• Xanthium strumarium;
  • flowering time;
  • geographic range;
  • latitude;
  • life-history trade-offs;
  • local adaptation;
  • marginal populations;
  • phenology;
  • reciprocal transplant;
  • species’ border


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Adaptation to local environments may be an important determinant of species’ geographic range. However, little is known about which traits contribute to adaptation or whether their further evolution would facilitate range expansion. In this study, we assessed the adaptive value of stress avoidance traits in the common annual Cocklebur (Xanthium strumarium) by performing a reciprocal transplant across a broad latitudinal gradient extending to the species’ northern border. Populations were locally adapted and stress avoidance traits accounted for most fitness differences between populations. At the northern border where growing seasons are cooler and shorter, native populations had evolved to reproduce earlier than native populations in the lower latitude gardens. This clinal pattern in reproductive timing corresponded to a shift in selection from favouring later to earlier reproduction. Thus, earlier reproduction is an important adaptation to northern latitudes and constraint on the further evolution of this trait in marginal populations could potentially limit distribution.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

The ability of plant species to adapt to local environmental conditions is an important determinant of their geographic range. Populations are often adapted to local conditions along geographic gradients (examples in Levin, 1993; Joshi et al., 2001) and introduced species can undergo rapid evolution accompanying their range expansion (Weber & Schmid, 1998; Davis & Shaw, 2001; Maron et al., 2004). Recent theoretical work has demonstrated that evolutionary constraints on one or more traits necessary for local adaptation may limit a species’ geographic distribution (Kirkpatrick & Barton, 1997; Case & Taper, 2000; García-Ramos & Rodríguez, 2002; Holt, 2003). However, despite the potential for local adaptation to shape species’ ranges, very little is known about what types of traits contribute to local adaptation across broad geographic gradients and whether further evolution of these traits in marginal populations would result in range expansion (Hoffmann & Blows, 1994; Jonas & Geber, 1999; Eckhart et al., 2004). Identifying traits important for local adaptation across geographic gradients is an initial step in understanding how constraints on the evolution of such traits may ultimately determine geographic distribution (Hoffmann & Blows, 1994; Sih & Gleeson, 1995; Jenkins & Hoffmann, 1999; Eckhart et al., 2004).

Latitudinal range limits of many plant species are often associated with decreases in the length of the growing season and average temperature (e.g. Qian et al., 2003; Root et al., 2003). This presents two problems for plants: firstly, growth and reproduction must be completed in a shorter time period and secondly, growth conditions during this period are less favourable. In general, a species can adapt to such stresses by altering its life-history to complete its life-cycle during less stressful (in this case warmer) conditions, and/or by increasing its ability to tolerate the stress (Grime, 1979; Chapin, 1980; Hoffmann & Parsons, 1991; Stanton et al., 2000).

Life-history traits associated with stress avoidance have evolved along latitudinal and similar altitudinal gradients within many species’ ranges (for reviews, see Grime, 1990; Heide, 1994; Davis & Shaw, 2001). Marginal populations often reproduce earlier (Weber & Schmid, 1998; Jonas & Geber, 1999; Olsson & Ågren, 2002), and in some cases germinate later (Nishitani & Masuzawa, 1996; Jonas & Geber, 1999) than those from lower latitudes or altitudes. These changes may allow species to avoid damage from late spring frosts and to complete reproduction before the end of the growing season. Earlier reproduction can also incur fitness costs: plants that reproduce earlier have fewer resources available for subsequent growth and reproductive output (Reekie & Bazzaz, 1987; Geber, 1990; Dorn & Mitchell-Olds, 1991). As a result, evolution of early reproduction is predicted to be accompanied by evolution of greater prereproductive growth rates to offset these fitness costs (Arendt, 1997; Li et al., 1998; Stanton et al., 2000).

While population differences in life-history traits are well documented, few studies have determined the adaptive value of these differences along a broad latitudinal gradient (but for example, see Etterson, 2004). To do so, reciprocal transplant experiments are needed both to reveal the clinal pattern (direction) of phenotypic change along the gradient as well as to measure changes in the direction and strength of selection on traits (e.g. Bennington & McGraw, 1995; Donohue et al., 2000; Etterson, 2004). The clinal pattern of trait change is the result of environmental effects on phenotype as well as the genetic differences between populations that have evolved to enhance or offset these environmental effects (Conover & Schultz, 1995; Eckhart et al., 2004). Shifts in selection along the gradient indicate that corresponding trait change is adaptive (Wade & Kalisz, 1990; Bennington & McGraw, 1995; Etterson, 2004).

In this study, we performed a reciprocal transplant to determine the importance of stress avoidance traits for local adaptation in the common North American annual Cocklebur (Xanthium strumarium) along a latitudinal gradient extending from the interior to the northern edge of its range in central Michigan. Traits exhibiting an adaptive, clinal pattern of evolution along this gradient were considered important for latitudinal adaptation. Moreover, traits important for local adaptation to environmental conditions within the species’ current range may facilitate adaptation to further changes in these conditions beyond the range (Antonovics, 1976; Jenkins & Hoffmann, 1999; Jonas & Geber, 1999; Eckhart et al., 2004). Our hypothesis was that if populations at higher latitudes evolved to avoid the stress of cooler, shorter growing seasons, they would alter their phenology by germinating later but flowering and senescing earlier. In addition, more rapid initial growth would compensate for any resulting decrease in prereproductive growth period. We addressed three specific questions: (1) Are populations locally adapted to their respective latitudes? (2) Have traits evolved in a clinal pattern across this latitudinal gradient? (3) Are trait changes across this gradient adaptive? This study is one of the first attempts to directly measure the adaptive value of stress avoidance traits across a broad latitudinal gradient.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Study Species

Cocklebur (X. strumarium L. Asteraceae) is a common annual native to the Americas (Löve & Dansereau, 1959). In North America, it has a well-documented range that extends as far north as central Saskatchewan (Weaver & Lechowicz, 1982) (Fig. 1a). The northernmost distribution of Cocklebur in central Michigan occurs in Isabella County (∼43.8°N) (Voss, 1996; personal communications from Michigan State University Agricultural Extension Agents). Cocklebur grows in a variety of disturbed habitats, especially agricultural fields (Weaver & Lechowicz, 1982; Hocking & Liddle, 1986). It is a qualitative short day species and so flowering is induced when day length declines below a critical photoperiod (Salisbury, 1969). Separate male and female capitula (inflorescences) are produced; the latter developing into burrs with two seeds each. Plants predominantly self-fertilize (Weaver & Lechowicz, 1982). The hooked spines on the burrs facilitate dispersal via animals and water (Hocking & Liddle, 1986). Cocklebur is such a successful disperser and colonizer that it is now present on every continent except Antarctica (McMillan, 1971, 1974a,b).


Figure 1. The North American range of Cocklebur (a) and an expanded section of this map (b) showing the locations of the common gardens (square symbols, bold type) and source populations (round symbols, plain type) used in this experiment. (Range data from Weaver & Lechowicz, 1982; personal communications from Michigan State University Agricultural Extension Agents).

Download figure to PowerPoint

Locations of source populations and gardens

During the fall and winter of 1996–97, burrs were collected from six source populations, two from each of three locations along a latitudinal cline – central Tennessee (southern), south central Indiana (intermediate) and central Michigan (range edge) (see Fig. 1b for population locations). Source populations were located along the edges of corn or soya bean fields and thus evolved under similar competitive environments, since these crops are usually rotated. Common gardens were established in the three latitudinal regions from which populations were collected (see Fig. 1b for garden locations). Garden locations had progressively shorter growing seasons and lower monthly temperatures at higher latitudes (Fig. 2) (Koss et al., 1988, and National Climatic Data Center, All gardens were established in fields that were cultivated in the past, but had been fallow for several years prior to this experiment.


Figure 2. Average monthly temperatures from 1971 to 2000 (solid lines and symbols), and during the year of this study, 1997 (dashed lines, open symbols) at the three National Climatic Data Center sites closest to each of the common gardens. Dates of 50% frost probability at the beginning and end of the growing season are indicated below the horizontal axis (Data from Koss et al., 1988;

Download figure to PowerPoint

Sowing and transplanting into common gardens

On 3 May 1997, burrs from all source populations were soaked for 24 h in tap water to break dormancy, then sown into flats filled with sterilized soil. Flats were placed outdoors at the Indiana University Botany Experimental Field in Bloomington, Indiana. Seedlings began to emerge 6 days after sowing. Emergence dates in the flats corresponded to the emergence period of naturally occurring seedlings in agricultural fields adjacent to the intermediate (Indiana) garden (T. M. Griffith, pers. obs.), and the range of germination times was similar to that for a naturally occurring population (Zimmerman & Weis, 1984).

The design of the three gardens consisted of six blocks each with seven plants per population (total n = 3 gardens × 6 populations per garden × 42 plants per population = 756). The columns and rows of each block were spaced 1.5 m apart. Prior to planting, each garden was disked to simulate the agricultural conditions in which source populations had grown. Seedlings were transplanted into the intermediate and range edge gardens on 27 May–1 June. Rain at the southern garden delayed disking and transplanting by 26 days. However, during this time, temperature at the intermediate site (where flats with seedlings were kept) was very similar to that at the southern garden (Fig. 2) and seedling development was not affected (T. M. Griffith, unpublished). Surrounding vegetation was mowed as needed to maintain a similar competitive environment in each garden.

Data Collection

Germination in the flats was recorded daily. Following transplantation, the flowering and senescence status of each plant was recorded once per week until all plants were dead. Flowering was defined as the appearance of developing anthers from beneath the involucral bracts of the staminate head. Senescence was defined as plant death resulting from either frost or endogenous (internal) control (Watson & Lu, 1999). Senescence at the southern garden was entirely endogenous (by 27 October), however a hard frost at the intermediate garden (24 October) and a snowstorm at the range edge (26 October) killed plants in those gardens that had not yet senesced. Primary branch number was measured as an indicator of plant size twice during growth: once prior to flowering (13–19 July) and once after all plants had flowered (22–29 September) (correlation between branch number and above-ground plant mass r = 0.810 and 0.862 pre- and post-flowering, respectively, T. M. Griffith unpublished). Plants for which the stem or branches had been accidentally damaged (n = 54) were removed from all subsequent data sets. After all plants died, burrs were stripped from each plant and weighed. Plant burr mass provided a better approximation of fitness than burr number because many burrs, especially among southern populations at the range edge, were below the threshold size at which their seeds would germinate (T. M. Griffith, unpublished).

Statistical analysis

Population differentiation along the latitudinal gradient was assessed in three ways. Firstly, to determine whether population differences existed between source latitudes, a univariate, mixed model anova was performed for each hypothesized stress avoidance trait (germination, flowering and senescence time, and preflowering primary branch number) and fitness (burr mass). The terms in the models were source latitude, source population (nested within source latitude), garden, block (nested within garden), and the interaction among all terms except block. Source latitude refers to the latitudinal region (southern, intermediate, or range edge) from which populations were collected while source population refers to the specific locations of the original populations. Source population and block were treated as random factors. Terms were tested over their appropriate denominators for this mixed model (SPSS v.10 UNIANOVA). Germination time was inverse transformed and burr mass was log(x + 1) transformed to increase homogeneity of variances. Secondly, to determine whether there was a latitudinal trend in the differences between source populations, each trait was regressed against source population latitude (six source population latitude, measures from all three gardens were included). Because there were multiple individuals for each source population latitude, the significance of the regression coefficient was tested over the mean squares of the population deviations from the regression line (Sokal & Rohlf, 1995, box 14.4). Thirdly, to determine whether there was a latitudinal trend in the differences between native populations, each trait was again regressed against source population latitude, but only measures from the source populations in their native gardens were included (i.e. southern source populations in the southern garden, etc.). This analysis indicates how genetic differences between source populations would translate into phenotypic differences between populations at their native latitude along the gradient.

Post hoc tests were used to further assess adaptation of populations to their source latitudes. Fitness differences between populations were compared with a Tukey's HSD test to see whether native populations had the highest fitness in each garden. Then, fitness differences between populations were again compared after ancova was used to remove the effect of the stress avoidance traits on fitness in each garden. Any reduction in population fitness differences was thus attributable to the effect of measured or closely correlated traits on fitness (e.g. Donohue et al., 2000). Each ancova included source population as a fixed factor and all avoidance traits as covariates.

We assessed the adaptive value of trait changes along the latitudinal gradient by comparing changes in the strength of selection using a combination of selection differentials, selection gradients and path analyses (e.g. Mitchell-Olds & Bergelson, 1990; Dudley, 1996; Weinig, 2000; Volis et al., 2004). Selection differentials (s) indicate the total strength of linear selection on traits in each garden while selection gradients (β) and path analyses can be used to partition selection into direct and indirect effects. Selection gradients indicate the direct effect of each trait on fitness independent of other traits (Lande & Arnold, 1983; Conner & Hartl, 2004). When traits are expressed at different developmental times, as is the case in this study, traits expressed earlier in development can also causally affect fitness indirectly by altering traits expressed later in development (Mitchell-Olds & Bergelson, 1990; Kingsolver & Schemske, 1991; Scheiner et al., 2000). In such cases, direct, adaptive selection is appropriately measured as both the direct and indirect causal effects of each trait on fitness (Scheiner et al., 2000). Thus, we used path analysis to measure the direct effects as well as the sequential indirect effects of stress avoidance traits on fitness (e.g. Mitchell-Olds & Bergelson, 1990; Jordan, 1992; Weinig, 2000). The model was constructed to show the sequential effect of phenology (germination, flowering and senescence time) on growth (preflowering and post-flowering primary branch number) and fitness. Paths from germination indicated how emergence influenced initial growth (preflowering branch number) and the timing of reproduction (flowering). Paths from preflowering branch number to post-flowering branch number and flowering time indicated how initial growth influenced reproductive timing (e.g. through competency to day length cues, Salisbury, 1981) and subsequent growth. The effect of reproductive timing on fitness was considered via the effects of flowering time on post-flowering size (branch number) and longevity (senescence time). All analyses were performed using pooled data from populations in each garden to expand the range of phenotypes and thus better detect difference in selection and the strength of trait interactions between gardens (Wade & Kalisz, 1990; e.g. Weinig, 2000; Etterson, 2004).

For each stress avoidance trait in each garden, selection differentials were calculated as the covariance between the standardized trait and relative fitness and selection gradients were calculated as the multiple regression coefficient of the standardized trait on relative fitness (SPSS v.10 REGRESSION) (Lande & Arnold, 1983). Selection differentials and gradients were considered significant if the regression coefficient differed significantly from 0 at α = 0.05 using a sequential Bonferroni correction (Rice, 1989). Changes in differentials and gradients between each pair of gardens were assessed using ancova. Because all possible pair-wise comparisons between gardens were made, a strict Bonferroni test (α = 0.05) was used to control the among-garden error rate. Path coefficients were calculated as the multiple regression coefficients for each successive set of independent traits on a dependent trait in the model (SPSS v.10 REGRESSION). Traits other than relative fitness were standardized.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Latitudinal adaptation

Our results confirmed that populations were adapted to their native latitudes. The interaction between source latitude and common garden was highly significant for plant burr mass (Table 1), with native populations tending to have the highest average burr mass in each garden (Fig. 3a). The only exception to this pattern occurred in the intermediate garden, where one southern population had a significantly higher average burr mass than either of the native intermediate populations. In contrast to reproduction, there was little variation in survival – only nine plants died in all six populations.

Table 1.  ANOVA results for stress avoidance traits and fitness (burr mass).
Model termGermination timeFlowering timeSenescence timePreflowering 1° branch numberBurr mass
  1. In each cell, top number is the F ratio, lower numbers are numerator and denominator d.f. Bold values are statistically significant; *P < 0.05; **P < 0.01; ***P < 0.001.

Source latitude1.927.4*36.7**741.9***1.1
2, 3.02, 3.02, 3.02, 3.02, 3.0
Source pop. (source lat.)13.4***8.4*
3, 7443, 6.03, 6.03, 6.03, 6.0
2, 5.92, 6.62, 15.72, 10.0
Block (garden)***2.7**
15, 65315, 64815, 71515, 647
Source lat. × garden1.97.0*2.019.6**
4, 6.04, 6.04, 6.04, 6.0
Source pop. (source lat.) × garden13.2***12.9***1.64.3***
6, 6536, 6486, 7156, 647

Figure 3. Average population burr mass (a) and least squares (LS) mean burr mass from ancova (b) for source populations in each of the three common gardens. In b, the LS means indicate population burr mass after removing the effects of measured stress avoidance traits. Within each garden, populations not sharing a letter differ significantly (Tukey's HSD test α = 0.05).

Download figure to PowerPoint

Latitudinal adaptation largely overcame the effects of climatic variation along the latitudinal gradient. Despite shorter and cooler growing seasons at higher latitudes, there was no latitudinal trend in the average fitness (burr mass) of native populations (Table 2, column B). Average fitness of the native populations in the southern garden was strikingly similar to that of the native populations at the range edge (71.5 and 72.2 g, respectively). Observed fitness differences among populations within gardens were largely explained by differences in the measured stress avoidance traits. When phenotype differences among populations were removed using an ancova, few significant fitness differences remained (Fig. 3b).

Table 2.  Latitudinal trends (standardized regression coefficients) among all source populations (column A) and among only source populations in their native gardens, i.e. southern populations in the southern garden, etc. (column B). Column A indicates genetic differences between populations while column B indicates the net change in trait phenotype along the gradient.
TraitLatitudinal trends
(A) Trait v. source population latitude (all populations, all gardens)(B) Trait v. source population latitude (native populations only)
  1. * P  < 0.05; **P < 0.01.

Germination time −0.146
Flowering time −0.820** −0.775**
Senescence time −0.615** −0.042
Preflowering 1° branch number0.221*0.253
Burr mass −0.0370.000

Clinal patterns in trait change

Of the four stress avoidance traits measured in this study (germination, flowering and senescence time and preflowering branch number), only flowering time changed in a clinal pattern along the latitudinal gradient (Table 2, column B). Populations from higher source latitudes flowered significantly earlier than those from lower source latitudes in all gardens (Table 1, source latitude term; Table 2, column A). The earlier flowering of higher latitude source populations more than offset the delay in flowering induced by higher latitude gardens (Table 1, garden term; Fig. 4a) (Delayed flowering at higher latitudes is consistent with the effect of longer day lengths on a short day species.). Thus as hypothesized, higher latitude source populations in their native latitude actually flowered significantly earlier than lower latitude source populations in their native latitude (Table 2B). This earlier flowering resulted in a longer interval between flowering and senescence (from 55.8 days for native southern populations compared with 70.7 days for native range edge populations), thereby giving native populations at the range edge more time for reproduction during the relatively warm temperatures of August.


Figure 4. Population means of stress avoidance traits for all six source populations in each of the three common gardens. Symbols are as in Fig. 3– squares, solid lines = southern (Tennessee) source populations; circles, dashed lines = intermediate (Indiana) source populations; triangles, dotted lines = range edge (Michigan) source populations. Error bars ± 1 SE.

Download figure to PowerPoint

There was no clinal pattern among native populations for other traits (Table 2, column B). Contrary to our predictions, native populations at the range edge did not produce significantly more branches prior to flowering or senesce significantly earlier than native populations in the lower latitude gardens (Fig. 4b,c; Table 2, column B). There was no significant difference between source populations for germination time (Table 1, source latitude term).


There was a significant change in the strength of both selection differentials (total selection) and gradients (direct effects) between gardens for all stress avoidance traits except germination time, for which selection was not significant in any garden (Table 3). However, the direction of selection gradients changed only for flowering time: later flowering increased fitness in the intermediate garden while earlier flowering increased fitness at the range edge. Furthermore, flowering time was the only trait for which selection gradients differed between the edge and intermediate gardens, indicating that of the measured traits, a change in flowering time was most likely to be adaptive at the range edge.

Table 3.  Linear selection differentials (s) and gradients (β) for traits in each garden.
TraitsCommon garden
Southernmost (Spring Hill, TN, USA)Intermediate (Bloomington, IN, USA)Range edge (Isabella Co., MI, USA)
  1. *Significant at α = 0.05; **Significant at α = 0.01 with sequential Bonferroni test.

  2. For a trait, s or β with same letter does not differ significantly (α = 0.05, strict Bonferroni).

Germination time
  s0.030a −0.029a −0.156a
  β −0.044a,b0.044a −0.091b
Flowering time
  s0.427**a0.223**b −0.567**c
  β 0.023a0.138**a −0.663**b
Senescence time
 s0.551**a0.261**b −0.365**c
  β 0.549**a0.283**b0.240**b
Preflowering 1° branch number
  s −0.093a0.238**b0.491**c
  β 0.026a0.398**b0.389**b

Path analysis confirmed a shift in selection from favouring later reproduction in the lower latitude gardens to favouring earlier reproduction at the range edge. In the southern and intermediate gardens, selection for later flowering was primarily the result of the effect of flowering time on subsequent growth and development. Plants that delayed reproduction were bigger and longer-lived (Fig. 5a,b; flowering time[RIGHTWARDS ARROW]post-flowering branch number and flowering time[RIGHTWARDS ARROW]senescence time), and as a result had higher fitness (post-flowering branch number[RIGHTWARDS ARROW]relative fitness, and senescence time[RIGHTWARDS ARROW]relative fitness). Thus, the direct selection (sum of direct and indirect forward, causal paths on fitness) was positive (0.451 and 0.363 in the southern and intermediate gardens, respectively), indicating that earlier flowering was adaptive in both gardens. By contrast, at the range edge, there was a shift in the direction of direct selection (−0.509) so that earlier flowering was adaptive. This shift was the result of a substantial increase in the direct effect of flowering on fitness (Fig. 5c; flowering time[RIGHTWARDS ARROW]relative fitness). Although plants that delayed reproduction still grew bigger (Fig. 5c; flowering time[RIGHTWARDS ARROW]post-flowering branch number), they lacked time to produce and mature as many burrs. For example, the southern source populations were the last to flower and largest in post-flowering size of all the populations at the range edge (anova source latitude term, post-flowering branch number P < 0.05), but had the smallest average burr masses (Fig. 3a).


Figure 5. Path analyses showing direct and indirect sequential effects of traits on relative fitness in each garden. Line thickness indicates magnitude of path coefficients. Dashed lines indicate negative coefficients.

Download figure to PowerPoint

Path analysis also showed that changes in selection differentials (total selection) on traits other than flowering time were the result of their correlation with flowering time. The significant shift in total selection from favouring later senescence in the intermediate gardens to earlier senescence at the range edge (Table 3) was caused by selection for earlier flowering, which was strongly correlated with earlier senescence in all gardens (Fig. 5a–c; flowering[RIGHTWARDS ARROW]senescence). Likewise, the slight increase in the strength of total selection on preflowering size (branch number) at the range edge (Table 3) was not the result of an increase in the direct effect of preflowering branch number on fitness (Fig. 5b,c; preflowering branch number[RIGHTWARDS ARROW]relative fitness). Instead, the advantage of early growth at the range edge came at least partially from hastening flowering in an environment where earlier flowering was beneficial (Fig. 5c; preflowering branch number[RIGHTWARDS ARROW]flowering time[RIGHTWARDS ARROW]relative fitness).


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Results from this study clearly demonstrate the importance of stress avoidance, particularly early reproduction, for adaptation to cooler, shorter growing seasons at higher latitudes. There was a high degree of local adaptation of populations to their native latitudes and the hypothesized stress avoidance traits accounted for most of the fitness differences between populations. Native populations in the higher latitude gardens had evolved to reproduce earlier than native populations in the lower latitude gardens, thereby increasing the duration of the reproductive period. At the range edge, the earlier and longer reproductive period of the native populations enabled them to achieve an absolute fitness similar to that of the native populations in the two lower latitude gardens. This latitudinal cline in reproductive timing corresponded to a shift in direct selection from favouring later reproduction at lower latitudes to favouring earlier reproduction at the range edge. Together, the clinal change in flowering time and shift in selection indicate that earlier flowering is an important adaptation to higher latitudes for this species.

Latitudinal adaptation and geographic distribution

Cocklebur populations in their non-native latitude tended to have substantially lower fitness than native populations, particularly at the range edge where the fitness of every non-native population was significantly less than those of the two source populations native to that latitude. This pattern is consistent with a previous study in which Cocklebur populations collected in Mississippi and planted in a Michigan garden failed to produce any burrs at all (Ray & Alexander, 1966). Together, these studies indicate that local adaptation enables Cocklebur to persist as far north as it currently does. Such findings support recent suggestions that adaptation to new climatic conditions may be an important component of species’ ability to persist and spread (Huey et al., 2000; Davis & Shaw, 2001; García-Ramos & Rodríguez, 2002; Lee, 2002; Rice & Emery, 2003; Maron et al., 2004).

Reproductive timing, stress avoidance and latitudinal adaptation

Of the stress avoidance traits measured in this study, reproductive timing was the only one that exhibited a clearly adaptive, clinal pattern of evolution along the gradient. Native populations at higher latitudes flowered before native populations at lower latitudes. This pattern was the result of substantial evolution of earlier flowering in higher latitude source populations. The clinal trend in flowering time corresponded to a shift in selection. In the lower latitude gardens, later flowering caused plants to grow larger, as evidenced by more post-flowering branches, and achieve a higher reproductive output. This result is consistent with life-history theory, which predicts that earlier reproducing individuals will pay a cost in terms of having fewer resources to allocate to future growth and reproduction (Dorn & Mitchell-Olds, 1991; Sugiyama & Hirose, 1991; Kozlowski, 1992; Arendt, 1997). However, in the cooler, shorter growing seasons at the range edge, plants with delayed flowering still grew large but failed to achieve as high a reproductive output as earlier flowering plants by the end of the growing season. Thus, later flowering was adaptive in the two lower latitude gardens while earlier flowering was adaptive at the range edge.

Previous physiological studies on Cocklebur have shown that populations from higher source latitudes have evolved longer critical photoperiods (Ray & Alexander, 1966; McMillan, 1974a). These longer critical photoperiods speed flowering because Cocklebur is induced to flower when summer day lengths decrease below the critical photoperiod (Salisbury, 1969, 1981). However, growth under longer summer day lengths at the higher latitudes where these populations are native would have the opposite effect of delaying flowering. Thus, the extent to which the effect of longer critical photoperiods was offset by growth at higher latitudes was not known. Our reciprocal transplant demonstrated that the evolution of earlier flowering time in higher latitude source populations actually overcame the delayed flowering at these latitudes and confirms that the earlier flowering at higher latitudes is adaptive.

The adaptive value of early reproduction at higher latitudes was enhanced by the surprising lack of fitness costs. Although a tradeoff between reproductive timing and fitness was evident within gardens (southern and intermediate), no such tradeoff was evident between gardens. Despite flowering earlier, the range edge populations in their native garden did not have a significantly lower absolute fitness than southern populations in their native garden. This lack of fitness costs between gardens was even more surprising because there was no evidence that size costs associated with early reproduction were offset by greater prereproductive growth: native populations did not differ in their preflowering size (branch number). There are two possible explanations for this lack of fitness costs. Firstly, populations adapted to higher latitudes may allocate a larger proportion of their resources to reproduction and a smaller proportion to growth than populations adapted to lower latitudes. For instance, although the absolute fitness of the native populations was similar, native populations in the range edge garden were smaller on average than native populations in the intermediate and southern gardens (T. M. Griffith, unpublished). Secondly, at higher latitudes, there were more hours of light per day. As a result, the daily growth rate of plants at these latitudes may have been greater than that of plants at lower latitudes, thus partially compensating for the shorter growing period (Hay, 1990).

In contrast to flowering time, the other measured stress avoidance traits in this study did not contribute substantially to adaptation at higher latitudes. There was neither a clinal trend in preflowering growth (branch number) nor a significant change in the direct effect of this trait on fitness at the range edge. Although these results are contrary to predictions from life-history theory (Arendt, 1997), they are consistent with another study of stress avoidance in which selection for, and evolution of, greater preflowering growth was not observed under a variety of stressful conditions (Stanton et al., 2000). Similarly, there was no significant change in the direct effect of senescence on fitness at the range edge. There was no significant direct effect or total selection on germination time in any garden, but in situ germination experiments will be required to fully understand any adaptive role this trait may have.

In general, early reproduction may be an important adaptation for many species along latitudinal gradients. Like Cocklebur, many annuals and perennials that flower at the end of the summer are short day species that would be similarly effected by day length and climate at higher latitudes (Vince-Prue, 1975; Salisbury & Ross, 1992; Heide, 1994). The importance of reproductive timing for latitudinal adaptation is further suggested by examples of rapid evolution of flowering time following the introduction of species to new continents or to shifts in climate (Weber & Schmid, 1998; Davis & Shaw, 2001). The results of this study indicate that such evolution may be a critical component for adaptation to these new ranges.

Potential range-limiting traits

Clinal patterns of evolution in functional traits along an environmental gradient are often interpreted as indicating the importance of such traits for local adaptation and, consequently, geographic distribution (Antonovics, 1976; Jonas & Geber, 1999; Eckhart et al., 2004). Constraints on the evolution of these traits in marginal populations potentially limit further range expansion along this gradient (Hoffmann & Blows, 1994; Kirkpatrick & Barton, 1997; Jenkins & Hoffmann, 1999; Case & Taper, 2000). Given the general importance of stress avoidance for local adaptation in Cocklebur, and the adaptive value of reproductive timing in particular, could a constraint on the further evolution of earlier flowering at the range edge limit its northern range? Beyond the species border in Isabella Co., there is a rapid further decrease in average temperature and length of the growing season. For example, between the range edge and Grayling, Michigan (∼90 km further north), the difference in average August temperatures (the period of initial burr formation) is comparable with that between the intermediate and edge gardens while the difference in frost date exceeds that between the southern and edge gardens (data from National Climate Data Center, Consequently, an additional decrease in flowering time, similar in magnitude to that observed between source latitudes in this study, may be necessary for adaptation to this locale. Thus, any constraint on the evolution of earlier flowering phenotypes could contribute substantially to limiting the northern range of Cocklebur in Michigan.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

We thank E. Brodie III, L. Delph and K. Clay for their tremendous help throughout this study. This manuscript was improved by the thoughtful comments of S. Sultan, S. Halpern, M. Evans, B. Libman and two anonymous reviewers. The University of Tennessee Middle Tennessee Experiment Station, the Indiana University Department of Biology and the P. Gross family generously provided field sites. J. Griffith and L. Miller helped enormously with fieldwork. Funding for this research came from the Indiana University Department of Biology, Floyd Summer Stipends and a Floyd Fellowship to TMG.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  • Antonovics, J. 1976. The nature of limits to natural selection. Ann. Mo. Bot. Gard. 63: 224247.
  • Arendt, J. 1997. Adaptive intrinsic growth rates: an integration across taxa. Quart. Rev. Biol. 72: 149177.
  • Bennington, C.C. & McGraw, J.B. 1995. Natural selection in ecotypic differentiation in Impatiens pallida. Ecol. Monogr. 65: 303323.
  • Case, T.J. & Taper, M.L. 2000. Interspecific competition, environmental gradients, gene flow, and the coevolution of species’ borders. Am. Nat. 155: 583605.
  • Chapin, F.S. 1980. The mineral nutrition of wild plants. Annu. Rev. Ecol. Syst. 13: 229259.
  • Conner, J.K. & Hartl, D. 2004. A Primer of Ecological Genetics. Sinauer Associates Inc., Sunderland, MA.
  • Conover, D.O. & Schultz, E.T. 1995. Phenotypic similarity and the evolutionary significance of countergradient variation. Trends Ecol. Evol. 10: 248252.
  • Davis, M. & Shaw, R.G. 2001. Range shifts and adaptive responses to quaternary climate change. Science 292: 673679.
  • Donohue, K., Messiqua, D., Pyle, E.H., Heschel, H.S. & Schmitt, J. 2000. Evidence of adaptive divergence in plasticity: density- and site-dependent selection on shade avoidance responses in Impatiens capensis. Evolution 54: 19561968.
  • Dorn, L.A. & Mitchell-Olds, T. 1991. Genetics of Brassica campestris. I. Genetic constraints on the evolution of life-history characters. Evolution 45: 371379.
  • Dudley, S.A. 1996. Differing selection on plant physiological traits in response to environmental water availability: a test of adaptive hypotheses. Evolution 50: 92102.
  • Eckhart, V.M., Geber, M.A. & McGuire, C.M. 2004. Experimental studies of adaptation in Clarkia xantiana. I. Sources of trait variation across a subspecies border. Evolution 58: 5970.
  • Etterson, J.R. 2004. Evolutionary potential of Chamaecrista fasciculata in relation to climate change. I. Clinal patterns of selection along an environmental gradient in the great plains. Evolution 58: 14461458.
  • García-Ramos, G. & Rodríguez, D. 2002. Evolutionary speed of species invasions. Evolution 56: 661668.
  • Geber, M.A. 1990. The cost of meristem limitation in Polygonum arenastrum: negative genetic correlations between fecundity and growth. Evolution 44: 799812.
  • Grime, J.P. 1979. Plant Strategies and Vegetation Processes. John Wiley, Chichester, UK.
  • Grime, J.P. 1990. Ecological effects of climate change on plant populations and vegetation composition with particular reference to British flora. In: Climate Change and Plant Genetic Resources (M. T.Jackson, B. U.Ford-Lloyd & M. L.Parry, eds), pp. 4060. Belhaven Press, New York.
  • Hay, R.M.K. 1990. Tansley review no. 26. The influence of photoperiod on the dry-matter production of grasses and cereals. New Phytol. 116: 233254.
  • Heide, O.M. 1994. Control of flowering and reproduction in temperate grasses. New Phytol. 128: 347362.
  • Hocking, P.J. & Liddle, M.J. 1986. The biology of Australian weeds: 15. Xanthium occidentale Bertol. complex and Xanthium spinosum L. J. Austral. Inst. Ag. Sci. 52: 191221.
  • Hoffmann, A.A. & Blows, M.W. 1994. Species borders: ecological and evolutionary perspectives. Trends Ecol. Evol. 9: 223227.
  • Hoffmann, A.A. & Parsons, P.A. 1991. Evolutionary Genetics and Environmental Stress. Oxford University Press, Oxford, UK.
  • Holt, R.D. 2003. On the evolutionary ecology of species’ ranges. Evol. Ecol. Res. 5: 159178.
  • Huey, R.B., Gilchrist, G.W., Carlson, M.L., Berrigan, D. & Serra, L. 2000. Rapid evolution of a geographic cline in size in an introduced fly. Science 287: 308309.
  • Jenkins, N.L. & Hoffmann, A.A. 1999. Limits to the southern border of Drosophila serrata: cold resistance, heritable variation, and trade-offs. Evolution 53: 18231834.
  • Jonas, C.S. & Geber, M.A. 1999. Variation among populations of Clarkia ungulata (Onagraceae) along altitudinal and latitudinal gradients. Am. J. Bot. 86: 333343.
  • Jordan, N. 1992. Path analysis of local adaptation in two ecotypes of the annual plant Diodia teres Walt. (Rubiaceae). Am. Nat. 140: 149165.
  • Joshi, J., Schmid, B., Caldeira, M.C., Dimitrakopoulis, P.G., Good, J., Harris, R., Hector, A., Huss-Dannell, K., Jumpponen, A., Minns, A., Mulder, C.P.H., Pereira, J.S., Prinz, A., Scherer-Lorenzen, M., Siamantziouras, A.-S.D., Terry, A.C., Troumbis, A.Y. & Lawton, J.H. 2001. Local adaptation enhances performance of common plant species. Ecol. Lett. 4: 536544.
  • Kingsolver, J.G. & Schemske, D.W. 1991. Path analyses of selection. Trends Ecol. Evol. 11: 276280.
  • Kirkpatrick, M. & Barton, N.H. 1997. Evolution of a species’ range. Am. Nat. 150: 123.
  • Koss, W.J., Owensby, D.R., Steurer, P.M. & Ezell, D.S. 1988. Freeze/Frost Data. Climatography of the U.S. No. 20. Supplement No.1. National Climatic Data Center, Asheville, North Carolina.
  • Kozlowski, J. 1992. Optimal allocation of resources to growth and reproduction: implications for age and size at maturity. Trends Ecol. Evol. 7: 1519.
  • Lande, R. & Arnold, S.J. 1983. The measurement of selection on correlated characters. Evolution 37: 12101226.
  • Lee, C.E. 2002. Evolutionary genetics of invasive species. Trends Ecol. Evol. 17: 386391.
  • Levin, D.A. 1993. Local selection in plants: the rule, not the exception. Syst. Bot. 18: 197208.
  • Li, B., Suzuki, J.-I. & Hara, T. 1998. Latitudinal variation in plant size and relative growth rate in Arabidopsis thaliana. Oecologia 115: 293301.
  • Löve, D. & Dansereau, P. 1959. Biosystematic studies on Xanthium: taxonomic appraisal and ecological status. Can. J. Bot. 37: 173208.
  • Maron, J.L., Vilà, M., Bommarco, R. & Elmendorf, S. 2004. Rapid evolution of an invasive plant. Ecol. Monogr. 74: 261280.
  • McMillan, C. 1971. Photoperiod evidence in the introduction of Xanthium (Cocklebur) to Australia. Science 171: 10291031.
  • McMillan, C. 1974a. Photoperiodic adaptation of Xanthium strumarium in Europe, Asia Minor, and northern Africa. Can. J. Bot. 52: 17791791.
  • McMillan, C. 1974b. Photoperiodic responses of Xanthium strumarium L. (Compositae) introduced and indigenous in eastern Asia. Bot. Mag. Tokyo 87: 261269.
  • Mitchell-Olds, T. & Bergelson, J. 1990. Statistical genetics of an annual plant, Impatiens capensis. II. Natural selection. Genetics 124: 417421.
  • Nishitani, S. & Masuzawa, T. 1996. Germination characteristics of two species of Polygonum in relation to their altitude and distribution on Mt. Fuji, Japan. Arct. Alp. Res. 28: 104110.
  • Olsson, K. & Ågren, J. 2002. Latitudinal population differentiation in phenology, life history and flower morphology in the perennial herb Lythrum salicaria. J. Evol. Biol. 15: 983996.
  • Qian, H., Song, J.-S., Krestov, P., Guo, Q., Wu, Z., Shen, X. & Guo, X. 2003. Large-scale phytogenographical patterns in east Asia in relation to latitudinal and climatic gradients. J. Biogeog. 30: 129141.
  • Ray, P.M. & Alexander, W.E. 1966. Photoperiodic adaptation to latitude in Xanthium strumarium. Am. J. Bot. 53: 806816.
  • Reekie, E.G. & Bazzaz, F.A. 1987. Reproductive effort in plants. 3. Effect of reproduction on vegetative activity. Am. Nat. 129: 907919.
  • Rice, W.R. 1989. Analyzing tables of statistical tests. Evolution 43: 223225.
  • Rice, K.J. & Emery, N.C. 2003. Managing microevolution: restoration in the face of global change. Front. Ecol. Environ. 1: 469478.
  • Root, T.L., Price, J.T., Hall, K.R., Schneider, S.H., Rosenzweig, C. & Pounds, A.J. 2003. Fingerprints of global warming on wild animals and plants. Nature 421: 5760.
  • Salisbury, F.B. 1969. Xanthium strumarium L. In: The Induction of Flowering: Some Case Histories (L. T.Evans, ed.), pp. 1463. Cornell University Press, Ithaca, New York.
  • Salisbury, F.B. 1981. Twilight effect: initiating dark measurements in photoperiodism of Xanthium. Plant Physiol. 67: 12301238.
  • Salisbury, F.B. & Ross, C.W. 1992. Plant Physiology. Wadsworth Publishing Company, Belmont, CA.
  • Scheiner, S.M., Mitchell, R.J. & Callahan, H.S. 2000. Using path analysis to measure natural selection. J. Evol. Biol. 13: 423433.
  • Sih, A. & Gleeson, S.K. 1995. A limit-oriented approach to evolutionary ecology. Trends Ecol. Evol. 10: 378382.
  • Sokal, R.R. & Rohlf, J.F. 1995. Biometry: The Principles and Practice of Statistics in Biological Research. W.H. Freeman and Company, New York.
  • Stanton, M.L., Roy, B.A. & Thiede, D.A. 2000. Evolution in stressful environments. I. Phenotypic variability, phenotypic selection and response to selection in five distinct environmental stresses. Evolution 54: 93111.
  • Sugiyama, H. & Hirose, T. 1991. Growth schedule of Xanthium canadense: does it optimize the timing of reproduction? Oecologia 88: 5560.
  • Vince-Prue, D. 1975. Photoperiodism in Plants. McGraw Hill, New York.
  • Volis, S., Verhoeven, J.F., Mendlinger, S. & Ward, D. 2004. Phenotypic selection and regulation of reproduction in different environments in wild barley. J. Evol. Biol. 17: 11211131.
  • Voss, E.G. 1996. Michigan Flora: A Guide to the Identification and Occurrence of the Native and Naturalized Seed-Plants of the State. Part III Dicots (Pyrolaceae – Compositae). Cranbrook Institute of Science Bulletin 61. University of Michigan Herbarium, Ann Arbor, MI.
  • Wade, M.J. & Kalisz, S. 1990. The causes of natural selection. Evolution 44: 19471955.
  • Watson, M.A. & Lu, Y. 1999. Timing of shoot senescence and demographic expression in the clonal perennial, Polypodium peltatum (Berberidaceae). Oikos 86: 6778.
  • Weaver, S. & Lechowicz, M.J. 1982. The biology of Canadian weeds. 56. Xanthium strumarium L. Can. J. Plant Sci. 63: 211225.
  • Weber, E. & Schmid, B. 1998. Latitudinal population differentiation in two species of Solidago (Asteraceae) introduced into Europe. Am. J. Bot. 85: 11101121.
  • Weinig, C. 2000. Differing selection in alternate competitive environments: shade avoidance responses and germination timing. Evolution 54: 124136.
  • Zimmerman, J.K. & Weis, I.M. 1984. Factors affecting survivorship, growth, and fruit production in a beach population of Xanthium strumarium. Can. J. Bot. 62: 21222127.