Pilar Angélica Gómez-Ruiz·Cuauhtémoc Sáenz-Romero·Roberto Lindig-Cisneros

Abstract Assisted migration has been proposed as a strategy for adaptive management of forest species in response to expected effects of climate change,but it is controversial for several reasons.Tropical dry forests are among the most threatened ecosystems in the world.In Mexico,historically,land-use change and deforestation have been decreasing forest cover,and climate change is shifting the potential distribution of different forest types,exacerbating the risk of local extinctions.Assisted altitudinal migration could be a feasible strategy for reducing local extinctions in response to climate change and lack of landscape connectivity. Our objective was to evaluate survival and growth of Albizia plurijuga and Ceiba aesculifolia,two tropical deciduous forests species in Mexico.We transplanted 4-month-old seedlings to experimental raised beds at three altitudes(2100,2400 and 2700 m a.s.l.),exceeding their upper regional limit of distribution(2000 m a.s.l.).We also tested seed germination at each altitude.We monitored the experiment for 10 months.For both species,as altitude increased and cold weather was more prevalent,plant performance declined.Within species, differences in individual growth were significant among altitudes.Overall survival was 18.5% for A.plurijuga and 24.5% for C.aesculifolia.Both species had higher survival and better growth at lower altitude,and no seedling emergence at any altitude.We conclude that assisted migration can be implemented for each species by an upward attitudinal shift within,and not exceeding,400 m beyond their present upper altitudinal limit of distribution.Our results indicate that for many species that show altitudinal gradients at regional scales,unless current climate conditions change,the potential to establish outside their range is minimal.

Keywords Albizia plurijuga·Ceiba aesculifolia·Climate change·Fabaceae·Forest management·Mitigation strategy·Range expansion

Introduction

Climate change will aggravate the current environmental crisis,driven by land-use change and other human impacts on the biosphere,due to remarkable changes in climate patterns estimated by model predictions(IPCC 2013).An increase in land and ocean temperatures,a rise in sea level,a major incidence of extreme events,and high variability in precipitation regimes are the major climatic changes predicted(Harris et al.2006).On average,projections for Mexico are an increase in mean annual temperature of 1.5°C by 2030,2.7°C by 2060 and 3.7°C by 2090,and a decrease in annual precipitation of 6.7% by 2030,9% by 2060 and 18.2% by 2090,compared to contemporary climate(average for 1961-1990)(Sáenz-Romero et al.2010).Projected impacts of climate change predict vegetation shifts toward the north in the northern hemisphere(Rehfeldt et al.1999;Rehfeldt 2006;Tchebakova et al.2005;Hamann and Wang 2006).However,the main shift for areas with a complex topography and climate is toward higher altitudes,as is the case for the the North Central Plateau of Mexico,given its complex climate,geography and large semiarid region(Sáenz-Romero et al.2010).

These potential changes in suitable habitat pose a challenge for many plant species in terms of their future survival,performance and persistence.Species responses to changes in local climate can involve acclimation via individual phenotypic plasticity,genetic/evolutionary adaptation,or natural migration to meet their requirements.If none of these are possible,extinction may occur(Rehfeldt 1994;Davis and Shaw 2001;Reusch and Wood 2007;Aitken et al.2008;Gienapp et al.2008;Ledig et al.2010;Feeley et al.2012;Bellard et al.2012).

Because the velocity of climate change exceeds the natural rates of adaptation and migration of many species(Corlett and Westcott 2013),many species are expected to become extinct,especially tree species(Davis and Shaw 2001)with a long generation time that will have a longer adaptation lag than short-lived species(Jump and Penuelas 2005). However, some evidence indicates that certain species from different taxonomic groups are already shifting their geographic range of distribution to higher latitudes and altitudes(Lenoir et al.2008;Chen et al.2011).A species ability to shift its range depends on several factors,in addition to those already mentioned:a suitable space to migrate,age at maturity,landscape connectivity and the nature of biotic interactions in new localities; thus, migration could be more difficult or impossible in the absence of any one or more of the required conditions(Christmas et al.2016).

Because a gradual decoupling is expected between populations and the climate to which they are adapted(Castellanos-Acu?a et al.2015),managed movement of species could be provide a proactive response to climate change.Such assisted migration is generally defined as the movement of genotypes within the current range of distribution of one or more species to sites where they are currently absent(Hewitt et al.2011;Kreyling et al.2011;Christmas et al.2016).The purpose of assisted migration is to facilitate the changes in distribution ranges necessary to accommodate for climate change(Vitt et al.2010)or provide an adaptive strategy to manage and restore endangered ecosystems facing climate change(Lunt et al.2013;Williams and Dumroese 2013).Although assisted migration has been considered mainly for temperate forest species of commercial use(Winder et al.2011;Pedlar et al.2012;Rehfeldt et al.2014),it is also recognized as a feasible tool for species conservation and restoration programs to overcome dispersion limitations and reduce the risk of extinction(Seddon 2010;Kreyling et al.2011;Frascaria-Lacoste and Fernández-Manjarrés 2012).

Some forms of assisted migration have been strongly criticized.For example,Ricciardi and Simberloff(2009)asserted that‘‘a(chǎn)ssisted colonization’’should not be considered as a conservation strategy because of the perceived risk of a translocated species becoming invasive.Others stated a similar concern in terms of the conflict involved in favoring the conservation of a single species(translocated species)versus the protection of full ecological communities(Schwartz 2016).In part,the debate depends on scale; Mueller and Hellmann (2008) concluded, after reviewing data on invasive species,that species from the same continent will be less likely to produce invasive events than species from other continents after assisted migration;thus,the risk might be even lower if the managed species is already present within the watershed.

The main assumption for assisted migration is that local populations will become maladapted as the climate changes,but that new locations having climatic conditions closer to the physiological optimum might provide sites for viable populations in the future(Bucharova et al.2016).Implementation of assisted migration as a management option requires experimental testing of the establishment potential of the selected species in the new locality(Thomas et al.2004;Hoegh-Guldberg et al.2008;Richardson et al.2009),which is actually testing whether the migrated species can establish under current conditions,and if followed in the long term,how they respond to climate change.

There are few empirical studies on assisted migration or colonization(Schwartz 2016).Our aim in this study was to investigate the feasibility of assisted altitudinal migration of tropical deciduous forest species growing at lower altitudes in the watershed to higher sites that currently are dominated by pine-oak forests but are expected in the future to have climatic conditions appropriate for this type of vegetation.Such field tests are important for obtaining reliable data on plant performance in natural conditions and for recommending appropriate management strategies.In a field test of assisted migration of Ceiba aesculifolia(Malvaceae)in an urban heat island(Valle-Díaz et al.2009),plant performance improved after the shift to a higher altitude compared to the natural populations at the lower altitude.

In the present study of assisted migration,we aimed to provide data on the performance of tree species that are moved to a different plant community where the climate is expected to become suitable in the future. We chose Albizia plurijuga(Fabaceae)and C.aesculifolia,which are characteristic of the tropical deciduous forests vegetation in central-west Mexico in the Bajío region(Rzedowski and Calderón de Rzedowski 1987;Rzedowski et al.2014).Assisted migration of C.aesculifolia has been tested under conditions of urban heat island(Valle-Díaz et al.2009),and results showed that an improvement in performance occurred when shifted upward in altitude,compared to natural populations at lower altitude.Our aim was to provide data on performance of tree species that are moved to a different plant community where the climate would become suitable in the future.The empirical data from such trials will help us foresee possible consequences of assisted migration strategies(Hewitt et al.2011;Schwartz 2016).

Materials and methods

Target species and propagation

A.plurijuga(syn.A.occidentalis;Fabaceae)is a deciduous tree species with a known altitudinal range between 1600 and 1900 m a.s.l.in western Mexico(Andrade et al.2007).It is cited as endangered B1 in the IUCN Red List of threatened species(González-Espinosa 1998),so all necessary permits had been obtained previously because the Campus is registered as a botanical garden (SGPA/DGGFS/712/2912/17). Its dispersal mechanism is by gravity.C.aesculifolia is also a deciduous species,commonly distributed between 1600 and 2000 m(Carranza and Blanco-García 2000),with gravity as its dispersal mechanism.They are characteristic elements of tropical deciduous forests in central-west Mexico,at the Bajío region(Rzedowski and Calderón de Rzedowski 1987).

Seeds were collected from different individuals from one single provenance of each of the species,A.plurijuga(19.9051°N,101.1207°W,1875 m a.s.l.)and C.aesculifolia(19.7021°N,101.1388°W,1950 m a.s.l.)in January 2013. Damaged seeds were discarded. All apparently healthy seeds were sown in plastic containers(380 cm3)filled with vermiculite.Seeds of C.aesculifolia required no pre-germination treatment,but seeds of A.plurijuga were immersed in sulfuric acid for 20 min before sowing.Germination in growth chamber conditions(25°C,12 h light)was 84% for C.aesculifolia,and 78% for A.plurijuga.Propagation was carried out in a shade house at the Campus of the National University of Mexico in Morelia(19.64900°N,101.22829°W,1970 m a.s.l.)where light and irrigation conditions were uniform.All permits necessary for working with this endangered species were obtained previously because the Campus is registered as a botanical garden(SGPA/DGGFS/712/2912/17).Seeds were sown in plastic containers of 380 cm3filled with vermiculite.We fertilized all individuals with a hydroponic solution,70 ml for each individual, which contained 11 macro- and micronutrients.Fertilization was done twice a week for 3 months before translocation and continued in the field twice a month for the duration of the experiment.

Assisted migration experiment

Experimental sites were located in the communal lands of the Nuevo San Juan Parangaricutiro Community(NSJP),in Michoacán,Mexico(Fig.1).Climate is mainly temperate and seasonal,with mean annual temperature of 15°C and mean annual precipitation of 1200 mm(Velázquez et al.2003).The dominant vegetation of this area is pine and pine-oak forests that are managed for timber extraction under sustainable forestry practices.In the local landscape,tropical deciduous forests has an upper altitudinal limit below 2000 m a.s.l. With increasing altitude, tropical deciduous forests is replaced by mixed pine-oak forests,and from 2100 to 2200 m a.s.l.pine-oak forests become dominant,with more diversity and abundance of the pine species.

In July 2013,432 4-month-old individuals(216 from each species)were planted in three sites of different altitudes:2100,2400 and 2700 m a.s.l.(Fig.1).The‘‘migration’’distances to the three sites were 875,525 and 225 m for A.plurijuga and 750,450 and 150 m for C.aesculifolia.These distances roughly correspond to the expected shifts in suitable habitat in the region for the years 2030,2060 and 2090(Sáenz-Romero et al.2012).Four monthold individuals were used because previous work with C.aesculifolia (Valle-Díaz et al. 2009) has shown that 6-month-old seedlings have high survival rates.In contrast,C.aesculifolia trees in the region produce seeds during the dry winter season(December and January),and seedlings emerge at the beginning of the rainy season(June).We therefore used 4-month-old seedlings as a compromise between having sturdy plants for transport and planting(important considerations for forestry practice),but still young enough to be as close as possible in age to plants recruiting under natural conditions.Since each field site had its own history of soil perturbation and management that could generate confounding effects,in addition to the main effect of altitude,we planted the seedlings into two raised wood beds 7.8 m long,1.5 m wide and 0.4 m high filled with homogenized soil obtained from a single site near the middle altitude(2400 m a.s.l.).Additionally,we manipulated the standing tree cover around each experimental site to ensure similar light conditions by removing some adult trees and pruning others.A metal mesh was placed below the soil in each raised bed to prevent underground intrusion by gophers.All sites were also fenced with barbed wire to keep out cattle and deer.The three field sites also had similar slopes,and rain did not differ significantly among them(Castellanos-Acu?a et al.2015).

Fig.1 Left,middle:Study site maps of Nuevo San Juan Parangaricutiro,Michoacan,Mexico.Right:raised beds at each test site at the three altitudes(S1:2700;S2:2400;S3:2100 m a.s.l.)

At each altitude,144 plants(72 of each species)were planted,divided between the two raised beds.Within each raised bed,plants of both species were arranged in a Latinsquare design repeated four times in each raised bed.In summary,we had three test sites;at each site were two raised beds,with 72 plants of each species transplanted(3×2×72=432 individuals).Also,to test the ability to germinate in the new climatic conditions,we sowed 100 seeds from each species in a different section of each experimental raised bed(300 seeds per species in three test sites).Seeds came from the same stock of propagated seedlings.

The experiment was monitored every month over 10 months(July 2013 to May 2014).We assessed survival(alive or dead)and growth(height of each individual from base to apex stem)until November 2013.The number of individuals with green stems and sprouts for each species was recorded when each species started to resprout(March 2014 for C.aesculifolia;May 2014 for A.plurijuga).Air temperature at each site was recorded with data loggers throughout the experiment.We also recorded the number of seeds germinated of each species at each altitude.

Statistical analysis

Data for each species were analyzed independently,and monthly data were not analyzed by means of repeated measures ANOVA because the assumption of circularity of the covariances matrices for this test was not met.Effect of altitude on growth was analyzed using height data for only one evaluation(November 2013)with linear mixed-effects models and the nlme package(Pinheiro et al.2016),with initial height(August)and altitude as fixed variables,and raised bed as the random effect,and multiple comparisons were made using multcomp with R(R Core Team 2013).Survival,green stems and sprouts were analyzed using binomial error structures with the glmer command in the lme4 package(Bates et al.2015)in R 3.1.1.Cox proportional hazards analysis of survival data was done separately for species and altitude using the package survival(Therneau and Grambsch 2000).Seed germination data were not statistically analyzed because seedling emergence was too low for valid comparison.

Results

We detected significant altitudinal performance differences in both species.Mortality increased after the fourth month of the experiment(November 2013),when cold temperatures were more frequent,reaching their coldest level in February 2014.This trend of continued mortality lasted until the end of the experiment(May 2014)when we registered an overall(regardless of altitude)survival of 18.5% for A.plurijuga and 24.5% for C.aesculifolia(Table 1).General mixed model analyses showed significant differences between altitudes for both species(A.plurijuga: z=-3.734, p ＜0.0005; C. aesculifolia:z=-3.813; p ＜0.0005). Cox analysis also showeddifferences between altitudes,mainly between 2700 and the other two altitudes,the model was significant for both species: A. plurijuga (likelihood ratio test=7.49,p ＜0.05; Wald test=7.58, p ＜0.05; score test=7.7,p ＜0.05)and C.aesculifolia(likelihood ratio test=14.9,p ＜0.005;Wald test=15.2,p ＜0.005;score test=15.7;p ＜0.005).Cox analysis made evident the moment when high mortality started (Fig.2), after which mortality remained more or less constant for C.aesculifolia;for A.plurijuga,mortality continued to increase until the end of monitoring.The trends in survival indicated better performance at the lowest site(2100 m).

Table 1 Average height(±SD)in November 2013 and number of individuals with green steams and sprouts for A.plurijuga and C.aesculifolia at three test sites

Fig.2 Cox survival analysis for A.plurijuga and C.aesculifolia during monitoring(July 2013 to May 2014,0 to 10 months)

Growth of A.plurijuga and C.aesculifolia was estimated by variation in height between periods. Young individuals lost their leaves progressively after the experiment started,and all plants lost all their leaves during the cold season(November to February).Height data were recorded until November 2013 because,after this month,mortality increased notably due to cold temperatures.Statistical analysis with linear mixed-effects models showed that altitude had a significant effect on height in November for both species(A.plurijuga:t=-5.77,p ＜0.005;C.aesculifolia:t=-2.98,p ＜0.005).Data for green stems and sprouts from both species are in Table 1.In the case of A.plurijuga,data in May 2014 showed a trend for more green stems (with no sprouts or leaves) as altitude increased but more sprouts as altitude decreased.The effect of altitude was significant for the number of green stems(z=3.171, p ＜0.005) and the number of resprouting plants (z=-3.172, p ＜0.005). For C. aesculifolia,sprouts were recorded in March 2014,with an increasing number as altitude decreased,but the effect of altitude was not significant(z=0.993,p=0.32).

For seeds of A.plurijuga sowed in the experimental beds,17.5% germinated at the beginning of the experiment in the middle and lowest sites,but as cold conditions started in November 2013,no more seeds germinated and mortality of young seedlings increased considerably.For C.aesculifolia,only 3% of the total seeds germinated at the lowest site,and none germinated at the higher sites.At the end of the experiment,all seedlings of both species died(Fig.3).

Fig.3 Number of germinated seeds of A.plurijuga and C.aesculifolia at the three altitudes from August 2013 to May 2014)

Temperature data at three altitudes throughout the experiment(August 2013 to May 2014)were divided into four periods. Table 2 shows maximum and minimum average temperatures at each altitude for each period.young plants have to be established now at higher altitudes.In the field,many variables influence plant performance,but in our experiment,we controlled the soil conditions by using the same substrate at all sites,thus reducing some of among-site variability.

For our model species,A.plurijuga and C.aesculifolia,after 10 months of monitoring,we observed that young individuals of both species had limited potential to establish and persist at high altitudes that are farthest from their upper limit of actual distribution.In other words, the potential of establishing viable populations decreases as altitude increases,becoming critical if the upward shift is greater than 400 m.General survival was low by the end of the experiment,following the pattern expected for the worse performance at higher altitudes as reported for other species in the same region(Castellanos-Acu?a et al.2015).But we note that at the lowest site(2100 m)some plants survived;at these elevations,die-offs of the native pine species attributed to drought have been observed by our work group in the last 5 years,but they do not appear to be correlated with the natural drought cycles in the region.Growth of the transplants also declined as the altitude increased for both species.

These results suggest that the chance for plant failure after they are transplanted to a new altitude is greater thanNovember 2013 to March 2014 was the coldest period among all sites in keeping with the season,and March 2014 to May 2014 was the warmest.As we expected,temperature differences between altitudes were remarkable,mainly between the lowest site and the other two.

Table 2 Average temperature(T),maximum(Max),and minimum(Min)(±SE)registered with data loggers from July 2013 to May 2014 in the three experimental sites at Nuevo San Juan Parangaricutiro

Discussion

This study investigated the response of tropical deciduous forests tree species in the field after they were transplanted beyond the upper limit of their current range of altitudinal distribution.Viable populations of reproductive age in the future depend on the trees to be matched with their climatic niche.For successful matching and establishment of viable individuals based on predictions of climate change models,the chance for success(Dalrymple et al.2012).Nevertheless,transplanting A.plurijuga and C.aesculifolia near to their upper limit could be considered as a management strategy in the face of climate change,especially in areas where other species are already stressed by climatechange-related drought conditions.Survival and growth decrease when species are relocated to areas that are colder(at higher elevations) in comparison to their original habitats(Castellanos-Acu?a et al.2015).This pattern is predictable because relocated individuals are more likely to experience abiotic and biotic conditions that differ more from their physiological optimum at their sites of origin.In the present study,survival was better for both species at 2100 and 2400 m a.s.l.,which for C.aesculifolia is similar to the trends observed in a site that suffers the effects of an urban heat island(Valle-Díaz et al.2009),where altitudinal displacement occurs because the regional climate around urban areas becomes dryer and hotter.

By assisting the migration of species such as those used in this study,we might be able to establish populations that avoid the high temperatures and low precipitations predicted for our study region in the next decades(Sáenz-Romero et al.2010).These environmental changes imply a high risk of drought due to elevated rates of leaf transpiration during the growing season(Woodward et al.1990).A shift in range distribution for both species will be necessary in the near future,considering that a reduction of more than 90% of natural habitat in their current distribution area is expected due to climate change in the following decades(Reyes-Abrego 2014).Areas currently outside but near their range of distribution could have climatic conditions appropriate for them in the next decades,making assisted migration a feasible strategy for species management.The paradox of assisted migration is that we expect warmer environments in the future at many locations,but currently movement to higher altitudes implies facing cold environments with a higher frequency of frost events than sites at lower altitudes,which might select for frost-resistant genotypes.

Under the unprecedented rate of environmental changes in all ecosystems,we need to consider and implement new strategies to conserve threatened species.Advocates of assisted migration suggest that the strategy could reduce the extinction risk for species that cannot adapt fast to climate change and that do not have good dispersal ability(Hunter 2007;Mueller and Hellmann 2008;Richardson et al.2009).The two species in our study do not have an efficient dispersal mechanism to reach long distances,not to mention high altitudes,which reduces their ability to migrate by their own as required by the rate of climate change.Sites expected to have climatic conditions similar to their current habitat are far away(Reyes-Abrego 2014),so reaching these places will likely be very difficult without assistance such as assisted migration. Moreover, their populations are very fragmented and dispersed,limiting their long-term persistence in current habitats.A.plurijuga has been reduced to small remnant populations in the wild(Rico-Arce et al.2008).Therefore,these species need assistance in extending their distribution.Assisted migration could be a valuable tool for long-lived,locally adapted species,especially those threatened by fragmentation and limited migration capacities(Dumroese et al.2015).With the seed experiment,we provide evidence that these species cannot become invasive because their rate of seed germination was very low,probably because of low temperatures at higher sites,so they would be unable to establish in new localities on their own.Our young seedlings were very susceptible to climatic variability in the new locations,but older seedlings could be more resistant(Valle-Díaz et al. 2009), so we strongly recommend planting seedlings that more than 6 months old to increase the likelihood of success of the assisted migration.Populations with the best probability of establishment at higher altitudes could be selected using common garden experiments and by testing their performance in natural soils from the receptor localities,because soil is another factor that could determine the success of assisted migration.We conclude that assisted migration can be implemented for each species if the upward attitudinal shift is within and not exceeding 400 m beyond their present upper altitudinal limit of distribution,because moving beyond this range will prevent successful establishment.Our results indicate that, many species that show altitudinal gradients at regional scales,unless current climate conditions change,have minimal establishment potential outside their range.

AcknowledgementsWe are grateful to Dante Castellanos and Mariela Gómez for field assistance and to Dr.Elise Buisson(Universitéd’Avignon,IMBE UMR CNRS IRD AMU),Dr.Carlos Martorell and Dr.Erick de la Barrera for reviewing the paper before submission.