Journal of Plant Ecology Research Article Structure, diversity and the conservation value of tropical dry forests in highly fragmented landscapes Oscar R. Lanuza1,2,3,*, , Fernando Casanoves4, Sergio Vílchez-Mendoza4, Josep Maria Espelta2, Downloaded from https://academic.oup.com/jpe/article/16/1/rtac046/6545844 by guest on 25 January 2023 Josep Peñuelas2,3 and Guille Peguero1,2,3,5 1 Facultad Regional Multidisciplinaria Estelí, Universidad Nacional Autónoma de Nicaragua (UNAN Managua/FAREM Estelí), 49, Estelí, Nicaragua, 2CREAF, E08193 Bellaterra, Cerdanyola del Vallès, Spain, 3Global Ecology Unit, CSIC, CREAF‐CSIC‐UAB, 08193 Bellaterra, Catalonia, Spain, 4CATIE-Centro Agronómico Tropical de Investigación y Enseñanza, 30501, Turrialba, Costa Rica, Spain, 5Departament de Biologia Animal, Biologia Vegetal i Ecologia, Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain *Corresponding author. E-mail: olanuza@unan.edu.ni Handling Editor: Michael O'Brien Received: 15 November 2021, First Decision: 15 December 2021, Accepted: 8 January 2022, Online Publication: 9 March 2022 Abstract Although tropical dry forests are among the most degraded and fragmented biomes in the world, we still have a poor understanding of their basic ecological features and conservation status, particularly in the Neotropics. Here, we assess the diversity, composition, structure and conservation value of tropical dry forests in a highly fragmented landscape in Nicaragua. We established 31 plots and transects in and along river corridors, secondary forests, living fences and pasture-woodlands. We recorded all trees with diameters at breast height ≥ 2.5 cm using Hill numbers (qD, where q = 0, 1 or 2) and estimated the richness and diversity of trees in each forest type. We calculated the Importance Value Index (IVI) to species and family levels and, finally, performed a Non-Metric Multidimensional Scaling (NMDS) ordination and an Analysis of Similarities (ANOSIM) using the Bray–Curtis index of similarity. Diversity (1D, 2D) but not species richness (0D) differed between forest types (P = 0.01 and 0.66, respectively). IVI was highest for the legume family Fabaceae, followed by the Moraceae and Malvaceae (27.8, 11.1 and 10.5, respectively). Vachellia pennatula, Guazuma ulmifolia and Bursera simaruba had IVIs >30%, the former two being the most abundant trees in all forest types. An analysis of community similarity revealed that each forest type had a distinct composition (P < 0.01, R2 = 0.30), thereby underlining the importance of conserving all these different types of land cover. Keywords Fabaceae, fragmentation, Hill numbers, matrix composition, Importance Value Index, river corridor, tropical dry forests 破碎化景观格局中热带干旱森林的结构、多样性与保护价值 摘要：尽管热带干旱森林是世界上退化和破碎化程度最严重的生物群落之一，但我们对它们的基本生 态特征和保护状况仍然知之甚少，尤其是在新热带地区。在本研究中，我们对尼加拉瓜破碎化景观格 局中的热带干旱森林的多样性、组成、结构和保护价值进行了评估。我们在河流廊道、次生林、绿 篱和林间牧地之中和沿线地区建立了31处样地和样带。我们使用希尔数(Hill number)对胸高处直径 ≥2.5 cm的所有树木进行了记录，并对每种森林类型中树木的丰富性和多样性进行了估算，接着计算 © The Author(s) 2022. Published by Oxford University Press on behalf of the Institute of Botany, Chinese Academy of Sciences and the Botanical Society of China. This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (https://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com JOURNAL OF PLANT ECOLOGY | https://doi.org/10.1093/jpe/rtac046 1 了种级分类和科级分类上的重要值指数(Importance Value Index, IVI)，最后则使用Bray-Curtis相似性指 数进行了非度量多维标度(Non-Metric Multidimensional Scaling, NMDS)排序和相似性分析(Analysis of Similarities, ANOSIM)。不同森林类型之间在多样性上存在差异，但物种丰富度却不存在差异。豆科植物 的IVI值最高，其次是桑科植物和锦葵科植物。Vachellia pennatula、Guazuma ulmifolia和Bursera simaruba的IVI值都>30%，其中前两种是所有森林类型中最为丰富的树木。对群落的相似性进行分 析后可知每种森林类型都有着不同的组成，从而凸显了对所有这些不同类型的土地覆盖开展保护 的重要性。 关键词：豆科植物，破碎化，希尔数(Hill number)，重要值指数，河流廊道，热带干旱森林 Downloaded from https://academic.oup.com/jpe/article/16/1/rtac046/6545844 by guest on 25 January 2023 INTRODUCTION the second highest rate of deforestation in the Central American region (Portillo-Quintero and Smith 2018). Despite the extremely rich diversity of organisms and Most TDFs are highly fragmented due, above all, high levels of endemism they harbor, tropical dry to the agricultural practices and economic activities forests (TDFs) are among the most threatened and that can easily sidestep the low protection level these degraded of all biomes on Earth (Banda et al. 2016). forests enjoy (Miles et al. 2006; Portillo-Quintero Moreover, worldwide around 97% of the remaining and Sánchez-Azofeifa 2010). Cattle grazing, invasive areas of TDFs are severely endangered due to a species and wildfires caused by humans are among variety of anthropogenic pressures (Miles et al. the main drivers of forest degradation (Dimson 2006). These forest ecosystems are characterized and Gillespie 2020). Domestic cattle in TDFs in the by deciduous plant species that are well adapted to Neotropics, however, are an efficient agent for the the extremely uneven distribution of the otherwise dispersal of many TDF tree species preadapted to limited precipitation—typically <1800 mm a year— endozoochory by large mammals, and fire is known and to the mean annual temperature of >25 °C (Gentry to trigger the germination of physically dormant 1995; Gillespie et al. 2000; Murphy and Lugo 1986; seeds (Janzen and Martin 1982; Peguero and Espelta Sánchez-Azofeifa et al. 2005). About 66.7% of the 1 2014). The combination of seed dispersal by cattle million km2 of TDF estimated still to exist lie within and frequent pasture fires is a powerful driver the tropical belt of the American continents (Miles that favors a specialized subset of TDF tree species et al. 2006). These forests provide many ecosystem (Peguero and Espelta 2014). Nevertheless, cattle services including carbon storage, the conservation grazing in early successional patches of TDFs seriously of nutrient cycles, and water storage and supply reduces the ability of new seedlings and saplings (Portillo-Quintero et al. 2015) that are indispensable to become established, thereby further degrading for human society (Balvanera et al. 2011; Becknell these already fragile ecosystems (Hester et al. 2006; et al. 2012; Calvo et al. 2017; Maass et al. 2005; Portillo- Quisehuatl‐Medina et al. 2020). Quintero et al. 2015). However, only 3.9% of the Landscape fragmentation modifies alpha and beta TDF remnants in the Americas are legally protected diversities since it leads to the establishment of new (Portillo-Quintero and Sánchez-Azofeifa 2010). environmental filters and assemblages, and the more Globally, land-use changes have already affected than likely disappearance of other assemblages from 48.5% of the world’s TDFs (Hoekstra et al. 2005) and, the landscape matrix (Halffter and Moreno 2005). for instance, 66% and 77% of their original surface Although habitat loss provokes important changes area in North and South America, and Nicaragua, in species richness and community structure (Pardini respectively, have been permanently converted et al. 2010), the composition and configuration of the to agricultural production (Portillo-Quintero and landscape matrix may also be key in the eventual Sánchez-Azofeifa 2010). Stevens et al. (2001) report recovery of many taxa. In a low-quality matrix that, as a consequence of human pressure, just <1% (e.g. roads, annual crops and built-up areas), the of primary TDFs remain in Nicaragua. Likewise, mortality of forest species is very high, whereas in recent reports have estimated that up to 661,307 ha high-quality landscape matrices (e.g., arboreal crops of TDFs were lost during 2001–2013 in Central and and agroforestry systems), the survival of forest North America, including 51,336 ha in Nicaragua, species is high (Arroyo-Rodríguez et al. 2020). Matrix 2 JOURNAL OF PLANT ECOLOGY | 2023, 16:rtac046 composition in farmland explains 30%–40% of the 2021). The study area has a long history of habitat great variation in biodiversity thresholds of mammals fragmentation due to the conversion of forests into and birds (Ochoa-Quintero et al. 2015). Boesing pasture for livestock. This trend can be traced back to et al. (2018) found that once the loss of forest cover the second half of the twentieth century and the rise reaches a critical threshold of around 20%, the fall in of extensive cattle ranching and coffee production avian taxonomic and phylogenetic diversity is abrupt. in the area, which peaked during the first decade of Thus, matrix quality and, particularly, forest cover the twenty-first century (Ravera 2007). A detailed play a major role in determining patterns of species study of land-use change showed that by 2011 only richness not only in plants but also in invertebrates, 37.7% of the area was covered by forest fragments, birds, amphibians and reptiles (Reider et al. 2018). with an additional 26.3% being extremely degraded Despite the great fragmentation of most TDFs or consisting only of early successional secondary and the many pressures they have to withstand, we patches (Ruiz et al. 2013). It should be noted, Downloaded from https://academic.oup.com/jpe/article/16/1/rtac046/6545844 by guest on 25 January 2023 lack any full evaluation of their conservation status however, that the study by Ruiz et al. (2013) was and the quality of the landscape matrix at local level carried out in a protected landscape, so the figures (Arroyo-Rodríguez et al. 2020; Gillespie et al. 2000). for forest loss and fragmentation are even more In particular, we still have little knowledge of the alarming in our study sites that lack any kind of richness and composition patterns of TDF patches in management regulation or legal protection. We used highly fragmented and degraded landscapes. Filling Google Earth satellite images (Gorelick et al. 2017) this gap is crucial if we are to successfully implement to select fragments that were representative of the plans to restore TDFs at a landscape scale. Efforts landscape. Additionally, we used a Landsat-8 satellite should be directed towards enhancing biodiversity image with 30 m resolution taken on 20 August conservation and recovering ecosystem functions 2020 to assess the different land-cover types in the and services including climate regulation, water area. We determined the Normalized Difference supplies, crop pollination and biological agricultural Vegetation Index (NDVI) using bands 4 and 5 with pest control (Arroyo-Rodríguez et al. 2020). Here, we ArcGis V.10 software. Of the 53 652 ha of our study address this issue by studying the structure, diversity area, currently 74.3% are devoted to pasturing and and conservation value of the main TDF types other agricultural purposes (39 871 ha), about 10% remaining in Nicaragua. We selected river corridors (5380 ha) is bare ground including roads and built-up (RC), secondary forests (SF), living fences (LF) and areas, 12.9% (6911 ha) is covered by sparse forest pasture woodlands (PW) because they are the TDF vegetation at different successional stages and only types most commonly found in highly human- 2.8% (1490 ha) is recognizably covered by dense modified landscapes (Arroyo-Rodríguez et al. 2020). forest vegetation (Fig. 1b). Specifically, we evaluated which forest types retain We established sampling plots for pasture structures and diversities that are similar to those of woodlands (PW: pasture grasslands with scattered other better-preserved TDFs and so could contribute trees, N = 8) and secondary forests (SF: closed forest in the future to the spread and restoration of these patches at different successional stages, N = 7) and valuable but endangered ecosystems. transects for living fences (LF: rows of trees used to delimit properties, N = 8) and river corridors MATERIALS AND METHODS (RC: the forest fragments that typically line water courses, both permanent and intermittent, N = 8). Study area and sampling The SF plots, set up following standard procedures This study was carried out in the upper part of the for floristic inventories (Gentry 1982, 1988), water catchment of the river Estelí between the consisted of 1000 m2 divided into ten 2 m × 50 m natural reserves of Tomabú (13°01′43″ N, 86°17′51″ transects, 2 m apart; the RC plots consisted of two W), Quiabúc Las Brisas (13°07′31″ N, 86°26′01″ W) 50 m × 10 m transects (1000 m2), one on either and Tisey La Estanzuela (12°59′30″ N, 86°22′46″ side of the river (Sánchez-Merlos et al. 2005); W) in northern Nicaragua at an altitudinal range of for the LF, we recorded all individuals along a 620–1470 m a.s.l. (Fig. 1a). Over the past 10 years, transect of 2 m × 350 m (750 m2), and for PW, 1-ha the average temperature and precipitation were (10 000 m2) plots were established. We then 23.1 °C and 892 mm year−1, respectively, and the measured and recorded all trees with diameters annual water deficit −385.4 mm year−1 (Gómez et al. at breast height (DBH) ≥2.5 cm in each plot. For JOURNAL OF PLANT ECOLOGY | 2023, 16:rtac046 3 Downloaded from https://academic.oup.com/jpe/article/16/1/rtac046/6545844 by guest on 25 January 2023 Figure 1: (a) Map with the location of the study area. (b) Map of the land-use cover in the study area. Red, blue, green and purple circles show the location of transects along river corridors (RC) and living fences (LF), and of plots in secondary forests (SF) and pasture woodlands (PW), respectively. Land cover types are as follows: dense forest vegetation (dark green), sparse forest vegetation (light green), pasture and agricultural areas (light brown) and bare ground built-up areas (dark brown). 4 JOURNAL OF PLANT ECOLOGY | 2023, 16:rtac046 atypical individuals with buttresses or more than (GLM) with the extrapolated values of the Hill one trunk or on steep slopes, DBH was measured numbers (q = 0, 1, 2) to identify differences in alpha following the recommendations in Aranda et al. diversity between forest types. We standardized (2005) and Camacho (2000). All recorded trees alpha diversity using an extrapolation-interpolation were identified to species and family levels using the of the Hill numbers (Chao et al. 2014) for each Flora de Nicaragua database (Stevens et al. 2001). sampling plot or transect. To extrapolate the effective species richness to a specific number of individuals, Data analysis we used 200 individuals as this was the maximum We compared the diversity of the different TDF types tree density recorded at plot or transect level. The using Hill numbers (Hill 1973; Jost 2006) and the forest type was included as a fixed effect term, the R library iNEXT (Hsieh et al. 2016). We used Hill distribution family was the negative binomial and numbers since they allow us to calculate the effective the link function was a logarithm. The assumption Downloaded from https://academic.oup.com/jpe/article/16/1/rtac046/6545844 by guest on 25 January 2023 number of species—or true diversity (qD)—present of normality was checked using Shapiro–Wilks’ tests in a community (Hill 1973) and are considered to and the models were fitted using the diagnostic be equivalent to diversity indices (Jost 2006). Hill plots of residuals. Differences between sites were numbers are defined for q ≠ 1 as determined using the Fisher LSD post hoc tests (P < S  1/(1−q)  0.05).  q (1) D= q Pi Tree species dominance was assessed by means of i = 1 Abi Di = (2) where D is the effective species richness, S the q E (ha) number of species in the assemblage and the ith ï ò species has a relative abundance Pi. The parameter Di DR (3) i =  q determines the sensitivity of the measure to i=1... n Di the relative abundances (Chao et al. 2014). Hill where Di is the absolute cover, DRi the relative cover numbers of order 0 (0D) correspond to species of species i with respect to the cover, Ab the basal richness without considering abundance and so area of species i, and E the surface (ha). award high values to rare species. Hill numbers The relative frequency was evaluated by of order 1 (1D) take abundance into account and reflect the effective number of common species Pi Fi = (4) in the assemblage, whereas Hill numbers of order NS 2 (2D) give a high value to abundance and thus ï ò Fi identify the effective number of dominant species FRi = n (5) × 100 in the assemblage (Chao et al. 2014; Hsieh et al. i=1 Fi 2016). where Fi is the absolute frequency, FRi the relative We performed individual-based rarefaction and frequency of species i with respect to the sum extrapolation curves of tree diversity based on Hill of the frequencies, Pi the number of sites where numbers (q = 0, 1, 2) (Chao et al. 2014) to compare species i is present and NS the total number of gamma diversity: species richness (q = 0), effective sampling sites. number of common species (q = 1) and effective We calculated the relative density, frequency number of dominant species (q = 2). This method and dominance of each species to determine their allows for the effective species richness to be contribution to the structure of each forest type extrapolated to a certain number of individuals; in and their Importance Value Index at the species and this study, the rarefaction/extrapolation curves were family levels (IVIsp and IVIf, respectively) (Graciano- estimated as the mean of 100 replicate bootstrapping Ávila et al. 2018; Linares and Fandiño 2009; Nebel runs with 95% confidence intervals and were et al. 2001; Vargas and Hidalgo-Mora 2013). The standardized to 800 individuals given that this was Importance Value Index (IVI) for family and species the maximum number of trees recorded in any one was defined as of the forest types. Whenever the 95% confidence intervals did not overlap, species numbers differed ARi + DRi + FRi (6) IVI = significantly at P < 0.05 (Colwell et al. 2012). 3 Additionally, we built generalized linear models The absolute and relative abundance were defined by JOURNAL OF PLANT ECOLOGY | 2023, 16:rtac046 5 RESULTS Ni Ai = (7) We recorded data from 2,489 trees belonging to 84 E species and 32 families at 31 sampling sites (Table 1). ï ò Ai The commonest RC species were Inga vera Kunth., ARi = n (8) × 100 i=1 Ai Ficus insipida Willd. and Vachellia collinsii (Saff.) Seigler & Ebinger; the commonest SF species were where Ai is the absolute abundance, ARi the Vachellia pennatula (Schltdl. & Cham.) Seigler & relative abundance of species i with respect to total Ebinger, Guazuma ulmifolia Lam. and Tecoma stans abundance, Ni the number of individuals of species i (L.) Juss. ex Kunth; the commonest LF species were and E the sampling area (ha). Bursera simaruba L, V. pennatula and T. stans; and, We also built rank-abundance curves (RAC) to finally, the commonest PW species were V. pennatula, Downloaded from https://academic.oup.com/jpe/article/16/1/rtac046/6545844 by guest on 25 January 2023 determine the patterns of abundance distribution V. collinsii and Lysiloma auritum (Schltdl.) Benth. for each type of forest community (Whittaker 1965). (Supplementary Fig. S1). Species richness did not RACs and model fitting were done with R package differ significantly between forest types (Table 1). vegan 2.5-6 (Oksanen et al. 2019). We selected the The rarefaction curves indicated a complete overlap, best adjusted rank-abundance model (i.e., Null, thereby indicating that the effective species richness Preemption, Lognormal, Zipf or Mandelbrot) based did not differ substantially between forest types (Fig. on their Akaike Information Criterion (AIC). Finally, 2). Despite the lack of difference in the total effective we compared differences in composition between number of species, RC and SF were more diverse forest types using an analysis of similarity (ANOSIM) (Table 1; Fig. 2). The corresponding linear models and applied non-metric multidimensional scaling confirm that RC and SF were the most diverse forest (NMDS) to visualize the data. We used standardized types, with the diversity of RC being significantly values of tree density per area and the Bray– higher than the diversities of both PW and LF (Table Curtis index of similarity for both procedures. The 1; Supplementary Table S2 for complete model ANOSIM was conducted with 1000 permutations specifications and outputs). using non-standardized data, and P values were adjusted using the SidaKSS method (Benjamini and The legume family (Fabaceae) was represented Hochberg 1995) with the library vegan (Community by 20 species and 851 individuals, 34.3% of all trees Ecology Package) (Oksanen et al. 2013). The NMDS in the inventory, and was the most diverse family was conducted using a maximum of 50 iterations in all forest types. The legumes were followed by and a convergence tolerance threshold of 0.00001, Malvaceae (six species, 288 individuals; 11.6%), which was reached at 10 iterations with the libraries Burseraceae (two species, 288 individuals; 11.5%), MASS and the isoMDS functions (Venables and Moraceae (five species, 132 individuals; 5.3%) and Ripley 2002). Meliaceae (eight species,105 individuals; 4.2%; Fig. All analyses were performed using InfoStat v2019 3; Supplementary Table S1). The IVIf consequently (Di Rienzo et al. 2019) and Qeco (Di Rienzo et al. indicated that the Fabaceae (27.8%) was the most 2010) with R v3.5.1 (R Core Team 2019). important family in all forest types, followed by Table 1: Characteristics of the types of fragments of tropical dry forestsin Nicaragua River corridor Secondary forest Living fence Pasture woodland P Individuals sampled 872 624 531 462 – Number of species 51 51 46 40 – Tree density (no./ha) 108 ± 12 b 78 ± 12 ab 66 ± 12 a 58 ± 12 a 0.0278 Species richness (0D) 23.6 ± 3.3 a 18.7 ± 2.9 a 18.7 ± 2.7 a 18.0 ± 2.6 a 0.5262 Shannon diversity (1D) 16.4 ± 1.8 a 12.8 ± 1.7 ab 10.0 ± 1.3 b 9.2 ± 1.2 b 0.0104 Simpson diversity (2D) 12.8 ± 1.3 a 9.9 ± 1.2 ab 6.8 ± 0.9 bc 6.3 ± 0.9 c 0.0010 All values are means ± 1 standard error. Species richness and diversity indices per plot are estimated using Hill numbers. Different letters within a row denote significant (P < 0.05) differences between forest types in a Fisher LSD post hoc test. 6 JOURNAL OF PLANT ECOLOGY | 2023, 16:rtac046 Downloaded from https://academic.oup.com/jpe/article/16/1/rtac046/6545844 by guest on 25 January 2023 Figure 2: Individual-based rarefaction (solid line) and extrapolation curves (dashed lines) of gamma diversity based on Hill numbers (q = 0, 1 or 2) for the four tropical dry forest types in the study area. Moraceae (11.1%), Malvaceae (10.5%), Burseraceae 6.94%), Cordia alliodora (Ruiz & Pav.) Oken (121 (9.1%) and Meliaceae (5.9%), with IVIf <5% for the individuals, 4.88%), I. vera (110 individuals, 4.47%), other 27 families (Fig. 4a). L. auritum (102 individuals, 4.1%) and F. insipida (85 Of the 84 species found in the study area, a individuals, 3.4%). These 10 species represented group of 15 act as the most significant contributors 64.82% of all sampled individuals (Supplementary to the structure of the sampled sites: V. pennatula Table S1). (Fabaceae), G. ulmifolia (Malvaceae) and B. simaruba The number of individuals in the DBH (Burseraceae) had IVIsp >30%, whereas the other classes differed between forest types (F = 2.32, 12 species had IVIsp <25%, which demonstrates the P < 0.01, Fig. 5). Trees with DBHs of 10–20 cm were relative dominance of just a few species in these the most abundant (1,108 individuals, 44.7%), highly modified forests (Fig. 4b). This dominance was followed by trees with DBHs <10 cm (522 individuals, particularly clear when assessing species composition 21.1%) and ≥20 cm (508 individuals, 20.5%). These using rank-abundance curves, in which V. pennatula, three DBH classes represented 86.3% of all trees. RC V. collinsii and T. stan were dominant in all forest types had the most trees in all classes, usually followed (Supplementary Fig. S1). by SF. The most distinctive and dominant species of A preemption model best accounted for the each forest type were also the largest (Supplementary distribution of RC abundance, with I. vera and Table S1). F. insipida being the most dominant species. SFs and Finally, the analysis of similarity showed PWs were better fitted to a Zipf-Mandelbrot model, that the studied TDF differed in terms of species with V. pennatula as the most abundant species in both composition. This implies that each land cover forest types, whereas LF was best explained by a log- type was dominated by a distinct set of species normal function, with B. simaruba and V. pennatula (P < 0.01, R2 = 0.30). The post hoc tests indicated being the most dominant species. Overall, the most that RC species diverged significantly from the abundant species were V. pennatula (307 individuals, compositions of SF, LF and PW. The first two axes 12.38%), B. simaruba (259 individuals, 10.45%), of the NMDS represented 77% of the compositional G. ulmifolia (193 individuals, 7.78%), V. collinsii variation; NMDS2 also separated the TDF types (183 individuals, 7.38%), T. stans (172 individuals, well (Fig. 6). The RC plots were the most similar JOURNAL OF PLANT ECOLOGY | 2023, 16:rtac046 7 Downloaded from https://academic.oup.com/jpe/article/16/1/rtac046/6545844 by guest on 25 January 2023 Figure 3: Dominance models based on the number of tree species (spp.) and individuals (n) for each plant family and in each of the four tropical dry forest types in the study area. to each other, indicating a consistently distinctive fragments assessed still conserve a large and valuable tree species composition in this forest type. The degree of diversity (Gentry 1995; Gillespie et al. distribution of SF and PW plots across the NMDS 2000; Tarrasón et al. 2010). The number of species space notably overlapped, thereby indicating a high per plot was lower than reported in other well- level of compositional similarity between these two preserved TDFs with some kind of legal protection forest types. The great spread of the points of SF, (e.g. Santa Rosa = 75 and Palo Verde = 65 in Costa PW and also of LF across the first axis of the NMDS Rica; Chacocente = 54 and La Flor = 59 in Nicaragua, revealed, however, a high level of between-plot Gillespie et al. 2000). Our results were, however, dissimilarity in these forest types. consistent with the findings of similar studies carried out in fragmented landscapes at different successional stages with no legal protection like our study area DISCUSSION (Hernández-Ramírez and García-Méndez 2015). Our comprehensive floristic inventory found that the The IVI was highest for the legume family, which study area was representative of TDF ecosystems in is a clear reflection of the dominance of this family the Neotropics and that the different types of forest in many Neotropical dry forests at all conservation 8 JOURNAL OF PLANT ECOLOGY | 2023, 16:rtac046 Downloaded from https://academic.oup.com/jpe/article/16/1/rtac046/6545844 by guest on 25 January 2023 Figure 4: Importance Value Index (IVI) at family (a) and species (b) levels for the four tropical dry forest types in the study area. use of water thanks to their typically bipinnate leaves and the presence of defensive structures (Ceroni Stuva 2003; Gei et al. 2018). Other important plant families in our study including the Moraceae, Malvaceae, Burseraceae, Meliaceae, Euphorbiaceae and Anacardiaceae in similar areas throughout the Mesoamerican and Caribbean regions also have high reported IVI values (Ballesteros-Correa et al. 2019; Gillespie et al. 2000; Hernández-Ramírez and García- Méndez 2015; Hilje et al. 2015; Linares and Fandiño 2009). The 15 species with the highest IVI values at species level in our study area are also among the commonest tree species in Neotropical TDFs (Gillespie et al. 2000; Hernández-Ramírez and García- Méndez 2015; Sabogal 1992; Sánchez-Merlos et al. 2005; Vargas and Hidalgo-Mora 2013). However, the dominance of three species (V. pennatula, B. simaruba Figure 5: Distribution of tree abundances by diameter at breast height (cm) for the four tropical dry forest types in and G. ulmifolia), which represent nearly 30% of the study area. all individuals, indicates that the habitats in our study area are greatly altered, and that recurrent levels (Ballesteros-Correa et al. 2019; Gillespie et al. disturbances maintain these TDF remnants in early 2000; Herazo Vitola et al. 2017; Hernández-Ramírez succession stages (Ballesteros-Correa et al. 2019; Hilje and García-Méndez 2015; Mendoza 1999; Vargas et al. 2015; Sabogal 1992; Sánchez-Merlos et al. 2005; and Hidalgo-Mora 2013). Legumes are also the most Stevens et al. 2001; Vargas and Hidalgo-Mora 2013). dominant tree species (15%–21% of all individuals) Secondary forests (SF) and, particularly, river in TDFs during early and intermediate successional corridors (RC) had the highest diversity and species stages (Hilje et al. 2015). The success of this plant family richness, with 51 species in each and both within is based on several ecophysiological adaptations such the range of species richness previously reported for as the usual symbiotic N fixation, their conservative these types of TDF remnants (Sánchez-Merlos et al. JOURNAL OF PLANT ECOLOGY | 2023, 16:rtac046 9 Downloaded from https://academic.oup.com/jpe/article/16/1/rtac046/6545844 by guest on 25 January 2023 Figure 6: Non-metric multidimensional scaling (NMDS) ordination analysis of the differences in composition between each of the four tropical dry forest types in the study area. Only the most representative tree species of each forest type, i.e. those with correlation coefficients >0.20 on each axis, are shown. See Supplementary Table S1 for species abbreviations. 2005). The Niche-Preemption model selected in RC for SF and PW (Supplementary Fig. S2) demonstrates (Supplementary Fig. S2) shows that a few abundant that the presence of a species in such forest types species occupy a large proportion of the niche is dependent on previous physical conditions. hyperspace, whereas the remaining species—the Although pioneer species typically require few prior vast majority—are generally rare, thereby suggesting conditions, the needs of late-successional species are a lack of evenness in the distribution of resources usually greater due to the organization, time and compared to the other rank-abundance models energy accumulated in the ecosystem before their (Whittaker 1965). The most dominant species in RC establishment (Wilson 1991). Therefore, establishment (I. vera and F. insipida) are specifically adapted to this costs are usually lower for pioneer species (V. pennatula forest type, in which water availability is typically and V. collinsii), which allows them to be dominant greater than in the other assessed habitats (Stevens in early stages of succession. On the other hand, the et al. 2001). Thus, this dominance could also be dominance of V. pennatula and G. ulmifolia at these sites related to their hydrochoric dispersal syndrome and is associated with their edible fruits with many hard- the fact that they are not valued as timber. On other coated seeds that are well adapted to endozoochoric hand, V. collinsii is a common pioneer species in the dispersal by cattle (Janzen and Martin 1982; Peguero early stages of secondary succession in Neotropical and Espelta 2014). Germination in these species can dry forests (Hurka and Heinrich 2004), although its increase by up to 15% after the frequent pasture fires, presence in the RC could also be the result of seeds which enhances their colonization capacity in these having been dispersed by rain or birds. anthropized landscapes (Esquivel et al. 2008; Peguero The most abundant species in SF were also and Espelta 2011; Peguero and Espelta 2014). dominant in pasture woodlands (PW), which reveals Unlike RC, species richness was lowest in living a strong connection between these two types of forest fences (LF), where abundances are best explained by communities. The Zipf-Mandelbrot model selected a log-normal function; in this forest type, B. simaruba, 10 JOURNAL OF PLANT ECOLOGY | 2023, 16:rtac046 V. pennatula and T. stan are the most abundant species. priority in anthropogenic environments (Bennett This pattern was probably due to artificial selection et al. 2014). Therefore, improving the matrix quality by landowners, who favor these species since they by designing biodiversity-friendly landscapes with perform well during vegetative propagation (Zahawi greater coverage of riparian and secondary forests 2005). Nevertheless, B. simaruba is also an important (Arroyo-Rodríguez et al. 2020; Reider et al. 2018) resource for many bird species that eat its fruit and will provide species with additional resources and disperse its seeds (Graham et al. 2002). The additional refuges, and facilitate their dispersal and survival, habitat and increase in spatial connectivity provided thereby helping us to reduce the impact of the loss of by LF often play a key role in biodiversity in tropical native forests (Boesing et al. 2018). areas by increasing seed dispersal and fomenting faunal movement, and thus, its composition and CONCLUSIONS structure must be taken into account in fragmented Downloaded from https://academic.oup.com/jpe/article/16/1/rtac046/6545844 by guest on 25 January 2023 landscapes (Arroyo-Rodríguez et al. 2020; Harvey The SF fragments and RC were the most diverse et al. 2004; Medina et al. 2004). forest types at all the evaluated sites. Overall, the The distribution of the DBH classes, with a largest share of the local pool of species was found paucity of trees in the highest diameter classes, in RC. These habitats have key ecological roles suggested that all forest types were in early stages to play, even if their contributions to landscape- of succession, irrespective of their diversity and scale conservation will surely change as human composition (Sabogal 1992). This bias toward activity increases. The protection and restoration younger cohorts is usually associated with greater of these habitats—especially riparian vegetation— resilience in forest patches before repeated are therefore priorities in these anthropogenic disturbances (Lamprecht 1990; Oliver and Larson ecosystems as a means of increasing the structural 1990; Vargas and Hidalgo-Mora 2013). Although and functional connectivity between surrounding individuals with DBH <2.5 cm were not included in protected areas (Arroyo-Rodríguez et al. 2020; Reider our forest inventory, visual inspections revealed that et al. 2018). Finally, these forest ecosystems need to our plots had low rates of sapling recruitment. This be analyzed in greater detail, so they can be used as was likely due to the herbivorous pressure exerted by reference ecosystems in future restoration programs cattle, which typically move around freely in forest at landscape scale and to advance toward more patches. The lack of any control of cattle grazing mature successional states. This will enrich them and trampling are sources of forest degradation with late-successional and/or endangered species that jeopardize any possible ecological restoration and contribute to modifying the distribution of in this threatened tropical ecosystem (Hester et al. diametric classes by promoting the largest individual 2006; Quisehuatl‐Medina et al. 2020). Our results, tree sizes. nevertheless, indicate that both SF and RC conserve a high level of tree diversity, and that RC also had an especially distinct community composition. Supplementary Material Supplementary material is available at Journal of RCs connect aquatic and terrestrial systems and Plant Ecology online. play an essential ecological role in protecting water Table S1: Number of trees with a DBH ≥2.5 cm by resources, biodiversity and carbon stocks due to their forest remnant type, and proportions (%) relative to high net primary productivity (Bennett et al. 2014; the total number of trees recorded. Giese et al. 2003; Tibbets and Molles 2005). These RCs Table S2: Results of the GLMs to identify differences are of great conservation value as they can increase in alpha diversity among forest types using Hill connectivity in treeless matrices (Arroyo-Rodríguez numbers (q = 0, 1, 2). et al. 2020). However, the impact of human activities Figure S1: Rank-abundance curves for the four on land adjacent to riparian forests has seriously types of TDF remnants: (A) river corridor, (B) affected the ecological conditions and properties of secondary forest, (C) living fence and (D) pasture such habitats (Zermeño‐Hernández et al. 2020). The woodland. contribution of riparian forests to landscape-scale Figure S2: Rank-abundance curves of each habitat conservation will probably change as environments type adjusted to 1 of 5 model (Null, Preemption, become increasingly modified, so the protection and Lognormal, Zipf, and Mandelbrot) using the Akaike restoration of these communities should be a high Information Criterion (AIC). JOURNAL OF PLANT ECOLOGY | 2023, 16:rtac046 11 Acknowledgements and estimation in species diversity studies. Ecol Monogr We would like to thank Delver Francisco Pérez, 84:45–67. Hanier Roberto Herrera and Jorge Arlington Palacios Colwell RK, Chao A, Gotelli NJ, et al. (2012) Models and estimators linking individual-based and sample-based for their invaluable assistance with the fieldwork, as rarefaction, extrapolation and comparison of assemblages. well as the landowners who collaborated with the J Plant Ecol 5:3–21. setting up and running of this project. We thank Di Rienzo J, Casanoves F, Balzarini M, et al. (2019) Grupo infostaf Mike Lockwood for editing and revising the English FCA. Argentina: Universidad Nacional de Córdoba. version of the manuscript. Di Rienzo J, Casanoves F, Pla L, et al. (2010) Qeco-quantitative ecology software: a collaborative. Rev Lat Cons 1:73–75. Conflict of interest statement. The authors declare Dimson M, Gillespie TW (2020) Trends in active restoration of that they have no conflict of interest. tropical dry forest: methods, metrics, and outcomes. For Ecol Manage 467:118150. Downloaded from https://academic.oup.com/jpe/article/16/1/rtac046/6545844 by guest on 25 January 2023 Esquivel MJ, Harvey CA, Finegan B, et al. (2008) Effects of REFERENCES pasture management on the natural regeneration of Aranda UD, Dorado FC, Anta MB, et al. (2005) Prácticas de neotropical trees. J Appl Ecol 45:371–380. dasometría, España. Gei M, Rozendaal DMA, Poorter L, et al. (2018) Legume Arroyo-Rodríguez V, Fahrig L, Tabarelli M, et al. (2020) abundance along successional and rainfall gradients in Designing optimal human-modified landscapes for forest neotropical forests. Nat Ecol Evol 2:1104–1111. biodiversity conservation. Ecol Lett 23:1404–1420. Gentry A (1982) Patterns of Neotropical Plant Species Diversity Ballesteros-Correa J, Morelo-García L, Pérez-Torres J (2019) Evolutionary Biology. New York: Springer, 1–84. Composición y estructura vegetal de fragmentos de bosque Gentry AH (1988) Changes in plant community diversity and seco tropical en paisajes de ganadería extensiva bajo manejo floristic composition on environmental and geographical silvopastoril y convencional en córdoba, colombia. Caldasia gradients. Ann Mo Bot Gard 75:1–34. 41:224–234. Gentry A (1995) Diversity and floristic composition of Balvanera P, Castillo A, Martínez-Harms MJ (2011) Ecosystem neotropical dry forests. In Medina E, Mooney HA, services in seasonally dry tropical forests. In Dirzo R, Bullock SH (eds). Seasonally Dry Tropical Forests, Cambridge: Young HS, Mooney HA, et al. (eds). Seasonally Dry Tropical Cambridge University Press, 146–94. Forests: Ecology and conservation, London, UK: Island Press, Giese LA, Aust W, Kolka RK, et al. (2003) Biomass and 259–77. carbon pools of disturbed riparian forests. For Ecol Manage Banda K, Delgado-Salinas A, Dexter KG, et al. (2016) Plant 180:493–508. diversity patterns in neotropical dry forests and their Gillespie TW, Grijalva A, Farris CN (2000) Diversity, conservation implications. Science 353:1383–1387. composition, and structure of tropical dry forests in Central Becknell JM, Kucek LK, Powers JS (2012) Aboveground America. Plant Ecol 147:37–47. biomass in mature and secondary seasonally dry tropical Gómez VLR, Monserrat RS, Lanuza ORL, et al. (2021) forests: a literature review and global synthesis. For Ecol Evolución de la temperatura y precipitación en cuatro Manage 276:88–95. estaciones meteorológicas, ubicadas en la región norcentral Benjamini Y, Hochberg Y (1995) Controlling the false discovery de nicaragua, centroamérica. Revista Científica de FAREM- rate: a practical and powerful approach to multiple testing. Estelí 38:197–212. J R Stat Soc Series B (Methodol) 57:289–300. Gorelick N, Hancher M, Dixon M, et al. (2017) Google earth Bennett AF, Nimmo DG, Radford JQ (2014) Riparian engine: planetary-scale geospatial analysis for everyone. vegetation has disproportionate benefits for landscape‐scale Remote Sens Environ 202:18–27. conservation of woodland birds in highly modified Graciano-Ávila G, Alanís-Rodríguez E, Aguirre-Calderón O, environments. J Appl Ecol 51:514–523. et al. (2018) Estructura y diversidad postincendio en un área Boesing AL, Nichols E, Metzger JP (2018) Biodiversity del matorral espinoso tamaulipeco. Polibotánica 45:89–100. extinction thresholds are modulated by matrix type. Graham C, Martínez‐Leyva JE, Cruz‐Paredes L (2002) Use Ecography 41:1520–1533. of fruiting trees by birds in continuous forest and riparian Calvo S, Sanchez AG, Duran SM, et al. (2017) Assessing forest remnants in los tuxtlas, veracruz, mexico 1. Biotropica ecosystem services in neotropical dry forests: a systematic 34:589–597. review. Environ Conserv 44:34–43. Halffter G, Moreno C (2005) Significado biológico de las Camacho M (2000) Parcelas permanentes de muestreo en bosque diversidades alfa, beta y gamma. In Halffter G, Soberón J, natural tropical; guía para el establecimiento y medición, Koleff P, et al. (eds). Sobre Diversidad Biológica: El Significado Turrialba, C.R: CATIE. de las Diversidades alfa, beta y gamma, Zaragoza, España: Ceroni Stuva A (2003) Distribución de las leguminosas de la GORFI, 5–18. parte alta de la cuenca la gallega. Morropón. Piura. Ecol apl Harvey C, Tucker N, Estrada A (2004) Live fences, isolated 2:9–13. trees and windbreaks: Tools for conserving biodiversity in Chao A, Gotelli NJ, Hsieh T, et al. (2014) Rarefaction and fragmented tropical landscapes? In Schroth G, Izac A-MN, extrapolation with hill numbers: a framework for sampling Vasconcelos HL, et al. (eds). Agroforestry and Biodiversity 12 JOURNAL OF PLANT ECOLOGY | 2023, 16:rtac046 Conservation in Tropical Landscapes. Washington, DC: Island Ochoa-Quintero JM, Gardner TA, Rosa I, et al. (2015) Press, 261–89. Thresholds of species loss in amazonian deforestation Herazo Vitola F, Mercado Gómez J, Mendoza Cifuentes H frontier landscapes. Conserv Biol 29:440–451. (2017) Estructura y composición florística del bosque seco Oksanen J, Blanchet G, Friendly M, et al. (2019) vegan: tropical en los montes de maría (sucre-colombia). Ciencia en Community Ecology Package. R package version 2.5-6. http:// desarrollo 8:71–82. cran.r-project.org/package=vegan. Hernández-Ramírez AM, García-Méndez S (2015) Oksanen J, Blanchet FG, Kindt R, et al. (2013). vegan: Community Diversidad, estructura y regeneración de la selva tropical Ecology Package. R Package version 2.0. http://cran.r-project. estacionalmente seca de la península de yucatán, méxico. org/package=vegan. version 2:1–295. Rev Biol Trop 63:603–616. Oliver C, Larson B (1990). Forest Stand Dynamics. New York: Hester A, Bergman M, Iason G, et al. (2006) Impacts of large McGraw-Hill herbivores on plant community structure and dynamics. In Pardini R, Bueno AA, Gardner TA, et al. (2010) Beyond Danell K, Bergström R, Duncan P, et al. (eds). Large Herbivore the fragmentation threshold hypothesis: regime shifts Downloaded from https://academic.oup.com/jpe/article/16/1/rtac046/6545844 by guest on 25 January 2023 Ecology, Ecosystem Dynamics and Conservation. Cambridge: in biodiversity across fragmented landscapes. PLoS One Cambridge University Press, 97–141. 5:e13666. Hilje B, Calvo-Alvarado J, Jiménez-Rodríguez C, et al. (2015) Peguero G, Espelta JM (2011) Disturbance intensity and Tree species composition, breeding systems, and pollination seasonality affect the resprouting ability of the neotropical and dispersal syndromes in three forest successional stages in dry-forest tree acacia pennatula: do resources stored below- a tropical dry forest in mesoamerica. Trop Conserv Sci 8:76–94. ground matter? J Trop Ecol 27:539–546. Hill M (1973) Diversity and evenness: a unifying notation and Peguero G, Espelta JM (2014) Endozoochory and fire its consequences. Ecology 54:427–432. as germination triggers in neotropical dry forests: an Hoekstra JM, Boucher TM, Ricketts TH, et al. (2005) experimental test. Biotropica 46:83–89. Confronting a biome crisis: global disparities of habitat loss Portillo-Quintero C, Sánchez-Azofeifa GA (2010) Extent and and protection. Ecol Lett 8:23–29. conservation of tropical dry forests in the americas. Biol Hsieh T, Ma K, Chao A (2016) Inext: an r package for Conserv 143:144–155. rarefaction and extrapolation of species diversity (hill Portillo-Quintero C, Sanchez-Azofeifa A, Calvo-Alvarado J, numbers). Methods Ecol Evol 7:1451–1456. et al. (2015) The role of tropical dry forests for biodiversity, Hurka H, Heinrich A (2004) Species richness and composition carbon and water conservation in the neotropics: lessons during sylvigenesis in a tropical dry forest in northwestern learned and opportunities for its sustainable management. Costa Rica. Trop Ecol 45:43–27. Reg Environ Change 15:1039–1049. Janzen DH, Martin PS (1982) Neotropical anachronisms: the Portillo-Quintero C, Smith V (2018) Emerging trends of fruits the gomphotheres ate. Science 215:19–27. tropical dry forests loss in north & central america during Jost L (2006) Entropy and diversity. Oikos 113:363–375. 2001–2013: the role of contextual and underlying drivers. Lamprecht H (1990) Los Ecosistemas Forestales en Los Bosques Appl Geogr 94:58–70. Tropicales y Sus Especies Arbóreas-posibilidades y Métodos Para Quisehuatl‐Medina A, Averett JP, Endress BA, et al. (2020) un Aprovechamiento Sostenido. Eschborn, Germany. Removal of cattle accelerates tropical dry forest succession Linares R, Fandiño M (2009) Estado del bosque seco tropical in Northwestern Mexico. Biotropica 52:457–469. e importancia relativa de su flora leñosa, islas de la vieja Ravera F (2007) Degradación y gestión sostenible de los recursos: providencia y santa catalina, colombia, caribe suroccidental. Actores, conflictos y horizontes de diálogo. Tesis Doctoral. Revista de la Academia Colombiana de Ciencias 33:1–12. Bellaterra, España: Una investigación participativa en un área Maass JM, Balvanera P, Castillo A, et al. (2005) Ecosystem services protegida del trópico seco nicaragüense. of tropical dry forests: Insights from long-term ecological and R Core Team (2019) R: A Language and Environment for Statistical social research on the pacific coast of mexico. Ecol Soc 10:17. Computing. Vienna, Austria: R Foundation for Statistical Medina A, Harvey C, Sánchez D, et al. (2004) Diversidad y Computing. composición de chiropteros en un paisaje fragmentado de Reider IJ, Donnelly MA, Watling JI (2018) The influence bosque seco en rivas, nicaragua. Encuentro 68:24–43. of matrix quality on species richness in remnant forest. Mendoza H (1999) Estructura y riqueza florística del bosque Landscape Ecol 33:1147–1157. seco tropical en la región caribe y el valle del río magdalena, Ruiz V, Savé R, Herrera A (2013) Análisis multitemporal del colombia. Caldasia 21:70–94. cambio de uso del suelo, en el paisaje terrestre protegido Miles L, Newton AC, DeFries RS, et al. (2006) A global miraflor moropotente nicaragua, 1993–2011. Ecosistemas overview of the conservation status of tropical dry forests. J 22:117–123. Biogeogr 33:491–505. Sabogal C (1992) Regeneration of tropical dry forests in Murphy PLA (1986) Ecology of tropical dry forest. Annu Rev Central America, with examples from nicaragua. J Veg Sci Ecol Syst 17:67–88. 3:407–416. Nebel G, Kvist LP, Vanclay JK, et al. (2001) Structure and Sánchez-Azofeifa GA, Quesada M, Rodríguez JP, et al. (2005) floristic composition of flood plain forests in the peruvian Research priorities for neotropical dry forests. Biotropica amazon: I. Overstorey. For Ecol Manage 150:27–57. 37:477–485. JOURNAL OF PLANT ECOLOGY | 2023, 16:rtac046 13 Sánchez-Merlos D, Harvey CA, Grijalva A, et al. (2005) Venables W, Ripley B (2002) Modern Applied Statistics with S. Diversidad, composición y estructura de la vegetación en New York, NY: Springer. un agropaisaje ganadero en matiguás, nicaragua. Rev Biol Whittaker RH (1965) Dominance and diversity in land plant Trop 53:387–414. communities: numerical relations of species express the Stevens WD, Ulloa C, Pool A, et al. (2001) Flora de Nicaragua. importance of competition in community function and St. Louis: Missouri Botanical Garden Press. evolution. Science 147:250–260. Tarrasón D, Urrutia JT, Ravera F, et al. (2010) Conservation Wilson JB (1991) Methods for fitting dominance/diversity status of tropical dry forest remnants in nicaragua: do curves. J Veg Sci 2:35–46. ecological indicators and social perception tally? Biodivers Zahawi R (2005) Establishment and growth of living fence Conserv 19:813–827. species: an overlooked tool for the restoration of degraded Tibbets T, Molles M (2005) C:N:P stoichiometry of dominant areas in the tropics. Restor Ecol 13:92–102. riparian trees and arthropods along the middle Rio Grande. Zermeño‐Hernández I, Benítez‐Malvido J, Suazo‐Ortuño I, Freshwat Biol 50:1882–1894. et al. (2020) Impact of adjacent land use on the ecological Downloaded from https://academic.oup.com/jpe/article/16/1/rtac046/6545844 by guest on 25 January 2023 Vargas G, Hidalgo-Mora JE (2013) Sucesión de un bosque condition of riparian habitats: the relation between tropical seco en la isla san lucas, puntarenas, costa rica. condition and vegetation properties. Appl Veg Sci UNED Research Journal 5:261–269. 23:610–621. 14 JOURNAL OF PLANT ECOLOGY | 2023, 16:rtac046