| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ecology |
2Instituto de Ecología, Universidad Nacional Autonoma de México, C. P. 04510, México D. F., México 3Department of Biological Sciences, University of North Texas, Denton, Texas 76203-5218 USA
Received for publication June 27, 2000. Accepted for publication January 30, 2001.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: Chamela Biological Station • clonal integration • clonal plants • Convolvulaceae • Cuscuta • parasitic plants
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Alpert and Mooney, 1986
; Jónsdóttir and Callaghan, 1988
; Welker, Briske, and Weaver, 1991
; Evans and Whitney, 1992
; Alpert, 1996
; Charpentier, Melséard, and Thompson, 1998
; Pennings and Callaway, 2000
). Such effects are presumed to indicate limits to relative fitness of the genet under natural conditions. Beyond this, breakage can damage tissue, allow entry of pathogens (Agnos, 1978
), and create conditions that may attract or stimulate herbivores (Feeny, 1992
).
Nonetheless, fragments of clonal plants clearly may survive. Studies of genotype distribution in the field indicate that clonal individuals (genets) can persist as variously sized ramet groups (Maddox et al., 1989
; Berg and Hamrick, 1994
; Ke, Harris, and Kelly, unpublished data). However, such patterns of fragmentation may not be due to traumatic separation of ramets; clonal plants may have internally dictated schedules of degeneration of connecting structures in which abscission layers are formed, and many of the negative effects of breakage may be expected not to operate in such cases (Cook, 1978
; Room, 1983
; Kelly, 1995
). Alternatively, traumatic breakage may occur at low frequency or at a part of the plant such that it has little or no effect on the overall growth or survivorship of the genet. Furthermore, other environmental factors may be of greater importance or override effects of breakage, or the whole plant may possess responses that minimize the negative effects of natural breakage. Thus, although experimentally induced breakage may have a large negative effect, the role of traumatic breakage in the ecology of a clonal plant under natural conditions may not be so straightforward.
The goal of this study, therefore, was to determine the frequency and to explore the consequences of natural breakage in the ecology of clonal individuals growing under natural conditions. To do this, we first identified the effect of experimental severance on ramet performance in a naturally occurring individual of the parasitic clonal plant Cuscuta corymbosa to ascertain the potential effects of breakage and breakage position within an individual at this site. We then used single-genotype, simultaneously transplanted individuals of C. corymbosa to identify (1) the frequency and position of traumatic breakage under natural growing conditions, (2) the effect of breakage on ramet growth and survivorship, and (3) the effect of breakage on whole-plant growth and survivorship within a season together with the potential for an individual to survive to a second season. The data derived from these experimental and field observations were applied to a simulation of clonal growth in order to explore the implications of ramet-level changes for whole-plant ecology.
The results indicate that (1) breakage affects growth rates at the level of the ramet but has no discernable effect on biomass accumulation at the level of the individual, although plants that manage to persist into the dry season have fewer breaks per unit size than those that do not persist; (2) breakage-induced changes in stolon extension and stolon generation are sufficient to cause significant changes in whole-plant "shape" of a sort that are likely to affect probability of encounter with patchily distributed resources; correspondingly, breakage is negatively associated with the number of persistent hosts used by an individual parasite; (3) different patch types (here, host species) confer identifiably different advantages to the clonal individual, i.e., immediate biomass accumulation vs. long-term persistence; and (4) the capacity of ramets to survive disconnection significantly decreased the loss of tissue that would occur if ramets did not possess the potential to function independently. These findings are discussed in terms of a potential trade-off between immediate vs. long-term rewards, in light of the ubiquitous negative effect of traumatic breakage found in manipulative studies of other clonal species, and relative to the multispecies finding that clonal species have smaller range sizes than related nonclonal species (Kelly and Woodward, 1996
).
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
.
Species in the genus Cuscuta are commonly called dodder. All species are rootless, effectively leafless, only minimally photosynthetic, and totally dependent on their hosts. Cuscuta corymbosa Ruiz & Pavon, the species used in this study, at Chamela is approximately central in the north–south axis of its distribution in the tropical deciduous forest of coastal México, having been reported from the state of Nayarit in the north to the Isthmus of Tuantepec in the south (McVaugh, 1983–1992
). In the dodders, a ramet is initiated when a stolon comes into contact with a host branch or leaf, after which the tip of the parasite stolon coils about the host structure (a coiling bout). A dodder individual sends pegs of absorptive tissue (haustoria) into the host vascular system from the inner surface of these coils (MacLeod, 1961
). A ramet of dodder includes this coiling site plus its emergent stolon(s), and the branches emanating from the stolons.
Henceforth, the term "ramet" will refer to this entire system of structures, while "connection" will refer to the coil alone; "stolon" and "interconnection" will be applied interchangeably to the portion of stem joining two connections; the term "stem" includes both stolons and branches. Each growing tip, whether on a stolon growing from the coil itself, or on a branch of the stolon, has the potential to initiate a new ramet. A single dodder plant can have hundreds of ramets and can infest several different host individuals at any one time, with varying distances from one point of infestation to the next, both within and between host branches and individuals (Kelly, 1988, 1990
; Kelly, Venable, and Zimmerer, 1988
; Kelly and Horning, 1999
). Dodder lends itself well to the experiments and observations reported here, in that (1) all of dodder's parts are aboveground, hence total growth can be unequivocally assessed, (2) its lack of leaves or other elaborate structures allows easy quantification of morphology and morphological change, and (3) its rapid growth (up to 20 cm·stem–1·d–1) allows an experimental period encompassing significant amounts of vegetative growth, which need not be confounded by allocation to sexual reproduction.
Experimental severance of interconnections
Previous work with two closely related, ecologically similar species of Cuscuta, C. subinclusa, and C. europaea, has shown that individual growing tips can be very responsive to local conditions (Kelly, 1988, 1990, 1992, 1994
). Such sensitivity could be a possible indicator that dodder ramets normally function as physiologically independent subunits immune to the negative effects of traumatic breakage that have invariably been seen in studies using breakage as an experimental technique (e.g., Hartnett and Bazzaz, 1983
; Alpert and Mooney, 1986
; Jónsdóttir and Callaghan, 1988
; Welker, Briske, and Weaver, 1991
; Evans and Whitney, 1992
; Alpert, 1996
; Charpentier, Melséard, and Thompson, 1998
; Pennings and Callaway, 2000
). In order to determine the effect of experimentallyinduced traumatic breakage on ramet growth in Cuscuta corymbosa and to examine the possibility that breakage under natural conditions might be expected to be of importance in the ecology of C. corymbosa, we subjected 45 unbranched series of ramets (ramet "lines") to one of three treatments during the height of the growing season in mid-August. The treatments included (1) severing a one-ramet section from the main body of the clone, (2) severing a two-ramet section from the main body of the clone, and (3) unbroken (
control). Stolons were cut with a single-edge razor blade; cuts were positioned distal to (just following; in treatment 1) or proximal to (just before; in treatment 2) the second connection, in order to decrease the impact of differences in volume between the treatments (Fig. 1).
|
; Alpert and Mooney, 1986
; Jónsdóttir and Callaghan, 1988
; Welker, Briske, and Weaver, 1991
; Evans and Whitney, 1992
; Alpert, 1996
; Charpentier, Melséard, and Thompson, 1998
; Pennings and Callaway, 2000
). Sample size was determined by the number of unbranched, similarly vigorous stolons simultaneously available for use. Given the finite number of stems available for manipulation, simultaneity was necessary to factor out dodder's sensitivity to ambient conditions, which can produce significantly different responses from the same genetic individual on the same host individual within the space of days (Kelly, 1992
). All manipulations were performed in situ on a single, naturally occurring dodder individual growing on a single host individual (limón; Citrus aurantifolia Swingle). Replicates were grouped physically in threes, with each triplet containing one of each of the three treatments. Dodder tends to establish itself on fast-growing, more tender portions of its hosts (here, near the outer edge of the canopy of the host tree); the relative height of each triplet in the host canopy was recorded so that any effect due to this aspect of position could be taken into account if necessary. We were fortunate in having available a large dodder individual naturally growing on a large host individual, thus allowing experimental replicates of comparable vigor and size.
For each replicate, we measured at the beginning of the experiment stolon diameter at 5 and 25 cm proximal to the growing tip and just before the second connection (at the proximal end of the ramet line). Stolon length from the growing tip to the first connection was also noted, as was length from the first connection to the second connection. Stolons were marked at the 5 and 25 cm points with a small spot of India ink, so that growth between and beyond these points could be assessed at a later time. Twenty-four hours after beginning the experiment, we returned and measured for each replicate diameter at the former 5 and 25 cm points and at the point currently 5 cm proximal to the growing tip. Length from the marked points to the growing tip was also measured. Twenty-four hours was chosen as the observation period in order to contain the analysis to that of extensive growth; beyond 24 h, some but not all of the stolons within the experimental design would have begun coiling (ramet initiation) simply because of uncontrollable, differential proximity to coiling sites (host branches), thereby introducing a nontarget, potentially confounding variable into the experimental design. We note also that given the growth rate of C. corymbosa (up to 20 cm/d), this period of time relates to the period of observation in the second portion of this study in a manner similar to that of studies of experimentally induced breakage reported in the literature (e.g., Hartnett and Bazzaz, 1983
; Alpert and Mooney, 1986
; Jónsdóttir and Callaghan, 1988
; Welker, Briske, and Weaver, 1991
; Evans and Whitney, 1992
; Alpert, 1996
; Charpentier, Melséard, and Thompson, 1998
; Pennings and Callaway, 2000
).
Data were analyzed as an analysis of covariance in order to determine the measure best predicting stolon extension during the 24-hr experimental period. Cutting treatment was entered as a factor; covariates were volume, stolon diameter at 5 cm, at 25 cm, and distal to connection 2, all on day 1. Volume was calculated as the sum of two tubes. The volume of the "first tube" was calculated as that having a diameter equal to the width of the stem at 5 cm proximal to the growing tip and a length equal to that of the stem from the tip to the first connection. The "second tube" was taken to have a diameter equal to that of the stem immediately distal to the first connection, with a length equal to that of the stem remaining from that point to the point of breakage.
Unmanipulated breakage under natural conditions
In order to observe breakage and its effects under natural conditions, single-genotype, detached stems with growing tips of dodder were transplanted onto 16 host individuals of Baccharis salicifolia Gray once the rainy season was well under way, in mid-July. Baccharis salicifolia, unlike Limón, tends to be aggregated and one reason for using this species as the host for this portion of the study was that the physical juxtaposition of host individuals increased the comparability of replicates. See Kelly (1990, 1992)
for details on transplant techniques. Transfer of dodder stolons and stem tips in this manner is similar to the accidental dispersal of dodder stolons between agricultural fields by farm equipment and animals (Ashton and Santana, 1976). Incidental transfer of vegetative material by large animals or birds may be a means by which dodder disperses naturally (C. K. Kelly, personal observation), although the importance of such processes in the ecology of dodders currently is unknown. Previous work with Cuscuta subinclusa and with C. corymbosa has found that once established, dodder stems transplanted in this manner do not respond significantly differently to naturally occurring dodder individuals of the same size (number of ramets) in terms of both growth rate and probability and extent of coiling (Kelly, 1988
; C. K. Kelly, unpublished data).
Host individuals were chosen for similarity of size and vigor and for similarity to the host plant from which the dodder stems were collected. Baccharis salicifolia is a woody shrub, and the selected host individuals were similarly 1.5 m in height with a round or oval shaped cross section between 1.0 and 1.5 m in diameter. The low-lying area (arroyo) in which all host individuals occurred was open and, as with the host plant from which the transplant stolons were obtained, all transplant hosts were without adjacent woody shrubs or trees. All shrubs chosen as transplant hosts were within 50 m of the original host and thereby 100 m or less in distance from one another. Transplants hosts were uniformly vigorous, with a large number of branches emerging from the root crown, but with little secondary branching. Later events indicated that the aboveground tissue of these individuals is periodically removed by flooding and that the vigor and architectural similarity of these individuals is a function of that process. Thirteen of the 16 transplants became established. Each transplant was considered to be a discrete individual and will be referred to as a plant or individual.
Tissue can disappear rapidly in dodder following breakage (C. K. Kelly, personal observation) and with it the ability to differentiate between disconnections due to traumatic breaks and those due to natural degradation. Therefore, individual clones were measured and mapped at seven censuses over the next 7 wk, until mid-September. The first such census took place 10 d after the initial transplant; censuses 2–6 were made at 5–7 d intervals, with the final census covering a 10-d period of growth. All new connections were marked with numbered laundry tags at each census. Also noted at the same time was the length of stolons between new connections, and whether any stolon, old or new, was disconnected. Cause of disconnection was determined by inspection of the tissue at the site of disconnection. A disconnection was designated as a break if there was any evidence of sap leakage (a shiny surface on a remaining tip of a broken stolon) on either side of the disconnection, and as natural degeneration if not. All information for each individual was recorded on an ongoing map. Detailed measurement for all plants ceased when reaching into the depths of the host plant to make measurements on the largest dodder individual caused breakage in the parasite. Survivorship was recorded and biomass accumulation estimated (as plant volume) for the last time in the middle of the dry season in the following January. Subsequent observations were made impossible by an unseasonal flood in the second week of January, which swept away all hosts and parasites.
Statistical analyses
Patterns of growth and breakage at the level of the whole plant were compared using general linear models (GLM; MINITAB). The use of general linear models allowed evaluation of the effects of breakage on the various target variables independent of the effects of whole-plant size. For plants in general, many aspects of growth are dependent on plant size, and dodder is no exception. Breakage between ramets especially falls into this category, in that the number of ramets necessarily limits the number of effective breaks possible. The relationship between survival and breakage was thereby determined by first entering final plant size (number of ramets) into a general linear model comparing number of breaks in plants that survived to the last observation in January with those that did not. The effect of breakage on growth per ramet, with growth measured as total length of stem, was similarly assessed by first factoring out the number of ramets per individual. On the other hand, final number of ramets and total stem length at the end of the census period were necessarily compared directly for plants living and dead in January, to ascertain the effect of size on survival.
A general linear model allows a multifaceted analysis of the effects of the various factors being examined, through a comparison of two manners of apportioning out the variation among potential explanatory factors and variables. In the first instance, the explanatory variables are entered into the statistical model in an order in accord with a biological hypothesis or in the order in which, when considered individually, the variables are able to best explain variation in the target variable. As an explanatory variable is entered into the model, it sequesters a portion of the variation in the target variable; the next factor entered into the model is used to try to explain the remaining, unexplained variation in the target variable, and so on, until all the factors of interest have been entered into the model. The resulting significance level for each factor indicates the independent effect of the variable relative to the factors that were previously entered into the model and is derived from what is referred to as the sequential mean sums of squares. In the second instance, the significance level of a factor derives from the adjusted mean sums of squares, which is determined from the amount of the independent effect of the factor calculated after all of the other variables are first entered into the model. Please note, in a general linear model, a newly entered variable has the capacity to "explain" only that variation in the target variable that has not yet been explained by previously entered factors. This is in contrast to multiple or stepwise regression, where the amount of variation apportioned to each factor is recalculated, and changes, every time a new variable is entered into the model, and thus the variation explained is not independent of the other variables in the model. Comparison of the results of the sequential vs. adjusted models allows further assessment of the interactions between target variables (as in Pirie et al., 2000)
.
Early events may have more impact than events later in the life of a plant, and breaks in a census period, for each census period, against final plant size was also entered into a general linear model to test the effect of timing of breakage. For this analysis, both absolute number of breaks and cumulative number of breaks were used as measures of breakage. Similarly, the effect of a trauma may dissipate over time. Therefore number of breaks per day and percent breakage in a census period were compared to number of new ramets per ramet (relative growth rate) in the following census period.
Ramet lines (a ramet line = all ramets originating during the same census period and traceable back to one mother ramet) with and without breaks were compared in order to determine the effect of breakage on stem growth and mean interconnection distance at the level of the ramet. Stolon production by the ramet immediately proximal to a break, and by the line of ramets distal to a break, were compared to stolon production by ramets not associated with a break in order to establish the effect of breakage on stolon initiation.
Other effects of breakage
In the similar Cuscuta subinclusa, individuals that successfully survive from one growing season to the next show more points of infestation on hosts on which they may potentially overwinter than individuals that do not survive into a second season (Kelly, 1988, 1990
). That is, some host plants provide "refugia" from otherwise lethal conditions. During the final observation of C. corymbosa in January, living tissue was evident only on one host species, Baccharis salicifolia. Therefore, a general linear model was used to determine the relationship between number of B. salicifolia stems infested by dodder and dodder survivorship independent of any effect of whole-plant size on host infestation.
By definition, ramets of clonal plants differ from the subunits of nonclonal plants in their potential to survive independently of the main body of the plant. In order to investigate the potential selective advantage that this ability may confer upon clonal plants, we calculated the loss in tissue that would result if ramets were not able to persist when physically independent of the main body of the clone, i.e., if all tissue distal to a break were to die.
Simulation
The potential effects of breakage on whole-plant architecture were examined through simulations applying the observed responses of dodder to breakage. Dodder growth was simulated as a two-dimensional process, for the same reasons as those used in the two-dimensional simulation of rooted, ground-based clonal plants. These reasons were that (1) the resource environment, here a host branch, does not change in character over a vertical dimension (i.e., it is still a host branch whether encountered at upper or lower reaches of the branch), and (2) dodder, as do ground-based plants, tends to advance in a fairly narrow horizontal band within the depth of the resources theoretically of use to it. That being said, a dodder stolon can establish itself successfully on a host branch on which another ramet already exists, or even over an existing ramet, and thus stolons are less likely to interfere with one another than has been suggested may be the case for ground-based plants (e.g., Noble, Bell, and Harper, 1979
). In consequence, stolons were assumed not to interact within the framework of the simulation model.
For the simulation model, ten individuals were generated under each of four possible regimes: (1) growth with breakage patterns paralleling that of natural conditions; (2) growth with no breakage; (3) growth with breakage but with "nutrient limitation" simulated as biomass accumulation (length of stem tissue) capped at the average value observed under natural conditions, and (4) growth without breakage but with "nutrient limitation" simulated as in (3). This method of replicating nutrient limitation effectively mimics the progress of dodder growth on a host or host branch. On a host, and particularly annual hosts, initial dodder growth will be very vigorous until a point at which both host or host branch and dodder appear to descend rapidly into debility and sometimes death (C. K. Kelly, personal observation).
Each treatment was run for seven "cycles." A cycle constituted 7 d of growth, corresponding approximately to the schedule of the field observations. Each simulated individual, as with the real individuals, began as a single ramet. In all treatments, at the beginning of each cycle, a ramet was assigned a daily growth rate and number of stolons to be generated over the course of the cycle by random selection from a list of values for these responses for ramets of that "age" (number of cycles) as drawn from the field observations of individuals growing under natural conditions. This selection regime does not presuppose any particular shape for the distribution of values for growth responses, a concern for behavioral responses (Martin and Bateson, 1993
). Number of branches per stolon relative to stolon "age" were based on branching rates with and without breakage under natural conditions. Position and frequency of breaks were based on those observed in individuals growing under natural conditions, with extension rates assigned according to the decrement in extension recorded in the experimental severance of stolons (i.e., for fragments containing only one ramet, 0.39 that of a randomly selected value for an unbroken stolon in that cycle; for fragments containing two ramets, 0.64 that of a randomly selected value for an unbroken stolon within that cycle). Branch angle was assigned as a uniform random number between 90° and 270°, signifying the degree angle measured from the incoming stolon giving rise to that ramet, using the same algorithm with and without breakage. Angular measures are not a simple quantity in dodder, in that the angles of stolons both entering or exiting a coil change over time as the stolon elongates, which it will continue to do over several days both for the incoming and outgoing stolon. Hence, the value of the angle depends on when the measure is made relative to when the coil is established. In general, however, over the course of the census period dodder individuals advanced and expanded through the host shrubs on which the dodder occurred, and breakage was not observed to change that pattern. Thus, the above algorithm was applied across all treatments, and any error introduced by error in the algorithm would be similar across the simulation treatments.
Also drawing on the observations of individuals under natural conditions, probability of ramet mortality was assumed to be 0.01 over the course of the seven cycles for the condition without breakage; for ramets distal to a break 0.23 in the ramet immediately distal to a break, and 0.01 in ramets following that ramet. The relative "perimeter" of the two-dimensional projection of each plant was measured by drawing a line from tip to tip of stolon tips outside the body of the plant, with relative "area" represented by the area within that border.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
). Detached dodder stems became thinner over 24 h, while unmanipulated stems did not (F1,30 = 4.90, P = 0.035).
|
|
Breakage showed a significant positive relationship with final plant size, in terms of both biomass accumulation (F1,11 = 4.97, P = 0.048) and number of ramets (F1,11 = 6.25, P = 0.030). However, more connections (ramets) also signifies more interconnections, and the positive relationship between final plant size and number of breaks (Fig. 3) follows from the fact that more interconnections give more opportunities for breakage. Nonetheless, while final plant size did not predict plant survival into the dry season (F1,11 = 1.33, P = 0.273), plants with fewer breaks were more likely to survive than those with more breaks, independent of final plant size (F1,10 = 6.10; P = 0.031; Fig. 4).
|
|
Early breaks evidenced no more impact on final plant size than later breaks. Indeed, of the 28 comparisons possible of breakage with final plant size (absolute number of breaks and cumulative number of breaks per census period for seven census periods against final number of ramets and total stem length), only number of breaks in census period 7 had a significant, and positive, relationship with final number of ramets (F1,10= 7.67, r = 0.658, P = 0.02). Given 28 comparisons, this single significant correlation is no more than would be expected by chance alone. Similarly, breakage did not significantly alter subsequent growth from census period to census period; rate of ramet generation in the following census period was not significantly affected by number of breaks (y = 0.12 – 1.202x; F1,72 = 2.2229, P = 0.145, r = 0.176).
The negative effects of breakage on growth under natural conditions can be better seen at the ramet level, even if not always evident at the level of the whole individual. A number of breaks sufficient for comparison of stem extension and interconnection distance between ramet lines with and without breaks occurred only in census period 7. Because a break could have occurred at any time during the census period, it was not possible to differentiate between interconnection distances laid down before the break occurred in time, but physically following a break, and those that were laid down subsequent in time to the breakage. Therefore, the average of all interconnection distances distal to a break was used in the analysis as a conservative estimate of breakage-induced interconnection distances. Ramet lines with breaks had significantly shorter interconnection distances (F1,44 = 10.54, P = 0.002) than lines without breaks. Additionally, ramets immediately proximal to breaks developed significantly more stolons than ramets in lines not suffering breaks (Fisher F test, P < 0.05); ramets following a break produced fewer stolons than those proximal to the break (Fisher F test, P < 0.05), but not fewer than ramets in lines without breaks (Fisher F test, P > 0.05; overall F2,162 = 19.091, P < 0.0002; Fig. 5); this comparison was also restricted to census period 7.
|
2 = 49.24, P < 0.001). Unfortunately, 10 of the 14 ramets that died following breaks were in the same individual, disallowing statistical analysis of any possible additional effect of ramet death given breakage. Overall, the percentage of ramets that died relative to the total number of ramets in the population increased from 1 to 4%. In summary, at the level of the ramet, breakage was negatively associated with stem extension and ramet survivorship, shortened interconnection distances, and was positively associated with stolon initiation in ramets distal to the break.
Other effects of breakage
Contribution to biomass accumulation differed among resource types (Table 2). Assessing contribution to biomass accumulation after the effect of total number of ramets was statistically removed from the estimation of variation among parasites in final total stem length, individuals showed a steeper increase in final stem length per ramet from ramets on annual hosts than those on the one perennial host (Baccharis salicifolia), and ramets on stems of either annual or perennial hosts contributed more to parasite biomass accumulation than ramets connected to host leaves. Ramets attached to dodder stems actually showed a negative effect on biomass accumulation. Such a finding runs counter to the expectation that a greater degree of integration of resources within the body of the individual will be beneficial through increased efficiency of resource distribution (e.g., Alpert, 1996
).
|
Individuals that survived into the dry season used significantly more B. salicifolia branches than did those that did not survive into the dry season, independent of the effect of total number of ramets (F1,10 = 6.51, P = 0.029; Fig. 4b). After statistically factoring out the effect of the relationship between total number or ramets and number of ramets using B. salicifolia, breakage was negatively associated with number of ramets on B. salicifolia (F1,12 = 6.73, P = 0.025).
Simulation
Plants that suffered breakage covered significantly less area than those that did not suffer breakage (F1,37 = 22.54; P << 0.001), with no significant difference in mean area whether growth (stem length) was limited by resources ("growth" stopped at final stem length equal to average size of field population) vs. by time (number of cycles; F1,37 = 2.52; P = 0.121; Fig. 6). Further, plants that suffered breakage had significantly less perimeter per unit area than those that did not, although the direction of this effect differed depending upon whether the parasite was limited by size or by time (Table 3; Fig. 6).
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
The effect of breakage in C. corymbosa was associated instead with responses capable of changing whole-plant "shape," rather than with differences between individuals in biomass accumulation. Increased breakage was significantly associated with shorter interconnection distances at both the ramet and individual levels. Our experimental results showed that breaks induce thinning in the stolons of detached ramets; thinner stolons are associated with increased probability of coiling and ramet initiation in Cuscuta (Kelly, 1988, 1994
). Simultaneously, branch proliferation was greater in ramets physically proximal to a break than in ramets distant from breaks. Finally, a ramet immediately distal to a break did not have a significantly different probability of generating a new stolon than ramets not associated with a break, although probability of mortality of a ramet following a break increased from 1 in 100 to 1 in 4.
Simulations of C. corymbosa growth in which branching, interconnection distances, and ramet mortality were varied according to the responses observed under natural conditions indicate that such changes at the level of the ramet can translate into significant differences in whole-plant shape in such a manner as to alter resource acquisition patterns. Our simulations show that breakage responses of C. corymbosa were sufficient to decrease significantly the two-dimensional area encompassed by the individual. Further, the effect of breakage on area of the individual did not differ significantly regardless of whether or not resources were limiting (growth terminated at the average biomass value for the habitat). On the other hand, the effects of resource limitation on the capacity of the individual to acquire resources may be mitigated by the effects of a greater relative perimeter length when resources are limiting than when they are not.
It has been suggested that in plants, growth functions in "exploration" for patchily distributed resources in that extension into the habitat increases the frequency of contact with resource patches. This has been proposed particularly as an attribute of clonal plants (cf. Janzen, 1977
; Bell, 1980
; Cain, 1996
), although encounter and exploitation of multiple resource patches occur for any plant in which the spatial extent of the individual is greater than the size of a resource patch (Kelly, 1995
; Kelly and Horning, 1999
). The importance of breakage-induced changes in C. corymbosa shape relative to the advantages of "exploration" is indicated by work showing that, in a two-dimensional space such as that defined by the lateral spread of a branching organism, the single largest determinant of encounter rate with a patchily distributed resource is area, or horizontal projection of the individual. For objects of the same area, encounter rate is further enhanced by "legginess," i.e., increased length of perimeter relative to area of an object (Rubin, 1987
).
We suggest that breakage affected long-term survival in C. corymbosa through its impact on the resource-acquisition properties of whole-plant shape. Breakage appeared to decrease acquisition of hosts able to support persistence into the dry season for the parasite at this site and with it the chance for successful reproduction in more than one growing season. In our observation of parasite growth over the season, only one of the four host species exploited, Baccharis salicifolia, was able to support survival of ramets into the dry season. After factoring out the effect of whole-plant size, it is possible to see the negative relationship between breakage and number of branches of B. salicifolia supporting dodder ramets. In turn, a positive relationship between parasite survival into the dry season and number of branches of B. salicifolia supporting C. corymbosa ramets emerges when the effect of whole-plant size is first eliminated from the analysis, establishing the independent effect of acquisition of B. salicifolia on longer term parasite survival and identifying the route by which breakage decreases survivorship of individual plants. A similar dynamic of "refugia" necessary to long-term survival has been observed in C. subinclusa in the chaparral of southern California (Kelly, 1988, 1990
). Uptake sites in nonparasitic clonal species may also segregate into long-term vs. short-term values: Turkington's elegant work with Trifolium repens has shown that growth rate, growth patterns, and survivorship can differ significantly for ramets of a clonal individual depending upon the identity of neighboring plant species (Turkington and Cavers, 1978
; Turkington and Klein, 1993
; Mehrhoff and Turkington, 1996
).
However, much more surprising was our finding that breakage did not appear to impair growth and biomass accumulation of individual parasites within the course of the observation period. Our result does not appear to be an artifact of our experimental or observational design. Although our sample size was not large, it seems unlikely that the levels and types of damage were insufficient to induce measurable variation in the target variable, the greatest concern with small sample sizes. Under natural conditions, breakage levels ranged from 2 to 66% of all interconnections within an individual, with the large majority of those (78%) at sites shown experimentally to carry the greatest possible impact on growth. As outlined above, our simulations of changes in growth patterns associated with natural levels of breakage established the significant impact natural breakage has upon aspects of whole-plant shape relevant to resource acquisition, also with relatively small sample sizes. Most tangibly, that these changes can affect resource acquisition patterns is demonstrated by the negative relationship between breakage and number of ramets on persistent hosts noted above.
Our experiment and observations indicate that breakage in C. corymbosa has the capacity to interrupt natural patterns of flow from one ramet to the next of resources essential to growth of the individual, to the extent that growth and survivorship of the ramet were both negatively affected by breakage in this species. Additionally, return per ramet in terms of growth under natural conditions varied among host species, dictating an advantage to individual parasites with a greater proportion of high-return ramets able to allocate resources among ramets to maximize growth per unit of resource (e.g., Caraco and Kelly, 1991
; Oborny, 1994a, b
). To the extent that types and not just quantities of resources vary among host species, breakage would also disrupt mechanisms of uptake facilitation (Kelly and Horning, 1999
), or the balancing among subunits of disparities in the resource types available to each (sensu Liebig's Law of Limits; Pitelka and Ashmun, 1985
; Alpert, 1996
; Ryel and Caldwell, 1998
). Nonetheless, regardless of all the possible routes by which breakage may be expected to have an effect on biomass accumulation, naturally occurring breakage rates in C. corymbosa individuals did not have a significant effect on average return per ramet over the course of the observation period.
Rather, we are left with the question "how could breakage not have interfered with whole-plant biomass accumulation?" As outlined above, the conclusion that the observed changes in branching patterns induced by breakage were not sufficient to affect resource acquisition is contradicted by the evidence. However, some life history traits (such as biomass accumulation) may be invariable in the face of perturbation, not because they are unimportant to the organism, but because they are of such large importance that past selection insured their invariability (Charnov, 1993
; Caswell, 2000
). Biomass accumulation in C. corymbosa would seem a likely candidate for such a selective regime. As with many plants, biomass accumulation within a season in Cuscuta is correlated with seed set in that season (Kelly, 1988, 1990
). Further, biomass accumulation is of sufficient importance that C. subinclusa and C. europaea, ecologically similar congeners, both possess complex behavioral mechanisms that maximize growth during the course of the growing season (i.e., optimize costs and benefits of resource uptake through schedules of choice or investment gauged to expected reward levels; Kelly, 1990, 1992, 1994
). Lastly, as do most researchers, we have for logistical reasons chosen to study our organism at a site where it is common and robust, and we therefore can reasonably assume a fair match between the habitat in which we have found C. corymbosa and its responses to that habitat (Harvey and Pagel, 1991
). Therefore, we surmise that we found no effect of breakage on biomass accumulation in our study because the observed C. corymbosa growth responses serve to minimize the impact of breakage on resource acquisition and growth over the course of a season. Several responses observed here may be capable of "compensating" for breakage in this manner: for example, the increased stolon generation of ramets proximal to a break or the thinning and subsequent greater responsiveness of detached ramet segments.
The differing effects of breakage on within-season growth vs. the chance of perennation are also consistent with Caswell's thesis of the relationship between importance and variation of a trait. A corollary of Caswell's observation is that traits of lesser fitness value may vary greatly, or, at least, more than those of more fundamental importance to the individual. Theory has shown that for many species capable of persistence, perennation can be viewed as a fortuitous event, but not likely to be one worth investing in greatly and certainly not at the cost of lowered reproductive success in the current season (Charnov and Schaffer, 1973
; Fox, 1990a, b
). In this light, it is relevant that the relationship between biomass accumulation and perennation may represent a conflicting goal in C. corymbosa, in that the host species that allows persistence into the dry season is not the host species conferring the greatest growth return per ramet (Table 2). By mid-dry season, 5 of the 13 individuals had already died; from the similar pattern of mortality in C. subinclusa, 10% of the population might be expected to make it through to a second season in a normal year (Kelly, 1988
). Off-season flooding such as that that wiped out the parasite population at the end of this study is unusual, but not unheard of (Bullock and Solis-Magallanes, 1990
), further discounting the potential value of perennation. Adding to this, the almost 60% decrease in growth of a ramet exploiting B. salicifolia relative to the growth increment of a ramet on an annual host indicates that the fitness to be gained by a second reproductive season would have to be very large for long-term persistence to be favored at the cost of within-season reproductive success. It is relevant to this point that in C. subinclusa, individuals arising from overwintered tissue have been shown to support at most a two-fold advantage in biomass accumulation over individuals starting from seed in the same season, but more often much less than that (Kelly, 1988, 1990
).
Our hypothesis that C. corymbosa responses to breakage are a result of selection by naturally occurring levels of breakage within its native habitat is consistent with the observation that experimentally induced traumatic breakage has been invariably reported in the literature to have a negative effect on biomass accumulation. In this context, the negative effects of experimentally induced breakage observed in so many studies (e.g., Hartnett and Bazzaz, 1983
; Alpert and Mooney, 1986
; Jónsdóttir and Callaghan, 1988
; Welker, Briske, and Weaver, 1991
; Evans and Whitney, 1992
; Alpert, 1996
; Charpentier, Melséard, and Thompson, 1998
; Pennings and Callaway, 2000
) would be a result of breakage patterns outside the normal schedule of breakage frequency or position, or both; small or insignificant effects of experimental breakage would follow from treatments that deviated less from the breakage schedule an individual experiences under the natural conditions from which it is derived. The data capable of evaluation of our conjecture are not currently available, but our results suggest that such data may be of interest.
Lastly, this study shows that under natural conditions, the exploitation of either refugia or patchy resources was greatly enhanced by the capacity for subunits to persist independently of one another. In a nonclonal plant, loss of a segment of tissue necessarily results in loss of all tissue subsequent to that segment. For C. corymbosa, mortality of all post-break tissue would have meant a 53% decrease in average biomass of individuals to that observed. When the effect of breakage is combined with regular ramet mortality, 3 out of 13 of the individuals studied here would have been dead before the end of the observation period, well before flower and fruit set. Further, a dependence on a continuing physical link between subunits to the degree necessarily true in nonclonal plants would have meant an average loss of 44% of all ramets on B. salicifolia for the C. corymbosa individuals observed here, and a concomitant decrease in the probability of growth and seed set into additional growing seasons. Thus, in this species, the potential independence of subunits that is the defining characteristic of plant clonality directly enhances survival and reproduction.
In conclusion, given the common use of traumatic breakage as an experimental tool in the study of clonal plants (e.g., Hartnett and Bazzaz, 1983
; Alpert and Mooney, 1986
; Jónsdóttir and Callaghan, 1988
; Welker, Briske, and Weaver, 1991
; Evans and Whitney, 1992
; Alpert, 1996
; Charpentier, Melséard, and Thompson, 1998
; Pennings and Callaway, 2000
), the fit between our hypothesis of growth responses gauged to expected breakage rates and the evidence indicates that further investigation of this question may be appropriate. However, our hypothesis is also intriguing for other associations that it generates. That is, clonal plant species can be robust and dominant within the habitats in which they occur, with the grasslands and salt marshes of the world being noted examples (and see, e.g., Rodwell, 1991a, b, 1992, 1995
). Nonetheless, clonal species have smaller range sizes than phylogenetically related, nonclonal species (Kelly and Woodward, 1996
), indicating a pattern running counter to the general observation that locally abundant species will also be widely distributed (Hengeveld and Haeck, 1982
; Brown, 1984
). We note that a match of plant growth responses or "shape" to breakage rate such as that suggested here could augment the capacity to exploit a particular habitat, but may also impart a sensitivity to changes in ambient conditions that would limit movement between habitats to produce in angiosperms the observed, smaller range sizes of clonal species in comparison to those of nonclonal species.
|
FOOTNOTES |
|---|
4 Author for reprint requests (Current address: School of Biological Sciences, University of Southampton, Bassett Crescent East, Southampton SO16 7PX; (Fax +44 1865 271256; ckk{at}soton.ac.uk
).
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Alpert P. 1996 Nutrient sharing in natural clonal fragments of Fragaria chiloensis. Journal of Ecology 84: 395-406[CrossRef][ISI]
———, and H. A. Mooney 1986 Resource sharing among ramets in the clonal herb Fragaria chiloensis. Oecologia 70: 227-233[CrossRef][ISI]
Ashton F. M. D. Santana 1976 Cuscuta spp. (Dodder): a literature review of its biology and control. University of California Division of Agricultural Sciences
Bell A. D. 1980 Adaptive architecture in rhizomatous plants. Botanical Journal of the Linnean Society 80: 125-160
Berg E. E. J. L. Hamrick 1994 Spatial and genetic structure of two sandhills oaks: Quercus laevis and Quercus margaretta. American Journal of Botany 81: 7-14
Brown J. H. 1984 On the relationship between abundance and distribution of species. American Naturalist 124: 255-279[CrossRef][ISI]
Bullock S. H. J. A. Solis-Magallanes 1990 Phenology of canopy trees of a tropical deciduous forest in México. Biotropica 22: 22-35[CrossRef][ISI]
Cain M. 1996 Spatial models of foraging in clonal plant species. American Journal of Botany 83: 76-85[CrossRef][ISI]
Caraco T. C. K. Kelly 1991 On the adaptive value of physiological integration in clonal plants. Ecology 72: 81-93[CrossRef][ISI]
Caswell H. 2000 Prospective and retrospective perturbation analyses: their roles in conservation biology. Ecology 81: 619-627[ISI]
Charnov E. L. 1993 Life history invariants. Oxford University Press, Oxford, UK
———, and W. M. Schaffer 1973 Life history consequences of natural selection: Cole's result revisited. American Naturalist 107: 791-793[CrossRef][ISI]
Charpentier A. F. Melséard J. D. Thompson 1998 The effects of rhizome severing on the clonal growth and clonal architecture of Scirpus maritimus. Oikos 83: 107-116[CrossRef][ISI]
Cook R. E. 1978 Asexual reproduction: a further consideration. American Naturalist 113: 769–772
Evans J. P. S. Whitney 1992 Clonal integration across a salt gradient by a nonhalophyte, Hydrocotyle bonariensis. American Journal of Botany 79: 1344-1347[CrossRef][ISI]
Feeny P. 1992 The evolution of chemical ecology: contributions from the study of herbivorous insects. In G. A. Rosenthal and M. Berenbaum [eds.], Herbivores: their interactions with secondary plant metabolites, 1–44. Academic Press, New York, New York, USA
Fox G. A. 1990a Demographic stochasticity and the evolution of annual versus perennial life histories. Evolutionary Ecology 7: 1-15
———. 1990b Perennation and the persistence of annual life histories. American Naturalist 135: 829-840[CrossRef][ISI]
Hartnett D. C. F. A. Bazzaz 1983 Physiological integration among intraclonal ramets in Solidago canadensis. Ecology 64: 779-788[CrossRef][ISI]
Harvey P. H. M. D. Pagel 1991 The comparative method in evolutionary biology. Oxford University Press, Oxford, UK
Hengeveld R. R. Haeck 1982 The distribution of abundance. Journal of Biogeography 9: 303-316[CrossRef][ISI]
Janzen D. H. 1977 What are dandelions and aphids?. American Naturalist 111: 586-589[CrossRef][ISI]
Jónsdóttir I. S. T. V. Callaghan 1988 Interrelationships between interconnected tillers of Carex bigelowii. Oikos 52: 120-128[CrossRef][ISI]
Kelly C. K. 1988 Host use and foraging in the parasitic plant Cuscuta subinclusa. Ph.D. dissertation, University of Arizona, Tucson, Arizona, USA
———. 1990 Plant foraging: a marginal value model and coiling response in Cuscuta subinclusa. Ecology 71: 1916-1925[CrossRef][ISI]
———. 1992 Resource choice in Cuscuta europaea. Proceedings of the National Academy of Science, USA 89: 12 194-12 197
———. 1994 On the economics of plant growth: stolon length and ramet initiation in the parasitic clonal plant Cuscuta europaea. Evolutionary Ecology 8: 459-470
———. 1995 Thoughts on clonal integration: facing the evolutionary context. Evolutionary Ecology 9: 575-585[CrossRef][ISI]
———, and K. Horning 1999 Acquisition order and resource value in Cuscuta attenuata. Proceedings of the National Academy of Sciences, USA 96: 13 219-13 222
———, D. L. Venable K. Zimmerer 1988 Host specialization in Cuscuta costaricensis: an assessment of host use relative to host availability. Oikos
