Geographical variation in population demography and life history traits of Tecate cypress (Hesperocyparis forbesii) suggests a fire regime gradient across the USA–Mexico border
- First Online:
- Cite this article as:
- de Gouvenain, R.C. & Delgadillo, J. Plant Ecol (2012) 213: 723. doi:10.1007/s11258-012-0035-3
- 164 Downloads
Plant adaptations to fire often display spatial heterogeneity associated with geographical variation in fire regime. We examined whether populations of the Tecate cypress (Hesperocyparis forbesii Adams) in southern California and northern Baja, Mexico, exhibited spatial heterogeneity in cone serotiny, in other life history traits associated with fire-adaptation, and in population demographic structure, to assess a putative difference in fire regime across the USA–Mexico border. Demographic data, tree life history data, and tree ring series were used to compare the demographic structure and life history traits of three populations in southern California with three populations in northern Baja California. In Baja populations, a greater number of tree size classes were present (χ2 = 12,589; P < 0.05), cone serotiny was more facultative (Mann–Whitney U = 58, P < 0.05), and young adult trees had a higher reproductive output (Mann–Whitney U = 2.65, P < 0.05), suggesting that a difference in fire regime between southern California and northern Baja has existed long enough (ca 8000 years) to drive microevolutionary divergence between the two sets of populations, and is not solely the result of 20th century differences in fire management policies across the international border. The transitional area between the two different fire regimes does not appear to coincide with the border itself but may lie in a zone of ecological transition south of Ensenada. The range of phenotypic variation observed within the Tecate cypress metapopulation suggests this species has the capacity to adapt to future environmental changes.
KeywordsCalifornia Callitropsisforbesii Closed-cone cypress Cupressus forbesii Life history evolution Serotiny
Adaptations of plant life history traits to recurring fires in fire-prone ecosystems display spatial variability that has been attributed to variations in fire regime, a compound descriptor that includes variables such as fire size, fire frequency, fire intensity, and fire season (Bond and Keeley 2005; Sugihara et al. 2006; Wright and Bailey 1982). For instance, lodgepole pine (Pinus contorta) stands that originated from intense stand-replacement fires in Montana are predominantly made up of trees with serotinous cones (cones that remain alive and closed on branches of adult trees until a fire opens them and releases their seeds), whereas other stands that originated from low-intensity ground fires or from non-fire disturbances contain mostly trees with non-serotinous cones (Muir and Lotan 1985). Similarly, Gauthier et al. (1996) found that different fire regimes produced microevolutionary changes in the frequency of serotiny among populations of jack pine (Pinus banksiana) in the Canadian boreal forest, with stand-replacement fires promoting high serotiny and lower intensity “nonlethal” fires (typically ground fires) triggering the regeneration of non-serotinous tree cohorts. Some European pine species and several Proteaceae species in the genus Banksia are similarly polymorphic for cone or fruit serotiny and their populations exhibit spatial variation in the frequency of serotinous individuals that has been attributed to geographical variation in fire regime (Cowling and Lamont 1985; Ne’eman et al. 2004; Tapias et al. 2004).
Age at reproduction can also vary along fire regime gradients, and Schwilk and Ackerly (2001) proposed a model of fire-adaptation strategies for North American pines which predicted that populations with a “fire-embracing” strategy (adapted to high-intensity lethal fires) would contain trees that reproduce earlier than those in populations exhibiting a “fire-surviving” strategy (adapted to low-intensity nonlethal fires). However, Gill (1977) argued that if fires occur frequently within a fire-adapted ecosystem, production of seeds early is advantageous, while if fires are infrequent, investing in vegetative growth and delaying reproduction may increase relative fitness. Finally, fire regime can also influence the demographic structure of forest stands, nonlethal fires resulting in uneven-aged tree populations, while stand-replacement lethal fires promote even-aged populations (Gauthier et al. 1996).
Fire is the most important environmental disturbance in the Mediterranean chaparral ecosystems of southern California and northern Baja California (Hanes 1988; Keeley and Swift 1995; Keeley and Fotheringham 2001a; Sugihara et al. 2006), two regions that belong to the same California Floristic Province (Minnich and Franco-Vizcaíno 2005; Munz and Keck 1973; Ornduff et al. 2003) and to the same Martirense phytogeographical unit (Peinado et al. 2008), and therefore adaptation to fire is one of the most important life history requirements of chaparral plants, including the Tecate cypress (Fites-Kaufman et al. 2006). While plant adaptations are shaped by multiple environmental variables that influence fitness, and by interactive effects of some of these variables (Reznick and Travis 2001), suites of life history traits such as cone serotiny or age at maturity often represent ecological strategies related to fire (Borchert and Odion 1995; Hanes 1988; Keeley and Fotheringham 2001a; Keeley and Swift 1995; Schwilk and Ackerly 2001; Zedler 1995; Zedler et al. 1983). Although fire has probably been a component of the southern California and northern Baja landscapes since the middle of the Miocene epoch 10 million years ago (Keeley 2006), chaparral plant associations in southern California and northern Baja developed approximately 8000 years ago following the end of the Wisconsin glacial period when the climate of southwestern USA and northwestern Mexico went from cool with infrequent fires to dry and warm with frequent fires (Keeley 2006; Vogl et al. 1988) and previously abundant coniferous species such as the Tecate cypress became restricted in distribution to arboreal islands within the chaparral (Barbour 2007; Vogl et al. 1988).
Four populations of Tecate cypress are located in southern California (Barbour 2007; Dunn 1986; Minnich and Everett 2001), where the cypress typically regenerates following stand-replacement fires that release seeds from serotinous cones but also kill all adult trees, generating new even-aged cypress populations every 40 to 60 years (de Gouvenain and Ansary 2006; Vogl et al. 1988). More than a dozen populations are located in northern Baja (Fig. 1) (Dunn 1986; Minnich and Chou 1997). In California, the Tecate cypress is the obligate host species for the larvae of the rare Thorne’s hairstreak Butterfly (Callophrys thornei) (Hogan 2004) and has been listed by the California Native Plant Society as rare and seriously endangered, while in Mexico, the Secretaria de Medio Ambiente y Recursos Naturales has listed the cypress as rare.
If fire regime differences between southern California and northern Baja have developed from environmental causes during the last 8000 years, we would expect to find differences in the frequency of life history traits associated with fire adaptation and in the population demographic structure of Tecate cypress between the two regions, analogous to the differences found for other North American coniferous species by Gauthier et al. (1996); Gill (1977); Muir and Lotan (1985); Schwilk and Ackerly (2001). Failure to find such differences would suggest either that fire regime differences between the two regions have developed too recently to allow microevolutionary divergence, or that there is no difference. Understanding how different fire regimes may have influenced microevolutionary changes in the frequency of life history traits and in the population attributes of a fire-dependent plant species would contribute to a better understanding of the evolution of fire-adaptive traits (Bond and Keeley 2005) and would enhance our ability to project the future of its populations (Pausas et al. 2004). Furthermore, contrasting the microevolutionary response of the flora of two similar fire-prone ecosystems can help us predict how these ecosystems may respond to global human-caused environmental changes in the future (Montenegro et al. 2004).
Assuming a ca 8000 year old fire regime difference, and in line with the findings of Gauthier et al. (1996); Gill (1977); Muir and Lotan (1985); Schwilk and Ackerly (2001), we first expected that in California, infrequent but intense and severe fires should maintain the adaptive fitness of serotiny while in Baja, more frequent but less intense/less severe fires should decrease the adaptive fitness of serotiny. Incidence of serotiny should thus be higher in southern California than in northern Baja. Second, we predicted that low-intensity fires burning in a greater range of weather in Baja (Chou et al. 1993; Franco-Vizcaino and Sosa-Ramirez 1997; Keeley and Fotheringham 2001a; Minnich 2001), would allow some adult cypress trees to remain alive and would promote inter-fire seedling establishment under a more frequently disturbed and open chaparral canopy (Enright et al. 1998; Lamont and Enright 2000) unlike in southern California, where seedlings cannot survive under the undisturbed southern California chaparral canopy (Keeley 1992; Tyler 1995). Baja populations should thus contain multiple age classes (or stages), from seedlings to adults, while California populations should display two age structure types: only seedlings and/or saplings in recently burned populations or only adult trees in populations that have not burned within the last 20 years. Last, differing from Schwilk and Ackerly’s (2001) model for pines but agreeing with Gill (1977), we reasoned that higher survival of seedlings between fire events in Baja (for the reason explained above) would increase the fitness benefit of reproducing early since there is no advantage for adult trees to wait and synchronize seed production with lethal fire events, while low probability of seedling survival between fire events in California should increase the fitness benefit of delaying reproduction until a fire event occurs, typically every 40 to 60 years. Cone production for saplings and young adult cypresses should thus be higher in Baja than in California, and cone production for old adult cypresses should be higher in California than in Baja.
Natural features, demographic structure, and life history trait frequency for the six sampled Tecate cypress populations
Southern California (USA)
Region mean (SD)
Northern Baja (Mexico)
Region mean (SD)
Longitude and latitude
Elevation (m) (SD)
Approximate pop. size (ha)
Year of last fire
Age of oldest tree sampled, live or dead
# live stem/ha stage 1 (0–0.5 Øc)
# live stem/ha stage 2 (0.51–5.0 Ø)
Stage 2 mean tree age—by region (SD)
# live stem/ha stage 3 (5.1–20.0 Ø)
Stage 3 mean tree age—by region (SD)
# live stem/ha stage 4 (>20.0 Ø)
Stage 4 mean tree age—by region (SD)
Ave. # of cones/tree stage 2 (SD)
5 (11) nd = 64
12 (20) n = 36
Ave. # of cones/tree stage 3 (SD)
69* (85) n = 66
150* (202) n = 46
Ave. # of cones/tree stage 4 (SD)
586 (394) n = 20
525 (275) n = 4
Ave. % cones open—stages 3 + 4 (SD)
3.9* (7.1) n = 12
35.0* (34.9) n = 17
Ave. # seeds/cone (SD)
Based on information from de Gouvenain and Ansary (2006); Dunn (1986); and Zedler (1977), we described the life history of Tecate cypress with a four-stage life-cycle where each stage corresponds to the following diameter intervals: stage 1 = seedlings (0–0.5 cm), stage 2 = saplings (0.51–5.0 cm), stage 3 = young adults (5.1–20.0 cm), and stage 4 = old adults (>20.0 cm). At each population, two pseudo-replicate 20 × 30 m plots were randomly located in 2006 and 2009 (not allowing for overlap between plots), each plot being further subdivided into twenty-four 5 × 5 m quadrats. The use of two plots per population made it easier to sample small, dissected populations (e.g., San Rafael, Tecate Peak) where one large plot would have been impractical. Depending on the terrain and the size and outline of the population, the distance between the two plots ranged from 50 to 500 m for any population. In each plot, we (a) inventoried and recorded the basal diameter of all Tecate cypresses with a basal diameter ≥5 cm (diameter at breast height is not feasible with the Tecate cypress), corresponding to stages 3 and 4; (b) counted stages 1 and 2 seedlings and saplings within two randomly selected quadrats; (c) collected two cores from each of five (CA-3, BA-1, BA-3), seven (BA-2) or eight (CA-1, CA-2) trees ≥5 cm-diameter (or full cross-sections when trees were dead in populations that had burned in the recent past). Additionally, in populations that had not burned in the recent past, we (d) counted the number of cones on a random subsample of the live trees in stages 2, 3, and 4 at each population (number of trees subsampled ranged from 5 to 27 per population); (e) estimated the fraction of these cones (stages 3 and 4 trees only) that were open on live branches; and (f) collected 15 closed cones from each cored tree to estimate seed production. Since fires burned in 2003 and 2005 on Otay Mountain and on Tecate Peak, respectively, we supplemented our 2006 and 2009 field data collections with 2009 cone data from a stand on Otay Mountain that escaped the 2003 fire, and with 2004 cone data we collected on Tecate Peak for a previous study (de Gouvenain and Ansary 2006).
Tree cores and tree cross-sections were mounted on wood, sanded, polished, and growth rings were counted and dated under a stereo-microscope. Each year’s ring width was then measured to the nearest 1/100th mm using a Velmex® linear dovetail slide and Acu-Rite® encoder, and simultaneously entered into a MeasureJ2X® computerized database, each core or cross-section resulting in a ring-width series. Except for few very young trees, ring width was not significantly correlated with tree age (r = 0.11, n = 40 trees), and thus no adjustment was made for any slight decreasing ring-width trend with tree age. Lining up bar graphs of all tree ring-width series from each population in Excel® to match ring years revealed several short (two to four year) ring-width patterns common to all trees within a population; these were then used to cross-date the ring series (Stokes and Smiley 1996; Fulé and Covington 1999) within a population by filling in missing rings or removing false rings when a pattern mismatch occurred in one series relative to all others in a population. On average, 1 in 3 of the cored trees (38%) exhibited from one to two false and/or missing rings, but cross-dating corrections were straightforward. Cones were weighed, measured, and burned over Bunsen burners for approximately five minutes at 300°C until they opened, and seeds were then collected and counted for each cone.
Non parametric Mann–Whitney tests were used to compare number of cones per tree per stage for stages 2, 3, and 4, and percent of cones open on live branches of trees in stage 3 and 4 between the two regions. Stage 2 trees were excluded from the percent open cones contrast because they had very few cones and because conifer serotiny is often not expressed in the youngest fertile age classes (Gauthier et al. 1993a; Schoennagel et al. 2003). Population elevation and number of seeds per cone between the two regions were compared with t-tests. Demographic structure was contrasted by way of a Chi-square goodness-of-fit test using stages as categories and mean tree counts per stage as categorical proportions for the two regions to test Ho: Baja (“observed”) stage proportions = California (“expected”) stage proportions, after normalizing total Baja tree counts to total California tree counts, and using only the two populations with the same recent fire histories in both regions (CA-1 and CA-2, and BA-2 and BA-3, respectively).
The oldest Tecate cypresses, in the San Rafael and Otay Mountain populations, were 163 and 121 years, respectively (Table 1). The most abundant regeneration was observed on Otay Mountain (CA-1), which burned in 2003, with stage 2-size stems (0.51–5 cm diam.) averaging more than 120,000/ha. The next most abundant regeneration was observed at San Vicente (BA-2), a population that has not seen a stand-replacement fire in at least 50 years judging from the oldest cored live tree (Table 1). No fire scars were found from the twelve cross-sections collected from trees killed in 2003 and 2005 fires at CA-1 or CA-2, respectively, but a 36 year-old tree that died in ca 2005 at BA-2 was found to contain a fire scar dated to ca 1994 (indicating that it survived at least one fire). Tecate cypresses located in northern Baja produced more seeds per cone than those located in southern California (t = 6.16; df = 182; P < 0.05, corrected for unequal variance), but we cannot say if trees in BA-1 follow that trend since no live closed-cones were available there. Stage 3 and stage 4 mean tree age was not significantly different between the two regions (Table 1). The three southern California populations are all located at a higher elevations than the northern Baja populations (t = 4.3; df = 5; P < 0.05, corrected for unequal variance) (Table 1).
Two of the three populations sampled in northern Baja (BA-2 and BA-3) contained live cypresses in all four stages, from seedling size to old adult size. Southern California populations contained live stems in only two consecutive stages: stages 1 and 2 in populations that had burned within the last 20 years (CA-1, CA-2) and stages 3 and 4 in the population that had not burned in the last 20 years (CA-3) (Table 1). The San Antonio de las Minas population (BA-1), which burned in 2003, had a demographic structure similar to that of the southern California populations CA-1 and CA-2, even though it is located in northern Baja (Table 1). The demographic contrast between CA-1 and CA-2 on the one hand, and BA-2 and BA-3 on the other, was significant (χ2 = 12,589; df = 3; P < 0.05).
Life history traits associated with fire adaptation
The differences we observed between the California and Baja Tecate cypress populations in cone serotiny frequency and in population demographic structure are analogous to differences in serotiny frequency and demographic structure reported by Gauthier et al. (1993b, 1996); Givnish (1981); Muir and Lotan (1985), and Radeloff et al. (2004) for other coniferous species across landscapes with heterogeneous fire regimes, which suggests that a fire regime gradient exists across the USA–Mexico border. Higher seed production by young adult (stage 3) trees in Baja matches Gill’s (1977) observations that lodgepole pines growing in stands disturbed by frequent fires produce seeds at a young age and further supports the fire regime difference hypothesized by Keeley and Fotheringham (2001b); Minnich (2001); and Minnich and Franco-Vizcaíno (2005). While other environmental variables can influence population demography and tree life history trait frequency, fire regime is likely to be the most important variable that influences the population and tree characteristics of a fire-adapted tree such as the Tecate cypress. Radeloff et al. (2004) for instance found that serotiny gradients within P. banksiana forests of northwestern Wisconsin could be best explained by a gradient in fire intensity from stand-replacing fires to low-intensity ground fires. This strong association between fire regime and serotiny is also found across species within a genus; for instance Pinus halepensis and P. pinaster (both serotinous) grow in shrubby areas prone to high-intensity crown fires in Spain while P. nigra, which grows in low-nutrient sandy soils with sparse understory vegetation, is non-serotinous (Ne’eman et al. 2004; Tapias et al. 2004). The difference in elevation between the populations of the two regions may have had an indirect causal effect on the ecological differences found in this study through its influence on plant communities, fuel type, fuel moisture, and therefore on fire regime (Falk et al. 2011), but perhaps not a direct effect on tree serotiny since other studies have suggested that elevation is not a significant explanatory variable for variation in serotiny in other coniferous species. For instance Tapias et al. (2004) found that in P. pinaster, incidence of serotiny was unrelated to elevation, and neither elevation, aspect, nor slope was related to cone serotiny gradients for P. contorta in western Montana (Muir and Lotan 1985).
Microevolutionary changes in the frequency of cone serotiny in populations of several pine species, such as Pinus radiata in New Jersey and P. banksiana in Wisconsin, can take place over a few generations, making stands regenerating from different disturbance intensities diverge fairly rapidly for that character (Givnish 1981; Radeloff et al. 2004). If northern Baja and southern California fire regimes diverged during the development of the modern chaparral plant communities beginning ca 8000 years ago following the end of the Wisconsin Ice Age, this could have provided sufficient time for microevolution to produce different frequencies of life history traits associated with fire adaptation among the cypress populations of the two regions. It is unlikely that 60 years of dissimilar fire management policies implemented on the ground since WWII (Keeley and Zedler 2009), would have provided sufficient time to promote such microevolutionary divergence given the 40- to 60-year average generation time for Tecate cypress in California (de Gouvenain and Ansary 2006; Dunn 1986; Zedler 1977). Recent human-driven differences in fire regime between the USA and Mexico may have added further complexity to the natural difference in fire regimes (Minnich 2002) but are unlikely to be the primary drivers of that difference. The similarity in demographic structure of the most northern of the Baja populations we studied (San Antonio) with that of two southern California populations suggests that the geographical area where the natural shift in fire regime may occur is located south of the political border, in a transitional climate zone located just south of Ensenada and described by Keeley and Zedler (2009) and Peinado et al. (1995). At that latitude, the Gulf of California and the Sierra de Juarez weaken the dry easterly “Santa Ana” winds that drive fire regimes in southern California, leaving on-shore breezes to become the predominant winds influencing chaparral fires south of Ensenada (Keeley and Fotheringham 2001a; Keeley and Zedler 2009).
The limitations of this study are threefold: First, the limited number of populations north of the border restricts overall sample effort and replication, and thus our ability to generalize our inferences to all Tecate cypress populations. Second, dissimilarity in recent fire history among the sampled populations makes unambiguous comparison across regions difficult. Finally, we did not examine the possible association between variation in the frequency of fire-adapted life history traits and the Tecate cypress metapopulation phylogenetic structure. Differences we observed between the two regions may be due to phylogenetic distance rather than ecologically-driven microevolutionary divergence. Conversely, if the ecological differences our study suggests exist between cypress populations located north and south of Ensenada result from a long-standing difference in fire regime, genetic differences could exist between the two sets of populations that are the result of microevolutionary divergence.
While it may be difficult to draw inferences about a narrowly distributed species with a heterogeneous spatial distribution of recent fire histories across its few populations, or to discriminate between the role of an ancestral phylogenetic structure and that of natural selection on today’s metapopulation genetic structure, a genetic analysis similar to that conducted by Truesdale and McClenaghan (1998) in California and by Rosas Escobar et al. (2011) in Baja could help elucidate the role that fire regimes may have played in the evolution of a rare but resilient species that spans more than 400 km of natural and human-made boundaries. Since most climate projections forecast increasing fire activity trends in the temperate zones of the Earth (Falk et al. 2011; Gavin et al. 2007), understanding the evolution of fire-adaptive traits, even for a species with narrow distribution and limited economic importance like the Tecate cypress, can shed light on how future fire regimes may influence the health and resiliency of fire-prone ecosystems, and therefore on how best to manage them to conserve their biodiversity (Bond and Keeley 2005; Montenegro et al. 2004; Pausas et al. 2004). Comparison of the regeneration ecology and of the demographic patterns of Tecate cypress across a plausible fire regime gradient suggests that this cypress species, despite its narrow distribution, is plastic in its fire-adaptation strategies, and thus that its metapopulation still contains the genetic diversity needed to allow future evolution in the context of global environmental change.
We thank Jim Bartel, Ibes Fabian Davila Flores, Jocelyne and Trevonte de Gouvenain, and Edelyn Ramírez Espinoza for their help in the field, and Kristin Chauvin and Katherine D’Ovidio for their help in preparing and analyzing field samples and tree cores. We thank the Bureau of Land Management and the Forest Service for allowing us to conduct research on federal lands and Saul Martin del Campo for allowing us to conduct research on his property. We are especially grateful to Joyce Schlachter for her logistical assistance. This manuscript benefited from discussions with Jim Bartel, Jon Keeley, Richard Minnich, and Sula Vanderplank, and from comments from three anonymous reviewers. This research was supported by grants from Chapman University, Rhode Island College, the National Science Foundation-Rhode Island Experimental Program to Stimulate Competitive Research (EPSCoR), and in-kind support from the Universidad Autónoma de Baja California, Ensenada, Mexico.