Chile’s VII Region (Región Del Maule, capital Talca) and the VIII Region (Región Del Biobío, capital Concepción) together cover around 67338 km². Around 2.7 million people inhabit this zone [53] . The area has a warm-temperate, summer dry climate, the mean annual temperature is around 12.2˚C and precipitation is around 1330mm (Köppen-Geiger: Csb, [54] ). Parallel to the Pacific lies the coastal range, followed by the central depression in the inner side of the area. Towards the east, the depression is followed by the Andes. The coastal range is made of metamorphic rock (Paleozoic age) and some intrusive volcanic rock. This basement is covered with quaternary and tertiary material transported from the Andes [55] . The coastal range is mainly covered with deciduous and sclerophyllous forests, many of which are secondary [56] . The central depression is mainly covered with shrublands of Acacia caven [57] . The zone is an area of intensive landscape transformation since the arrival of the Spanish conquerors in the 16^th century. The most intensive change in modern times, however, has taken place under the economic transformation of the Pinochet era. The area has been opened for large-scale forestry by the establishment of exotic tree plantations for economic purposes. Within this area, we assessed plant biodiversity of the following plant formations.

In July of 2011, a field campaign in two regions of Central Chile was carried trough (VII Región del Maule, VIII Región del Biobío). The goal of the field campaign was to assess plant biodiversity in natural and managed ecosystems of Chile’s temperate zone. We assessed the species number, abundance and minimum areal according to the method of Braun-Blanquet [73] . A total of 62 vegetation relevés was assessed (13 native forest patches, seven degraded forest patches, 38 plantations and four shrublands of A. caven). Assessments were assisted by local vegetation experts from the University of Concepción (UdeC). No standard values from literature for the minimum areal were used since transferability to the biodiversity hotspot of Central Chile was not assumed. In order to be able to analyze β-diversity, ten coupled samples (e.g. a plantations relevé in direct vicinity to a native forests relevé) were assessed. Vegetation strata were identified and sampling was carried. According to Braun- Blanquet’s recommendations, we started with an area of 5 × 5 m, collecting and identifying all plant species present. Then, the number of species in the 5 × 5 m plot was determined. The coverage of each species according to Braun-Blanquet’s scale was estimated, usually by two scientist, who then agreed on a final value. Afterwards, the sampling area was sequentially increased (10 × 10 m, 20 × 20 m etc.). The species-area curve was determined and sampling stopped when this curve reached an asymptote. The respective size of the plots for which the asymptote was reached is considered the minimum areal. The abundance-dominance method of Braun-Blanquet has been subject to some review recently [74] . For instance, it is argued that its non-metric scale makes it difficult to analyze especially with multivariate statistics [75] . However, having performed some adaptations to the abundance-dominance data, data can indeed be analyzed statistically and valid results can be achieved [76] . In order to analyze data statistically, we transformed the abundance scores into relative scores according to [77] : A species with a score of e.g. 2 has a coverage between 25% and 50%. It is given a coverage value r_i of (0.25 + 0.5)/2 = 0.375. Other scores were transformed accordingly. Species with a score of + were given r_i = 0.025 and species with r were given r_i = 0.00625. Then, the relative coverage of a species was calculated as for each of the i, j = 1,・・・, S species present in one relevé.

As [28] [78] point out it is crucial to assess biodiversity issues via standardized quantitative figures in order to enable political stakeholders to develop adequate conservation strategies. Since finding commonly acceptable standards for biodiversity is a hard task, it is proposed to use a variety of indicators to evaluate biodiversity conditions [28] [79] . One of the measures we used in order to meet these requirements is the plantation biodiversity benefit score (PBBS) proposed by [80] . The PBBS is a method specifically designed to assess the value of plantations for biodiversity conservation which consists of two components. The site scale components measures predominantly habitat diversity (e.g. presence of paddock trees, dead wood, age structure, species composition, presence of native trees), at specific sites up to 75 points can be yielded. The landscape scale component measures landscape heterogeneity and the integration of a plantation into a landscape (e.g. surrounding with native vegetation, shape of the plantation), up to 25 points can be yielded. The PBBS is the sum of both which therefore can reach a maximum value of 100.

Several biodiversity indices are computed herein which are derived from relative abundance values. [81] proposes a new and consistent concept of defining biodiversity indices. α-diversity ^qD_α indices are derived from the basic sum of relative abundances using some exponent q, that is by computing. From this equation, most biodiversity indices can be derived. The first index used, ⁰D_α, i.e. using q = 0 equals S, the species richness. The second index employed herein ²D_α uses an exponent of q = 2. Note that with this ex-

ponent, the basic sum equals the original Simpson index and equals the

inverse Simpson index. Besides richness ⁰D_α and inverse Simpson diversity ²D_α an evenness index should be employed. In order to avoid biases and redundancy in the analysis, evenness should be independent of richness. [82] shows that this is only possible if evenness = diversity/richness is assumed. Thus, evenness is computed as.

For β-diversity, following [81] the Sørensen index ^qD_β of species turnover between two plots A and B is computed. Again, ^qD_β can be computed on the basis of different values for q. Herein, ⁰D_β with q = 0 is used. Thus, differences of species relative abundances p_i are ignored at the β-scale species presences and absences are considered for β-diversity. ⁰D_β is computed according to:. In order to facilitate inter pretation of the figures, we computed the maximum Sø rensen index max⁰D_{β possible} for each plantation/forest couple. We did so by assuming that all species of the site with the lowers pecies number are found on the other site as well, i.e.. We further more gave the percentage of ⁰D_β, i.e..

We analyzed our data with standard statistical test procedures. Some indices of α-diversity, like e.g. richness ⁰D_α, cannot be assumed to be Gaussian distributed. Instead of having to test for Gaussianness and homoscedasticity for each pair of variables, and employing a different test for each case, a consistent test procedure is employed. The nonparametric Mann-Whitney U test, which does not assume Gaussianness is employed for each pair of variables. In order to clarify the amount of reduction in ⁰D_α, we performed Mann-Whitney U test on the hypothesis that the values ⁰D_α^A of richness ⁰D_α in plant formation A are lower than some percentage of the values ⁰D_α^B in another plant formation B (i.e.). We evaluate the values by testing the assumption several times and by determining the critical value f_c, where H0 cannot be rejected anymore. is assumed to be the statistically significant percentage of reduction. Results were considered as significant on the basis of the p £ 0.05 confidence level. All tests were computed using Matlab R2014a.

The richness data ⁰D_α of plantations is assumed to partly depend on tree height. Young plantations with open canopies impose less harsh competition on understory species than plantations with higher trees and closed canopies. Thus, in younger plantations, higher ⁰D_α are expected than in plantations with higher trees. In very old plantations with the highest trees, more time for ecosystem development has passed, leaving chances for more understorey species to establish. Thus, in plantations with the highest trees, richness ⁰D_α is also expected to be higher. In total, these assumption theoretically motivate using 2^nd degree polynomial functions, which were used exclusively herein. The quality of the regression curved is assessed using the R² measure. Furthermore, a randomized procedure is employed to assess the quality of regressions. Regressions are computed N = 1000 times on random subsets including 75% of the field data of the respective plantations. For each subset, a regression curve is produced. Then, the standard deviation σ of the N = 1000 regressions is computed and plotted in the graphs. All regressions were computed using the statistical package Matlab R2014a.

At first, the characteristics of the four major plant formations (near-natural forests, degraded forests, plantations and shrublands) investigated were outlined. Near-natural forests are found in NPRs and as small remnants within plantations. All three types of near-natural forests which were assessed showed a heterogeneous and diverse aspect. They were composed of three to five strata (trees, shrubs and herbaceous strata, the latter one is missing in riparian forests). For most near-natural forests we sampled the minimum areal size required to reach an asymptote in the species-area curve was 160 ´ 160 m, however, in some forests, up to 640 ´ 640 m had to be assessed. Near-natural forests are composed of numerous herbaceous and woody plant species which grow so densely that accessibility is not assured in some places. Degraded forests also show a heterogeneous and diverse aspect. They consist of two to four vegetation strata and have a guideline minimum areal of 160 ´ 160 m. Under tree species which dissociate allelopathic compounds―e.g. C. alba and P. boldus [83] ―the richness of the herbaceous stratum can be reduced. All sites were more open than the near-natural forests and affected by human presence. The invasive bamboo species C. quila was found at sites were light availability and soil moisture were sufficiently high. We investigated one degraded forest of E. globulus and three of P. radiata. The sites had been established as plantations in the first place but had to remain unmanaged afterwards due to local governmental decisions. Richness within these sites was comparable to the one found under C. alba and P. boldus. However, sites were partly invaded by T. monspessulana or Rubus ulmifolius which both are far less abundant or absent in the near-natural forests. The A. caven shrublands are composed of a single tree (A. caven). Coverage of A. caven differs between them. They are composed of two to three vegetation strata and have a minimum areal of 20 ´ 20 m. The sites were used as pastures with differing intensity. Hence, the herbaceous stratum was affected by grazing. Exotic tree plantations show a very homogeneous pattern, which is not surprising given the fact they are managed as monocultures in Central Chile. An area of 20 ´ 20 m should be assumed as the minimum areal, though many plantations had minimum areas of only 5 ´ 5 m. Below the tree stratum, one to four shrub or herbaceous strata are found. In most cases, their coverage is very low. In the cases where coverage is higher, it is frequently composed of invasive shrubs (T. monspessulana, R. ulmifolius, Rosa rubiginosa and Acacia dealbata).

Boxplots of the richness values ⁰D_α are given in Figure 2(a), Figure 2(d) and Figure 2(g). The average richness ⁰D_α of native forests is 33.3 species. Degraded forests show the second highest richness ⁰D_α with 20.8 species, followed by shrublands with 13.5 species and plantations with 13.1. In the next step, the differences between richness ⁰D_α of different landscape units were tested for significance according to the scheme described in Sec. 2.5. The richness ⁰D_α of near-natural forests was significantly higher than the richness ⁰D_α of plantations (Mann-Whitney U Test p ≈ 8.402⁻⁷). Furthermore, the near-natural forests hosted significantly more species than degraded forests and shrublands (Mann-Whitney U Test p ≈ 0.032/0.0033). Other differences were found to be insignificant. Hence, a significant reduction of richness ⁰D_α below exotic tree plantations could be observed.

Next, we determined the amount of reduction in ⁰D_α.. We evaluated different values of f and found f_c = 0.50 to be the critical value. From there, we concluded that the richness ⁰D_α below plantations was reduced by p_c = 45% in comparison to near-natural forest.

For the three different types of forests, only differences in the richness ⁰D_α between mixed deciduous forests and riparian forests were significant (Mann-Whitney U Test p ≈ 0.004). For the three different types of plantations, only differences in the richness ⁰D_α between P. radiata and E. globulus were significant (Mann-Whitney U Test p ≈ 0.012). Since P. radiate covered the largest areas [67] , the plantation type predominant in Central Chile obviously is far less favourable in terms of biodiversity than less frequent plantation types.

Our field experience from plantations not sampled revealed a clear ranking of richness’s ⁰D_α, which were

highest below E. globulus, followed by P. nigra and P. radiata. For both diversity index ²D_α and evenness ²E_α values, P. radiata plantations yielded lower values than both E. globulus and P. nigra.

In the next step, the relationship between richness ⁰D_α and height of the exotic trees within plantations was investigated. Experience from the field survey revealed that the number of species in the shrub and herbaceous strata is higher below young trees of P. radiata, E. globulus and P. nigra. Once the trees grow higher and close their canopy, richness ⁰D_α in the strata below is reduced. In very old plantations of P. radiata and E. globulus, a slight rise of richness ⁰D_α can be recorded. However, this observation does not account for P. nigra. In order to investigate the relationships statistically, regression curves to the plots species number against tree height were computed in Figure 4. The richness ⁰D_α values of P. radiata and E. globulus were approximated via a 2^nd degree polynomial fit, due to the lower number of relevés, P. nigra was linearly fitted. A convincing fit can be recorded for P. radiate (Figure 4(a), R² ≈ 0.796). Richness ⁰D_α is highest² in P. radiata tree plantations of 2 m and lowest at 21 m. A similar, but far less strong tendency can be observed for E. globules (Figure 4(b), R² ≈ 0.274). For E. globulus plantations richness ⁰D_α is highest at 3 m and lowest at 17 m. The lower determinedness is explained by the large variance in species number at medium tree heights (around 10 m). Unlike P. radiata plantations, E. globulus plantations do differ much more at intermediate ages and heights. The distances between individual

trees differ much more and while some sites are densely covered with pulp E. globulus, others are not. These factors seem to impose variable pressure on other species. Therefore the slight rise of the fitting curve at bigger tree heights should be interpreted cautiously. Due to the fact that P. nigra is far less frequently planted in Central Chile, we only took four relevés. Therefore, a linear fit instead of a 2^nd degree polynomial fit has been computed (Figure 4(c), R² ≈ 0.379). Richness ⁰D_α is highest in plantations of 8 m and lowest at 25 m. This low value is explained by the fact that herbicides had been applied to the site a couple of months before our relevé. Keeping in mind this fact, a similar tendency as for P. radiata and E. globulus was expected and confirmed by our field experience.

In is frequently argued that plantations can contribute to biodiversity since they raise heterogeneity of landscapes, thus helping to creating diverse habitat conditions raising nichés for species [22] [30] [33] [34] [86] . If this assumption holds true, β-diversity between plantations and native species should be high; or β-similarity should be low, respectively. In the VII and VIII region of Central Chile, plantations cover vast areas and native forests are found almost exclusively as remnants within plantations, in riparian habitats along rivers or in national reserves [10] [88] . We chose ten specific sites were plantations actually appear closely linked to native forests. Five sites represented P. radiata plantation and five were E. globulus plantations. Individual figures for the index can be found in Table 1 . As can be seen, most plantation sites had a relatively high β-similarity with the native forests found nearby. The lowest value of % ⁰D_β yielded was 42.9%, the highest value was 100%, stating that all species found in the plantation also appear in the native forest (it should be noted that the opposite is not the case). The % ⁰D_β of all plantations was 64.5%, indicating that most plantations show a high β-similarity with native forest. Frequently, the only species found in the plantation but not in the native forest were the plantation trees and exotic invaders like T. monspessulana, or A. dealbata. The average % ⁰D_β of P. radiata plantations was 73.0% and 56.0% for E. globulus respectively. Thus, it appears that E. globulus contributes more to β-diversity than P. radiata does.

As [78] point out, it is important to not only assess biodiversity on the site scale. A scale jump to the landscape scale is crucial in order to draw valid conclusions on biodiversity impacts. By then, we had analyzed and compared the site specific α-diversity of different natural and managed plant formations and analyzed the β-diversity between them. Therefore, in a next step, we aimed to assess the plant biodiversity on the landscape scale via considerations on γ-diversity. Due to the vast extent of deforestation and plantation establishment in Central

Chile, it might be argued that the landscape scale of γ-diversity is the most important one. In fact, we found 165 plant species in 63 families in total. Among them, 69 species (~42%) appear exclusively in non-commercial ecosystems (near-natural forests, degraded forests and shrublands). Only 9 species (~5%) appear exclusively in plantations. Of these, five are introduced in Chile (Cirsium vulgare, Crysanthemum coronarium, P. nigra, Robinia pseudoacacia, and Stellaria media). Three are native species (Lobelia tupa, Tropaeolum leptophyllum and Vicia magnifolia). The last one, Escallonia revoluta, is endemic to Chile. Among the 165 species, 35 are endemic to Chile (~21%), 56 species have been introduced (~34%). Of the 35 endemics, 34 (~97%) were found in at least one relevé from a natural ecosystems and 14 (~40%) in plantations. In general, many species―endemic or not―are rarely found. On average, species are found in only 7 relevés. Introduced species have been shown to successfully invade natural ecosystems in Chile [89] [90] . Therefore, we aimed to analyze the effects of plantations’ invasive species in the next step by applying the manual of invasive plants of Chilean authorities as data basis [90] . From these manual, we selected 18 species which are expected to be found at this time of year. We calculated the relative abundances p_i of each species mentioned therein at each relevé site. We furthermore calculated the sum of relative abundances of invasive species to show to which extent each site is dominated by (potentially) invasive species. Of course, the plantation trees themselves were not included in this analysis since naturally high abundances would strongly bias results. Within plantations, 3.8 of the 18 invasive species were found on average, whereas in native forests, 3.6 invasive plants were found on average. The number of invasive species encountered within plantations was not significantly different from native forest (t-Test p = 0.7348). Subsequently, we tested the difference of the sum of relative abundances of the 18 invasive species for significance and revealed that is significantly higher in plantations (Mann-Whitney U Test p ≈ 0.009). Furthermore, we tested differences in the relative abundances pi of individual species for significance. For most species, no significant differences could be demonstrated, though T. monspessulana, Plantago lanceolata, A. dealbata and Verbascum thapsus were significantly more abundant in plantations (Mann-Whitney U Test p ≈ 0.0038/0.026/0.050/0.040). Furthermore, some invasive species featured significantly higher abundance values in specific types of plantations. C. vulgare, Eschscholzia californica, Lupinus angustifolius and Taraxacum officinale were significantly more abundant in P. nigra plantations, while P. lanceolata was clearly more abundant in P. radiata plantations (Mann-Whitney U Test p ≈ 0.008/0.032/0.009/0.030/0.001). Since Central Chile is considered an area rich in endemics [9] [91] [92] , an important question is how plantations influence the abundance of endemics. We assessed this influence in the same manner as for invasive species. The information on species origin was derived from references like [93] . The average number of endemic species was 8.30 in native forests and 1.73 in exotic tree plantations. Differences were tested positively for significance. It may be argued that a reduction of relative abundances of endemic trees in plantation is straightforward. Yet, despite the intense discussions on plant biodiversity decline in plantations [10] - [20] , to the knowledge of the authors, the scientific proof for the case of plants in Central Chile is still missing. Moreover, plantations managed in a more sustainable manner may maintain higher relative abundances of endemic trees. Nevertheless, in order to gain information not only on trees, but also on the understory, we analyzed the tree, shrub and herbaceous stratum separately. We computed the relative abundances of each endemic species and tested the sum of relative abundances for significant differences. The sum of relative abundances of endemic species in the tree and shrub strata turned out to be significantly reduced in plantations (Mann-Whitney U Test p ≈ 0.000/0.004. Differences in the herbaceous stratum were insignificant (Mann-Whitney U Test p ≈ 0.018). In the tree stratum, the relative abundances pi of C. alba, Myrceugenia obtusa, Nothofagus alessandrii, N. glauca and N. obliqua were significantly reduced (Mann-Whitney U Test p ≈ 0.004/0.017/0.000/0.020). The relative abundances pi of P. boldus, L. caustica and Quillaja saponaria were lower, yet differences were not significant. These species are frequently found as small shrubs in the understory of plantations. In the shrub stratum, the relative abundance of Pseudopanax valdiviensis was significantly reduced (Mann-Whitney U Test p ≈ 0.018). In the herbaceous stratum, the relative abundances of Hymenoglossum cruentum and Satureja multiflora were significantly reduced (Mann-Whitney U Test p ≈ 0.016/0.017). However, note that plantations tend to be composed of less vegetation strata, the same abundance value r_i will result in a higher relative abundance p_i.

Since the PBBS measures biodiversity benefits through plantations, plantations that are claimed to promote―or at least maintain―biodiversity should yield high PBBS values (attesting good management practices to forest companies). In order to evaluate a possible beneficial influence of Chilean plantations on plant biodiversity, the method was applied to three sites of each tree (note that the PBBS focuses on biodiversity in general, not only plant biodiversity). All plantations observed yield rather low values, cf. Figure 3(b), (a P. radiata plantation yielded the highest value of 37). There were no significant differences between the types of plantation according to the Mann-Whitney U test³. Since the management of the plantations is very similar, they were not expected either. The landscape scale component yielded rather poor values for all plantations observed. This is explained by the fact that plantations are almost completely surrounded by other plantations. The connection to native vegetation endorsed by the PBBS is very weak.