Vol.:(0123456789)1 3 Conserv Genet (2017) 18:467–478 DOI 10.1007/s10592-016-0921-7 RESEARCH ARTICLE Climatic stability and contemporary human impacts affect the genetic diversity and conservation status of a tropical palm in the Atlantic Forest of Brazil Carolina da Silva Carvalho1,2  · Liliana Ballesteros‑Mejia1 · Milton Cezar Ribeiro2 · Marina Corrêa Côrtes2 · Alesandro Souza Santos3 · Rosane Garcia Collevatti1  Received: 10 May 2016 / Accepted: 23 December 2016 / Published online: 12 January 2017 © Springer Science+Business Media Dordrecht 2017 of genetic diversity in E. edulis is the result of historical instability during the mid-Holocene and recent anthropo- genic impacts, mainly those that affect important ecological process such as seed dispersal. Thus, an efficient plan for species conservation must account for human impacts and environmental suitability and also assess the genetic diver- sity of seedlings and adults in fragmented landscapes. Keywords Ecological niche modeling · Defaunation · Human impacts · Conservation genetics · Euterpe edulis · Atlantic Forest Introduction Understanding how past and current environmental suit- ability and human impact factors affect patterns of genetic diversity and biodiversity distribution is a challenge for both ecologists and biogeographers (Gaston 2000). The maintenance of genetic diversity in natural populations is essential for species management planning (Hughes et  al. 2008), since species with low genetic diversity may be less likely to cope with environmental changes and therefore may become endangered or face extinction (Hoffmann and Willi 2008). Although genetic diversity has a fundamental role in species persistence and adaptation, few studies have reported the role of environmental suitability and human impacts in genetic diversity on a broad scale (Diniz-Filho et al. 2009; Ortego et al. 2012). Several historical and contemporary factors may influ- ence genetic diversity and inbreeding in natural popula- tions. Historical factors related with climate and geologi- cal events may change the habitat suitability for species establishment or persistence and may therefore imprint a signal in the species genetic pool over time (Hewitt 2000). Abstract Understanding how historical and current envi- ronmental suitability and human impacts affect the genetic diversity on a large scale is essential to species manage- ment planning. However, most studies in conservation genetics are carried out at a local or regional scale and rarely on broad spatial scales such as an entire biome. We evaluated the relative contribution of historical and current environmental suitability, current landscape features and human impacts to explain genetic diversity, allelic rich- ness and inbreeding variation among populations, using Euterpe edulis and the Atlantic forest as the model system. We fitted linear mixed models within a multiple compet- ing hypotheses approach with model selection based on Akaike’s Information Criteria. We showed that overall genetic diversity was lower in sites with absence of large seed dispersers and higher in sites with historically stable climate. Both seedling and adults showed to be negatively influenced by human impact factors; with adults mainly affected by the reduction of forest cover while seedlings by the loss of large seed dispersers. Thus, the current pattern Electronic supplementary material The online version of this article (doi:10.1007/s10592-016-0921-7) contains supplementary material, which is available to authorized users. * Carolina da Silva Carvalho carolina.carvalho@ymail.com 1 Laboratório de Genética & Biodiversidade, Universidade Federal de Goiás, Goiânia, Goiás 74001-970, Brazil 2 Departamento de Ecologia, Universidade Estadual Paulista (UNESP), Rio Claro, São Paulo 13506-900, Brazil 3 Laboratório de Marcadores Moleculares, Centro de Biotecnologia e Genética, Universidade Estadual de Santa Cruz, Rodovia Ilhéus-Itabuna, km 16, Ilhéus, BA CEP 45662-900, Brazil http://orcid.org/0000-0002-0063-2185 http://crossmark.crossref.org/dialog/?doi=10.1007/s10592-016-0921-7&domain=pdf http://dx.doi.org/10.1007/s10592-016-0921-7 468 Conserv Genet (2017) 18:467–478 1 3 The Quaternary climatic oscillation, for example, might have affected species genetic diversity due to the dynam- ics of population’s expansion and contraction (Gugger et al. 2013). Population range contraction may decrease effec- tive population size, leading to population bottlenecks and genetic drift, with the consequent loss of genetic diver- sity (Pauls et al. 2013). Loss of genetic diversity may also occur during population expansion into new habitats (Pauls et al. 2013) due to founder effects or recovery from popula- tion bottlenecks, since only a part of the original genetic diversity moves or persists into new habitats (Pauls et  al. 2013). The colonization of new habitats during expansion can result in a high genetic admixture of immigrants arriv- ing from different populations (Ortego et al. 2015), which could either increase genetic diversity due to heterosis or reduce genetic diversity if outbreeding depression occurs. Finally, sites with high environmental suitability and stable climate over time may preserve high genetic diversity in populations (Collevatti et al. 2014). Contemporary factors that may affect genetic diver- sity include processes occurring at ecological scales and relate strongly with human activities (Miraldo et al. 2016). Global changes induced by human activities including cli- mate change and conversion of land covers may affect spe- cies richness, abundance, and several key ecological pro- cesses (Fahrig 2003), such as pollination (González-Varo et al. 2009), seed dispersal (Uriarte et al. 2011) and animal movement (Eycott et al. 2012). Fragmentation and habitat loss are two factors identified as the main anthropogenic threats to genetic diversity (Lowe et  al. 2005; DiBattista 2007; Aguilar et  al. 2008). These anthropogenic threats may lead to a drastic reduction of population sizes, and consequently of effective population size (Frankham et al. 1999; Andersen et  al. 2004). The probability of mating between closely related individuals may increase in small populations and, together with genetic drift, lead to a sub- stantial reduction in genetic diversity (Breed et  al. 2012). In these human-dominated scenarios, local extinction of medium- and large-body sized animals (i.e., defaunation; Dirzo and Miranda 1991) affects the composition and eco- logical functions of animal communities (Wright 2003). Seed dispersal is amongst the many ecological roles these animals provide to the community. Large-sized frugivores are important for long distance seed dispersal because they can move seeds hundreds of meters away from the mother plant. Therefore, these animals can promote gene flow and increase genetic diversity in plant populations (García et al. 2007; Jordano et al. 2007). Although many studies focus separately on the effects of historical climate oscillation and human disturbances on genetic diversity, few evaluate the relative contribution of these historical and contemporary effects in the same system (Walter et al. 2012). Understanding how historical events impact genetic diversity is important to predict whether populations are able or not to persist under future scenarios of environmental change (Pauls et  al. 2013). In addition, evaluating the effects of contemporary factors like habitat fragmentation on genetic diversity helps identify- ing actions to prevent genetic erosion or to rescue genetic diversity of impacted populations. Therefore, quantifying the effects of historical and contemporary factors on spe- cies genetic pool may help design strategies for species and landscape conservation. Large spatial and temporal scale studies are necessary to better understand how the historical and current factors shape genetic diversity across the distribution of natural populations and contribute to efficient species conserva- tion planning (Huber et al. 2010). Large-scale studies can act as natural experiments to investigate historical climate effects on biodiversity because distinct sites may be dif- ferently affected by the oscillation in climate. Due to time lag effects, adult plants and seedlings may respond in a different way to human impacts (Van Geert et  al. 2008). When estimated using microsatellite markers, the genetic diversity of adults of perennial plants often responds to past landscape conditions but not to recent habitat changes (Collevatti et al. 2001; Kramer et al. 2008). When evaluat- ing ongoing processes, such as dispersal events culminating in gene flow, it is more appropriate to assess genetic vari- ability in early recruitment stages (e.g., seed and seedlings in plants) rather than adults (Kramer et al. 2008). The Atlantic forest biome has an extensive geographic range, presenting a wide variation in climate, topography and vegetation along its distribution (Ribeiro et al. 2011). It is also one of the most threatened ecosystems in South America, mainly due to anthropogenic disturbances (Rands et  al. 2010; Tabarelli et  al. 2010). Its current extension constitutes only 12% of its original 150  million hectares (Ribeiro et  al. 2009). The heart of palm (Euterpe edulis, Arecaceae), a threatened Atlantic forest species, was once one of the dominant palms in this ecosystem (Henderson et  al. 1997). The species is present in different vegetation types ranging from dense mountain rain forests to season- ally dry forests, where it is restricted to wet microhabitats. The edible meristem (heart of palm), obtained mainly by illegal exploitation, is used as source of raw material for food industries (Galetti and Fernandez 1998). Euterpe edu- lis is a monoic species with a predominantly outcrossed mating system with low amount of selfing (Gaiotto et  al. 2003). Its flowers are mainly pollinated by small bees (Reis et  al. 2000). The ripe fruits constitute an important food resource for animals, including more than 50 species of birds and mammals (Galetti et  al. 2013). Several popula- tion genetics studies of E. edulis have been carried out at local or regional scales. Metadata combining the already published genetic diversity data may contribute to the 469Conserv Genet (2017) 18:467–478 1 3 understanding of large-scale patterns (Lowe et  al. 2005; DiBattista 2007) of E. edulis. Here, we address the distribution of genetic diversity of E. edulis and evaluate whether past and current environ- mental variables and human-induced changes can explain variation among localities, using published data derived from microsatellite loci. First, we built ecological niche models for E. edulis in three different time periods (the Last Glacial Maximum, the mid-Holocene and the present-day) to understand how changes over time in habitat suitability for species establishment may have affected genetic diver- sity and inbreeding. Second, we evaluated the relative con- tribution of climatic habitat suitability in the past and in the present-day, together with current landscape features modi- fied by human activities to explain the patterns in genetic diversity of E. edulis. We expect to find high genetic diver- sity in areas of high climatic stability and high habitat suit- ability. In addition, we expect to find stronger effects of human-modified environmental factors on seedlings than in adults, given that adults may have not had enough time to respond to the recent intensification of environmental changes caused by humans, mainly forest fragmentation and functional extinction of seed dispersers in the Atlantic Forest. Methods Literature search We surveyed publications of population genetic studies on E. edulis using the following databases: Scielo (http:// www.scielo.org), Web of Science (http://www.webofsci- ence.bom), Scopus (http://www.scopus.com), and Google Scholar (http://www.scholar.google.com.br). The survey covered all publications available from 2001 to Novem- ber 2015. We used the most frequent keywords in the area: “population genetics”, “E. edulis”, “microsatellites”, “genetic diversity”, “conservation genetics”, and “heart of palm”. We included only publications comprising popu- lations from the northeast to southeast Brazilian Atlan- tic coast that used microsatellites as molecular markers (Fig. 1). Genetic diversity response variables Three genetic parameters were used as response variables in our analyses: the inbreeding coefficient (f); the genetic diversity, measured by expected heterozygosity (He) and allelic richness (AR). We chose these genetic parameters because they are present in most publications (Supple- mental Information, Table A1) and are frequently used in conservation genetics studies as surrogates for population evolutionary potential. Predictive models We used three groups of covariates to explain genetic diversity: (i) human-induced; (ii) climatic habitat suitabil- ity; and (iii) stability of climatic habitat suitability. Human- induced covariates were: (a) percentage of forest cover and (b) loss of large-body frugivores (i.e., defaunation). Cli- matic habitat suitability was measured through ecological niche modeling (ENM hereafter) for each of the time peri- ods (the Last Glacial Maximum, the mid-Holocene and the present-day). We also measured the cumulative suitability as the sum of climatic habitat suitability over the three peri- ods. The climatic stability was calculated as the difference between past and present climatic habitat suitability. To quantify the covariates for modeling genetic diver- sity, we drew a buffer area with a 2  km radius around each of the 57 selected study sites from the surveyed lit- erature. We chose this radius because the foraging distance of potential pollinators, such as Plebleia droryana and Trigona spinipes, can reach up to 540 and 840 m, respec- tively (Zurbuchen et al. 2010). Furthermore, seed dispersal distances by large-body bird species that are taxonomi- cally related to the dispersers of E. edulis are likely to be shorter than 600  m (Holbrook 2011). We used the SOS Mata Atlântica map (2005, http://mapas.sosma.org.br/) and the ArcGIS 9.2 software (ENRI) to calculate the total area and the percentage of forest inside each 2 km-radius buffer area. Defaunation was defined as the absence or very low abundance of Ramphastidae (e.g., toucans and toucanets) and Cotingidae (e.g., bellbirds) species within the buffer area and was classified as a binary variable: defaunated (0) or non-defaunated (1). Species of these families are impor- tant long distance dispersers for E. edulis seeds (Holbrook 2011; Galetti et  al. 2013). To assign the defaunation sta- tus of each site, we characterized the avian seed dispersal assemblage using the Wikiaves website (http://www.wiki- aves.com.br) and available literature (Galetti et  al. 2013; Morante-Filho et al. 2015). To assess the climatic suitability of the study area we carried out ecological niche modeling of E. edulis combin- ing species occurrences and climatic variables. A total of 229 occurrence records (Fig A1 in Supplementary mate- rial) were obtained from Species Link (http://www.splink. org.br), Jardim Botânico do Rio de Janeiro (http://www. jbrj.gov.br) and Biota Fapesp (http://www.biota.org.br/). All records were carefully checked for taxonomic reliability (i.e., synonymous and/or misidentification and duplicates). We carefully restricted our modeling area to encompass only Cerrado and Atlantic forest biomes (Fig. 2). http://www.scielo.org http://www.scielo.org http://www.webofscience.bom http://www.webofscience.bom http://www.scopus.com http://www.scholar.google.com.br http://mapas.sosma.org.br/ http://www.wikiaves.com.br http://www.wikiaves.com.br http://www.splink.org.br http://www.splink.org.br http://www.jbrj.gov.br http://www.jbrj.gov.br http://www.biota.org.br/ 470 Conserv Genet (2017) 18:467–478 1 3 The environmental space required for the ENMs was characterized by five bioclimatic variables: temperature seasonality, maximum temperature of the warmest month, precipitation of the wettest quarter, precipitation of the dri- est quarter, precipitation of the warmest quarter together with elevation and soil classes (http://www.embrapa.br). Fig. 1 Distribution of the 57 sampled sites with genetic infor- mation for Euterpe edulis across the Brazilian Atlantic forest biome in Brazil 40°0'W45°0'W50°0'W 15°0'S 20°0'S 25°0'S 0 250 Km Atlantic forest remnants Sites Brazil Fig. 2 Maps of mean climatic suitability across the ten ecological niche models (ENMs), five Atmosphere–Ocean General Circulation Models (AOGCMs) and three time periods (at the LGM, mid-Holocene and present, respectively) for Euterpe edulis in Brazil http://www.embrapa.br 471Conserv Genet (2017) 18:467–478 1 3 Climatic variables were selected out of the 19 bioclimatic variables (Worldclim; http://worldclim.org; Hijmans et  al. 2005) after performing a factor analysis with Varimax rota- tion to minimize collinearity among the variables when building the distribution models (For details on selection of variables see Supplementary material, Table  A2). We derived distribution models using layers from the pre- industrial period (representing current conditions; 0  ka), the mid-Holocene (6 ka) and the Last Glaciation Maxima (LGM, 21  ka). Current models were then projected onto three Atmosphere–Ocean General Circulation Models (AOGCMs; CCSM4, MIROC-ESM and MPI-ESM-P) obtained from Coupled Model Intercomparison Project Phase 5 (CMIP5; http://cmip-pcmdi.llnl.gov/cmip5/). We assumed soil and elevation as constraint variables (constant throughout time). All layers were used in a spatial resolu- tion of 2.5 arc-min (ca. 5 × 5 km). A total of 200 spatially independent localities were used for the models. ENM was carried out with biomod2 in R 3.1.2 (Thuiller et  al. 2009). Six different algorithms were used (see Supplementary material for model param- eterization): Artificial Neural Networks (ANN), General- ized Additive Models (GAM), Generalized Linear Models (GLM), Generalized Boosted Models (GBM), Random Forest (RF), and maximum entropy (MAXENT; Phillips and Dudík 2008). Usually ENM requires presence-absence data for model fitting and testing; however, there are not reliable absences available. Therefore, we chose to gener- ate pseudo-absences randomly across the whole area (Fer- rier et  al. 2002), excluding cells with presences to ensure a complete sampling of the environmental space and mini- mum extrapolation when projecting to different historical period. To maximize model performance, pseudo-absences were weighted to ensure maintaining prevalence equal to 0.5 (Barbet-Massin et al. 2012). We used random data par- titioning (80% training, 20% testing) and five repetitions for each model to assess model stability due to the absence of independent testing data. We assessed model accuracy through the calculation of the True Skill Statistic–TSS (Allouche et  al. 2006) and the area under the receiver curve–AUC (Fielding and Bell 1997) (Table  A3 in Sup- plementary material). Since results might differ among different algorithms, we performed an ensemble modeling (Araújo and New 2007) to calculate a consensus map that included only models with a TSS > 0.7 and an AUC > 0.75 and the resulting map was weighted by the individual model TSS value. Thus we generated a surface represent- ing the suitability values for the occurrence of the species across our study area for both current and past climatic periods. To access climatic stability we calculated the difference between the past (6 and 21 ka) and current (0 ka) suitability within the 2 km radius buffers for each site. Positive values indicate that current climatic suitability is lower than in the past (contraction population range) and negative values show that the current climatic suitability is higher than in the past (expansion population range). Zero value indicates that the site is climatically stable, which could either be associated with high environmental suitability or not. To account for the cumulative suitability over time, we also calculated the sum of suitability in the three different time periods. Statistical analyses We investigated the effects of the explanatory variables (i.e., defaunation status, percentage of forest cover, cli- matic habitat suitability in the present-day, cumulative cli- matic habitat suitability, stability from the mid-Holecene to present-day and stability from the LGM to present-day) on the three genetic parameters (i.e., f; He, AR). We fitted full models only for adults, only for seedlings and adults and seedlings combined for each genetic parameter using a linear mixed effects and the maximum-likelihood method using the R package lme4 (Bates et al. 2015), which is an adequate way to assess the influence of fixed effects given random structure. Defaunation status, percentage of for- est cover, current climatic habitat suitability, cumula- tive climatic habitat suitability, climatic stability from the mid-Holocene to present-day and climatic stability from the LGM to present-day were considered as fixed effects. Ontogenetic stage (adults or seedlings) and “study” were included as random effects while sample size of each site was included as a weight parameter in all models. The “study” was included to take into account the difference in the set of microsatellites markers and in the number of pop- ulations and individuals sampled (Table A1). We assumed that these factors might influence the estimation of genetic parameters. The ontogenetic stage (adults and seedlings) was included as random effects only in the model that included all individuals. After fitting the full model, we fitted nested models with all possible combinations of the fixed effects. Because the effect of historical and current factors on genetic variables tends to be complex and dependent on the interaction of diverse factors (Boscolo and Metzger 2011), we used a model selection with multiple compet- ing hypotheses approach (Burnham and Anderson 2002). To select the best model, we used the Akaike Information Criteria (AIC), where the model with the lowest AICc (AIC corrected by sample size and number of parameters) (Burnham and Anderson 2002) was considered as the more plausible to explain observed patterns. Delta AICci (ΔAICci, where i represents each model) was calculated as the difference between AICc for the model i and the smallest observed AICc. We determined Akaike’s weight http://worldclim.org http://cmip-pcmdi.llnl.gov/cmip5/ 472 Conserv Genet (2017) 18:467–478 1 3 of evidence (wAICc) representing the relative contribution of model i to explain the observed pattern, given a set of competing models (Burnham and Anderson 2002). Mod- els with ΔAICc < 2 were considered as equally plausible to explain the observed pattern (Burnham and Anderson 2002). We also carried out a model averaging approach to compute the parameter average from all equivalent models (ΔAICc < 2) and the significance of the estimated parame- ter values associated with each explanatory variable (Burn- ham and Anderson 2002). All analyses were performed using the MuMIn packages implemented in R version 3.0.2 (Barton 2016). We also analyzed if seedlings have on aver- age lower levels of genetic diversity than adult populations. To account for the presence of spatial autocorrelation, we ran linear mixed models with the restricted maximum- likelihood method with different spatial covariance struc- tures (Gaussian, exponential and spherical) and compared them to a model without a spatial covariance structure. We ran these models by fitting all explanatory covariates as fixed effects and the study and ontogenetic stage as random effects. Results Literature search We obtained eight publications, which accounted for 61 study sites. However, four sites were excluded from the analysis because they were outside of the Atlantic Forest biome; in the end, we used 57 sites (Fig. 1 and Supplemen- tal Information Table  A1). Most populations of E. edulis in the Atlantic forest biome showed high genetic diversity (Supplemental Information Table A1). He ranged from 0.57 to 0.86, and AR varied from 3.6 to 15.1. However, many populations showed high f value (Supplemental Informa- tion Table A1). Climatic suitability and stability models A total of 30 models were generated by the ENMs. From those, 83% were considered good under AUC criteria (Mean AUC, Table  S3). The map of the modeled distri- bution under present conditions mostly agrees with the known current distribution for the species. The consensus map reveals suitable areas for the occurrence of the species across the coastal part of Brazil, from Pernambuco state in the northeast to Santa Catarina state in the south (Fig. 2). The region of higher climatic suitability coincides with the core area of Atlantic rain forests (Fig.  2) and shows the preference of E. edulis for hot and humid climates. Hindcasting to the LGM and the mid-Holocene showed a retraction of suitable areas through time, especially in areas across central Brazil. The differences between past and cur- rent suitability values show a strong variation along the geographic distribution of the Atlantic Forest, with a reduc- tion in habitat suitability in central Brazil during the mid- Holocene (i.e., retraction) and high stability and suitabil- ity since the LGM in areas along the coastal part of Brazil (Fig. 2; Table A1). Effects of historical and current factors on genetic diversity The results of the model selection indicated that models without a spatial covariance structure were considerably better for modeling He, AR and f than those with spatial covariance structures (Gaussian, exponential and spheri- cal) (Supplementary material Table  A4). Considering all individuals, variation in He and AR were best explained by the lack of large seed dispersers (Table 1). Defaunated sites presented lower He and AR than non-defaunated sites (Fig.  3a, b). The inbreeding coefficient (f) was best explained by climatic stability from the mid-Holocene to present-day (Table 1). The inbreeding coefficient was lower in areas with decreasing suitability values over time (habi- tat contraction) and was higher in areas under expansion (Fig. 3c). We found that seedlings have on average lower lev- els of allelic richness than adult individuals (t = −2.49, p = 0.01), however we did not find differences in He and f between ontogenetic stages (t = −0.03, p = 0.97; t = −1.5, p = 0.12, respectively). Considering only adults, varia- tion of AR could not be explained by any of our compet- ing models (Table 1). On the other hand, He was explained by climatic stability from the mid-Holocene to present-day (Table 1). Lower values of He were associated with popula- tions in areas where climatic suitability decreased from the mid-Holocene to present-day (Fig.  3d). Nevertheless, the genetic diversity (He) remained high in areas under expan- sion or that were consistently suitable over time (Fig. 3d). Variation of f was best explained by the percentage of for- est cover and by climatic stability from the mid-Holocene to present-day (Table  1). Populations in landscapes with higher percentage of forest cover showed lower f than pop- ulations in landscapes with lower forest cover (Fig. 3e). The model also shows that there is a decrease in the inbreeding coefficient when there is a retraction from suitable areas through time (positive climatic stability, Fig. 3f). Likewise, the inbreeding coefficient (f) increases in areas undergoing expansion over time (Fig. 3f). Considering only seedlings, variation in He was not explained by any of our compet- ing models (Table 1); however, f was best explained by the climatic stability from the mid-Holocene to present-day (Table 1), as found for adults (Fig. 3g). Finally, defaunation 473Conserv Genet (2017) 18:467–478 1 3 (Table 1) explained variation in AR. Non-defaunated sites presented higher AR than defaunated sites (Fig. 3h). Discussion Our results show a retraction in the suitable areas for E. edulis through time and that the current pattern of E. edu- lis genetic diversity is the result of such historical popula- tion retraction and the current human impact. Combined effects of climatic habitat suitability, lack of large seed dis- persers and habitat loss and fragmentation affected genetic diversity of heart of palm populations. Nevertheless, these observed patterns were not consistent across ontogenetic stages, as adults and seedlings are affected by different anthropogenic factors. The complex patterns found between different genetic parameters with historical environmental variables and human impacts indicate that it may be diffi- cult to disentangle the processes involved in the origin and maintenance of genetic diversity. The consensus map of climatic habitat suitability in the present-day (0 ka), mid-Holocene (6 ka BP) and LGM (21 ka) revealed a contraction in climatically suitable areas for E. edulis from central sites towards the Brazilian Atlan- tic coast. Our ENM results showed that the potential dis- tribution of E. edulis is currently narrower than it was in the mid-Holocene and in the LGM. The strong reduction in suitability during the mid-Holocene might have caused population extinction at least in part of the E. edulis dis- tribution. Nevertheless, some areas, mainly in the Brazil- ian Atlantic coast, have been climatically stable through time and might have been important climatic refugia for E. edulis. In fact, we found that climatically stable sites can harbor a high amount of genetic diversity in heart of palm populations. Other studies also found that the Brazil- ian Atlantic coast may have been refuge for natural popula- tions during the Quaternary climatic oscillation (Carnaval et  al. 2009; Leite et  al. 2016). Euterpe edulis populations in these refugia with stable or increasing suitability over time presented higher diversity compared to populations in areas with decreasing suitability since the mid-Holocene. Higher genetic diversity in populations under recent expan- sion probably reflects the admixture of immigrants arriv- ing from different populations and also the maintenance of Table 1 Summary statistics for the model-averaging used to explain the genetic diversity pattern in populations of Euterpe edulis from 57 sites, distributed along the Brazilian Atlantic Rainforest Model averaged indicates the model-averaged estimates for fixed effects and SE the standard errors, p-value, p value for estimate coefficient values; He expected heterozygozity; AR allelic richness; f inbreeding coefficient. Effects where p value < 0.05 are highlighted by stars (*) and in bold numbers Parameter He f AR Model averaged (SE) p-value Model averaged (SE) p-value Model averaged (SE) p-value All individuals  Defaunation* 0.29 (0.13) 0.03 – – 0.24 (0.08) 0.00  % Forest cover – – −0.13 (0.09) 0.18 −0.05 (0.05) 0.28  Current climatic suitability −0.02 (0.23) 0.93 −0.12 (0.08) 0.17 −0.22 (0.11) 0.06  Climatic stability Holocene-present* – – −0.29 (0.14) 0.04 – –  Climatic stability LGM-present 0.15 (0.09) 0.10 −0.11 (0.11) 0.31 – –  Cumulative suitability 0.22 (0.14) 0.12 −0.09 (0.08) 0.29 0.27 (0.16) 0.09 Adults  Defaunation 0.48 (0.26) 0.09 – – 0.28 (0.19) 0.17  % Forest cover* – – −0.57 (0.14) 0.00 −0.16 (0.12) 0.21  Current climatic suitability – – – – −0.36 (0.17) 0.06  Climatic stability Holocene-present* −0.28 (0.13) 0.04 −0.33 (0.14) 0.03 – –  Climatic stability LGM-present – – – – – –  Cumulative suitability 0.22 (0.12) 0.08 – – 0.35 (0.20) 0.09 Seedlings  Defaunation* 0.15 (0.18) 0.43 – – 0.39 (0.11) 0.00  % Forest cover 0.09 (0.10) 0.39 – – −0.01 (0.06) 0.86  Current climatic suitability 0.11 (0.08) 0.23 −0.15 (0.09) 0.11 0.08 (0.13) 0.55  Climatic stability Holocene-present* −0.19 (0.18) 0.31 −0.45 (0.17) 0.01 0.07 (0.11) 0.53  Climatic stability LGM-present – – −0.22 (0.12) 0.08 −0.03 (0.07) 0.65  Cumulative suitability 0.11 (0.10) 0.29 −0.12 (0.10) 0.24 −0.01 (0.17) 0.92 474 Conserv Genet (2017) 18:467–478 1 3 large effective population sizes (Ortego et al. 2015). On the other hand, range retractions due to low environmental suit- ability might lead to population bottlenecks, with loss of individuals and genetic diversity (Pauls et al. 2013). Conversely, our results show that populations of both adults and seedlings in areas under recent expansion of cli- matic suitability have higher inbreeding coefficients. The high expected heterozygosity (He) despite high inbreeding coefficients (f) observed in some populations may indicate assortative mating between E. edulis palms most likely due to short-distance pollination, spatial distribution of plants, fine-scale spatial genetic structure or flowering patterns (Ghazoul 2005). Another possible explanation is that self- ing may occur more frequently to favor reproductive assur- ance, which may be advantageous in expanded populations (Arnaud-Haond et  al. 2006; Pujol et  al. 2009). Thus, we Fig. 3 Relationship between genetic parameters and the main explanatory factors in 57 study sites of Euterpe edulis across the Atlantic forest in Brazil. a genetic diversity (He) and defaunation for all individu- als, b allelic richness (AR) and defaunation for all individu- als, c inbreeding coefficient (f) and climatic stability from mid-Holocene to present-day for all individuals, d genetic diversity (He) and climatic stability from mid-Holocene to present-day for adults, e inbreeding coefficient (f) and percentage of forest cover for adults, f inbreeding coefficient (f) and climatic stability from mid-Holocene to present-day for adults, g inbreeding coefficient (f) and climatic stability from mid-Holocene to present-day for seedlings, and h allelic rich- ness (AR) and defaunation for seedlings He Faunated 14 Defaunated 0. 65 0. 80 0. 70 0. 75 0. 60 FaunatedDefaunated A R 0. 85 12 10 8 6 4 f 0. 30 0. 20 0. 10 01.0- 00.0 -0 .2 0 mid-Holocene to present day 0.00 0.15-0.10-0.20 He mid-Holocene to present day 0. 90 07.0 08.0 0. 60 0.000.10 0.100.05-0.05 0. 20 0. 10 0. 00 20 10006 0840 Forest cover (%) 0.15 mid-Holocene to present day 0.00 0.100.05-0.05 f 0. 20 0. 10 0. 00 f f 0. 00 0. 20 -0 .2 0 mid-Holocene to present day 0.00 0.10-0.10-0.20 FaunatedDefaunated 14 A R 12 10 8 6 4 A B C D F G E H 475Conserv Genet (2017) 18:467–478 1 3 suggest that populations in areas with expanded range are an admixture of individuals from different populations, yet with production of inbred descendants. It could be the result of increased reproductive assurance due to self- ing and increased biparental inbreeding due to assortative mating. On the other hand, we found low inbreeding coefficients in populations that have recently contracted. The reduction of habitat suitability may lead to a reduction of popula- tion size and density. Some authors argue that population density is critical when the distance between individuals leads to changes in the foraging behavior of pollinators (Ghazoul 2005; Côrtes et al. 2013). Low population density and higher distance among individuals may increase pol- len dispersal distance due to high mobility of pollinators in search for resources (Côrtes et  al. 2013). Low inbreeding coefficients may also be the result of loss of homozygous individuals due to lower survival success (i.e., inbreeding depression, Keller and Waller et  al. 2002), mainly in low suitable habitats. Thus, we suggest that while contracting populations lose genetic diversity due to bottleneck effects, inbreeding remains low due to higher mobility of pollina- tors in low density populations or higher loss of homozy- gous individuals in unsuitable habitats. Along with climatic suitability, anthropogenic factors may also affect the genetic diversity of E. edulis. Despite the high fragmentation and habitat loss E. edulis popula- tions still present high genetic diversity within the Atlan- tic forest (Gaiotto et al. 2003; Conte et al. 2008; Carvalho et  al. 2015). The high historical effective population size (Galetti et  al. 2013) and wide historical climatic habitat suitability distribution may have led to high genetic diver- sity in currently small and isolated populations. However, we did find significant and negative effects of defaunation on the genetic diversity (He) and allelic richness of this spe- cies. The loss of large-body seed dispersers may result in the loss of very important ecological functions for plants that include the colonization of new habitats and connec- tivity between isolated populations (Pacheco and Simonetti 2000; Markl et al. 2012). Connectivity between populations is important for maintaining gene flow and, consequently, large effective population size and genetic diversity (Young et  al. 1996). Moreover, habitat loss increased inbreeding levels in E. edulis. Habitat loss may affect the composition, abundance and behavior of pollinators communities, which in turn can limit pollination by increasing the probability of selfing and mating between closely related individuals. In addition, decrease in the amount of forest cover is com- monly associated with higher fragmentation (Ribeiro et al. 2009), with the creation of smaller and more isolated habi- tat remnants. In this scenario, it is also possible that gene flow among patches within these fragmented landscapes was also negatively impacted, resulting in smaller effective population sizes and higher inbreeding. Adults and seedlings can show distinct responses to anthropogenic disturbances (Van Geert et  al. 2008). Due to time-lag effects it is expected that genetic diversity of adult perennial plants respond strongly to past landscape conditions rather than to recent habitat changes (Collevatti et  al. 2001; Kramer et  al. 2008). This is because changes in the composition and spatial configuration of landscapes occur faster than changes in the genetic diversity of arbo- real plants, which can take many generations to respond genetically to human impacts (Landguth et al. 2010). Our results partially corroborate this hypothesis, as seedling and adults behave differently. Seedlings and adults inbreeding coefficients (f) were both influenced by climate variation from the mid-Holocene to present-day. Adults He was also influenced by the same historical condition. Both seedling and adults showed to be negatively influenced by human impacts; however adults were mainly affected by the reduc- tion of forest cover while seedling by the loss of large seed dispersers. Climatic models forecast an increase in mean global temperature due to anthropogenic factors (IPCC 2007). Species can cope with global climate changes if their pop- ulations have the capacity to adapt to new environmental conditions or to disperse and track suitable habitats (Daw- son et al. 2011). Unfortunately, most natural habitats have been replaced by anthropogenic landscapes dominated by agricultural fields and urban settlements (Foley et al. 2005), which may reduce the capacity of individuals to disperse and find suitable habitat. The amount of genetic diversity is essential for species to cope with environmental changes (Jump et al. 2009). Although E. edulis still has high genetic diversity, we showed that the spatial distribution of this species, as well as its genetic composition, is affected by climate changes. Therefore, an efficient species conserva- tion planning should include the conservation of climati- cally stable areas that also harbors large extensions of for- est and diversified animal community to preserve species interactions and ecological functions such as pollination and seed dispersal. Finally, our results suggest that the pattern of genetic diversity in E. edulis is the outcome of historical instabil- ity during the mid-Holocene and current anthropogenic factors, mainly those that affect important ecological pro- cesses such as seed dispersal. Current human impacts may enhance historical processes, as well as shape the genetic diversity of heart of palm. Thus, an efficient plan for spe- cies conservation is urgent and must account for human impacts and environmental suitability, as well as assess genetic diversity of seedlings and adults in fragmented landscapes. 476 Conserv Genet (2017) 18:467–478 1 3 Acknowledgements This work was supported by the competi- tive grants from CNPq (Project No. 445353/2014-7) and CAPES (PROCAD Project No. 88881.068425/2014-01) which we gratefully acknowledge. CSC received a CNPq grant (Project No. 401258/2012- 2) and a scholarship from Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP, No. 2014/01029-5). LBM receives a fellow- ship from CAPES Ciências sem Fronteira (Project CSF-PAJT/CAPES No. 88881.030318/2013-01). RGC and MCR have been continuously supported by grants and scholarships from CNPq and CAPES. MCR is funded by FAPESP (Project No. 2013/50421-2). Author Contributions C.S.Carvalho conceptualized and per- formed the study, analyzed the data and wrote the paper. M.C.Ribeiro, R.G.Collevatti and L. Ballesteros-Mejia participated in the study design, analyzed the data and contributed to writing the paper. M.C.Côrtes assisted in the data analyses and contributed to writing the paper. A.S.Santos contributed to the study conception and to writ- ing the paper. References Aguilar R, Quesada M, Ashworth L et  al (2008) Genetic conse- quences of habitat fragmentation in plant populations: suscepti- ble signals in plant traits and methodological approaches. Mol Ecol 17:5177–5188. doi:10.1111/j.1365-294X.2008.03971.x Allouche O, Tsoar A, Kadmon R (2006) Assessing the accu- racy of species distribution models: prevalence, kappa and the true skill statistic (TSS). J Appl Ecol 43:1223–1232. doi:10.1111/j.1365-2664.2006.01214.x Andersen LW, Fog K, Damgaard C (2004) Habitat fragmentation causes bottlenecks and inbreeding in the European tree frog (Hyla arborea). Proc Biol Sci 271:1293–1302. doi:10.1098/ rspb.2004.2720 Araújo MB, New M (2007) Ensemble forecasting of species dis- tributions. Trends Ecol Evol 22:42–47. doi:10.1016/j. tree.2006.09.010 Arnaud-Haond S, Teixeira S, Massa SI et  al (2006) Genetic struc- ture at range edge: low diversity and high inbreeding in South- east Asian mangrove (Avicennia marina) populations. Mol Ecol 15:3515–3525. doi:10.1111/j.1365-294X.2006.02997.x Barbet-Massin M, Jiguet F, Albert CH, Thuiller W (2012) Select- ing pseudo-absences for species distribution models: how, where and how many? Methods Ecol Evol 3:327–338. doi:10.1111/j.2041-210X.2011.00172.x Barton, K (2016) MuMIn: multi-model inference, R package version 1.15.6. Bates D, Mächler M, Bolker B, Walker S (2015) Fitting linear mixed- effects models using lme4. J Stat Softw 67:1–48. doi:10.18637/ jss.v067.i01 Boscolo D, Metzger JP (2011) Isolation determines patterns of spe- cies presence in highly fragmented landscapes. Ecography 34:1018–1029. doi:10.1111/j.1600-0587.2011.06763.x Breed MF, Gardner MG, Ottewell KM et  al (2012) Shifts in repro- ductive assurance strategies and inbreeding costs associated with habitat fragmentation in Central American mahogany. Ecol Lett 15:444–452. doi:10.1111/j.1461-0248.2012.01752.x Burnham KKP, Anderson DRD (2002) Model selection and mul- timodel inference: a practical information-theoretic approach. Springer, New York Carnaval ACOQ, Hickerson MJ, Haddad CFB et  al (2009) Stability predicts genetic diversity in the Brazilian Atlantic forest hotspot. Science 323:785–789. doi:10.1126/science.1166955 Carvalho CS, Ribeiro MC, Côrtes MC et  al (2015) Contemporary and historic factors influence differently genetic differentia- tion and diversity in a tropical palm. Heredity 115:216–224. doi:10.1038/hdy.2015.30 Collevatti RG, Grattapaglia D, Hay JD (2001) Population genetic structure of the endangered tropical tree species Caryocar bra- siliense, based on variability at microsatellite loci. Mol Ecol 10:349–356. doi:10.1046/j.1365-294X.2001.01226.x Collevatti RG, Lima-Ribeiro MS, Terribile LC et al (2014) Recov- ering species demographic history from multi-model infer- ence: the case of a Neotropical savanna tree species. BMC Evol Biol 14:1–13. doi:10.1186/s12862-014-0213-0 Conte R, dos Reis SM, Mantovani A, Vencovsky R (2008) Genetic structure and mating system of Euterpe edulis Mart. Popula- tions: a comparative analysis using microsatellite and allozyme markers. J Hered 99:476–482. doi:10.1093/jhered/esn055 Côrtes MC, Uriarte M, Lemes MR et al (2013) Low plant density enhances gene dispersal in the Amazonian understory herb Heliconia acuminata. Mol Ecol 22:5716–5729. doi:10.1111/ mec.12495 Dawson TP, Jackson ST, House JI et  al (2011) Beyond predic- tions: biodiversity conservation in a changing climate. Science 332:53–58. doi:10.1126/science.1200303 DiBattista JD (2007) Patterns of genetic variation in anthropo- genically impacted populations. Conserv Genet 9:141–156. doi:10.1007/s10592-007-9317-z Diniz-Filho JAF, Nabout JC, Bini LM et al (2009) Niche modelling and landscape genetics of Caryocar brasiliense (Pequi tree: Caryocaraceae) in Brazilian Cerrado: an integrative approach for evaluating central-peripheral population patterns. Tree Genet Genomes 5:617–627. doi:10.1007/s11295-009-0214-0 Dirzo R, Miranda A (1991) Altered patterns of herbivory and diver- sity in the forest understory: a case study of the possible con- sequences of contemporary defaunation. In: Price PW et  al (eds) Plant–animal interactions: evolutionary ecology in tropi- cal and temperate regions. Wiley, New York, pp 273–287 Eycott AE, Stewart GB, Buyung-Ali LM et  al (2012) A meta- analysis on the impact of different matrix structures on spe- cies movement rates. Landsc Ecol 27:1263–1278. doi:10.1007/ s10980-012-9781-9 Fahrig L (2003) Effects of habitat fragmentation on biodiversity. Annu Rev Ecol Evol Syst 34:487–515. doi:10.1146/annurev. ecolsys.34.011802.132419 Ferrier S, Watson G, Pearce J, Drielsma M (2002) Extended statis- tical approaches to modelling spatial pattern in biodiversity in northeast New South Wales. I. Species-level modelling. Bio- divers Conserv 11:2275–2307. doi:10.1023/A:1021302930424 Fielding AH, Bell JF (1997) A review of methods for the assessment of prediction errors in conservation presence/absence models. Environ Conserv 24:38–49. doi:10.1017/S0376892997000088 Foley JA, DeFries R, Asner GP et al (2005) Global consequences of land use. Science 309:570–574. doi:10.1126/science.1111772 Frankham R, Lees K, Montgomery ME et al (1999) Do population size bottlenecks reduce evolutionary potential? Anim Conserv 2:255–260. doi:10.1111/j.1469-1795.1999.tb00071.x Gaiotto FA, Grattapaglia D, Vencovsky R (2003) Genetic structure, mating system, and long-distance gene flow in heart of palm (Euterpe edulis Mart.). J Hered 94:399–406. doi:10.1093/ jhered/esg087 Galetti M, Fernandez JC (1998) Palm heart harvesting in the Brazilian Atlantic forest: changes in industry struc- ture and the illegal trade. J Appl Ecol 35:294–301. doi:10.1046/j.1365-2664.1998.00295.x Galetti M, Guevara R, Côrtes MC et al (2013) Functional extinction of birds drives rapid evolutionary changes in seed size. Science 340:1086–1090. doi:10.1126/science.1233774 http://dx.doi.org/10.1111/j.1365-294X.2008.03971.x http://dx.doi.org/10.1111/j.1365-2664.2006.01214.x http://dx.doi.org/10.1098/rspb.2004.2720 http://dx.doi.org/10.1098/rspb.2004.2720 http://dx.doi.org/10.1016/j.tree.2006.09.010 http://dx.doi.org/10.1016/j.tree.2006.09.010 http://dx.doi.org/10.1111/j.1365-294X.2006.02997.x http://dx.doi.org/10.1111/j.2041-210X.2011.00172.x http://dx.doi.org/10.18637/jss.v067.i01 http://dx.doi.org/10.18637/jss.v067.i01 http://dx.doi.org/10.1111/j.1600-0587.2011.06763.x http://dx.doi.org/10.1111/j.1461-0248.2012.01752.x http://dx.doi.org/10.1126/science.1166955 http://dx.doi.org/10.1038/hdy.2015.30 http://dx.doi.org/10.1046/j.1365-294X.2001.01226.x http://dx.doi.org/10.1186/s12862-014-0213-0 http://dx.doi.org/10.1093/jhered/esn055 http://dx.doi.org/10.1111/mec.12495 http://dx.doi.org/10.1111/mec.12495 http://dx.doi.org/10.1126/science.1200303 http://dx.doi.org/10.1007/s10592-007-9317-z http://dx.doi.org/10.1007/s11295-009-0214-0 http://dx.doi.org/10.1007/s10980-012-9781-9 http://dx.doi.org/10.1007/s10980-012-9781-9 http://dx.doi.org/10.1146/annurev.ecolsys.34.011802.132419 http://dx.doi.org/10.1146/annurev.ecolsys.34.011802.132419 http://dx.doi.org/10.1023/A:1021302930424 http://dx.doi.org/10.1017/S0376892997000088 http://dx.doi.org/10.1126/science.1111772 http://dx.doi.org/10.1111/j.1469-1795.1999.tb00071.x http://dx.doi.org/10.1093/jhered/esg087 http://dx.doi.org/10.1093/jhered/esg087 http://dx.doi.org/10.1046/j.1365-2664.1998.00295.x http://dx.doi.org/10.1126/science.1233774 477Conserv Genet (2017) 18:467–478 1 3 García C, Jordano P, Godoy JA (2007) Contemporary pollen and seed dispersal in a Prunus mahaleb population: pat- terns in distance and direction. Mol Ecol 16:1947–1955. doi:10.1111/j.1365-294X.2006.03126.x Gaston KJ (2000) Global patterns in biodiversity. Nature 405:220– 227. doi:10.1038/35012228 Ghazoul J (2005) Pollen and seed dispersal among dispersed plants. Biol Rev Camb Philos Soc 80:413–443. doi:10.1017/ S1464793105006731 González-Varo JP, Arroyo J, Aparicio A (2009) Effects of fragmenta- tion on pollinator assemblage, pollen limitation and seed produc- tion of Mediterranean myrtle (Myrtus communis). Biol Conserv 142:1058–1065. doi:10.1016/j.biocon.2009.01.017 Gugger PF, Ikegami M, Sork VL (2013) Influence of late Quaternary climate change on present patterns of genetic variation in val- ley oak, Quercus lobata. Mol Ecol 22:3598–3612. doi:10.1111/ mec.12317 Henderson A, Galeano-Garces G, Bernal R (1997) Field guide to the palms of the Americas. Princeton University Press, Princeton Hewitt G (2000) The genetic legacy of the Quaternary ice ages. Nature 405:907–913. doi:10.1038/35016000 Hijmans RJ, Cameron SE, Parra JL et al (2005) Very high resolution interpolated climate surfaces for global land areas. Int J Climatol 25:1965–1978. doi:10.1002/joc.1276 Hoffmann AA, Willi Y (2008) Detecting genetic responses to envi- ronmental change. Nat Rev Genet 9:421–432. doi:10.1038/ nrg2339 Holbrook KM (2011) Home range and movement patterns of tou- cans: implications for seed dispersal. Biotropica 43:357–364. doi:10.1111/j.1744-7429.2010.00710.x Huber PR, Greco SE, Thorne JH (2010) Spatial scale effects on conservation network design: trade-offs and omissions in regional versus local scale planning. Landsc Ecol 25:683–695. doi:10.1007/s10980-010-9447-4 Hughes AR, Inouye BD, Johnson MTJ et  al (2008) Ecologi- cal consequences of genetic diversity. Ecol Lett 11:609–623. doi:10.1111/j.1461-0248.2008.01179.x IPCC (2007) Climate change 2007: synthesis report. In: Core Writing Team, Pachauri RK, Reisinger A (eds) Contribution of working groups I, II and III to the Fourth Assessment Report of the Inter- governmental Panel on Climate Change. IPCC, Geneva Jordano P, García C, Godoy JA, García-Castaño JL (2007) Dif- ferential contribution of frugivores to complex seed dispersal patterns. Proc Natl Acad Sci 104:3278–3282. doi:10.1073/ pnas.0606793104 Jump AS, Marchant R, Peñuelas J (2009) Environmental change and the option value of genetic diversity. Trends Plant Sci 14:51–58. doi:10.1016/j.tplants.2008.10.002 Keller LF, Waller DM (2002) Inbreeding effects in wild popu- lations. Trends Ecol Evol 17:230–241. doi:10.1016/ S0169-5347(02)02489-8 Kramer AT, Ison JL, Ashley MV, Howe HF (2008) The Paradox of forest fragmentation genetics. Conserv Biol 22:878–885. doi:10.1111/j.1523-1739.2008.00944.x Landguth EL, Cushman S a, Schwartz MK et al (2010) Quantifying the lag time to detect barriers in landscape genetics. Mol Ecol 19:4179–4191. doi:10.1111/j.1365-294X.2010.04808.x Leite YLR, Costa LP, Loss AC et  al (2016) Neotropical forest expansion during the last glacial period challenges refuge hypothesis. Proc Natl Acad Sci 113:201513062. doi:10.1073/ pnas.1513062113 Lowe AJ, Boshier D, Ward M et al (2005) Genetic resource impacts of habitat loss and degradation; reconciling empirical evidence and predicted theory for neotropical trees. Heredity 95:255–273. doi:10.1038/sj.hdy.6800725 Markl JS, Schleuning M, Forget PM et al (2012) Meta-analysis of the effects of human disturbance on seed dispersal by animals. Con- serv Biol 26:1072–1081. doi:10.1111/j.1523-1739.2012.01927.x Miraldo A, Li S, Borregaard MK et al (2016) An Anthropocene map of genetic diversity. Science 353:1532–1535. doi:10.1126/sci- ence.aaf4381 Morante-Filho JC, Faria D, Mariano-Neto E, Rhodes J (2015) Birds in anthropogenic landscapes: the responses of ecological groups to forest loss in the Brazilian Atlantic forest. PLoS ONE 10:1– 18. doi:10.1371/journal.pone.0128923 Ortego J, Riordan EC, Gugger PF, Sork VL (2012) Influence of envi- ronmental heterogeneity on genetic diversity and structure in an endemic southern Californian oak. Mol Ecol 21:3210–3223. doi:10.1111/j.1365-294X.2012.05591.x Ortego J, Gugger PF, Sork VL (2015) Climatically stable landscapes predict patterns of genetic structure and admixture in the Cali- fornian canyon live oak. J Biogeogr 42:328–338. doi:10.1111/ jbi.12419 Pacheco LF, Simonetti JA (2000) Genetic structure of a mimosid tree deprived of it seed disperser, the spider monkey. Conserv Biol 14:1766–1775. doi:10.1046/j.1523-1739.2000.99182.x Pauls SU, Nowak C, Bálint M, Pfenninger M (2013) The impact of global climate change on genetic diversity within populations and species. Mol Ecol 22:925–946. doi:10.1111/mec.12152 Phillips SJ, Dudík M (2008) Modeling of species distributions with Maxent: new extensions and a comprehensive evaluation. Ecog- raphy 31:161–175. doi:10.1111/j.0906-7590.2008.5203.x Pujol AB, Zhoua S, Vilasa JS, Pannella JR (2009) Reduced inbreed- ing depression after species range expansion. Proc Natl Acad Sci 106: 15379–15383. doi:10.1073/pnas.0902257106 Rands MRW, Adams WM, Bennun L et al (2010) Biodiversity con- servation: challenges beyond 2010. Science 329:1298–1303. doi:10.1126/science.1189138 Reis MS, Fantini AC, Nodari RO, Reis A, Guerra MP, Manto- vani A (2000) Management and conservation of natural populations in Atlantic rainforest: the case study of palm heart (Euterpe edulis Martius). Biotropica 32:894–902. doi:10.1111/j.1744-7429.2000.tb00627.x Ribeiro MC, Metzger JP, Martensen AC et  al (2009) The Brazilian Atlantic forest: how much is left, and how is the remaining for- est distributed? Implications for conservation. Biol Conserv 142:1141–1153. doi:10.1016/j.biocon.2009.02.021 Ribeiro MC, Martensen AC, Metzger JP et  al (2011) The Brazil- ian Atlantic forest: a shrinking biodiversity hotspot. In: Zachos FE, Habel JC (eds) Biodiversity hotspots. Springer, New York, pp 405–434 Tabarelli M, Aguiar AV, Ribeiro MC et al (2010) Prospects for biodi- versity conservation in the Atlantic Forest: Lessons from aging human-modified landscapes. Biol Conserv 143:2328–2340. doi:10.1016/j.biocon.2010.02.005 Thuiller W, Lafourcade B, Engler R, Araújo MB (2009) BIOMOD—a platform for ensemble forecasting of species distributions. Ecog- raphy 32:369–373. doi:10.1111/j.1600-0587.2008.05742.x Uriarte M, Anciães M, da Silva MTB et  al (2011) Disentangling the drivers of reduced long-distance seed dispersal by birds in an experimentally fragmented landscape. Ecology 92:924–937. doi:10.1890/10-0709.1 Van Geert A, Van Rossum F, Triest L (2008) Genetic diversity in adult and seedling populations of Primula vulgaris in a frag- mented agricultural landscape. Conserv Genet 9:845–853. doi:10.1007/s10592-007-9409-9 Walter RP, Cena CJ, Morgan GE, Heath DD (2012) Historical and anthropogenic factors affecting the population genetic structure of ontario’s inland lake populations of walleye (Sander vitreus). J Hered 103:831–841. doi:10.1093/jhered/ess066 http://dx.doi.org/10.1111/j.1365-294X.2006.03126.x http://dx.doi.org/10.1038/35012228 http://dx.doi.org/10.1017/S1464793105006731 http://dx.doi.org/10.1017/S1464793105006731 http://dx.doi.org/10.1016/j.biocon.2009.01.017 http://dx.doi.org/10.1111/mec.12317 http://dx.doi.org/10.1111/mec.12317 http://dx.doi.org/10.1038/35016000 http://dx.doi.org/10.1002/joc.1276 http://dx.doi.org/10.1038/nrg2339 http://dx.doi.org/10.1038/nrg2339 http://dx.doi.org/10.1111/j.1744-7429.2010.00710.x http://dx.doi.org/10.1007/s10980-010-9447-4 http://dx.doi.org/10.1111/j.1461-0248.2008.01179.x http://dx.doi.org/10.1073/pnas.0606793104 http://dx.doi.org/10.1073/pnas.0606793104 http://dx.doi.org/10.1016/j.tplants.2008.10.002 http://dx.doi.org/10.1016/S0169-5347(02)02489-8 http://dx.doi.org/10.1016/S0169-5347(02)02489-8 http://dx.doi.org/10.1111/j.1523-1739.2008.00944.x http://dx.doi.org/10.1111/j.1365-294X.2010.04808.x http://dx.doi.org/10.1073/pnas.1513062113 http://dx.doi.org/10.1073/pnas.1513062113 http://dx.doi.org/10.1038/sj.hdy.6800725 http://dx.doi.org/10.1111/j.1523-1739.2012.01927.x http://dx.doi.org/10.1126/science.aaf4381 http://dx.doi.org/10.1126/science.aaf4381 http://dx.doi.org/10.1371/journal.pone.0128923 http://dx.doi.org/10.1111/j.1365-294X.2012.05591.x http://dx.doi.org/10.1111/jbi.12419 http://dx.doi.org/10.1111/jbi.12419 http://dx.doi.org/10.1046/j.1523-1739.2000.99182.x http://dx.doi.org/10.1111/mec.12152 http://dx.doi.org/10.1111/j.0906-7590.2008.5203.x http://dx.doi.org/10.1073/pnas.0902257106 http://dx.doi.org/10.1126/science.1189138 http://dx.doi.org/10.1111/j.1744-7429.2000.tb00627.x http://dx.doi.org/10.1016/j.biocon.2009.02.021 http://dx.doi.org/10.1016/j.biocon.2010.02.005 http://dx.doi.org/10.1111/j.1600-0587.2008.05742.x http://dx.doi.org/10.1890/10-0709.1 http://dx.doi.org/10.1007/s10592-007-9409-9 http://dx.doi.org/10.1093/jhered/ess066 478 Conserv Genet (2017) 18:467–478 1 3 Wright SJ (2003) The myriad consequences of hunting for vertebrates and plants in tropical forests. Perspect Plant Ecol Evol Syst 6:73–86. doi:10.1078/1433-8319-00043 Young A, Boyle T, Brown T (1996) The population genetic conse- quences of habitat fragmentation for plants. Trends Ecol Evol 11:413–418. doi:10.1016/0169-5347(96)10045-8 Zurbuchen A, Landert L, Klaiber J et  al (2010) Maximum foraging ranges in solitary bees: only few individuals have the capabil- ity to cover long foraging distances. Biol Conserv 143:669–676. doi:10.1016/j.biocon.2009.12.003 http://dx.doi.org/10.1078/1433-8319-00043 http://dx.doi.org/10.1016/0169-5347(96)10045-8 http://dx.doi.org/10.1016/j.biocon.2009.12.003 Climatic stability and contemporary human impacts affect the genetic diversity and conservation status of a tropical palm in the Atlantic Forest of Brazil Abstract Introduction Methods Literature search Genetic diversity response variables Predictive models Statistical analyses Results Literature search Climatic suitability and stability models Effects of historical and current factors on genetic diversity Discussion Acknowledgements References