Unexpected Long-Term Forest Experiments: A Case on the Island of Porto Santo, Madeira

Abstract

The aim of this study was to assess the long-term adaptation and growth performance of 50 species introduced in 1991 on the island of Porto Santo, Madeira Archipelago, in order to guide afforestation and soil restoration under the island’s arid conditions, especially in biosphere reserves. The experiment was conducted in Alentejo, Pico Juliana and Matinho, three sites with different types of elevation, soil and exposure. A total of 502 experimental units (five plants each) were established with a completely randomized design in the three sites in 1991 to test the adaptation of 50 species from Mediterranean, African, Australian and American dry climates. Plants were grown in local nursery conditions and planted in rows with 1 × 4 m spacing. Soil properties were analyzed, and survival and growth (height and stem diameter) were monitored in 1991, 1992 and 2025. An analysis of variance was performed for the whole experiment, with the three sites showing significant differences in survival and height among species and sites thirty-four years after the planting. Some species showed high survival and growth, such as Pinus halepensis, Eucalyptus sideroxylon and Casuarina cunninghamiana. Others, like Schinus terebinthifolius and Thevetia neriifolia, showed good adaptation, and invasive behavior at the best sites, but their performance was strongly dependent on site conditions, with Alentejo being the most limiting site. This study demonstrates the long-term value of forest experiments and of long-term monitoring, providing rare data on species adaptation under semi-arid insular conditions. The findings support future afforestation strategies focusing on ecological suitability and invasiveness risk.

1. Introduction

Many experiments on the adaptation of woody species have the objective to select species for the afforestation of areas where natural forests no longer exist. This was the case on the island of Porto Santo, located in the Atlantic Ocean, c. 45 km northeast of Madeira. The island was formed from submarine volcanic activity which started 18 million years ago, and human influence on the vegetation started with the discovery of the island by the Portuguese Gonçalves Zarco and Tristão Vaz in 1418. The early descriptions of the island by Frutuoso (1522–1590) [1] indicated that the island was dry but with some forest cover, the main woody species being the “dragoeiros”, the dragon trees (Dracaena draco L.), the “zimbros” (Juniperus turbinata subsp. canariensis (Guyot & Mathou) Rivas Mart., Wildpret & P.Pérez) and the “urzes” (Erica scoparia subsp. scoparia L.). However, the author also refers to the existence, on the way to the main “Vila” (Vila Baleira), of other woody species such as the “barbuzanos” (Apollonias barbujana (Cav.) Bornm), the “zambujos” (Olea europaea ssp. madeirensis (Lowe) Rivas Mart. & del Arco) and the “marmulanos” (Sideroxylon mirmulans R.Br.). The main objective of this study was the diversification of woody species used for afforestation. However, within the same objective, some of the native species were used in the experiment for possible use in restoration projects.

This initial vegetation was severely impacted by human activities, including the expansion of agriculture, grazing, and use of wood for timber and fuel. The extent of the use of the trees of Porto Santo was well described by the priest António Cordeiro, in 1717, who wrote that the dragon trees of Porto Santo were so large that fishing boats, capable of containing six or seven men, were made out of the trunks, and that the inhabitants fattened their pigs on the fruit [2]. However, as the use of the dragon trees was so intensive, “there was scarcely a dragon tree to be seen in the island” at that time. And in 1825, the British Thomas Edward Bowdich, visiting the island, wrote that “there are not twenty trees of any kind left standing in the island at present, and the inhabitants are obliged to make fires with cow dung when they cannot afford to import fire wood from Madeira” [3]. There is no doubt that the various uses of wood, including for fuel, had a severe impact on the forest cover.

Together with the direct use of trees by humans, there was also an indirect effect, probably just as detrimental to the Porto Santo forests, in particular affecting the regeneration of the trees: the introduction of the rabbit on the island. The impact of the exponential increase in the population of rabbits (Oryctolagus cuniculus L.), introduced to the island in the XV century at the time of the first governor Bartolomeu Perestrelo, was also dramatic for the vegetation on the island [4]. The writings of the Portuguese chronicler Gomes Eanes de Zurara, sent to the King in 1453, and those of the Venetian explorer Alvise Cadamosto, who visited Porto Santo in 1455, are clear in the effects of “an infinity of rabbits” that could not be controlled and “destroy all crops”. The combined effects of humans and rabbits for many centuries were responsible for the degradation of the soil and vegetation of the island [4], where the native woody species have almost disappeared, surviving only on inaccessible cliffs.

The forest cover at the beginning of the XX century was thus very scarce. The soil erosion was widespread. There was an urgent need for afforestation efforts. A first phase started around 1905 with the extraordinary work of the forester Schiappa de Azevedo, covering around 14 hectares at the top of the hill, shown in Figure 1, named Pico Castelo, that was later extended to around 50 hectares in the 1950s. Plantation forestry gained momentum and the afforestation efforts from before this experiment continued (

Table 1. The afforestation efforts in Porto Santo. Data adapted from public literature [4,5] and reports made available by the Madeiran Forest Services (Polícia Florestal, Direcção dos Serviços Florestais).

Years	Location	Area (Hectares)	Species
1905–1920	Pico do Castelo	14	Cupressus macrocarpa, Pinus sp., others
1955–1970	Pico do Castelo	36	Cupressus macrocarpa, Dracaena draco, Pinus halepensis, Pinus radiata, Phoenix canariensis, Quercus ilex, others
	Terra Chã	38
	Dunas da Fonte da Areia	10
	Picos do Facho e Gandaia	15
1975–1990	Picos do Facho e Gandaia	100	Mainly Pinus halepensis, Cupressus macrocarpa
	Morenos	80
	Pico Ana Ferreira	40
	Pedregal de Fora	10
	Pico Juliana	6

). The species used were mostly Pinus halepensis Mill. and Cupressus macrocarpa Hartw. ex Gordon, but other exotic species and some native species, such as Appolonias barbujana, were also included, however in minor quantities. Table 1 shows the years of the afforestation efforts, the locations, the areas planted and the species used.

In 1990, before the experiment, forests covered around 300 hectares of the 4.2 thousand hectares of the island, but diversity was low since Pinus halepensis alone accounted for more than 90% of the trees [6]. The need to increase species diversity of woody plants on the island and to have an experimental basis for the selection of the tree species to use in the afforestation efforts was clear.

The dominance of P. halepensis on Porto Santo—almost a monoculture—meant that the completely afforested area was at risk from potential attacks by diseases or pests, or from forest fires. Consequently, the local government encouraged research on a broader variety of species in view of their possible use for afforestation on Porto Santo. That research was not to be limited to tree species, but was to include various types of shrubs for use in a general restoration of the island’s vegetation, including the objective to halt the widespread and serious soil erosion. Consequently, a project was agreed in 1989 between the Regional Government of Madeira and the Danish Land Development Service (Hedeselskabet), funded by the European Community [7].

The objectives of the project were described as “an improved environmental situation on Porto Santo, specifically in terms of soil and water conservation” by the “provision of a list of indigenous and exotic trees and shrub species which can be recommended for future planting activities, with emphasis on soil and water conservation” [8].

Søndergaard [8] described the project layout and the experiment with 50 woody species. Field work started the same year and we found unpublished reports from visits to the island by Sondergaard [9,10], together with data on the survival and the condition of the plants at that time. Mendonça [11] published similar intermediate results as a thesis in the Agronomy course at the University of Évora. After the end of the project, in 1994, Søndergaard [7] published the results obtained by the end of 1993.

After the end of the project, in 1996, there was still a small survey on the experiment by the forester, Maria Gorete Freitas, with the support of Luís Silva, from the Madeiran Forest Services [12]. Since then, there have been no other measurements or reports on the experiment.

The reports and the published paper by Søndergaard already indicated the possible interest in long-term monitoring of the experiment to assess the survival and growth of the different species [6,7,9,10].

At the beginning of 2025, 3 decades later, revisiting the areas of the experiment, it was concluded that it was still possible to recognize the original layout and to measure the surviving plants at the three sites of the experiments: Alentejo, Pico Juliana and Matinho.

The objective of this study is, therefore, to test the hypothesis that there were differences in long-term survival and growth among the 50 species used in the experiment at the three sites after 3 decades. It is the results of this unexpected long-term forest experiment that are presented in this paper.

2. Materials and Methods

2.1. Species and Provenances Tested

A preliminary list of species to be tested was established, including species known to be or have been present on Porto Santo (both indigenous, such as D. draco, or already introduced with success, such as P. halepensis). However, new species of trees and shrubs from regions with similar climates were considered. These conditions included those of the adjacent parts of West Africa, with West Morocco as the first choice, but also those of the dry tropics south of the Sahara and from other regions with a dry Mediterranean-type climate, such as parts of the Mediterranean region, parts of Australia, New Zealand, South America and southwestern North America [8]. Seeds were thus obtained from various sources, such as

• The Australian Tree Seed Centre, Canberra;
• Centre National de Semences Forestières, Burkina Faso;
• Danida Forest Seed Centre, Denmark;
• CIRAD Forêt, Paris;
• Henry Doubleday Research Association, Coventry;
• Institut Senegalais de Recherches Agricoles/Direction des Recherches sur les Productions Forestieres (ISRA/DRPF), Senegal;
• Société Versepuy, France;
• Station de Recherches Forestieres, Rabat, Morocco.

Seeds and cuttings were also collected by the Madeiran Forest Service in Madeira and Porto Santo, and by Poul Søndergaard in Morocco [7]. Propagation of plant material was started at the local nursery belonging to the Madeiran Forest Service. Many of the seeds obtained from the various sources were not used in the production of plants in the experiment. The origins of seeds of the plants used are indicated in Table S1 (Supplementary Materials).

2.2. Study Area Conditions: Climate and Soils

The experimental plots were initially established in different research areas in four main locations on the island, of approximately 0.5 ha each, covering different ecological situations, at different elevations and exposures (Figure 2). The site of Morenos (1–3), with a low elevation (60–100 m) and very dry conditions, was abandoned in the first years due to the very low survival rate. The experiment in the areas of Alentejo (4), at 100 m of elevation, Pico Juliana (5), at 300 m, and Matinho (5), at 200 m, was followed until 1993 and resampled in 2025 in this study. More details on the sites are presented in this section.

The climate of Porto Santo is characterized by mild temperatures, with annual averages between 16 °C and 22 °C, very favorable for plant growth. Water is the limiting factor. Average annual precipitation is low, around 400 mm, but with strong variability, varying from less than 200 mm in 1994 and 2004 to more than 600 mm in 2010 and 2011 and strongly concentrated in the winter months, whereas in the summer months precipitation is generally absent. Annual precipitation and maximum and minimum monthly precipitation are shown in a graphic (Figure 3). Yearly fluctuations in total precipitation are important, but fluctuations in precipitation throughout the year are at least equally important. Because of these fluctuations, Sondergaard [7] recommended that a trial period of at least ten years should be considered in order to experience the main fluctuations in climate and to obtain reliable results to guide future afforestation and arboriculture on Porto Santo. He further recommended maintaining and developing the initial trial plots as small arboreta, where the long-term development of the various species could be analyzed. This study follows his recommendation.

The climate limitations were clear in the Morenos areas (1–3). This plot was established on existing plantations of P. halepensis in the southwestern part of the island, between 60 and 100 m of elevation, on eroded soils consisting of sandy loam with a high proportion of gravel and small stones. However, due to the very difficult conditions, especially the dryness and the exposure to the winds, the survival of plants was recorded to be very poor, with no specific data on survival and the experiment being abandoned very early in this location.

In the other experimental areas, at higher elevations, water was still scarce but the soil conditions also played an important role. The soils of the island are very variable, from loamy aeolian sands to silty clay with stones and boulders, and they are often alkaline. This was the case with the soils in the three areas of the experiments (Alentejo, Pico Juliana and Matinho), where analysis of soil from 4 composite samples, each from 5 subsamples randomly collected at each site at 0–20 cm depth, was conducted. The average indicative results are shown in

Table 2. Results of soil analyses (averages and standard deviations in parentheses) of composite soil samples collected at 0–20 cm depth at the 3 experimental sites. Significance was assessed with the Kruskal–Wallis test in SPSS.

Experimental Area	pH H2O	pH KCl	Phosphorus (ppm)	Potassium (ppm)	Nitrogen (Nitrates, ppm)
Alentejo	7.83 (0.39)	7.45 (0.48)	595.5 (98.9)	81.0 (33.1)	3.13 (0.39)
Pico Juliana	7.17 (0.15)	6.80 (0.10)	748.0 (579.9)	348.0 (63.5)	2.73 (0.84)
Matinho	7.88 (0.29)	7.34 (0.28)	1374.0 (687.0)	352.8 (239.8)	2.20 (0.28)
Significance (p)	0.042	0.040	0.029	0.030	0.071

. The analyses were performed at the Laboratory of the Direção Regional de Agricultura e Desenvolvimento Rural. The results confirm the alkaline nature of the soils (pH above 7) and the average concentrations of phosphorus, potassium and nitrates in parts per million (ppm). The evaluation of the significance of the differences between sites was determined using the Kruskal–Wallis test, which does not depend on several assumptions required for ANOVA. From these results, it is apparent that all areas have basic soils, with Pico Juliana being the closest to neutrality, resulting in significant differences for pH(H₂O), with p = 0.042, and for pH(KCl), with p = 0.040. The experiment in Alentejo, at the lowest elevation, also shows weaker concentrations of phosphorus and potassium, resulting in significant differences between sites for phosphorus (p = 0.029) and for potassium (p = 0.030). For nitrates the highest values are for Alentejo, and the differences between the three sites are almost statistically significant (p = 0.071).

2.3. Experimental Design

The plantings in the three experimental areas started in January and February 1991 mainly with plants propagated in the nursery at Porto Santo. Each experimental unit consisted in a group of 5 plants of the same species planted in 40–60 cm deep ditches (rows), with a distance between rows of around 4 m and a distance between plants in the row of 1 m. A number of 45–50 species (or provenances) was planned for each location, with 3–5 replications per site, established with a completely randomized design at the three sites.

This ideal layout was not fully implemented, as there were some minor deviations, with not necessarily the same number of replications per species. In Alentejo, 50 species were used in 200 experimental units, in Matinho, 48 species in 209 experimental units, and in Pico Juliana, the smaller area, 47 species in 93 experimental units. The list of species with names as presented in the project and the distribution of the number of experimental units in the three locations are shown in

Table 3. List of species with corresponding numbers of experimental units (5 plants each) used at the 3 different experimental sites.

Species	Alentejo (F)	Matinho (G)	Pico Juliana (J)
Acacia cyanophylla	5	5	2
Acacia gummifera	5	5	2
Acacia melanoxylon	5	5	2
Adansonia digitata	4	5	2
Albizzia lebbeckoides	5	5	2
Albizzia lophanta	5	5	2
Apollonias barbujana	5	5	2
Araucaria excelsa	3	4	2
Argania spinosa	1	0	0
Bauhinia rufescens	5	5	2
Cassia siamea	4	5	2
Cassia sieberiana	5	5	2
Casuarina cunninghamiana	5	5	2
Casuarina equisetifolia	5	5	2
Casuarina stricta	4	5	2
Ceratonia siliqua	7	7	4
Combretum aculeatum	4	4	2
Combretum micranthum	3	5	2
Cupressus atlantica	2	2	2
Dodonaea viscosa	5	5	2
Dracaena draco	5	5	2
Eucalyptus calophylla	2	2	2
Eucalyptus citriodora	5	5	2
Eucalyptus ficifolia	3	5	2
Eucalyptus sideroxylon	5	5	2
Eucalyptus tesselaris	2	5	2
Grewia bicolor	1	1	0
Juniperus cedrus	5	5	2
Lycium europaeum	5	5	2
Lygos monosperma	5	5	2
Lygos sphaerocarpa	5	5	2
Melia azedarach	5	5	2
Myoporum acuminatum	5	5	2
Nerium oleander	5	5	2
Parkia clappertoniana	5	5	2
Parkinsonia aculeata	5	5	2
Pinus brutia	3	2	1
Pinus canariensis	2	1	1
Pinus halepensis	5	5	2
Pistacia atlantica	5	5	4
Populus alba	1	0	0
Prosopis africana	1	2	1
Prosopis juliflora	5	5	2
Prosopis spicigera	5	5	2
Retama monosperma	2	1	1
Schinus terebinthifolius	1	1	1
Simmondsia chinensis	2	2	2
Tamarindus indica	5	5	2
Thevetia nereifolia	3	5	2
Zizyphus mauritiana	5	5	2
Total	200	209	93

.

The area of Alentejo (4, later designated as F) is located on relatively flat cultivated land (cereal) west of the airport on sandy soils at an elevation of around 100 m. In spite of a small stone wall on the west, there is no protection against the strong winds in the area (Figure 4).

Further to the northeast, the experimental area of Pico Juliana (5, later J) was established on the northern slope of Pico Juliana, at approximately 300 m elevation, on formerly cultivated terraces on a loamy soil with high content of clay and many stones (Figure 5).

Finally, the experimental area of Matinho (6, later designated as G) was located on the southern slope of Pico Castelo, at approximately 200 m of elevation, established on formerly cultivated terraces on loamy soils with a high content of clay and many stones. The Pico Castelo and Pico do Facho massifs protect the area from northern winds (Figure 6).

2.4. Field Work

The experiments at the 3 sites were followed, with plants recorded for survival and condition classes until 1993, with the results published by Sondergaard [7]. In this study we only use survival data from the records of October 1991 and October 1992, as the number of plants alive per each experimental unit is a good measure of their initial adaptation. The results of survival range from zero (all plants died) to five (all five plants from all the experimental units of that species survived). Survival is the main key variable to assess the adaptation of the different species at the sites.

In February 2025 the experiments were revisited and the number of plants that had survived in each experimental unit was recorded. The survival indicator was therefore the average number of plants that survived, ranging from zero when all plants had died to five when all five plants of all the experimental units of that species survived.

Total height, defined as the maximum vertical distance from the ground level to the treetop, was measured for all the trees using a hipsometer, and stem diameter was measured with a steel diameter tape at 30 cm from the ground. As many plants were multi-stemmed, we used only height in this analysis as a measure of adaptation and growth.

2.5. Data Analysis

Due to the imbalance in the initial representation of the different species at the different sites, we performed an analysis of variance to evaluate the differences between species separately at each of the sites and in each year. The variable used to evaluate differences in survival between species was the number of plants present in each year in each experimental unit (maximum 5 if all plants in the experimental unit survived). In 2025, the variable used to evaluate differences in growth of the surviving species was the average height of the surviving plants in each experimental unit.

In all cases, the statistical analyses used a univariate analysis of variance under the General Linear Model option in SPSS, with species as the only source of variation included. As Levene’s tests of homogeneity of variances indicated in some cases significant departures from the hypothesis of equal variances and the modified Breusch–Pagan and F tests indicated in some cases a departure from the hypothesis of homoscedasticity of residuals, we used a non-parametric test equivalent to one-way ANOVA, the Kruskal–Wallis test using SPSS (version 28). We considered differences to be significant when p < 0.05. The Kruskal–Wallis test does not require the assumptions required to use results derived from ANOVA. Results of both the parametric and non-parametric tests are provided.

3. Results

The analysis of variance for the Alentejo (J) area indicated that there were very high statistical differences in both the parametric and non-parametric tests (p < 0.001) in the survival of plants of the 50 different species used in the 200 experimental units for all the three surveys performed, in October 1991, October 1992 and February 2025. The same applies for both the sites Matinho (G), with 48 species in 209 experimental units, and Pico Juliana (J), with 47 species in 93 units, always with very highly significant differences (p < 0.001 in both parametric and non-parametric tests) between species for 1991, 1992 and 2025.

The average survival values, ranging from zero (all plants died) to five (all plants survived), for the three locations and the three dates are shown in Figure 7. In all cases it is very visible that already in 1991 and 1992 there were many plants that did not survive, and in 2025, 35 years from the planting, only a few of those species continued to show good survival. A summary of these results, averaging for all locations, is shown in Figure S1 (Supplementary Materials).

In order to better visualize and quantify the differences between species and sites, we present the data from 2025 in Figure 8. It should be noted that Argania spinosa (L.) Skeels was only used in Alentejo. Grewia bicolor Juss. and Populus alba L. were not used in the three locations, but in any case no survivors of these species were found in any location in 2025.

A similar analysis was made for the average heights measured in the plants surviving in 2025. Differences in height between species at the same site are due to the characteristics of the species (trees are naturally taller than shrubs) but also to their adaptation. However, differences in height for the same species at the different sites are clear indications of the adaptation of that species at the different sites.

Due to the reduced number of survivors, only a few experimental units were used in the statistical analyses, reducing the statistical power of the tests. In Alentejo it was only possible to measure 23 experimental units with 15 species in total. The differences in height between species were highly significant in the ANOVA test (p = 0.007), but failed to be significant in the non-parametric test (p = 0.139). In Matinho, with 20 species in 49 experimental units, the analysis was much more powerful, indicating a very high significance of the differences in height in the ANOVA results (p < 0.001) and highly significant differences in the Kruskal–Wallis test (p = 0.002). Finally, in the smaller experiment in Pico Juliana, with only 16 valid experimental units with 13 species, the ANOVA indicated that the differences were not statistically significant (p = 0.148), which was confirmed by the results of the Kruskal–Wallis test (p = 0.333). The reduced number of replicates for Alentejo and Pico Juliana did not allow us to conclude on the statistical significance of the differences. However, the average values of height for the different species in Alentejo and Pico Juliana are also shown, as are those from Matinho, in Figure 9.

In general, as expected, differences in species height reflect their adaptation to sites but they also clearly reflect the nature of the different species, which attain different heights with age. Tree species of the genera Casuarina, Eucalyptus or Pinus are naturally taller than smaller species such as Nerium oleander or Thevetia nereifolia. Some comments are made on some of the species measured for height.

Plants from the genus Acacia show some success in survival, and growth of Acacia cyanophylla Lindl. is not negligible in the three locations.

A. spinosa survived in Alentejo, the only location where it was planted, but with slow growth.

Plants from the genus Casuarina seem to be able to survive, with C. cunninghamiana also showing very good height growth at the more protected location of Matinho.

C. siliqua, Combretum micranthum G.Don, Cupressus atlantica Gaussen and D. draco showed some survival but slow height growth.

Plants from the genus Eucalyptus showed variable survival rates, generally better in Matinho, but very good growth in height, especially Eucalyptus sideroxylon A.Cunn. ex Woolls, a good survivor in all three locations, and Eucalyptus citriodora Hook. (now renamed Corymbia citriodora (Hook.) K.D.Hill & L.A.S.Johnson) in Matinho and Pico Juliana.

Lycium europaeum L., Myoporum acuminatum R.Br., Nerium oleander L. and Parkinsonia aculeata L. all had some survivor plants, with generally low heights except for P. aculeata.

From the genus Pinus, three species were tested. Pinus brutia Ten. and Pinus canariensis C.Sm. survived better in the higher elevation and northern exposure of Pico Juliana while P. halepensis showed a remarkable capacity for survival and growth in all three locations, even in the more difficult situation of Alentejo. In the conditions of Pico Juliana, regeneration of P. halepensis is easy to observe. Pinus halepensis is known to be remarkably adapted to soils with high pH, as in the calcareous mountains of the Mediterranean region of its origin [13].

P. atlantica had a reasonable survival and growth in all three locations.

Retama monosperma (L.) Boiss and Simondsia chinensis (Link) C.K.Schneid survived and grew reasonably well in Alentejo, whereas Thevetia neriifolia showed remarkable survival, growth and condition at Alentejo and Matinho.

Finally, Schinus terebinthifolius Raddi showed very strong survival and condition with very fast growth in all three locations. The adaptation of this species to the conditions of the island, with strong regeneration, makes it a difficult woody invader on the island.

4. Discussion

To our knowledge this study on the long-term experiment is unique in the context of the Macaronesian islands and therefore no comparison is possible to make with other similar studies. However, we believe that these results can provide useful information for the use of woody species in similar areas. They can also inspire similar works in different conditions in other regions.

From the analysis of the results from Figure 7, it is obvious that the early survival rate was already very different between species in 1991 and that these differences increased in 1992. However, only a few species survived the 33 years from 1992 to 2025. It is also clear, from Figure 8 and the corresponding analyses, that plant survival during the 34 years of the experiment was different for the different species at the different locations. However, a few species seemed to have performed relatively well in all three locations.

It became apparent, from Figure 9 and the corresponding analyses, that there were very strong differences in species height after 34 years, from the taller species of Eucalyptus sp. and the Casuarina cunninghamiana Miq. to the smaller Pistacia atlantica Desf. and Thevetia neriifolia Juss. ex Steud., with intermediate values for other species such as P. halepensis. However, the height of the species was also very dependent on location.

In general, it can be concluded, from this unexpected long-term experiment, that a few species are worth considering using in Porto Santo. However, it is clear that performance of all species in terms of both survival and height growth was strongly dependent on location, with Alentejo, the lower-elevation and drier experimental area, showing the most difficult conditions for plant survival and growth.

It is also concluded that elevation seems to be the most important factor determining the survival and growth of the species. In fact, the early abandonment of the Morenos site at an elevation below 100 m and the more difficult performance of vegetation in the Alentejo site indicate that the woody species considered have a strong preference for elevations above 200 m, which is related to precipitation. Also, as water is probably the most limiting factor, the capture of water by trees (horizontal precipitation) at higher elevations could be a crucial factor for survival and growth. Also, at the Alentejo site the poor soil nutrients with low phosphorus and potassium content could have contributed to the lower performance of the plants at this site.

The perspectives and objectives of the experiment should be observed in light of the distance of 35 years that have elapsed since the beginning of the experiment in 1990. For example, the very good adaptation of S. terebinthifolius is not currently viewed as an advantage but as a threat, due to its known invasive behavior, already demonstrated in other areas where it was introduced [14]. In addition, the relatively good adaptation of other species, such as Acacia melanoxylon R.Br. or Tamarix gallica L., is not an indicator of the possibility of their use in Porto Santo, due to their potential as invasive species, as shown by their inclusion in the list of invasive species in Medeiros et al. [15]. This indicates the particular attention that is now devoted to the problem of potentially invasive species, especially in the protected areas of the region.

In fact, since 1990, several changes have occurred in the context of Porto Santo, especially after the approval, in 2020, of the application of Porto Santo Island to be part of UNESCO’s World Biosphere Reserve Network, under the UNESCO Program “The Man and the Biosphere—MAB” [16]. The interest in native species increased but, interestingly, this is in line with some of the main recommendations already made by Sondergaard [7].

First, Sondergaard [7] highlighted the importance of including “all of the autochthonous Porto Santo woody species in the trials before definite conclusions are drawn”. At that time, seeds and plant material of some species were not available. Today, some experiments are in place but without the dimensions and statistical design of the experiments of 1990.

Second, Sondergaard [7] suggested the possibility of maintaining and developing “the current trial plots as small arboreta, where the long-term development of the various species can be studied”. This would also be an opportunity to show some trees and shrub species from various areas of the world and of great ornamental interest. This is the case with E. citriodora (now C. citriodora), with its particular citron smell, or E. sideroxylon, with its beautiful black bark, or the shrub T. neriifolia, used as an ornamental feature in many gardens. The hypothesis of an arboretum seems particularly adequate in the case of Matinho, where plant development was better and which shows continuity with Pico do Castelo, the area that was planted by the pioneer forester Schiappa de Azevedo at the beginning of the XX century, demonstrating continuity in time and space.

Third, in some areas where restoration of the natural vegetation is not the objective, Sondergaard [7] indicated that “interesting mixed plantations, could be established by using some well surviving but slow-growing introduced species such as C. siliqua, P. atlantica, or A. spinosa”. Our results also show that species such as Q. ilex were not included in the trial, but it has adapted well to the conditions of Porto Santo and is important in soil conservation because of its deep and strong root system. T. gallica should also be borne in mind, particularly in the fixation of moving sands, where it already plays an important role on Porto Santo.

Finally, Sondergaard [7] recommended that the well-established plantations of P. halepensis, which has proven its qualities as a reliable pioneer species on Porto Santo, “could be transformed into mixed and, hopefully, less vulnerable stands. Indigenous species such as D. draco, A. barbujano, O. europaea subsp. maderensis, and S. marmulano should be planted in the shelter of the pine trees, and such mixed plantations might eventually develop into a forest type reminiscent of the vanished natural forests of Porto Santo”. The author also indicates that the efforts to develop a “pure” Porto Santo forest could also include, beyond the native tree species such as D. draco, some “shrub species such as Echium nervosum Dryand. ex Aiton and Euphorbia piscatoria Aiton”. In fact, the pioneer work with P. halepensis and the physical protection it provides, even after death, seems to have been instrumental for the expansion of native species to the interior of the pine stands. The foresters of the Instituto das Florestas e Conservação da Natureza are now studying this process with the expansion of E. piscatoria into dead P. halepensis stands in the dry area of the Pico Ana Ferreira.

As a general final consideration, it appears that it was possible, by revisiting the experimental areas and measuring the surviving plants 35 years later, to transform a 4-year project into a long-term forest experiment. This was in line with the objectives and wishes of the initial project as indicated by its author, Poul Sondergaard, in his writings from 1990 to 1997. The afforestation objectives might slightly change—the emphasis on autochthonous species is now different—but the scientific value of these experiments is enormous and represents an excellent basis to support decisions that, in the forest area, influence future decades.

A strong forestry program, including a component of research, should be developed again on Porto Santo Island. The forestry challenges posed by an island that is now a Reserve of the Biosphere [17] and the history of the pioneer work of humans and trees since the beginning of the XX century deserve such a program.

5. Conclusions

This 35-year-period forest experiment on the island of Porto Santo gave us rare and valuable insight into species resilience and survival under semi-arid insular conditions. There are certainly limitations with this long-term experiment. Due to very low survival rates, one of the initial sites had to be abandoned in the first years, and many of the species in the other three sites did not survive, reducing the statistical power of the tests, namely the results for height. On the other hand, the longevity of the experiment allows for better conclusions on the adaptation of the species. The findings clearly demonstrate the strong influence of site-specific factors on species survival and growth, with notable differences across the three experimental sites. While a few species, such as Pinus halepensis, Eucalyptus sideroxylon and Casuarina cunninghamiana, showed promising long-term results, the risk of invasiveness must now be considered in all sustainable and restoration planning. The island’s designation as a UNESCO Biosphere Reserve and the evolution of environmental policies highlight the growing importance of using native and ecologically appropriate species. This study shows once more the scientific value of maintaining long-term trials and supports the idea of renewed afforestation programs integrating biodiversity conservation and sustainable land management tailored to island ecosystems.