLA changes.

327
Therefore, while families compensate strongly for leaf loss, we did not find evidence for strong 328 within-population genetic variation in this response. Biomass growth was determined from direct measurements, the other parameters with an iterative growth model.

333
Variance components were estimated using mixed effects models with REML estimation. RGR = Relative growth 334 rate; f lam = fraction of newly assimilated mass that is allocated to lamina growth; γ = fraction in daily change in 335 mean specific leaf area

336
To estimate genetic variation in tolerance and compensation itself, we paired defoliation 337 treatment individuals with control individuals from within the same family. By doing this, we 338 were obtaining replicated estimates of tolerance and compensation and could therefore estimate 339 the heritability of these parameters. Even though we found large variation between family mean 340 values of tolerance and compensation (e.g. family mean compensation in biomass growth 341 ranged from 0.16 to 1.03, i.e., 16 -~100% of potential loss being mitigated), within-family 342 variance was much larger. Therefore, estimations of heritability of tolerance and compensation 343 were low (the highest estimated heritability was for compensation in biomass growth, which 344 was only 0.01, Table 3).
345 Table 3. Estimated within and among family variance components and heritability of tolerance to defoliation, and 346 compensation after repeated defoliation events in a greenhouse experiment, performed seedlings of the understorey 347 palm Chamaedorea elegans. To be able to estimate tolerance and compensation, individuals from the defoliation 348 treatment were coupled to individuals from the control treatment based on their estimated biomass at the start of 349 the experiment. Compensation was calculated by using an iterative growth model that allowed estimation of a 350 hypothetical non-compensating individual. genetically correlated between treatments (Fig 2). The correlation coefficient for biomass growth rate and RGR in the control treatment (Fig 3). This did not yield clear evidence for any 367 positive or negative relation between tolerance/compensation and biomass growth/RGR. The 368 only significant correlation that we found was between tolerance and RGR. However, this 18 369 relationship was heavily pulled by two outlying data points; without these outliers there was no 370 longer a significant correlation. Therefore, we did not find evidence that would suggest costs 371 to tolerance in terms of growth. This study showed that genetic variation in tolerance and compensatory responses to 381 defoliation is limited within a population of a long-lived tropical forest species. We also 382 showed that genetic variation in growth potential was much larger than values usually 383 detected for small populations [14,30]. These results suggest that the studied population 384 might have limited ability to adapt in terms of tolerance to environmental changes that entail 385 leaf loss but does have the ability to adapt to environments that require different growth rates.

386
Furthermore, this is one of the first studies that has analyzed genetic variation in 387 compensatory growth responses to defoliation. We found large within-population genetic variation in growth rate, with estimations of narrow- Relation between growth and tolerance 427 We did not detect a genetic correlation between growth and tolerance or compensation, even 428 though it has been shown that such correlation exists at least at the ecotype level in short-lived 429 plants [32]. Therefore, the strong differences in growth that we detected among families cannot 430 be explained by a growth-tolerance trade-off. In contrast, we found that 'super-performing' 431 families that grew relatively fast under undisturbed conditions also grew fast when exposed to 432 defoliation. These types of superior genotypes could play a key role in population resistance 433 when the population is being disturbed by, for example, a storm (