The Contributions of Neotropical Tree Families to the Structure of Common Amazon Forest-Types

: In order to investigate how familial biodiversity structures forests in the critically important Amazon, I combined past plot samplings to investigate the contributions of tree families in those sam-plings to the structure of common Amazon forest types. I found that the families Arecaceae, Fabaceae, Clusiaceae, and Malvaceae had the most stems; Staphyleaceae, Caricaceae, and Anacardiaceae had the largest stems; Arecaceae, Fabaceae, Lecythidaceae, and Malvaceae had the largest basal area; Fabaceae, Malvaceae, and Sapotaceae had the most genera; Annonaceae, Euphorbiaceae, Fabaceae, Lauraceae, Malvaceae, Moraceae, and Sapotaceae had the most species, and the maximum Fisher’s α diversity index was found for many families. Together, results suggest that Fabaceae and Malvaceae are the most important families structuring these forests, but also that Arecaceae and Sapotaceae may be important. Thus, conservationists and managers may help sustain structure in these forests by propagating and maintaining species in these families. Finally, correlations between total number of stems and basal area, and between total number of genera and total number of species, suggest a causal relationship between them as they structure these forests, but the lack of correlations with Fisher’s α suggest it has little structural utility for these forests.


Introduction
The effect of biodiversity on ecosystem structure and function is a central ecological research topic [1], as the global loss in biodiversity [2] may be leading to a decline in ecosystem-level processes [3]. For example, productivity may be reduced by a decrease in biodiversity because studies have shown a positive relationship between productivity and species richness [4,5]. Ecosystem stability-which can be defined as resistance to, or resilience to, perturbation-could also deteriorate because of a reduction in biodiversity [6]. However, perhaps most critically for our shared human future, loss of biodiversity might diminish key ecosystem "services" that we need, such as air and water purification and maintenance of soil fertility [7].
The Neotropics has long been known to be the most speciose place on Earth. It includes the very biodiverse Amazon [8,9], which has been an excellence place to investigate how biodiversity controls and affects the structure, function, and dynamics of ecosystems [10,11]. Surveys of Amazonian biodiversity have shown that there are at least 6727 tree species in lowland Amazon forested ecosystems (for trees at least 10 cm in diameter at breast height [12]) in at least 188 families, and the families with the most species are Fabaceae (1042 species), Lauraceae (400), Myrtaceae (393), Annonaceae (388), Rubiaceae (338), Melastomataceae (263), Chrysobalanaceae (256), Sapotaceae (244), Malvaceae (214), and Ochnaceae (166) [12][13][14][15]. Most of the sampling in these surveys has been in the most common Amazon forest types of terra firme, palm, white sand, várzea, and igapó, and has suggested that examining biodiversity at higher, more inclusive taxa levels than species (such as families) could facilitate ecological research on the effects of biodiversity on ecosystem structure and function [16].
Therefore, because of the importance of the Amazon and because it presents an exceptional opportunity to examine relationships between biodiversity and ecosystem structure, function, and dynamics, I combined past samplings in one ha plots of the most common Amazon forest types to investigate these questions: (1) What are the families and most common species in these Amazon forested ecosystems? (2) How do these families individually contribute to the key structural parameters of total number of stems, mean stem size, total basal area, total number of genera, total number of species and Fisher's alpha (α) diversity index of these forests? and (3) What do correlations among those structural parameters reveal about how these forests are structured at the familial level?

Study Sites
The first study site was the Estación Científica Yasuní (ECY: 0 • 41 S, 76 • 24 W), operated by the Pontificia Universidad Catolica of Ecuador (PUCE) (www.puce.edu.ec, accessed on 11 February 2023) and located within the Yasuni National Park of eastern Ecuador [17]. The mean annual rainfall of ECY is 3081 mm and the wettest months are April to May and October to November, with August as the driest month. The mean monthly temperature varies between 22 • C and 35 • C. Soils in the National Park are clayey, low in most cations, but rich in aluminum and iron [18]. Most of the station is in terra firme forest but várzea forest-located next to the white-water Tiputini River, which floods for a few weeks every year to a maximum depth of 3 m-is also common. ECY is the site of a long-term 50 ha plot in terra firme forest, maintained by PUCE as part of the ForestGEO network (www.forestgeo.si.edu/sites/neotropics/yasuni, accessed on 11 February 2023).
The second study site was the Área de Conservación Regional Comunal de Tamshiyacu-Tahuayo (ACRCTT: 4 • 18 S, 73 • 13 W; www.perujungle.com, accessed on 11 February 2023), located in Loreto Province 130 km south of Iquitos Peru [19], and named for two whitewater rivers-the Tamshiyacu and the Tahuayo-that form boundaries to the north and west. ACRCTT is part of one of the largest (270, 654 ha) protected areas in the Amazon where local black-water runoff creates igapó forests, and where terra firme forest and várzea forest are also common [20]. The precipitation ranges from 2400-3000 mm per year and the average temperature is relatively constant at 26 • C.
The third study site was the Sabalillo Forest Reserve (SFR: 3 • 20 S, 72 • 18 W) established in 2000 and operated by Project Amazonas (www.projectamazonas.org, accessed on 11 February 2023) [21]. SFR is located on both sides of the upper Rio Apayacuo, 172 km east of Iquitos, Peru. The reserve is part of 25,000 hectares set aside over the last decade and consists of low, seasonally inundated river basins of the upper Amazon. Annual precipitation is 3297 mm per year and the rainy season is between November and April [22]. Within the reserve, terra firme forest, white sand forest, palm forest, and igapó forest are often found.

Setting Up and Sampling of the One Ha Plots
A one ha plot was sampled (1) in várzea forest (under water one month per year) at ECY in May 2010, (2) in igapó forest (under water two months per year) at ACRCTT in May 2011, (3) in terra firme forest, (4) in palm forest, and (5) in white sand forest at SFR in June 2013. Each of these plots were the same shape (200 m × 50 m) and used the same sampling protocols, including measuring the diameter at breast height (dbh) of all trees and palms at least 10 cm dbh at the nearest/lowest point when the stem was cylindrical, or just above the buttresses if the tree was buttressed. Trees and palms were also identified to family (using the familial taxonomic nomenclature found in [23]), to species, or in a few cases to genus if identification to species was not possible, using [24,25] as taxonomic sources. The Missouri Botanical Garden website (www.mobot.org, accessed on 11 February 2023), and herbaria at the Universidad Nacional de la Amazonia Peruana (UNAP) in Iquitos and at ECY, were also consulted. Voucher samples are kept at the UNAP and the ECY herbaria. The plot data from várzea forest and from igapó forest are archived at the Luquillo Experimental Forest in Puerto Rico (LTERDBAS#172) as part of its Long-term Ecological Research (LTER) program (www.luq.lternet.edu, accessed on 11 February 2023) funded by the US National Science Foundation.

Data Analyses
The data for the plots were combined together to generate for each family found in any of the plots (1) the number of tree stems, (2) the average dbh (in cm) for those stems, (3) the sum of the basal areas of all individual tree stems in each family using the formula ∑πr 2 ; where r = the dbh of each individual stem/2, (4) the number of genera for those stems, (5) the number of species for those stems, and (6) Fisher's alpha (α) diversity index [26] using a Javascript algorithm based on Newton's method at (www.groundvegetationdbweb.com/ground_veg/home/diversity_index, accessed on 11 February 2023). Fishers α uses both number of stems and number of species in its calculation [26]. The most common species were also compiled. Finally, a Pearson product moment correlation coefficient (PPMCC) [27,28] was computed between all 6 parameters taken pair-wise with the number of stems found in each family (i.e., the columns of Table 1), using the algorithm at (www. socscistatistics.com/tests/pearson/default2.aspx, accessed on 11 February 2023). The data from each of the one ha plots has been previously peer-reviewed and published [8,[29][30][31][32][33][34][35] where the families and the ranges for their tree stem sizes were found to be similar and comparable to one ha plot samplings of the same forest types across the Amazon, as discussed and referenced in those published papers. Table 1. All families found in the five one ha plots sorted alphabetically with the total number (#) of tree stems sampled, mean stem dbh (in cm), basal area (in m 2 per 5 ha area), total number (#) of genera, total number (#) of species and Fisher's α index of diversity for each family (data taken from [8,[29][30][31][32][33][34][35]

Discussion
Three of the four most common families (Arecaceae, Fabaceae, Malvaceae) had the highest basal area, but none of the most common families had the biggest stems or the largest Fishers α, and only Fabaceae and Malvaceae had the most genera and the most species. Among the rare families, Staphyleaceae had the largest stem size and many of the rare families had a large Fisher α. Among the six structural parameters, Fabaceae and Malvaceae were among the families that had the largest values in four of them, and Arecaceae and Sapotaceae were among the families that had the largest values in just two of them. All other families had the largest values in at most one of the structural parameters. Among the genera with more than one species, only Inga was common in other Amazon samplings [12,13,[36][37][38]. When we combined families that had large basal areas (which has been used as a key to structural dominance elsewhere [39]) with these familial results, the families Fabaceae and Malvaceae were implicated as most important in structuring common Amazon forests, but Arecaceae and Sapotaceae may also be important. Therefore, conservationists and managers may want to focus on the preservation and propagation of tree species in these families in order to best maintain the structure (which enables function) of common Amazon forests. The families found here were also found in past Amazon tree surveys [12][13][14][15] and in past Amazon palm forest surveys [39][40][41], and are commonly found throughout the rest of the Neotropics as well [9,25]. Correlation analysis showed that as the number of stems increases (or decreases), basal area increases (or decreases), and that as number of genera increases (or decreases), the number of species increases (or decreases), which suggests causality between them as they structure Amazon forests. The lack of correlation between Fisher's α and any of the structural parameters, however, casted doubt on the necessity or meaning of computing Fisher's α [42] as a way to investigate structure.
Previous examinations of the five plots showed the common Amazon forest types had these three most common families: in terra firme Malvaceae, Fabaceae and Myristicacea (49 families total), in palm Arecaceae, Clusiaceae, and Fabaceae (6), in white sand Clusiaceae, Malvaceae, and Myrtaceae (15), in várzea Cecropiacea, Fabaceae and Meliaceae (41) and in igapó, Fabaceae, Moraceae and Sapotaceae (16), and that palm forests were dominated by just a few species [8,[29][30][31][32][33][34][35]. Any common families and species between terra firme and várzea and between terra firme and igapó may be due, in part, to terra firme tree species establishing ecotypes in flooded forests [43] where the regularity of the flood pulse facilitates adaptation and speciation [29]. Additionally, in those previous examinations, all the forest types except white sand had these unique families: in terra firme Capparidacea, Caricaceae, Malpighiacea, Memecyle, Quiinaceae, Staphyleacea, Malvaceae, Urticaceae and Verbenace, in palm Humiriaceae, in várzea Bixaceae, Ebenaceae, Ochnaceae, Opiliaceae, Pichramnaceae, and Vochyciaceae, and in igapó Salicaeae. These familial differences among forest types may be due, in part, to the unflooded forest types (terra firme, palm, white sand) having soil differences (which can explain up to 50% of floristic differences [44]) and different micro-topological relief, and to the flooded forest types (várzea, igapó) having different water characteristics such as sediment load and ion concentrations, and differences in both maximum depth and duration of flooding.

Conclusions
Two of the four common families Fabaceae and Malvaceae dominated the structural parameters of number of stems, basal area, number of genera and number of species, but not stem size or Fishers α. Another common family Arecaceae and the family Sapotaceae were also implicated in affecting structure. In addition, families in the plots were commonly found in other Amazon samplings, but few genera were in common. While there were large correlations between the number of stems and basal area and between the number of genera and number of species, which suggested that there was a causal relationship between these parameters as they structure these forests at the familial level, correlation results casted doubt on the necessity and meaning of computing Fisher's α as a way to investigate structure. Finally, more sampling is needed to make these conclusions more robust for these Amazon forest types.

Data Availability Statement:
The data analyzed in this study are available at: www.luq.lternet.edu.